Measurement Techniques Sensors and Platforms

Editor-in-Chief
John H. Steele
Marine Policy Center, Woods Hole Oceanographic Institution, Woods Hole,
Massachusetts, USA

Editors
Steve A. Thorpe
National Oceanography Centre, University of Southampton,
Southampton, UK
School of Ocean Sciences, Bangor University, Menai Bridge, Anglesey, UK

Karl K. Turekian
Yale University, Department of Geology and Geophysics, New Haven,
Connecticut, USA

Subject Area Volumes from the Second Edition

Climate & Oceans edited by Karl K. Turekian

Elements of Physical Oceanography edited by Steve A. Thorpe
Marine Biology edited by John H. Steele
Marine Chemistry & Geochemistry edited by Karl K. Turekian
Marine Ecological Processes edited by John H. Steele
Marine Geology & Geophysics edited by Karl K. Turekian
Marine Policy & Economics guest edited by Porter Hoagland, Marine Policy Center,
Woods Hole Oceanographic Institution, Woods Hole, Massachusetts
Measurement Techniques, Sensors & Platforms edited by Steve A. Thorpe
Ocean Currents edited by Steve A. Thorpe
 ENCYLOPEDIA
OF
OCEAN SCIENCES:
MEASUREMENT
TECHNIQUES,
SENSORS AND
PLATFORMS

Editor

STEVE A. THORPE

Boston • Heidelberg • London • New York • Oxford

Paris • San Diego • San Francisco • Singapore • Sydney • Tokyo
Academic Press is an imprint of Elsevier
Academic Press is an imprint of Elsevier
32 Jamestown Road, London NW1 7BY, UK
30 Corporate Drive, Suite 400, Burlington, MA 01803, USA
525 B Street, Suite 1900, San Diego, CA 92101-4495, USA

Copyright ^ 2009 Elsevier Ltd. All rights reserved

Material in the work originally appeared in Encyclopedia of Ocean Sciences (Elsevier Ltd., 2001) and Encyclopedia of Ocean Sciences,
2nd Edition (Elsevier Ltd., 2009), edited by John H. Steele, Steve A. Thorpe and Karl K. Turekian.

The following articles are US government works in the public domain and are not subject to copyright:
Satellite Oceanography, History and Introductory Concepts
Satellite Passive-Microwave Measurement of Sea Ice

No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means electronic,
mechanical, photocopying, recording or otherwise without the prior written permission of the publisher

Permissions may be sought directly from Elsevier’s Science & Technology Rights Department in Oxford, UK: phone (+44)(0)
1865 843830; fax (+44)(0) 1865 853333; email: [email protected]. Alternatively you can submit your request online by visiting
the Elsevier website at (http://elsevier.com/locate/permissions), and selecting Obtaining permissions to use Elsevier material

Notice
No responsibility is assumed by the publisher for any injury and/or damage to persons or property as a matter of products liability,
negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein,
Because of rapid advances in the medical sciences, in particular, independent verification of diagnoses and drug dosages should be made

British Library Cataloguing in Publication Data

A catalogue record for this book is available from the British Library

Library of Congress Control Number: 2009907113

ISBN: 978-0-08-096487-4

For information on all Elsevier publications

visit our website at www.elsevierdirect.com

PRINTED AND BOUND IN ITALY

09 10 11 12 13 10 9 8 7 6 5 4 3 2 1
 CONTENTS

Measurement Techniques, Sensors and Platforms: Introduction ix

RESEARCH VESSELS
Ships R P Dinsmore 3

LOWERED, EXPENDABLE, OR TOWED INSTRUMENTS

Continuous Plankton Recorders A John, P C Reid 15
CTD (Conductivity, Temperature, Depth) Profiler A J Williams, III 25
Deep-Sea Drilling Methodology K Moran 35
Expendable Sensors J Scott 43
Grabs for Shelf Benthic Sampling P F Kingston 50
Towed Vehicles I Helmond 60
Zooplankton Sampling With Nets and Trawls P H Wiebe, M C Benfield 70

MOORINGS, FLOATS, AND CURRENT METERS

Drifters and Floats P L Richardson 89
Moorings R P Trask, R A Weller 97
Profiling Current Meters A J Plueddemann 110
Single Point Current Meters P Collar, G Griffiths 115

UNDERWATER VEHICLES
Deep Submergence, Science of D J Fornari 125
Gliders C C Eriksen 140
Manned Submersibles, Deep Water H Hotta, H Momma, S Takagawa 148
Manned Submersibles, Shallow Water T Askew 156
Platforms: Autonomous Underwater Vehicles J G Bellingham 162
Platforms: Benthic Flux Landers R A Jahnke 174
Remotely Operated Vehicles (ROVs) K Shepherd 183
Seismology Sensors L M Dorman 189
Vehicles for Deep Sea Exploration S E Humphris 197

SENSORS: METEOROLOGY
Sensors for Mean Meteorology K B Katsaros 211
Sensors for Micrometeorological and Flux Measurements J B Edson 218

v
vi CONTENTS

SENSORS: TURBULENCE
Turbulence Sensors N S Oakey 229

SENSORS: OPTICAL
Absorbance Spectroscopy for Chemical Sensors R Narayanaswamy, F Sevilla, III, 239
Bio-optical Models A Morel 247
Fluorometry for Biological Sensing D J Suggett, C M Moore 257
Fluorometry for Chemical Sensing S Draxler, M E Lippitsch 265
Inherent Optical Properties and Irradiance T D Dickey 272
Nepheloid Layers I N McCave 282
Optical Particle Characterization P H Burkill, C P Gallienne 293
Photochemical Processes N V Blough 302
Radiative Transfer in the Ocean C D Mobley 313
Transmissometry and Nephelometry C Moore 323

SENSORS: CHEMICAL
Wet Chemical Analyzers A R J David 335

ACOUSTIC METHODS
Acoustic Measurement of Near-Bed Sediment Transport Processes P D Thorne, P S Bell 343
Acoustic Noise I Dyer 357
Acoustic Scattering by Marine Organisms K G Foote 367
Acoustic Scintillation Thermography P A Rona, C D Jones 376
Acoustics in Marine Sediments T Akal 380
Acoustics, Arctic P N Mikhalevsky 397
Acoustics, Deep Ocean W A Kuperman 406
Acoustics, Shallow Water F B Jensen 417
Bioacoustics P L Tyack 425
Seismic Reflection Methods for Study of the Water Column I Fer, W S Holbrook 432
Sonar Systems A B Baggeroer 442
Tomography P F Worcester 451

SATELLITE, AIRCRAFT, OR SHIP-BORNE REMOTE SENSING

Aircraft Remote Sensing L W Harding Jr., W D Miller, R N Swift, C W Wright 471
IR Radiometers C J Donlon 480
Remote Sensing of Coastal Waters N Hoepffner, G Zibordi 492
Satellite Altimetry R E Cheney 502
 CONTENTS vii

Satellite Oceanography, History and Introductory Concepts W S Wilson, E J Lindstrom,

J R Apel 509
Satellite Passive-Microwave Measurements of Sea Ice C L Parkinson 524
Satellite Remote Sensing Microwave Scatterometers W J Plant 535
Satellite Remote Sensing of Sea Surface Temperatures P J Minnett 548
Satellite Remote Sensing SAR A K Liu, S Y Wu 560
Satellite Remote Sensing: Ocean Colour C R McClain 571
Satellite Remote Sensing: Salinity Measurements G S E Lagerloef 584
Sea Surface Exchanges of Momentum, Heat, and Fresh Water Determined by Satellite Remote
Sensing L Yu 590
Gravity M McNutt 600

APPENDICES
Appendix 1. SI Units and Some Equivalences 611
Appendix 5. Properties of Seawater 614

INDEX 619
 MEASUREMENT TECHNIQUES, SENSORS AND
PLATFORMS: INTRODUCTION

Measurement is the foundation of any branch of science, and no less so in oceanography. New instruments
have led to greatly improved understanding of the ocean, sometimes to radical changes in knowledge of the
way in which it works. The challenges of extremely high pressures — at the mean ocean depth of 3750 meters
the pressure is almost 370 times that at the surface — corrosion, and the severe motions induced on
instruments lowered from the undulating sea surface into a sometimes violently moving sea, have resulted in
many failures and have been overcome only by instruments designed by some of the most ingenious and
talented engineers. The history of devices to measure the ocean is a major part of the history of the science
itself, matching, as time has advanced, developments in basic technology and other sciences, so that whereas
the early instruments those prior to the first World War, were largely mechanical with moving parts, those
made in recent years have taken advantage of the immense and rapid developments in electronics and
particularly in electronic data storage. Compare, for example, the 1904 Ekman propeller current meter
lowered on a wire, its data storage limited to fewer than about 50 measurements and direction measurement
relying on the falling of small balls into cups on a compass-mounted disc with 36 segments (a 10o resolution),
with a present-day free-fall turbulence probe with eight (or more) different sensors to measure current shear,
pressure, temperature, conductivity, and so density, each sampled to high precision more than 100 times a
second and with a capacity to continue sampling for many hours to depths of 5000 meters.
The two World Wars stimulated notable advances in instrumentations, particularly underwater sonar to
detect submarines during WW I. The ability to track neutrally buoyant floats using acoustic methods that
eventually followed in the 1950s, led to the discovery that in most parts of the ocean the kinetic energy is
dominated by mesoscale eddies with scales of many tens of kilometres and periods of several months. The
thermistor chain developed in WWII has led to much better understanding of these eddies and of the
structure of the ocean and its shorter-term variations such as internal waves.
Whilst only salinity was once sampled to a useful accuracy (by collection of water in water bottles, closed
mechanically through the impact of a ‘messenger’ sliding down the supporting wire followed, on recovery, by
analysis on board ship), a range of chemicals can now be routinely sampled in situ. The articles in this volume
provide an extensive coverage of the instruments, sensors and platforms used today. Many instruments
designed before the electronic revolution of the early 1960s are no longer in use, having been made re-
dundant by recent advances and, although very important in their time, are not described here. Nor are some
essential details that hindered or delayed improvements in instruments, such as reliable underwater seals, O-
rings, water tight electrical connections and dependable devices to release buoyant instruments from their
holding weights on the seabed.
Ocean measurement provides a story of the search for improved accuracy, precision and resolution, higher
sampling rates and data storage capacity, the endless need for quality control and precision manufacture.
New instruments and ways to measure, sample or ‘image’ the ocean often lead to the development of
understanding of the ocean’s dynamics and properties, more measurements stimulating advances in under-
standing that demand further accuracy and the ability to broaden the area coverage of measurements. In this
respect, satellite observation developed since the early 1970s has radically extended knowledge of the ocean,
both of the surface and also of some aspects of its motion at depth. Much has been learnt about the
distribution of plankton in the upper ocean, of the seasonal and area variations in sea-surface properties,
temperature, wave amplitude, mesoscale eddies and surface currents, which could never have been obtained
by other means.
Included in this volume are some articles on the development of subsurface vehicles capable of carrying
scientists into the depth of the ocean. Whilst operations in mid-water have resulted in discoveries of or-
ganisms so delicate and often so large that they cannot be caught without damage in nets or trawls, the most
remarkable and important studies have been those of the seabed, especially in the observations of hydro-
thermal vents and the prolific life that surrounds them. Much greater use is now being made of autonomous
vehicles that can remain submerged for many days, pre-programmed to follow a set course or under com-
mand from a mother vessel at the surface, sampling the water through which they pass and enabling access to

ix
x MEASUREMENT TECHNIQUES, SENSORS AND PLATFORMS: INTRODUCTION

regions previously little explored, such as those below the ice shelves around Antarctica. Ships are still
essential, but many of the observations once only made from ships can now be more efficiently and eco-
nomically made from autonomous underwater vehicles or floats that periodically communicate their data
back to shore.
It is regretted that some important instruments or measurement techniques of ocean properties that are
useful in prediction are not included. Most notably are those relating to waves, tides and sea level, the first
(waves) important in ship routing and required in the second World War to provide means of predicting
conditions during military land invasions from the sea, the last (sea level) now vital in estimating the effects
of climate change. Tidal elevation is one of the oldest measurements made of the ocean. Continuously
recording coastal tide gauges had been invented by the 1830s but ways to measure tides in deep water were
not devised until the 1960s. (Cartwright, 1999, gives a comprehensive description of the history of tidal
theory and measurement.) Although research vessels are the subject of an article, there is no reference to
FLIP, the Floating Instrument Platform operated by the Scripps Institution of Oceanography in La Jolla,
California, that can be towed into deep water, and which then ‘flips’ to become a vertical and very stable spar,
accommodating scientists and from which the properties of the upper ocean can be measured. GLORIA, the
Geological Long Range Inclined Asdic, that has contributed much to the mapping and discovery of features
of scientific and economic importance on the seabed around the continents and the oceanic island chains, is
another major instrument that is not described. (Knowledge of ocean bathymetry is fundamental to geo-
logical and physical oceanography, and instruments for measuring depth, or of ‘sounding’ the ocean, were
some of the earliest oceanographic instruments to be developed. Robert Hooke’s depth sounding device was
described in 1667 but, like some of its successors, was not necessarily successful — wooden floats became
waterlogged at depth, losing their buoyancy and, in deep water, the instrument failed to return to the
surface!)
Much reference is naturally made to the properties of seawater on which the sensing methods depend for
their response. Where it has seemed useful to do so, for example where the method and complexity of
measurement depend critically on the seawater properties, articles describing these properties are included as
well as articles on the measurement sensors themselves. The volume might therefore be regarded and used as
an introduction to the physical, biological, chemical and geological properties of the ocean, as well an
account of the methods of measurement.
Each author of the articles is an expert in the subject addressed. All the authors are distinguished re-
searchers who have given time to write concisely and lucidly about their subjects, and the Editors are
indebted to them all for the time given and care taken in preparing these accounts.
The articles in this volume would not have been produced without the very helpful advice and en-
couragement of the several members of the Encyclopedia’s Editorial Advisory Board, particularly Gwyn
Griffiths, Ken MacDonald and Robert Spindel. Each provided suggestions about the content and authorship
of particular subject areas. In addition to thanking the authors of the articles in this volume, the Editors wish
to thank the members of the Editorial Board for the time they gave to identify and encourage authors, to read
and comment on (and sometimes suggest improvements to) the written articles, and to make this venture
possible.

Steve A. Thorpe
Editor

REFERENCES
Cartwright DE (1999) Tides: A Scientific History, 292 pp. Cambridge: Cambridge University Press.
RESEARCH VESSELS
 SHIPS
R. P. Dinsmore, Woods Hole Oceanographic Military, defense, mapping and charting, acoustics
Institution, Woods Hole, MA, USA Academic, basic research, student practicum
Copyright & 2001 Elsevier Ltd. Commercial, exploration, petroleum, mining and
fisheries
According to information available from the
International Research Ship Operators Meeting
(ISOM), in 2000 there were approximately 420 RVs
Introduction (over 25 m length overall) operated by 49 coastal
Oceanographic research vessels are shipboard plat- nations.
forms which support the conduct of scientific re-
search at sea. Such research may include mapping
and charting, marine biology, fisheries, geology and History
geophysics, physical processes, marine meteorology,
The history of RVs is linked closely to the cruises and
chemical oceanography, marine acoustics, under-
expeditions of early record. Most of these were
water archaeology, ocean engineering, and related
voyages seeking new lands or trade routes. Oceanic
fields.
studies at best were limited to the extent and
Unlike other types of vessels (i.e., passenger, cargo,
boundaries of the seas. Little is known of the ships
tankers, tugs, etc.) oceanographic research vessels
themselves except it can be assumed that they were
(RVs) are a highly varied group owing to the diverse
typical naval or trading vessels of the era. A known
disciplines in which they engage. However, charac-
example is an expedition sent by Queen Hatshepsut
teristics common to most RVs are relatively small
of Egypt in 1500 BC. Sailing from Suez to the Land of
size (usually 25–100 m length overall); heavy outfit
Punt (Somaliland) to seek the source of myrrh, the
of winches, cranes, and frames for overboard work;
fleet cruised the west coast of Africa. Pictures of the
spacious working decks; multiple laboratories; and
ships can be seen on reliefs in the temple of Deir-el-
state-of-the-art instrumentation for navigation, data
Bahri.
acquisition and processing.
These ships represent the high-water mark of the
Categories of RVs may vary according to the
Egyptian shipbuilder and although handsome ves-
geographic areas of operations as well as the nature
sels, they had serious weaknesses and did not play a
and sponsorship of work. Examples are:
role in the development of naval architecture.
1. By Region: The earliest known voyage of maritime explor-
Coastal, usually smaller vessels of limited en- ation comes from Greek legend and sailors’ tales. It is
durance and capability that of Jason dispatched from Iolcus on the north-
Ocean going, larger vessels usually with multi- east coast of Greece to the Black Sea c.1100 BC with
purpose capabilities, ocean-wide or global range the ship Argo and a celebrated crew of Argonauts in
Polar, ice reinforced with high endurance and quest of the Golden Fleece. Although embellished by
multipurpose capability Greek mythology, this may have a factual basis; by
2. By discipline: sixth century BC there were Greek settlements along
Mapping and charting, emphasis on bathymetry the shores explored by Jason, and the inhabitants
Fisheries, stock surveys, gear research, environ- traditionally collected gold dust from rams fleeces
mental studies lain in river beds.
Geophysics, seismic and magnetic surveys Better recorded is an expedition commissioned by
Support, submersibles, buoys, autonomous ve- Pharaoh Necho II of Egypt in 609 BC. A Phoenician
hicles, diving fleet sailed south from Suez and circumnavigated
Multiple purpose, biological, chemical, and Africa on a four-year voyage. Information from this
physical oceanography; marine geology; acous- and other Phoenician voyages over the next 300
tics; ocean engineering; student training; may also years (including Hanno to the west coast of Africa in
include support services. 500 BC, and Pytheas to the British Isles in 310 BC),
3. By sponsor: although not specifically oceanographic research,
Federal, usually mission oriented and applied contributed to a database from which philosophers
research and scholars would hypothesize the shape and extent

3
4 SHIPS

of the world’s oceans. In today’s terminology these basis. In the mid to late eighteenth century the arts of
vessels would be characterized as ships of navigation and cartography were amply demon-
opportunity. strated in the surveys and voyages of Captains James
Under the Roman Empire, sea-trade routes were Cook and George Vancouver. The latter carried out
forged throughout the known world from Britain to surveys to especially high standards, and his vessel
the Orient. Data on ocean winds and currents were HMS Discovery might well be considered one of the
compiled in early sailing directions which added to first oceanographic research vessels.
the growing fund of ocean knowledge. Aristotle, al- Hydrographic departments were set up by sea-
though not a seafarer, directed much of his study to faring nations: France in 1720, Britain in 1795, Spain
the sea both in physical processes and marine ani- in 1800, the US in 1807. Most were established as a
mals. From samples and reports taken from ships he navy activity, and most remain so to this day; ex-
named and described 180 species of fish and in- ceptions were the French Corps of Hydrographic
vertebrates Pliny the Elder cataloged marine animals Engineers, and the US Coast Survey (now the Na-
into 176 species, and searched available ocean tional Ocean Survey). A date significant to this his-
soundings proclaiming an average ocean depth of 15 tory is 1809 when the British Admiralty assigned a
stadia (2700 m). Ptolemy (AD 90–168), a Greek vessel permanently dedicated to survey service.
mathematician and astronomer at Alexandria em- Others followed suit; the first such US ship was the
ployed ship reports in compiling world maps which Coast Survey Schooner Experiment (1831). By now
formed the basis of mapping new discoveries through hydrographic survey vessels were a recognized type
the sixteenth century. of vessel.
With the fall of the Roman Empire came the dark In the early to mid-nineteenth century expeditions
ages of cartography and marine exploration slowed were setting to sea with missions to include
to a standstill except for Viking voyages to Green- oceanographic investigations. Scientists (often
land and Newfoundland, and Arab trade routes in termed ‘naturalists’) were senior members of the
the Indian Ocean and as far as China. Arabian sea ships’ complement. These include the vessels:
tales such as ‘Sinbad the Sailor’ or the ‘Wonders of Astrolobe, 1826–29 (Dumont D’Urville); Beagle,
India’ by Ibn Shahryar (905) rival those of Homer 1831–36 (Charles Darwin); US Exploring Exped-
and Herodotus. The art of nautical surveying began ition, 1838–42 (James Wilkes and James Dana);
with the Arabs during medieval times. They had the Erebus & Terror, 1839–43 (Sir James Ross); Beacon,
compass and astrolabe and made charts of the 1841 (Edward Forbes); Rattlesnake, 1848–50 (Tho-
coastlines which they visited. mas Huxley). Most of the vessels participating in
In the West the city-state of Venice became in the these expeditions were navy ships, but the nature of
ninth century the most important maritime power in the work, their outfitting, and accomplishments
the Mediterranean. It was from Venice that Marco mark them as oceanographic vessels of their time.
Polo began his famous journeys. In the fifteenth A new era in marine sciences commenced with the
century, maritime exploration resumed, much of it voyage of the HMS Challenger, 1872–76, a 69 m
under the inspiration and patronage of Prince Henry British Navy steam corvette. Equipped with a cap-
of Portugal (1394–1460), known as the Navigator, able depth sounding machine and other instruments
who established an academy at Sagres, Portugal, and for water and bottom sampling, the Challenger
attracted mathematicians, astronomers, and cartog- under the scientific direction of Sir Charles Wyville
raphers. This led to increasingly distant voyages Thomson obtained data from 362 stations world-
leading up to those of Columbus, Vasco da Gama, wide. More than 4700 new species of marine life
and Magellan. With the results of these voyages, the were discovered, and a sounding of 8180 m was
shape of the world ocean was beginning to emerge, made in the Marianas Trench. Modern ocean-
but little else was known. From the tracks of ships ography is said to have begun with the Challenger
and their logs, information on prevailing winds and Expedition. This spurred interest in oceanographic
ocean currents were made into sailing directions. research, and many nations began to field worldwide
Magellan is reported to have made attempts at voyages. These included the German Gazelle (1874–
measuring depths, but with only 360 m of sounding 76); Russian Vitiaz (1886–89); Austrian Pola; USS
line, he achieved little. Blake (1887–1880); and the Arctic cruise of the
By the late seventeenth and early eighteenth cen- Norwegian Fram (1893–96).
turies instruments were being devised for deep The late 1800s and early 1900s saw a growing
soundings, water samples, and even subsurface interest in marine sciences and the founding of both
temperatures. However, any ship so engaged either a government- and university-sponsored oceanographic
naval or trading vessel remained on an opportunity institutions. These included: Stazione Zoologia,
 SHIPS 5

Naples; Marine Biological Laboratory, Woods Hole, continuous position fixing. Universities and govern-
USA; Geophysical Institute, Bergen, Norway; ment agencies both embarked on new marine in-
Deutsche Seewarte, Hamburg; Scripps Institution of vestigations, the former concentrating on basic
Oceanography, La Jolla, USA; Oceanographic Mu- science and the latter on applied science. In addition,
seum, Monaco; Plymouth Laboratory of Marine commercial interests were becoming active in re-
Biology, UK. These laboratories acquired and out- source exploration. The ships employed were mostly
fitted ships, which although mostly still conversions ex-wartime vessels: tugs, minesweepers, patrol boats,
of naval or commercial vessels, became recognized salvage vessels, etc. A 1950 survey of active research
oceanographic research vessels. The first vessel de- vessels showed 155 ships operated by 34 nations.
signed especially for marine research was the US Fish The first International Geophysical Year, 1957–58,
Commission Steamer Albatross built in 1882 for the brought together research ships of many nations
new laboratory at Woods Hole. This was a 72 m iron- working on cooperative projects. This decade also
hulled, twin-screw steamer. It also was the first vessel saw the formation of international bodies including:
equipped with electric generators (for lowering arc the International Oceanographic Congress; Inter-
lamps to attract fish and organisms at night stations). governmental Oceanographic Commission (IOC);
Based at the Woods Hole Laboratory, the Albatross the Special Committee on Oceanic Research (SCOR)
made notable deep sea voyages in both the Atlantic of the International Council of Scientific Unions, all
and Pacific Oceans. As with the Challenger, the Al- of which added to the growing pace of oceanic
batross became a legend in its own time and con- investigations.
tinued working until 1923. The decade of the 1960s is often referred to as the
In the early 1900s research voyages were aimed at ‘golden years of oceanography.’ Both public and
strategic regions. Nautical charting and resources scientific awareness of the oceans increased at an
exploration were concentrated in areas of political unprecedented rate. Funding for marine science both
significance: the Southern Ocean, South-east Asia, by government and private sources was generous,
and the Caribbean. Many of the ships were designed and the numbers of scientists followed suit. New
and constructed as research vessels, notably the shipboard instrumentation included the Chlorinity–
Discovery commanded by R. F. Scott on his first Temperature–Depth sounder which replaced the old
Antarctic voyage 1901–04. Another was the non- method of lowering water bottles and reversing
magnetic brig Carnegie built in 1909 which carried thermometers. Computers were available for ship-
out investigations worldwide. The Carnegie was the board data processing. Advances in instrumentation
first research vessel to carry a salinometer, an elec- and the new projects they generated were making the
trical conductivity instrument for determining the existing research ships obsolete; and the need for
chlorinity of sea water thus avoiding arduous new and more capable ships became a pressing issue.
chemical titrations. Research voyages by now were As a result, shipbuilding programs were started by
highly systematic on the pattern set by the Chal- most of the larger nations heavily involved in
lenger expedition. oceanographic research. One of the most ambitious
Another era of oceanic investigations began in was the construction of fourteen AGOR-3 Class
1925 with the German Atlantic Expedition voyage of ships by the US Navy. These ships, especially de-
the Meteor, a converted 66 m gunboat, that made signed for research and surveys, were 70 m in length,
transects of the South Atlantic Ocean using acoustic 1370 tonnes, and incorporated quiet ship operation,
echo sounders, modern sampling bottles, bottom centerwells, multiple echo sounders, and a full array
corers, current meters, deep-sea anchoring, and me- of scientific instrumentation. Of the fourteen in the
teorological kites and balloons. Unlike the random class, 11 were retained in the US and three were
cruise tracks of most earlier expeditions, the Meteor transferred to other nations. During this period, the
worked on precise grid tracklines. Ocean currents, Soviet Union also embarked on a major building
temperatures, and salinities, and bathymetry of the program which resulted by the mid-1970s in prob-
Mid-Atlantic Ridge were mapped with great ably the world’s largest fleet – both by vessel size and
accuracy. numbers. In 1979 there was an estimated total of
After World War II interest in oceanography and 720 research vessels being operated by 72 nations,
the marine sciences increased dramatically. As echo the USSR (194 ships), USA (115), and Japan (94)
sounding after World War I was a milestone in being the leading three.
oceanography, the advent of electronic navigation in By the mid-1980s, the shipbuilding boom which
the 1950s was another. Loran (LOng RAnge Navi- started in the late 1950s had dwindled, but many of
gation), a hyperbolic system using two radio stations those vessels themselves were becoming obsolete.
transmitting simultaneously, provided ships with New ships were planned to meet the growing needs
6 SHIPS

of shipboard investigators. This resulted in larger differences, there are commonalities that distinguish
ships with improved maneuverability, seakeeping, a research vessel from other ships. These are defined
and data-acquisition capabilities. The new ships built in the science mission requirements which set forth
to meet these requirements plus improvements to the operational capabilities, working environment,
selected older vessels constitute today’s oceano- science accommodations and outfit to meet the sci-
graphic research vessels. The worldwide fleet is now ence role for which the ship type is intended. The
smaller in terms of numbers than 25 years ago, but science mission requirements are the dominant fac-
the overall tonnage is greater and the capability tors governing the planning for a new vessel or the
vastly superior. conversion of an existing one. The requirements can
vary according to the size, area of operations, and
type of service, but the composition of the require-
The Nature of Research Vessels ments is a product of long usage.
The term ‘oceanographic research vessel’ is relatively A typical set of scientific mission requirements for
new; earlier ships with limited roles were ‘hydro- a large high-endurance general purpose oceano-
graphic survey vessels’ or ‘fisheries vessels.’ As mar- graphic research ship is as follows.
ine science evolved to include biological, chemical, 1. General: the ship is to serve as a large general
geological, and physical processes of the ocean and purpose multidiscipline oceanographic research ves-
its floor and the air–sea interface above, the term sel. The primary requirement is for a high endurance
‘oceanography’ and ‘oceanographic research vessel’ ship capable of worldwide cruising (except in close
has come to include all of these disciplines. pack ice) and able to provide both overside and la-
When research vessels became larger and more boratory work to proceed in high sea states. Other
numerous it was inevitable that regulations gov- general requirements are flexibility, vibration and
erning their construction and operation would come noise free, cleanliness, and economy of operation
into force. Traditionally, ships were either com- and construction.
mercial, warships, or yachts. Research ships with 2. Size: size is ultimately determined by require-
scientific personnel fit into none of these, and it be- ments which probably will result in a vessel larger
came necessary to recognize the uniqueness of such than existing ships. However, the length-over-all
ships in order to preclude burdensome and in- (LOA) should not exceed 100 m.
applicable laws. Most nations have now established 3. Endurance: sixty days; providing the ability to
a definition of an oceanographic research vessel. The transit to remote areas and work 3–4 weeks on sta-
United Nations International Maritime Organization tion; 15 000 mile range at cruising speed.
(IMO) has established a category of ‘Special Purpose 4. Accommodations: 30–35 scientific personnel in
Ship’ which includes ‘ships engaged in research, ex- two-person staterooms; expandable to 40 through the
peditions, and survey’. Scientific personnel are de- use of vans. There should be a science-library lounge
fined as ‘y all persons who are not passengers or with conference capability and a science office.
members of the crew y who are carried on board in 5. Speed: 15 knots cruising; sustainable through
connection with the special purpose y’ sea state 4 (1.25–2.5 m); speed control 70.1 knot in
United States law is more specific; it states the 0–6 knot range, and 70.2 knot in the range of
6–15 knots.1
The term oceanographic research vessel means a vessel that 6. Seakeeping: the ship should be able to maintain
the Secretary finds is being employed only in instruction in science operations in the following speeds and sea
oceanography or limnology, or both, or only in oceano-
states:
graphic or limnological research, including those studies
about the sea such as seismic, gravity meter, and magnetic
exploration and other marine geophysical or geological
• 15 knots cruising through sea state 4 (1.25–
2.5 m);
surveys, atmospheric research, and biological research.
• 13 knots cruising through sea state 5 (2.5–4 m);
The same law defines scientific personnel as those • 8 knots cruising through sea state 6 (4–6 m);
persons who are aboard an oceanographic research • 6 knots cruising through sea state 7 (6–9 m).
vessel solely for the purpose of engaging in scientific 7. Station keeping: the ship should be able to
research, or instructing or receiving instruction in maintain science operations and work in sea states
oceanography, and shall not be considered seamen. through 5, with limited work in sea state 7. There
The specific purposes of research vessels include:
hydrographic survey (mapping and charting); geo-
physical or seismic survey; fisheries; general purpose 1
1 (Nautical) mile ¼ 1.853 km, 1 knot ¼ 1.853 km h1 ¼ 0.515
(multidiscipline); and support vessels. Despite their m s1.
 SHIPS 7

should be dynamic positioning, both relative and ab- • a winch complex capable of handling 12 000 m of
solute, at best heading in 35-knot wind, sea state 5, 1.5 cm trawling or coring wire, and 10 000 m of
and 3-knot current in depths to 6000 m, using satellite 1.75 cm electromechanical conducting cable (up
and/or bottom transponders; 751 heading and 50 m to 10 kVA power transmission and fiberoptics);
maximum excursion. It should be able to maintain a this can be two separate winches or one winch
precision trackline (including towing) at speeds as low with two storage drums.
as 2 knots with a 451 maximum heading deviation
from the trackline under controlled conditions (satel- Additional special purpose winches may be in-
lite or acoustic navigation) in depths to 6000 m, in 35 stalled temporarily at various locations along work-
knot wind; and 3-knot current. Speed control along ing decks. Winch sizes may range up to 40 mtons and
track should be within 0.1 knot with 50 m of max- have power demands up to 250 kW. (See also
imum excursion from the trackline. multichannel seismics.) Winch control station(s)
8. Ice strengthening: ice classification sufficient to should be located for optimum operator visibility
transit loose pack ice. It is not intended for iceb- with communications to laboratories and ship con-
reaking or close pack work. trol stations.
9. Deck working area: spacious stern quarter area – 12. Overside handling: various frames and other
300 m2 minimum with contiguous work area along one handling gear able to accommodate wire, cable, and
side 4  15 m minimum. There should be deck loading free-launched arrays; matched to work with winch
up to 7000 kg m2 and there should be overside and crane locations but able to be relocated as ne-
holddowns on 0.5 m centers. The area should be highly cessary. The stern A-frame must have 6 m minimum
flexible to accommodate large, heavy, and portable horizontal and 10 m vertical clearances, 5 m inboard
equipment, with a dry working deck but not more than and outboard reaches, and safe static working load
2–3 m above the waterline. There should be a usable up to 60 mtons. It must be able to handle, deploy and
clear foredeck area to accommodate specialized towers retrieve very long, large-diameter piston cores up to
and booms extending beyond the bow wave. All 50 m length, 15 mtons weight and 60 mtons pullout
working decks should be accessible for power, water, tension. There should be provision to carry add-
air, and data and voice communication ports. itional overside handling rigs along working decks
10. Cranes: a suite of modern cranes: from bow to stern. (See also multichannel seismics).
13. Towing: capable of towing large scientific
(a) to reach all working deck areas and offload vans packages up to 4500 kg tension at 6 knots, and
and heavy equipment up to 9000 kg; 10 000 kg at 2.5 knots in sea state 5; 35 knots of
(b) articulated to work close to deck and water
wind and 3 knot current.
surface;
14. Laboratories: approximately 400 m2 of la-
(c) to handle overside loads up to 2500 kg, 10 m
boratory space including: main lab (200 m2) flexible
from side and up to 4500 kg closer to side;
for subdivision providing smaller specialized labs;
(d) overside cranes to have servo controls and mo-
hydro lab (30 m2) and wet lab (40 m2) both located
tion compensation;
contiguous to sampling areas; bio-chem analytical
(e) usable as overside cable fairleads at sea.
lab (30 m2); electronics/computer lab and associated
users space (60 m2); darkroom (10 m2); climate-
The ship should be capable of carrying portable
controlled chamber (15 m2); and freezer(s) (15 m2).
cranes for specialized purposes such as deploying and
Labs should be arranged so that none serve as
towing scanning sonars, photo and video devices,
general passageways. Access between labs should be
remotely operated vehicles (ROVs), and paravaned
convenient. Labs, offices, and storage should be
seismic air gun arrays.
served by a man-rated lift having clear inside di-
11. Winches: oceanographic winch systems with
mensions not less than 3  4 m.
fine control (0.5 m min1; constant tensioning and
Labs should be fabricated of uncontaminated and
constant parameter; wire monitoring systems with
‘clean’ materials. Furnishings, doors, hatches, venti-
inputs to laboratory panels and shipboard data sys-
lation, cable runs, and fittings should be planned for
tems; local and remote controls including laboratory
maximum lab cleanliness. Fume hoods should be
auto control.
installed permanently in wet and analytical labs.
Permanently installed general purpose winches
Main lab should have provision for temporary in-
should include:
stallation of fume hoods.
• two winches capable of handling 10 000 m of wire Cabinetry should be of high-grade laboratory
rope or electromechanical conducting cables quality with flexibility for arrangements through the
having diameters from 0.6 mm to 1.0 cm. use of bulkhead, deck and overhead holddown fittings.
8 SHIPS

Heating, ventilation, and air conditioning (HVAC) chest (1.5  3 m) located at the optimum acoustic
should be appropriate to labs, vans and other science location for afloat installation and servicing of
spaces being served. Laboratories should be able to transducers and transponders.
maintain a temperature of 20–231C, 50% relative 19. Multichannel seismics: all vessels shall have
humidity, and 9–11 air changes per hour. Filtered air the capability to carry out multichannel seismic
should be provided to the analytical lab. Each lab profiling (MCS) surveys using large source arrays
should have a separate electric circuit on a clean bus and long streamers. Selected vessels should carry an
and continuous delivery capability of at least 250 VA MCS system equivalent to current exploration in-
m2 of lab area. Total estimated laboratory power dustry standards.
demand is 100 kVA. There should be an un- 20. Navigation/positioning: There shall be a Glo-
contaminated seawater supply to most laboratories, bal Positioning System (GPS) with appropriate
vans, and several key deck areas. interfaces to data systems and ship control pro-
15. Vans: carry four standardized 2.5  6 m port- cessors; a short baseline acoustic navigation system;
able vans which may have laboratory, berthing, a dynamic positioning system with both absolute and
storage, or other specialized use. With hook-up relative positioning parameters.
provision for power, HVAC, fresh water, un- 21. Internal communications: system to provide
contaminated sea water, compressed air, drains, high quality voice communications throughout all
communications, data and shipboard monitoring science spaces and working areas. Data transmission,
systems. There should be direct van access to ship monitoring and recording systems should be avail-
interior at key locations. There should be provision able throughout science spaces including vans and
to carry up to four additional vans on working and key working areas. There should be closed circuit
upper decks with supporting connections at several television monitoring and recording of all working
locations. The ship should be capable of loading and areas including subsurface performance of equip-
off-loading vans with its own cranes. ment and its handling. Monitors for all ship control,
16. Workboats: at least one and preferably two environmental parameters, science and overside
inflatable boats located for ease of launching and equipment performance should be available in most
recovery. There should be a scientific workboat 8– science spaces.
10 m LOA fitted out for supplemental operation at 22. Exterior communications: reliable voice chan-
sea including collecting, instrumentation and wide- nels for continuous communications to shore stations
angle signal measurement. It should have 12 h en- (including home laboratories), other ships, boats and
durance, with both manned and automated oper- aircraft. This includes satellite, VHF and UHF. There
ation, be of ‘clean’ construction and carried in place should be: facsimile communications to transmit
of one of the four van options above. high-speed graphics and hard copy on regular
17. Science storage: total of 600 m3 of scientific schedules; high-speed data communications links to
storage accessible to labs by lift and weatherdeck shore labs and other ships on a continuous basis.
hatch(es). Half should have suitable shelving, racks 23. Satellite monitoring: transponding and re-
and tiedowns, and the remainder open hold. A sig- ceiving equipment including antennae to interrogate
nificant portion of storage should be in close prox- and receive satellite readouts of environmental re-
imity to science spaces (preferably on the same deck). mote sensing.
18. Acoustical systems: ship to be as acoustically 24. Ship control: the chief requirement is max-
quiet as practicable in the choice of all shipboard imum visibility of deck work areas during science
systems and their installation. The design target is operations and especially during deployment and
operationally quiet noise levels at 12 knots cruising retrieval of equipment. This would envision a bridge-
in sea state 5 at the following frequency ranges: pilot house very nearly amidships with unobstructed
stem visibility. The functions, communications, and
• 4 Hz–500 Hz seismic bottom profiling
layout of the ship control station should be designed
• 3 kHz–500 kHz echo sounding and acoustic
to enhance the interaction of ship and science oper-
navigation
ations; ship course, speed, attitude, and positioning
• 75 kHz–300 kHz Doppler current profiling
will often be integrated with science work requiring
The ship should have 12 kHz precision and control to be exercised from a laboratory area.
3.5 kHz sub-bottom echo sounding systems, and These science mission requirements are typical of a
provision for additional systems. There should be a large general-purpose research vessel. Requirements
phased array, wide multibeam precision echo for smaller vessels and specialized research ships can
sounding system; transducer wells (0.5 m diam) one be expected to differ according to the intended cap-
located forward and two midships; pressurized sea ability and service.
 SHIPS 9

Design Characteristics of deployed at sea has increased dramatically over the

Oceanographic Vessels past half century. As a result, the size of general
purpose vessels has increased significantly. A re-
In general the design of an oceanographic ship is search vessel of 60 m in length was considered to be
driven by the science mission requirements described large in 1950; the same consideration today has
above. In any statement of requirements an ordering grown to 100 m and new vessels are being built to
of priorities is important for the guidance of the that standard. Even existing vessels have been
design and construction of the ship. In the case of lengthened to meet the growing needs.
research vessels the following factors have been The majority of oceanographic research vessels,
ranked by groups of practicing investigators from all however, are smaller vessels, 25–50 m in length, and
disciplines. limited to coastal service.
1. Seakeeping: station keeping
2. Work environment: lab spaces and arrangements; Mapping and Charting Vessels
deck working area; overside handling (winches Mapping and charting ships were probably the
and wire); flexibility earliest oceanographic vessels, usually in conjunction
3. Endurance: range; days at sea with an exploration voyage. Incident to the estab-
4. Science complement lishment of marine trade routes, nautical charting of
5. Operating economy coastal regions became routine and the vessels so
6. Acoustical characteristics engaged were usually termed hydrographic survey
7. Speed: ship control ships. Surveys were (and still are) carried out using
8. Pay load: science storage; weight handling wire sounding, drags, and launches. Survey vessels
These priorities are not necessarily rank ordered are characterized by the number of boats and laun-
although there is general agreement among ocean- ches carried and less deck working space than gen-
ographers that seakeeping, particularly on station, eral purpose vessels. Modern survey vessels,
and work environment are the two top priorities. however, are often expected to carry out other sci-
The remaining are ranked so closely together that entific disciplines, and winches, cranes and frames
they are of equal importance. The science mission can be observed on these ships.
requirements set for each of these areas become Recent developments have affected the role (and
threshold levels, and any characteristic which falls therefore design) of this class of vessel. As a result of
below the threshold becomes a dominant priority. the International Law of the Sea Conferences, coastal
states began to exercise control over their continental
shelves and economic zones 200 miles from shore.
General Purpose Vessels This brought about interest in the resources (fishery,
bottom and sub-bottom) of these newly acquired
Ships of this type (also termed multidiscipline) con-
areas, and research vessels were tasked to explore
stitute the classic oceanographic research vessels and
and map these resources. The usual nautical charting
are the dominant class in terms of numbers today.
procedures are not applicable in the open ocean re-
They have outfitting and laboratories to support any
gions, and modern electronic echo-sounding instru-
of the physical, chemical, biological, and geological
ments have supplanted the older wire measurements.
ocean science studies plus ocean engineering. The sci-
This involves large hull-mounted arrays of acoustic
ence mission requirements given above describe a large
projectors and hydrophones which can map a swath
general purpose ship. Smaller vessels can be expected
of ocean floor with great precision up to five miles in
to have commensurately reduced requirements.
width, and at cruising speeds. The design of vessels
Current and future multidiscipline oceanographic
to carry this equipment requires a hull form to op-
ships are characterized as requiring significant open
timize acoustic transmission and reception, and to
deck area and laboratory space. Accommodations
minimize hull noise from propulsion and auxiliary
for scientific personnel are greater than for single
machinery. Further, such new ships also may be
purpose vessels due to the larger science parties car-
outfitted to perform other oceanographic tasks inci-
ried. Flexibility is an essential feature in a general
dent to surveys. Their appearance, therefore, may
purpose research vessel. A biological cruise may be
come closer to a general purpose vessel.
followed by geology investigations which can require
the reconfiguration of laboratory and deck equip-
Fisheries Research Vessels
ment within a short space of time.
In addition to larger scientific complements, the Fisheries research generally includes three fields of
complexity and size of instrumentation now being study: (1) environmental investigations, (2) stock
10 SHIPS

assessment, and (3) gear testing and development. often carry compressors and portable streamer reels
The first of these are surveys and analyses of sea sufficient for limited seismic profiling.
surface and water column parameters; both synoptic
and serial. These are biological, physical, and
chemical investigations (as well as geological if bot- Polar Research Vessels
tom fisheries are considered); and can be accom- Whereas most oceanographic research vessels are
plished from a general purpose oceanographic classed by the discipline in which they engage, polar
research vessel. research vessels are defined by their area of oper-
Ships engaged in fish stock assessment and ex- ations. Earlier terminology distinguished between
ploratory fishing, or development work in fishing polar research vessels and icebreakers with the for-
methods and gear, fish handling, processing, and mer having limited icebreaking capability, and the
preservation of fish quality on board, are specialized latter with limited or no research capability. The
types of vessels closely related to actual fishing ves- more current trend is to combine full research cap-
sels. Design characteristics include a stern ramp and ability into new icebreaker construction.
long fish deck for bringing nets aboard, trawling Arctic and Antarctic research ships in the nine-
winches, and wet labs for analyses of fish sampling. teenth and early twentieth centuries were primarily
Newer designs also include instrumentation and la- ice reinforced sealing vessels with little or no iceb-
boratories for environmental investigations, and ex- reaking capability. World War II and subsequent
tensive electronic instrumentation for acoustic fish Arctic logistics, and the International Antarctic
finding, biomass evaluation, and fish identification Treaty (1959) brought about increased interest in
and population count. polar regions which was furthered by petroleum
As with mapping and charting ships, most fishery exploration in the Arctic in the late twentieth cen-
research vessels are operated by government agencies. tury. Icebreakers with limited research capability
early in this period became full-fledged research
vessels by the end of the century.
Geophysical Research Vessels The special requirements defining a polar research
vessel include increased endurance, usually set at 90
The purpose of marine geophysical research vessels is days, helicopter support, special provisions for cold
to investigate the sea floor and sub-bottom, oceanic weather work, such as enclosed winch rooms and
crust, margins, and lithosphere. The demanding de- heated decks, and icebreaking capability. Other sci-
sign aspect for these ships is the requirement for a ence mission requirements continue the same as for a
MCS system used to profile the deep geologic struc- large general purpose RVs. Of special concern is
ture beneath the seafloor. The missions range from seakeeping in open seas. Past icebreaker hull shapes
basic research of the Earth’s crust (plate tectonics) to necessarily resulted in notoriously poor seakindli-
resources exploration. ness. Newer designs employing ice reamers into the
The primary components of an MCS system are hull form offer improved seakeeping.
the large air compressors needed to ‘fire’ multiple Ice capability is usually defined as the ability to
towed airgun sound source arrays and a long towed break a given thickness of level ice at 3 knots con-
hydrophone streamer which may reach up to 10 km tinuous speed, and transit ice ridges by ramming.
in length. The supporting outfit for the handling and Current requirements for polar research vessels have
deployment of the system includes large reels and varied from 0.75 m to 1.25 m ice thickness in the
winches for the streamer, and paravanes to spread continuous mode, and 2.0 to 3.0 m ridge heights in
the sound source arrays athwart the ship’s track. This the ramming mode. These correspond to Polar Class
latter results in the need for a large stern working 10 of Det Norske Veritas or Ice Class A3-A4 of the
deck close to the water with tracked guide rails and American Bureau of Shipping.
swingout booms. Electronic and mechanical work-
shops are located close to the working deck. The
Support Vessels
design incorporates a large electronics room for
processing the reflected signals from the hydro- Ships that carry, house, maintain, launch and retrieve
phones and integrating the imagery with magnetics, other platforms and vehicles have evolved into a
gravity and navigation data. class worthy of note. These include vessels that
The highly specialized design requirements for a support submersibles, ROVs, buoys, underwater
full-scale marine geophysics ship usually precludes habitats, and scientific diving. Earlier ships of this
work in other oceanographic disciplines. On the class were mostly converted merchant or fishing
other hand, large general purpose research ships vessels whose only function was to launch and
 SHIPS 11

retrieve and supply hotel services. Recent vessels, Table 1 Common measurements and observations made from
especially those dedicated to major programs such as RVs
submersible support, are large ships and fully out-
Underway On station
fitted for general purpose work.
Single channel echo sounding Echo sounding
Multichannel echo sounding Sub-bottom profiling
Other Classes of Oceanographic Research Vessels Sub-bottom profiling (3.5 kHz) Acoustic Doppler profiling
In addition to the above types, there are research Acoustic Doppler profiling Surface to any depth
Sea surface: Temperature
ships which serve other purposes. These include Temperature Salinity
ocean drilling and geotechnical ships, weather ships, Salinity Sound velocity
underwater archaeology, and training and education Fluorometry Dissolved oxygen
vessels. The total number of these ships is relatively Dissolved oxygen Water sample
small, and many of them merge in and out of the Towed magnetometer Bottom sampling
Gravimeter Bottom coring
category and serve for a limited stretch of time. Meteorological Bottom photography and video
Often ships will take on identification as a research Wind speed and direction Bottom dredging
vessel for commercial expediency or other fashion Barometric pressure Geothermal bottom probe
not truly related to oceanographic research. Such Humidity Biological net tows and trawls
roles may include treasure hunting, salvage, whale Solar radiation Biological net tows
Towed plankton recorder
watching, recreational diving, ecology tours, etc.
These vessels may increase the popular awareness of
oceanography but are not bona fide oceanographic
research vessels. Cooperative projects among research vessels in-
cluding different nations have become commonplace.
These share a common scientific goal, and cruise
tracks, times, methodology, data reduction, and
Research Vessel Operations archiving are assigned by joint planning groups.
Oceanographic cruises are usually the culmination of A significant factor affecting oceanographic research
several years of scientific and logistics planning. cruises today is the permission required by a research
Coastal vessels, typically 25–50 m length, will usu- vessel to operate in another nation’s 200-mile eco-
ally remain in a home, or adjacent regions on cruises nomic zone. As a result of the United Nations Law of
of 1–3 weeks’ duration. Ocean-going vessels, 50– the Sea Conventions and the treaties resulting there-
100 m length, may undertake voyages of 1–2 years from (1958–1982), coastal nations were given juris-
away from the home port, with cruise segments of diction over the conduct of marine scientific research
25–35 days working out of ports of opportunity. extending 200 nautical miles out from their coast
New scientific parties may join the ship at a port call (including island possessions). This area is termed ex-
and the nature of the following cruise leg can change clusive economic zone (EEZ). As of 2000 there were
from, for instance biological to physical ocean- approximately 151 coastal states (this number varies
ography. This involves complex logistics and careful according to the world political makeup), and 36% of
planning and coordination, within a typical 4–5 day the world ocean falls within their economic zones.
turnaround. Most coastal nations have prescribed laws governing
From the time of the Meteor expedition of 1925, research in their zones. The rules include requests for
cruise plans are usually highly systematic with work permission, observers, port calls, sharing data, and
concentrating along preset track lines and grids, or penalties. This often poses a burden on the operator of
confined to a small area of intense investigation. a research vessel when permission requirements are
Work can take place continuously while underway arbitrary and untimely. It is not, however, an un-
using hull-mounted and/or towed instrumentation, workable burden if done in an orderly manner, and
or the ship will stop at a station and lower instru- does have desirable features for international cooper-
ments for water column or bottom sampling. Typical ation. Problems arise when requests are not submitted
stations are at 15–60-mile intervals and can last within the time specified, or resulting data are not
1–4 h. Measurements or observations that are com- forthcoming. These are complicated by unrealistic re-
monly made are shown in Table 1. In addition, work quirements on the part of the coastal state, delay in
may include towed vehicles along a precise trackline acknowledging or acting on a request, or ignoring it
at very slow controlled speeds making many of the totally. These can have a profound effect on scientific
observations, chiefly acoustic, photographic, and research and need to be addressed in future Law of the
video. Sea Conventions.
12 SHIPS

Table 2 Countries operating 10 or more oceanographic Future Oceanographic Ships

research ships
The interest in, and growth of, marine science over
Russia 86 the past half century shows little or no indication of
United States 84 diminishing. The trend in oceanographic investi-
Japan 66 gations has been to carry larger and more complex
China 17
instrumentation to sea. The size and capability of
Ukraine 14
Korea 11 research vessels in support of developing projects has
Germany 11 also increased.
United Kingdom 10 Future oceanographic research ships can be ex-
Canada 10 pected to become somewhat larger than their coun-
All others (39) 111
terparts today. This will result from demands for
Total 420 more sizeable scientific complements and laboratory
spaces. Workdeck and shops will be needed for larger
equipment systems such as buoy arrays, bottom
stations, towed and autonomous vehicles. Larger
World Oceanographic Fleet overside handling systems incorporating motion
The precise number of oceanographic research ves- compensation will make demands for more deck
sels worldwide is difficult to ascertain. Few nations space.
maintain lists specific to research vessels, and num- There will be fewer differences between basic sci-
bers are available chiefly by declarations on the part ence, fisheries, and hydrographic surveying vessels so
of the operator. Some ships move in and out of a that one vessel can serve several purposes. This may
research status from another classification, e.g., result in fewer vessels, but overall tonnage and cap-
fishing, passenger, yacht, etc. Also, some operators acity can be expected to increase.
keep hydrographic survey, seismic exploration, and New types of craft may take a place alongside
even fisheries research ships as categories separate conventional ships. These include submarines, ‘flip’-
from oceanographic research; here they are included type vessels which transit horizontally and flip ver-
within the general heading of oceanographic re- tically on station, and small waterplane-area twin
search vessels. hull ships (SWATH). SWATH, or semisubmerged
Based on the best available information, 48 na- ships, are a relatively recent development in ship
tions or international agencies operated 420 design. Although patents employing this concept
oceanographic research vessels of size greater than show up in 1905, 1932, and 1946, it was not until
25 m LOA in 2000. Of these, 310 ships were from 1972 that the US Navy built a 28 m, 220 ton proto-
nine nations operating 10 or more ships each type model. The principle of a SWATH ship is that
(Table 2). submerged hulls do not follow surface wave motion,
A significant step in international cooperation af- and thin struts supporting an above water platform
fecting oceanographic research vessels and the mar- which have a small cross-section (waterplane) are
ine sciences they support has been the International nearly transparent to surface waves, and have longer
Ship Operators Meeting, an intergovernmental as- natural periods and reduced buoyancy force changes
sociation, founded in 1986, comprising represen- than a conventional hull. The result is that SWATH
tatives from various ship operating agencies that ships, both in theory and performance, demonstrate
meets periodically to exchange information on ship a remarkably stable environment and platform con-
operations and schedules, and work on common figuration which is highly attractive for science and
problems affecting research vessels. In 1999, 21 ship- engineering operations at sea.
operating nations were represented and extended
membership by other states is ongoing.
LOWERED, EXPENDABLE, OR TOWED
INSTRUMENTS
 CONTINUOUS PLANKTON RECORDERS
A. John, Sir Alister Hardy Foundation for Ocean a break for the Second World War, the survey re-
Science, Plymouth, UK started in 1946 and expanded into the eastern North
P. C. Reid, SAHFOS, Plymouth, UK Atlantic. Extension of sampling into the western
Copyright & 2001 Elsevier Ltd. North Atlantic took place in 1958. The survey
reached its greatest extent from 1962 to 1972 when
CPRs were towed for at least 120 000 nautical miles
annually. Sampling in the western Atlantic, which
had been suspended due to funding problems in
Introduction 1986, recommenced in 1991 and is still ongoing.
Figure 2A shows the extent of the survey in 1999.
The Continuous Plankton Recorder (CPR) survey is Initially based at the University College of Hull,
a synoptic survey of upper-layer plankton covering the survey moved to Leith, Edinburgh in 1950 under
much of the northern North Atlantic and North Sea. the management of the Scottish Marine Biological
It is the longest running and the most geographically Association (now the Scottish Association for Mar-
extensive of any routine biological survey of the ine Science). In 1977 it finally moved to Plymouth as
oceans. Over 4 000 000 miles of towing have re- part of the Institute for Marine Environmental Re-
sulted in the analysis of nearly 200 000 samples and search (now Plymouth Marine Laboratory). After a
the routine identification of over 400 species/groups short period of uncertainty in the late 1980s, when
of plankton. Data from the survey have been used to the continuation of the survey was threatened, the Sir
study biogeography, biodiversity, seasonal and inter- Alister Hardy Foundation for Ocean Science (SAH-
annual variation, long-term trends, and exceptional FOS) was formed in November 1990 to operate the
events. The value of such an extensive time-series survey. Since 1931 more than 200 merchant ships,
increases as each year’s data are accumulated. Some ocean weather ships, and coastguard cutters – known
recognition of the importance of the CPR survey was as ‘ships of opportunity’ – from many nations have
achieved in 1999 when it was adopted as an integral towed CPRs in a voluntary capacity to maintain the
part of the Initial Observing System of the Global survey. The Foundation is greatly indebted to the
Ocean Observing System (GOOS). captains and crews of all these towing ships and their
shipping and management companies, without
History whom the survey could not continue.
The CPR prototype was designed by Alister Hardy During the 1990s CPRs were towed by SAHFOS
for operation on the 1925–27 Discovery Expedition in several other areas, including the Mediterranean
to the Antarctic, as a means of overcoming the (1998–99), the Gulf of Guinea (1995–99), the Baltic
problem of patchiness in plankton. It consisted of a (1998–99), and the Indian Ocean (1999). A separate
hollow cylindrical body tapered at each end, survey by the National Oceanic and Atmospheric
weighted at the front and with a diving plane, hori- Administration/National Marine Fisheries Service
zontal tail fins, and a vertical tail fin with a buoyancy using CPRs along the east coast of the USA off
chamber on top (Figure 1A). Hardy designed a more Narragansett has been running since 1974; CPRs are
compact version with a smaller sampling aperture currently towed on two routes in the Middle Atlantic
for use on merchant ships and this was first deployed Bight. Following a successful 2000 mile trial tow in
on a commercial ship in the North Sea in September the north-east Pacific from Alaska to California in
1931 (Figure 1B). During the 1980s the design was July–August 1997, a 2-year survey by SAHFOS using
modified further to include a box-shaped double tail- CPRs in the north-east Pacific started in March 2000.
fin that provides better stability when deployed on In addition to five tows per year on the Alaska–
the faster merchant ships of today (Figure 1C). The California route, there is one 3000 mile tow annually
space within this tail-fin is used in some machines to east–west from Vancouver to the north-west Pacific
accommodate physical sensors and flowmeters. The (Figure 2B). A ‘sister’ survey, situated in the Southern
normal maximum tow distance for a CPR is ap- Ocean south of Australia between 601E and 1601E, is
proximately 450 nautical miles (834 km). operated by the Australian Antarctic Division. In this
By the late 1930s there were seven CPR routes in survey CPRs have been deployed since the early
the North Sea and one in the north-east Atlantic; in 1990s on voyages between Tasmania and stations in
1938 CPRs were towed for over 30 000 miles. After the Antarctic.

15
16 CONTINUOUS PLANKTON RECORDERS

V R
H J E F K
T
G

A
D B
P

C L
P
(A) (C)

Formalin tank

Propeller shaft Wire

Gears driven
fusee
Water
Water & plankton tunnel

Water exit

(B) Plankton, between two silk bands,stored in tank of formalin

Figure 1 (A) Diagram of the first Continuous Plankton Recorder used on ‘Discovery’. (Reproduced with permission from Hardy,
1967). (B) The ‘old’ CPR, used up to around 1983, showing the internal filtering mechanism. (C) The CPR in current use, with the ‘box’
tail-fin.

As the operator of a long-term international sur- supporting information, including sample locations,
vey, which has sampled in most of the world’s dates and times of samples, a taxon catalog, and
oceans, SAHFOS regularly trains its own staff in analyst details. In the near future it will also hold
plankton identification. In recent years SAHFOS has additional conductivity, temperature, and depth
also trained scientists from the following 10 coun- (CTD) data. Routine processing procedures ensure
tries: Benin, Cameroon, Côte d’Ivoire, France, Fin- that, despite various operational difficulties, the
land, Ghana, Italy, Nigeria, Thailand, and the USA. previous year’s data are usually available in the
database within 9 months.
In 1999 SAHFOS adopted a new open access data
The Database and Open Access Data Policy
policy, i.e. data are freely available to all users
The CPR database is housed on an IBM-compatible worldwide, although a reasonable payment may be
PC and stored in a relational Microsoft Access incurred for time taken to extract a large amount of
DATABASE system. Spatial and temporal data are data. The only stipulation is that users have to sign a
stored for every sample analysed by the CPR survey SAHFOS Data Licence Agreement. Details of the
since 1948. This amounts to 4175 000 samples, database can be found on the web site: http://
with around 400 more samples added per month. www.npm.ac.uk/sahfos/. This site advertises the
There are more than two million plankton data availability of data and allows requests for data to be
points in the database, which also contains made easily.
 CONTINUOUS PLANKTON RECORDERS 17

60˚N 60˚N

50˚N 50˚N

40˚N 40˚N

(A) 60˚W 40˚W 20˚W 0˚ 20˚E

_ 150 _ 145 _ 140 _ 135 _ 130 _ 125 _ 120

60

55

x 5 per year

50

60
45

50 x 1 per year 40

150 160 170 180 190 200 210 220 230 35

(B)

Figure 2 (A) CPR routes in 1999/2000. (B) CPR routes in the Pacific Ocean towed in 2000 and 2001.

The CPR Bibliography, which is available on the depth of 6–10 m. Water enters the CPR through a
SAHFOS web site, lists over 500 references using re- 12.7 mm square aperture and travels down a tunnel
sults from the survey. During the early years many of that expands to a cross-section of 50  100 mm,
the papers based on CPR data were published in the where it passes through a silk filtering mesh with a
‘in-house’ journal Hull Bulletins of Marine Ecology, mesh size of approximately 280 mm. The movement
which from 1953 onwards became the Bulletins of of the CPR through the water turns a propeller that
Marine Ecology; this was last published in 1980. drives a set of rollers and moves the silk across the
tunnel. At the top of the tunnel the filtering silk is
joined by a covering silk and both are wound onto a
Methods spool located in a storage chamber containing for-
Merchant ships of many nations tow CPRs each maldehyde solution. The CPRs are then returned to
month along 20–25 standard routes (Figure 2A) at a SAHFOS in Plymouth for examination. The green
18 CONTINUOUS PLANKTON RECORDERS

coloration of each silk is visually assessed by refer- Another possible application, in the context of the
ence to a standard color scale; this is known as new Pacific CPR programme, is an inter-comparison
‘Phytoplankton Color’ and gives a crude measure of with data from the CalCOFI Program, the only other
total phytoplankton biomass. The silks are then cut existing decadal-scale survey in the world sampling
into sections corresponding to 10 nautical miles marine plankton. This survey has taken monthly or
(18.5 km) of tow and are distributed randomly to a quarterly net samples from 1949 to the present over
team of 10–12 analysts. The volume of water filtered an extensive grid of stations off the west coast of
per 10-nautical-mile sample is approximately 3 m3. California. In the majority of samples the zoo-
Phytoplankton, small zooplankton (o2 mm in size) plankton has been measured only as displacement
and larger zooplankton (42 mm) are then identified volume, rather than being identified to species, but
and counted in a three-stage process. Over 400 dif- concurrently measured physical and chemical data
ferent taxa are routinely identified during the an- are more extensive.
alysis of samples and the recent expansion of the
survey into tropical waters and the Pacific Ocean will
Biogeography of Marine Plankton
certainly increase this figure.
A detailed and thorough quality control examin- Much of the early work of the survey focused on
ation is carried out by the most experienced analyst biogeography. Using Principal Component Analysis,
on the completed analysis data. Apparently anom- Colebrook was able to distinguish five main geo-
alous results are rechecked by the original analyst graphical distribution patterns in the plankton –
and the data are altered accordingly where necessary. northern oceanic, southern oceanic, northern inter-
This system ensures consistency of the data and acts mediate, southern intermediate, and neritic. Two
as ‘in-service’ training for the less experienced closely related species of calanoid copepod – Calanus
analysts. finmarchicus and C. helgolandicus – which co-occur
in the North Atlantic and are morphologically very
Instrumentation similar, show very different distributions (Figure 3).
C. finmarchicus is a cold-water species whose center
On certain routes CPRs carry additional equipment of distribution lies in the north-west Atlantic gyre
to obtain physical data. In the past temperature has and the Norwegian Sea (‘northern oceanic’). In
been recorded on certain routes in the North Sea contrast, C. helgolandicus is a warm–temperate
using BrainconTM recording thermographs, proto- water species occurring in the Gulf Stream, the Bay
type electronic packages, and AquapacksTM. Aqua- of Biscay and the North Sea (‘southern intermedi-
packs record temperature, conductivity, depth, and ate’). These different distribution patterns are re-
chlorophyll fluorescence. These are now deployed on flected in their life histories; C. finmarchicus
CPR routes off the eastern coast of the USA, in the overwinters in deep waters off the shelf edge,
southern Bay of Biscay and, until November 1999, in whereas C. helgolandicus overwinters in shelf
the Gulf of Guinea. VemcoTM minilogger tempera- waters.
ture sensors are used on routes from the UK to Ice- A new species of marine diatom, Navicula plana-
land, and from Iceland to Newfoundland. In order to membranacea Hendey, was first described from CPR
measure flow rate through the CPR, electromagnetic samples taken in 1962. The species was found to
flowmeters are used on some routes. Such recording have a wide distribution in the western North At-
of key physical and chemical variables simul- lantic from Newfoundland to Iceland.
taneously with abundance of plankton enhances our
An atlas of distribution of 255 species or groups
ability to interpret observed changes in the plankton. (taxa) of plankton recorded by the CPR survey be-
tween 1958 and 1968 was published by the Edin-
burgh Oceanographic Laboratory in 1973. An
Results and Applications of the Data
updated version of this atlas, covering more than 40
The long-term time-series of CPR data acts as a years of CPR data and over 400 taxa, is in
baseline against which to measure natural and an- preparation.
thropogenic changes in biogeography, biodiversity,
seasonal variation, inter-annual variation, long-term
Phytoplankton, Zooplankton, Herring, Kittiwake
trends, and exceptional events. The results have ap-
Breeding Data, and Weather
plications to studies of eutrophication and are in-
creasingly being applied in statistical analysis of A study in the north-eastern North Sea found that
plankton populations and modeling. Some examples patterns of four time-series of marine data and
are given below. weather showed similar long-term trends. Covering
 CONTINUOUS PLANKTON RECORDERS 19

60°W 40°W 20°W 0°

60°N 60°N

50°N 50°N

40°N 40°N

30°N 30°N
60°W 40°W 20°W 0°
Calanus finmarchicus
0.018 2.928 21.547

60°W 40°W 20°W 0°

60°N 60°N

50°N 50°N

40°N 40°N

30°N 30°N

60°W 40°W 20°W 0°

Calanus helgolandicus
0.000 0.200 1.293

Figure 3 Distribution of Calanus finmarchicus and C. helgolandicus recorded in CPR samples from 1958 to 1994.

the period 1955–87, these trends were found in the The mechanisms behind the parallelism in these data
abundance of phytoplankton and zooplankton (as over the 33-year period are still not fully understood.
measured by the CPR), herring in the northern North
Sea, kittiwake breeding success (laying date, clutch
Calanus and the North Atlantic Oscillation
size, and number of chicks fledged per pair) at a
colony on the north-east coast of England, and The North Atlantic Oscillation (NAO) is a large-
the frequency of westerly weather (Figure 4). scale alternation of atmospheric mass between
20 CONTINUOUS PLANKTON RECORDERS

3
Westerly weather
2
1
0
_1
_2
1950 1960 1970 1980 1990

3 3
Phytoplankton Zooplankton
2 2
1 1
0 0
_1 _1
_2 _2
1950 1960 1970 1980 1990 1950 1960 1970 1980 1990

3 3
Herring Kittiwake laydate
2 2
1 1
0 0
_1 _1
_2 _2
1950 1960 1970 1980 1990 1950 1960 1970 1980 1990

3 3
Kittiwake clutch Kittiwake chicks
2 2
1 1
0 0
_1 _1
_2 _2
1950 1960 1970 1980 1990 1950 1960 1970 1980 1990

Figure 4 Standardized time-series and 5-year running means for frequency of westerly weather, and for abundances of
phytoplankton, zooplankton, herring, and three parameters of kittiwake breeding (laying date, clutch size, and number of chicks
fledged per pair), from 1955 to 1987. (Reproduced with permission from Aebischer NJ et al. (1990) Nature 347: 753–755.)

subtropical high surface pressure, centred on the reverse is true. However, since 1995 this strong
Azores, and subpolar low surface pressures, centred Calanus/NAO relationship has broken down and the
on Iceland. The NAO determines the speed and dir- causes of this are presently unknown. It suggests a
ection of the westerly winds across the North At- change in the nature of the link between climate and
lantic, as well as winter sea surface temperature. The plankton in the north-east Atlantic.
NAO index is the difference in normalized sea level
pressures between Ponta Delgadas (Azores) and
North Sea Ecosystem Regime Shift
Akureyri (Iceland). There is a close association be-
tween the abundance of Calanus finmarchicus and Recent studies have shown changes in CPR Phyto-
C. helgolandicus in the north-east Atlantic and this plankton Color, a visual assessment of chlorophyll,
index (Figure 5). At times of heightened pressure for the north-east Atlantic and the North Sea. In the
difference between the Azores and Iceland, i.e. a central North Sea and the central north-east Atlantic
high, positive NAO index, there is low abundance of an increased season length was strikingly evident
C. finmarchicus and high abundance of C. helgo- after the mid-1980s. In contrast, in the north-east
landicus; during a low, negative NAO index the Atlantic north of 591N Phytoplankton Color
 CONTINUOUS PLANKTON RECORDERS 21

0.8 and more southerly tracks of the westerly winds and

0.7
Log Abundance

0.6 higher temperatures in western Europe. These

0.5 changes coincided with a series of other changes that
0.4 affected the whole North Sea ecosystem, affecting
0.3
0.2
many trophic levels and indicating a regime shift.
0.1 1999
1996 1997 1998
0_ _4 _2 North Wall of the Gulf Stream and Copepod
6 0 2 4 6
Numbers
NAO index
Zooplankton populations in the eastern North At-
Figure 5 Annual log abundance of Calanus finmarchicus in the lantic and the North Sea show similar trends to
north-east Atlantic Ocean against the NAO winter index for the
period 1962–99. (Adapted with permission from Fromentin JM,
variations in the latitude of the north wall of the Gulf
and Planque B (1996) Marine Ecology Progress Series 134: Stream, as measured by the Gulf Stream North Wall
111–118.) (GSNW) index, which is statistically related to the
NAO 2 years previously. Figure 7 shows the close
correlation between total copepods in the central
declined after the mid-1980s (Figure 6). These
North Sea and the GSNW index. This relationship is
changes in part appear to be linked to the recent high
also evident in zooplankton in freshwater lakes and
positive phase of the NAO index and reflect changes
in the productivity of terrestrial environments, indi-
in mixing, current flow, and sea surface temperature.
cating a possible climatic control.
The increase in Phytoplankton Color and phyto-
plankton season length after 1987 coincided with a
Biodiversity
large increase in catches of the western stock of horse
mackerel Trachurus trachurus in the northern North Analyses of long-term trends in biodiversity of zoo-
Sea, apparently connected with the increased trans- plankton in CPR samples indicate increases in diver-
port of Atlantic water into the North Sea. From 1988 sity in the northern North Sea. This may be related to
onwards the NAO index increased to the highest distributions altering in response to climatic change as
positive level observed in the twentieth century. geographical variation in biodiversity of the plankton
Positive NAO anomalies are associated with stronger shows generally higher diversity at low latitudes than

C North Sea CNE Atlantic NNE Atlantic

1995 1995 1995

1990 1990 1990

1985 1985 1985

1980 1980 1980

1975 1975 1975

1970 1970 1970

1965 1965 1965

1960 1960 1960

1955 1955 1955

1950 1950 1950

0 4 8 12 0 4 8 12 0 4 8 12
Months Months Months

Figure 6 Contour plots of mean monthly Phytoplankton Color during 1948–95 for the central North Sea, and for the central and
northern north-east Atlantic. (Reproduced with permission from Reid PC et al. (1998) Nature 391: 546.)
22 CONTINUOUS PLANKTON RECORDERS

Total copepods Central North Sea

1.1
2 r = 0.64
1.0

Log10 abundance
1 0.9

Arbitrary units
0.8
0
0.7
_1
0.6

_2 0.5
1965 1970 1975 1980 1985 1990 1995 2000

Figure 7 The latitude of the Gulf Stream (the GSNW index ‘arbitrary units’, broken line) compared with the abundance of total
copepods in the central North Sea (solid line). Adapted with permission from Taylor AH et al. (1992) Journal of Mar. Biol. Ass. UK 72:
919–921.

60˚
N

50˚
N

40˚
N

40˚W 20˚W

Taxonomic richness per CPR sample

1 1.5 2 2.5 3 3.5 4

Figure 8 The biodiversity (taxonomic richness) of calanoid copepods in the CPR sampling area. (Adapted with permission from
Beaugrand et al. (2000) Marine Ecology Progress Series 204: 299–303.)

at high latitudes. Calanoid copepods are the dominant diversity was found in the Gulf Stream extension, the
zooplankton group in the North Atlantic and the large Bay of Biscay, and along the southern part of the
data set from the CPR survey has been used to map European shelf. Cold water south of Greenland, east
their diversity. This has demonstrated a pronounced of Canada, and west of Norway was found to have
local spatial variability in biodiversity. Higher the lowest diversity (Figure 8).
 CONTINUOUS PLANKTON RECORDERS 23

Monitoring for Nonindigenous and Harmful Area 1

Algal Blooms 0.11
The regularity of sampling by the CPR enables it to 0.04
detect changes in plankton communities. Few case _ 0.03
histories exist that describe the initial appearance
46
and subsequent geographical spread of non- 50
indigenous species. In 1977 the large diatom Cosci- 54
58
nodiscus wailesii was recorded for the first time off 62
66
Plymouth, when mucilage containing this species Area 2
Ye 70 7
ar 4
was found to be clogging fishing nets. C. wailesii was

th 1
on 1
78

M 79
previously known only from southern California, the 0.30 82

5
86

3
1
Red Sea, and the South China Sea and it is believed 0.18
that it arrived in European waters via ships’ ballast
0.06
water. Since then the species has spread throughout
north-west European waters and has become an _ 0.06
important contributor to North Sea phytoplankton 46
50
biomass, particularly in autumn and winter. Such 54
58
introduced species can, on occasions, have con- 62
siderable ecological and economic effects on regional Area 3
66
Ye 70 7
ecosystems. ar 4

th 1
on 1
78

M 79
There has been an apparent worldwide increase in 0.07 82

5
86

3
the number of recorded harmful algal blooms and

1
0.03
Presence

the CPR survey is ideally placed to monitor such

_ 0.01
events. The serious outbreak of paralytic shellfish
46
poisoning that occurred in 1968 on the north-east 50
54
coast of England was shown by CPR sampling to 58
have been caused by the dinoflagellate Alexandrium 62
66
tamarense. Area 4 Ye 70
ar 74
Increased nutrient inputs into the North Sea since

th 1
78

on 1
M 79
0.39 82
the 1950s have been linked with an apparent in-

5
86

3
1
crease in the haptophycean alga Phaeocystis, par- 0.17
ticularly in Continental coastal regions where it _ 0.05
produces large accumulations of foam on beaches. In 46
contrast, long-term records (1946–87) from the CPR 50
54
survey, which samples away from coastal areas, 58
62
show that Phaeocystis has declined considerably in 66
the open-sea areas of the north-east Atlantic and the Area 5 Ye 70
0.11 ar 74 th 1
North Sea (Figure 9). It is notable that the decline 78
on 1
M 79

82
occurred both in areas not subject to anthropogenic 0.07 86
5
3
1

nutrient inputs (Areas 1 and 2, west of the UK) and 0.03

in the most affected area (Area 4, the southern North _ 0.01
Sea). This decrease in Phaeocystis up to 1980 is also 46
50
shown by many other species of plankton, suggesting 54
58
a common causal relationship. 62
66
Ye 70
ar 74
Exceptional Events
th 1
on 1

78
M 79

82
5

Doliolids are indicators of oceanic water and in CPR 86

3
1

samples are normally found to the west and south-

west of the British Isles; they occur only sporadically Figure 9 Presence of Phaeocystis in five areas of the north-
in the North Sea and are rarely recorded in the east Atlantic Ocean and the North Sea. Data are plotted for each
month for 1946–87 inclusive. (Reproduced with permission
central or southern North Sea. On two occasions in
from Owens NJP et al. (1989) Journal of Mar. Biol. Ass. UK 69:
recent years, in October–December 1989 and Sep- 813–821.)
tember–October 1997, the doliolid Doliolum natio-
nalis was recorded in CPR samples taken in the
24 CONTINUOUS PLANKTON RECORDERS

German Bight, accompanied by other oceanic indi- problems and now plays a significant role in our
cator species, suggesting a strong influx of north-east understanding of global ocean and climate change.
Atlantic water into the North Sea. Both these occa-
sions coincided with higher than average sea surface
temperature and salinities.
See also
Satellite Remote Sensing of Sea Surface Tempera-
Summary and the Future tures.
The long-term time-series of CPR data have been
used in many different ways:
• mapping the geographical distribution of Further Reading
plankton
• a baseline against which to measure natural and Colebrook JM (1960) Continuous Plankton Records:
anthropogenically forced change, including eu- methods of analysis, 1950–59. Bulletins of Marine
trophication and climate change Ecology 5: 51--64.
• linking of plankton and environmental forcing Gamble JC (1994) Long-term planktonic time series as
monitors of marine environmental change. In: Leigh RA
• detecting exceptional events in the sea
and Johnston AE (eds.) Long-term Experiments in
• monitoring for newly introduced and potentially
Agricultural and Ecological Sciences, pp. 365--386.
harmful species.
Wallingford: CAB International.
In the future new applications of CPR data may Glover RS (1967) The continuous plankton recorder
include: survey of the North Atlantic. Symp. Zoological Society
of London 19: 189--210.
• use as ‘sea-truthing’ for satellites Hardy AC (1939) Ecological investigations with the
• regional assessment of plankton biodiversity Continuous Plankton Recorder: object, plan and
• regional studies of responses to climate change methods. Hull Bulletins of Marine Ecology 1: 1--57.
• as input variables to predictive modeling for fish Hardy AC (1956) The Open Sea: Its Natural History. Part
1: The World of Plankton. London: Collins.
stock and ecosystem management
Hardy AC (1967) Great Waters. London: Collins.
• for construction and validation of new models
IOC and SAHFOS (1991) Monitoring the Health of the
comparing ecosystems of different regional seas.
Ocean: Defining the Role of the Continuous Plankton
The CPR survey has gathered nearly 70 years of Recorder in Global Ecosystem Studies. Paris: UNESCO.
data on marine plankton throughout the North At- Oceanographic Laboratory, Edinburgh (1973) Continuous
lantic Ocean, and has recently extended into the plankton records: a plankton atlas of the North Atlantic
North Pacific Ocean. Alister Hardy’s simple concept and the North Sea. Bulletins of Marine Ecology 7:
1--174.
in the 1920s has succeeded in providing us with a
Reid PC, Planque B, and Edwards M (1998) Is observed
unique and valuable long-term data set. There is in- variability in the observed long-term results of the
creasing worldwide concern about anthropogenic Continuous Plankton Recorder survey a response to
effects on the marine ecosystem, including eu- climate change? Fisheries Oceanography 7: 282--288.
trophication, overfishing, pollution, and global Warner AJ and Hays GC (1994) Sampling by the
warming. The data in the CPR time-series is being Continuous Plankton Recorder survey. Progress in
used more and more widely to investigate these Oceanography 34: 237--256.
 CTD (CONDUCTIVITY, TEMPERATURE, DEPTH)
PROFILER
A. J. Williams, III, Woods Hole Oceanographic surface along the wire releases a latch, which causes
Institution, Woods Hole, MA, USA the thermometer to invert, breaking the column of
& 2009 Elsevier Ltd. All rights reserved. mercury in the capillary tube, thus capturing the
volume of mercury expanded into the capillary
tube (Figure 1). When the reversing thermometer is
returned to the surface, this length of mercury in the
capillary tube can be measured, the change in length
Introduction: The In Situ Measurement due to the change in temperature from the sample
of Salinity, Temperature, and Density depth to the surface corrected for, and the in situ
in the Ocean temperature of the seawater at the sampled depth
computed. The technique is sensitive and reliable
One of the most useful instruments developed for with well-characterized reversing thermometers
determining seawater properties during the last four
decades has been the CTD (conductivity, tempera-
ture, depth). This device has supplanted the tradi-
tional hydrocast using Nansen bottles and reversing
thermometers that was standard physical oceano-
graphic practice from about 1910 to 1970. The
CTD, although an electronic instrument, has its
origin in the older technology. The computations of
properties such as depth, salinity, density, speed of
sound, and potential temperature have been greatly
facilitated by having the measurements of conduct-
ivity, temperature, and pressure in digital format for
direct entry into standard formulas, originally in
FORTRAN but now in Matlab and other computa-
tional engines.

Temperature
The temperature of seawater is directly important
because many physical properties depend upon
temperature and indirectly important because cal-
culations of salinity from conductivity measurements
are dominated by the temperature dependence of
conductivity. Temperature is a conservative property
of seawater. It is generally modified only when the
fluid is at the surface where it can exchange heat with
the atmosphere or rarely when it is in contact with
another body of water where exchanges of heat may
occur by mixing. Temperature has been sampled in Figure 1 The reversing thermometer has a constriction in the
capillary tubing that causes the mercury column to break when the
hydrocasts (hydrographic stations) with reversing thermometer is returned to its upright orientation after being
thermometers since the late nineteenth century. In a deployed upside down. This allows the temperature to be
reversing thermometer, mercury in a glass bulb ex- measured at depth with the thermometer in the normal,
pands or contracts filling a capillary tube to a greater connected column, but then when the thermometer is inverted
or less extent much as occurs in a normal fever by the agency of a messenger sent down the hydrographic wire,
the mercury beyond the constriction is captured and can be read
thermometer. However, when the reversing thermo- upon recovery to the surface. An auxiliary thermometer allows the
meter has equilibrated with the seawater at its depth surface reading temperature to be applied to correct for the
along the hydrowire, a messenger falling from the expansion of the trapped mercury column. General Oceanics, Inc.

25
26 CTD (CONDUCTIVITY, TEMPERATURE, DEPTH) PROFILER

where a long history of their calibration has been A continuous profile of temperature became pos-
kept. Resolution of temperature in deep expanded- sible first with the bathythermograph (BT) and sub-
range thermometers is typically 1 or 2 millidegree. sequently by the STD (salinity, temperature, depth)
However, hydrostatic pressure at depth would com- profiler. The BT, developed by Athelstan Spillhaus in
press the glass and cause the mercury to move along the 1930s, was a valuable tool during World War II
the capillary a greater distance than if the pressure for US destroyers and submarines in determining
were kept at 1 atm, so in situ temperature is meas- the acoustic refraction in the upper water column
ured with a pressure-protected reversing thermo- with consequences for acoustic submarine detection.
meter inside a pressure-resisting glass tube. Sharp thermal interfaces were revealed at places
The depth of the measurement can be determined where sound was refracted by the difference in speed
even if the hydrowire is not a straight vertical line the of sound of the water on the two sides of the inter-
length of which can be measured. Current shear in face. In the BT, a glass slide plated with carbon
the water column bends the hydrowire into a curve (smoked) or a thin gold layer was scratched by a
and causes the sample depth to be less than that stylus that was moved along one arc by expansion of
determined from the meter wheel reading: the dis- a fluid-filled coiled tube responding to temperature
tance that was paid out from that when the reversing and along another arc by a second coiled tube
thermometer entered the water until the lowering (Bourdon tube) empty of fluid and acted upon by
was stopped. So for more than a century it has been pressure. The fine scraped line on the slide could be
possible to determine the temperature profile from read in a viewer that was calibrated for the BT with
the surface to the bottom but only at discrete inter- which it was used. In practice, temperatures at
vals of depth. A heavily instrumented hydrographic regular depths were read off and transcribed to a
cast may have had a dozen reversing thermometers paper chart, thus missing the promise of a continu-
on it but these only gave the temperature at a dozen ous profile of temperature. In fact, there were fre-
depths. In some stations, several casts are made with quently apparent jiggles in the traces that were
instruments only in, say, the top 1000 m for higher initially ignored and only much later were discovered
resolution on one and deeper instruments with more to be real indications of physical phenomena and
widely spaced depths on another (Figure 2). termed fine structure.
Electronic instrumentation, accepted grudgingly at
first, made the measurement of temperature with a
platinum resistance thermometer a practical option
at sea. This sensor in which a temperature-sensitive
element of fine platinum wire is placed inside a
pressure-protecting metal sleeve was stable and
could be calibrated to a few millidegrees or even to
sub-millidegree precision. The near absence of work
hardening of the platinum wire and its freedom from
corrosion made the platinum resistance thermometer
a standard in temperature measurement in the lab
and in the sea. It did, however, require some as-
sumptions to define the specific resistance at tem-
peratures other than the critical points of boiling
water (100 1C) at standard pressure (1013 millibar)
and normal isotopic composition, and the melting
point of ice (0 1C) under the same conditions. The ice
melting point is sensitive to trace contamination with
electrolytes and to the state of the water in thermal
equilibrium with the ice, so a more reliable critical
point has been used to pin down the low-temperature
end of the platinum resistance thermometer scale,
the triple point of water. When standard isotopic
composition water is in equilibrium with all three
phases, liquid (water), vapor (steam), and solid (ice),
Figure 2 The photograph shows a hydrographic cast being and there are no other fluids or vapors present, the
taken with a scientist on the hero platform attaching a bottle to the temperature is 0.010 1C, a very well characterized
hydrographic wire. Woods Hole Oceanographic Institution. temperature. Sealed triple point cells have been
 CTD (CONDUCTIVITY, TEMPERATURE, DEPTH) PROFILER 27

Thermometer
Chilled liquid
Thermometer

Water vapor Thermometer well

Metal bushing
Ice and water
Ice mantle
Thermometer Inner melt
well

Ice mantle
Pure water

Pure water
Ice and water

Container

Triple point of water cell

Figure 3 The triple point cell is made with a film of ice in a sealed finger of glass containing only pure degassed water. At the point
where the ice surrounding the internal cavity starts to melt, the temperature is 0.010 1C. Sea-Bird Electronics, Inc.

constructed to establish ice/water/vapor equilibrium example, global warming, so the exact temperature
for precise temperature calibration. A triple point scale used for calibration and interpretation is critical.
cell has an internal cavity into which a sensor to be Thermistors are semiconductors where the resist-
tested can be placed (Figure 3). Before its use, an ice ance changes much more radically with a change in
layer is grown on the outside of the cavity in contact temperature than the platinum resistance thermo-
with the water by dropping dry ice into the cavity meter but they are not as stable. Thermistors are
and freezing the water to a thickness somewhat less heterogeneous ceramic beads exhibiting a negative or
than the clearance of the cavity from the outer wall a positive temperature coefficient of conductivity; the
of the triple point cell. Then a little warming is per- latter are generally used in this application as they
mitted until the ice covering can be spun around the become more conductive as they get warmer. The
cavity without sticking. This is the point where the beads, however, frequently contain defects that grow
cavity wall is at the triple point, 0.010 1C. or heal and change the thermistor resistance at a
Two points may define a linear relation between fixed temperature, so ultrastability is not a charac-
resistance and temperature but in reality, this rela- teristic of thermistors unlike platinum resistance
tionship is not exactly linear. The temperature scale is thermometers. Thermistors are useful as a secondary
in practice defined thermodynamically as that which sensor of temperature but are calibrated against a
an ideal gas would obey: PV ¼ nRT, where P is abso- platinum resistance thermometer which in turn is
lute pressure, V is specific volume, n is the number of calibrated against a triple point cell. The XBT or
moles of gas, R is the Boltzman constant, and T is the expendable bathythermograph uses a thermistor to
absolute temperature. Even though there is no ideal profile temperature from a sinking body that is
gas that obeys this relation due to van der Waals forces temporarily connected to the launcher by a fine
between molecules of gas at low temperatures, helium electrical wire (Figure 4).
approximates it better than any other gas. A practical There is one other enhancement of the calibration
temperature scale based upon resistance of a platinum of temperature in the oceanographic temperature
resistance thermometer was defined in 1978 but range. A gallium triple point cell can be used to es-
careful work led to an improved temperature scale tablish a critical point, 29.764 6 1C, closer to the
being established in 1983 that retained the critical oceanic temperature range than the boiling point of
endpoints but adjusted the nonlinear shape between water. In any case, the calibration of thermometers is
the endpoints with a difference of several millidegrees critical for long-duration monitoring of oceanic
(about 5 millidegree) near 35 1C. This is a serious error water masses to discern warming trends but is less
to introduce into long-term measurements of, for important for gradient measurements.
28 CTD (CONDUCTIVITY, TEMPERATURE, DEPTH) PROFILER

Deck-mounted
launcher
model LM 2A

Hand-held
launcher
model LM 3A
Launcher cable −
supplied with
launcher

Ship’s
hull

MK12 mating cable −

connects to
J1 on MK12 board

Thru-hull PC with MK12

launcher interface Connector box −
Probe model LM4A board supplied wired
to mating cable

Figure 4 XBT or expendable bathythermograph is a continuous recording temperature profiler that can be dropped from a moving
ship or an aircraft. A thermistor in the nose of the probe is sensed through a fine wire that is streamed by both the falling probe and the
moving vessel. Depth is assumed from time elapsed after probe contact with the water and is calibrated for fall rate. Lockheed Martin
Sippican, Inc.

As electronic temperature measurements were been determined with great care by laboratory
installed upon profilers, in particular the STD pro- methods. Of almost as great interest as the density
filer (Bissett-Berman, c. 1968), observations of fine (and more for water watchers) is the salinity. Salinity
structure in temperature came to be accepted as real, is, if anything, more conservative than temperature
which led to an understanding of internal thermal and only changes upon exposure of the water to the
mixing and stratification. Microstructure, as the atmosphere or another body of water at a different
layering at scales of meters first became known, salinity where mixing may occur. Hydrothermal
turned out to result from double-diffusive convection vents on the seafloor have been suspected of also
in many cases and produced layers on the order of changing the salinity of deep water masses by in-
tens of meters thick, which became known as fine jection of dissolved materials. The present theory
structure and which are now known to be nearly even suggests that hydrothermal vent circulation es-
ubiquitous in the ocean (Figure 5). tablishes the overall salinity of the sea, not river
runoff followed by evaporation. So determination of
salinity is important, not just for the equation of
Salinity state where salinity affects density.
The equation of state of seawater, which relates the The direct determination of salinity is awkward. It
density to temperature, salinity, and pressure, has is defined as the weight of solids in grams in 1 kg of
 CTD (CONDUCTIVITY, TEMPERATURE, DEPTH) PROFILER 29

Salinity (%°) The resulting instrument was the lab salinometer and
38.50 38.60
it allowed faster, more precise determinations of sal-
600
inity than the older titration method. But it did re-
quire a supply of standard seawater for comparison.
Standard seawater is a commodity required by the
S case for oceanographic cruises where many salinity
samples are expected. On a typical hydrographic cast,
Depth (m)

T each reversing thermometer was attached to a salinity

700 sample bottle. Fridtjof Nansen perfected metal sample
bottles used on a hydrowire at the turn of the twen-
tieth century and these, which were triggered with the
same messenger that caused the reversing thermo-
meters to reverse, closed valves at the top and bottom
capturing a water sample that could later be titrated
800 or compared to standard seawater in a salinometer
13.4 13.6 13.8 (Figure 6). Reversing thermometers were commonly
Temperature (°C) attached to Nansen bottles so that a single triggering
action might capture a temperature and a salt sample
Figure 5 The profile of sheets and layers from an instrument at one instant. Standard seawater is collected at an
named self-contained imaging microprofiler (SCIMP) shows at open ocean site of which several were used initially.
very high resolution the microstructure in the Tyrrhenian Sea.
Currently the IAPSO standard seawater is of North
The sensor in SCIMP was the first CTD to use internal recording.
Freedom from a cable allowed very smooth sinking and high Atlantic origin. After being filtered through a 200-nm
spatial resolution of temperature and salinity. Molcard R and filter, this water is evaporated and diluted to a specific
Williams AJ, III (1975) Deep stepped structure in the Tyrrhenian conductivity to serve as a standard, the principle
Sea. Memoires Societe Royale des Sciences de Liege VII: being that for many thousands of samples the ionic
191–210.
composition is essentially the same. By bringing its
conductivity to a standard value with the addition or
seawater when the water has been evaporated and all subtraction of pure water, all samples are assumed to
the carbonates have been converted to oxides, be interchangeable. An earlier standard, Copenhagen
bromine and iodine converted to chlorine, and all standard seawater, was unfortunately not represen-
organic matter completely oxidized. Direct evapor- tative of all seawater of the world being somewhat
ation does not work because chlorides are lost. But a anomalous in having a different ionic composition,
simpler indirect measure can be based on the almost so that the constant 0.03 had to be added to the
constant composition of seawater (same ratios of chlorinity determined by titration. However, as a
major ions nearly everywhere, only the water content conductivity standard rather than as a chlorinity
varies). This involves titrating the chloride (and other standard, Copenhagen water served well.
halogens) with silver nitrate and indicating with The temperature bath-controlled salinometer was
potassium chromate. The relation is a boon to laboratory analyses of salinity measure-
ments from bottles but it was not sufficient for an
salinity ¼ 0:03 þ 1:805  chlorinity ocean-going salinity-profiling instrument. Two more
steps were required. The first was to determine the
Titrating is slow and awkward, so an attempt was temperature coefficient of conductivity and this per-
made to determine the salinity by electrical conduct- mitted correction of the conductivity measurement
ivity measurements. The comparison was made be- without a thermostatic bath. The principal variable
tween conductivity of diluted standard seawater and responsible for conductivity changes in seawater is
full-strength standard seawater at a common tem- temperature, not salinity, so the temperature had to
perature. The relation was fairly linear even though be measured very accurately and the lab work done
seawater is more than a very dilute solution. Once the very carefully. Actually, the conductivity ratio could
relations were worked out from laboratory measure- still be used as long as the standard seawater was at
ments, it became possible to measure salinity by the same temperature as the unknown sample. But, it
putting the unknown sample in a temperature bath was also possible to just calibrate the salinometer
and measuring the ratio of its conductivity to that occasionally with standard seawater and to calculate
of a known sample in the same temperature bath. the difference in conductivity expected from the
Schleicher and Bradshaw at Woods Hole Oceano- temperature of the sample, which was different from
graphic Institution did this work in the 1960s. the standard seawater calibration temperature.
30 CTD (CONDUCTIVITY, TEMPERATURE, DEPTH) PROFILER

Figure 6 A case of standard seawater flasks is shown as carried on research cruises where many samples will be taken and
analyzed at sea. The standard seawater is used to calibrate the salinometer.

Salinometers permitted salinities to be run at sea CTD Sensors

from Nansen bottles, which improved accuracy
somewhat because saltwater samples can sometimes Temperature can be measured to about 2 millidegree
get spoilt if kept too long. But the observations were with reversing thermometers and salinity can be relied
made from only a few points in the profile. Then Neil upon to a few parts per million. To improve on this,
Brown in 1961 combined pressure measurements Brown aimed for resolution of salinity to 1 ppm
with conductivity and temperature measurements to which required resolution of temperature to 0.5 mil-
make an in situ sampler, the STD profiler. Schleicher lidegree Celsius. Stability had to be very good
and Bradshaw determined the pressure effect on to make calibrations to this standard meaningful.
conductivity (by now a three-variable problem) and For standards work, the platinum thermometer is
Brown and Allentoft extended the conductivity ratio used and Brown chose that for the CTD. To minimize
measurements. size and retain high stability with the conductivity
The STD opened a new window on the ocean and measurement, Brown designed a ceramic, platinum,
immediately presented problems for physical oceano- and glass conductivity cell. For pressure, he used a
graphers by showing fine structure in a way that could strain gauge bridge on a hollow cylinder. Using tem-
no longer be ignored. The STD converted the con- perature, conductivity, and pressure, salinity can be
ductivity measurement to salinity with in situ analog calculated and from temperature, salinity, and pres-
circuitry using temperature and pressure. However, sure, density can be calculated. Depth can then be
only a few years later, computers began to go to sea determined from pressure, the integral of density to
and Brown realized a better algorithm could be ap- the surface, and a local value for gravity. The cor-
plied to raw digital conductivity, temperature, and rection from pressure to depth is small and for many
pressure measurements by shipboard-based computers purposes inconsequential so that profiles of tempera-
than by using the analog conversions in the in situ ture and salinity against pressure are used in place of
instrumentation. Furthermore, if recorded digitally, profiles against calculated depth in most cases.
the original data could always be reprocessed at a later
time upon the improvement of the algorithm. Finally,
the precision and accuracy of the measurement could
Pressure
be improved and the size of the sensors reduced to Originally Brown planned to build each sensor him-
push the scale of observations from the meter to the self but technology in the commercial world pro-
centimeter scale. It was the latter that drew Brown to vided him with an adequate pressure sensor initially
Woods Hole Oceanographic Institution in 1969 to in a pressure transducer produced by Paine Instru-
develop the microprofiler. This was the first CTD. ments, Inc., subsequently improved with temperature
 CTD (CONDUCTIVITY, TEMPERATURE, DEPTH) PROFILER 31

Signal + reasonably stable, this technique did not offer the

Excitation high spatial resolution desired in the CTD nor was it
the only route to removing the difficulties with
electrodes, so Brown elected to design a miniature
stable conductivity cell for the CTD.
The conductivity cell of Brown’s CTD had four
electrodes, two for current and two to measure
voltage, to minimize electrode effects with a sym-
metry that made it insensitive to local contamination
of the electrodes. It was only 3 mm in diameter and
8-mm long, so it was hoped that this small size
Excitation − would be able to resolve centimeter-scale structure.
Signal −
Measurements were made at 10 kHz to circumvent
Figure 7 This drawing of a strain gauge pressure sensor polarization at the current electrodes. Other geom-
shows four sets of wires around the body of the sensor. Internal etries for electrode-type conductivity cells have been
pressure from the port at the left stretches the center of the used, a three-electrode cell from Sea-Bird Electronics,
cylinder more than the heavy ends but the thermal expansion is Inc., for example (Figure 8). Conductivity cell
similar in all four coils. Driven as a resistance bridge, the sensing
leads experience a doubling of the effect of a single coil but
design is a present occupation of sensor technolo-
cancel the thermal effect on stretching of the wires. gists. For example, long-duration, fast cells are being
developed by Ray Schmitt of Woods Hole Oceano-
graphic Institution for deployment on gliders
correction for reducing hysteresis during a profile. (autonomous underwater vehicles that profile along
Typical lowering speeds for deep CTD profiles are oblique paths by gliding between two depths).
between 30 and 100 m min  1, limited by the need to Original plans by Brown to make his own ther-
prevent the cable from going slack and jumping the mometer, in a helium-filled ceramic capillary tube,
sheave at the top or getting a loop in the cable near were discarded when it was discovered how hard it
the bottom. But this speed causes the temperature in was to work with ceramics. Endless difficulties in
the instrument housing to vary rapidly, especially glass to ceramic and glass to metal seals developed
while transiting the thermocline, and temperature and overcoming these in the conductivity cell which
gradients inside the instrument are a problem for had no voids was hard enough. A commercial plat-
sensors that were designed for constant temperature. inum resistance thermometer was chosen from
The hysteresis in pressure has remained a problem up Rosemont Inc., with a time constant of 300 ms and a
to the present time and has only been tolerated be- guaranteed stability of 10 millidegree in a year but in
cause the errors are not serious for the computation practice somewhat better.
of salinity. They are principally of concern when The 300-ms response time of the thermometer
trying to measure motion of a surface of constant meant that for 1-cm resolution, his original target,
salinity or temperature between profiles or between descent rates of 0.3 m min  1 would be the limit. This
the down profile (clean because the instrument is at was a bitter result; however, Brown added a fast re-
the leading edge of the insertion) and the up profile a sponse thermistor to correct the temperature meas-
few minutes to an hour later (Figure 7). urement at faster descent rates. The correction
technique added the derivative of the thermistor
temperature measurement in an analog circuit to the
Conductivity
stable platinum resistance thermometer temperature
Direct measurement of conductivity presents prob- measurement to replace the high-frequency variations
lems because of polarization of seawater at the that were lost, without affecting the overall accuracy
electrodes, so the salinity measurements of the STD of the platinum resistance temperature measurement
made electronically used an inductive cell without in less dynamic regions. Later, he increased the size of
electrodes. In this cell, made with dimensionally the conductivity cell to facilitate manufacture. This
stable materials to fix the geometry, a toroidal potentially degraded spatial resolution. Flushing of
transformer was constructed in which seawater the conductivity cell has been an issue and the original
formed a single shorted secondary turn through the cell, although only 8-mm long, had a flushing length
hole. Electric current was induced in this shorted at speeds above 10 cm s  1 of about 3.5 cm. The new
turn and the conductivity of the seawater in this cell flushed in about 8 cm of descent during lowering
path measured as a transformed conductivity in on a hydrographic station. With a thermistor response
the toroidal primary winding of the cell. While time of 30 ms, a 10-cm vertical resolution was
32 CTD (CONDUCTIVITY, TEMPERATURE, DEPTH) PROFILER

Platinum electrodes Borosilicate Current field

(3 places) glass cell between electrodes

Seawater Seawater
flow in flow out

Cell terminals

Figure 8 This drawing of a conductivity cell shows three electrodes, reducing the external electric field from the current between the
outer electrodes and the central electrode. Sea-Bird Electronics, Inc.

possible at descent speeds of 50 cm s  1 or 30 m

min  1, a reasonable winch speed. The requirement of
resolving structure to 10 cm at a descent rate of
30 m min  1 meant a sample rate of 10 Hz. (The ori-
ginal resolution target was higher and the first
microprofiler had three channels running at 32 ms
each in parallel.) The subsequent Neil Brown Instru-
ment Systems Mark III CTD successively digitized
conductivity, pressure, and temperature at 32 ms in
each channel so that it obtained a complete sample
every 96 ms, which was fast enough to resolve 10-cm
thick features at a lowering speed of 30 m min  1.
Practical solutions to the requirements of very high
accuracy and reasonable speed are hallmarks of the
very best instruments and the Mark III CTD estab-
lished a standard. The range in temperature is about
32 1C, from freezing to nearly the warmest surface
water. For packing efficiency, straight binary integers
were used and the value of 215 is 32 768. Thus a 16-bit
measurement of temperature gives 0.5 millidegree
resolution and 0–32.8 1C range. (For some work, a
 2 1C lower end is needed and this was later in-
corporated.) Conductivity varies over the same range
because it tracks temperature. Sixteen bits generally
permit salinities up to 38% to be measured to 0.001%
precision. The depth range needed for most of the
ocean is 6500 m (or a pressure range of 6500 decibar) Figure 9 A CTD with rosette sampler of large Nisken bottles
and, with a digitizer capable of 16-bit resolution, and reversing thermometers is being lowered off the side of a
10-cm depth resolution is permitted, again right on ship. Woods Hole Oceanographic Institution.
target for the resolution of the sensors. But to make a
measurement to a part in 65 536 (216) and have it
remain accurate and stable is not easy. Furthermore, profiler to be developed. Sea-Bird Electronics, Inc.,
the conductivity measurement must be made at about produced sensors for temperature and conductivity
10 kHz to minimize electrode polarization. Neil based upon a Wein bridge oscillator that were pre-
Brown’s solution to these problems was to make all of cise, compact, and easy to incorporate into instru-
the digitizations with transformer windings, weighted ments and soon Sea-Bird produced its own CTD
in a binary sequence and added electronically. These based upon these sensors. Sea-Bird’s conductivity cell
were driven at 10 kHz so that polarization effects were is a three-electrode cell with a slow natural flushing
minimal. Precision in the measurements was ensured time but it is generally flushed with an external pump
by the turns ratio in the transformer. to establish a constant flushing time irrespective of
While the Neil Brown CTD was the first of the lowering rate, thus improving the spatial resolution.
new profiling instruments, others soon followed. This device is now widely used on oceanographic
Guildline, Inc., produced an excellent lab salino- research vessels. Ocean Sensors, Inc., has produced a
meter and that technology similarly permitted a small CTD with internal recording for inclusion on
 CTD (CONDUCTIVITY, TEMPERATURE, DEPTH) PROFILER 33

Figure 10 Neil Brown (1927–2005), a long-time contributor to precise continuous measurements of physical properties of the ocean,
shown here confronting the CTD he invented, had a great sense of humor. His use of transformer-based analog-to-digital converters
provided the precision that permitted electronic measurements of temperature, conductivity, and pressure to be made in profilers.

autonomous instruments. Idronaut, S.r.l., has a CTD algorithm has been clumsy in that it sticks close to the
which can accommodate additional sensors for actual relations derived from the laboratory data sets.
oxygen, carbon dioxide, ambient light, pH, and op- These were derived by going from salinity to con-
tical backscatter, for example. In fact, the ability to ductivity, not the other way around. Dynamically one
add sensors to a digital instrument has been an ad- wants to know density and this is now computed as a
vantage not lost on instrument builders, so that a second step when it could be done directly. But com-
typical profiler package on a ship may contain a putational complexity is a negligible cost with even the
CTD with a suite of these additional sensors, some smallest PCs. Many CTDs export salinity directly,
sensors duplicated for redundancy, and generally having done the conversion internally. However, con-
includes a rosette sampler of Niskin bottles to cap- ductivity recording still permits enhanced processing if
ture water on trigger from the surface. raw values are retained. A direct measurement of
density might be the next step in sensor development.
However, the demand for sensors of light, chemistry,
Extended Deployments of the CTD and properties other than the classical ones of tem-
Battery-supplied power for the CTD and the devel- perature, salinity, and pressure has seemed to be more
opment in the early 1970s of digital magnetic tape significant and the modern CTD is often just the cen-
recorders freed the CTD from cable connection to tral element of a complex suite of sensors.
the ship (Figure 9). The salinity and temperature
profiles shown in Figure 5 were recorded on a Sea-
Data cassette tape from a free vehicle. Now with
See also
massive solid state memory such as compact flash, Gliders. Satellite Remote Sensing: Salinity Mea-
replacing hard disks of the 1980s and 1990s, digital surements. Satellite Remote Sensing of Sea
data storage is not a problem and CTDs are found on Surface Temperatures.
autonomous underwater vehicles, moored profilers,
and on gliders and floats.
The CTD is now a standard oceanographic instru-
ment and has replaced the Nansen cast as a hydro- Further Reading
graphic tool (Figure 10). The data are sent up
Bacon S, Culkin F, Higgs N, and Ridout P (2004) IAPSO
conducting cable as a frequency-shift-keyed signal.
standard seawater: Definition of the uncertainty in the
Multiconductor or fiber-optic armored cable has re- calibration procedure, and stability of recent batches.
placed the hydrographic wire for most hydrographic Journal of Atmospheric and Oceanic Technology 24:
surveys. Recording as an acoustic signal on tape has 1785--1799.
been replaced by direct storage of digital data on a PC Bradshaw A and Schleicher KE (1970) Direct measurement
or dedicated server but the decoding and computation of thermal expansion of seawater under pressure. Deep-
of salinity is still generally done on deck. The Sea Research 17: 691--706.
34 CTD (CONDUCTIVITY, TEMPERATURE, DEPTH) PROFILER

Bradshaw A and Schleicher KE (1980) Electrical conducti- Williams AJ, III (1974) Free-sinking temperature and
vity of seawater. IEEE Journal of Oceanic Engineering salinity profiler for ocean microstructure studies. IEEE
5: 50--56. International Conference on Engineering in the Ocean
Brown NL and Allentoft B (1966) Salinity, Conductivity Environment 2: 279--283.
and Temperature. Relationships of Seawater over the
Range of 0–50%. US ONR Contract Nr-4290(00), Relevant Websites
MJO 2003 Final Report, Washington, DC.
Cox RA, Culkin F, and Riley JP (1967) The electrical http://www.paineelectronics.com
conductivity/chlorinity relationship in natural seawater. – Downhole and Differential Pressure Transducer,
Deep-Sea Research 14: 203--220. Sensor, and Transmitter, Paine Electronics.
Dauphinee TM (1980) Introduction. Special Issue on the http://www.sippican.com
Practical Salinity Scale 1978. IEEE Journal of Oceanic – Expendable Probes, Sippican, Inc.
Engineering 2: 1--2. http://www.generaloceanics.com
Forch C, Knudsen M, and Sorensen SPL (1902) Berichte – General Oceanics, Inc.
uber die Konstantenbestimmungen zur Aufstellung der http://www.guildline.com
hydrographischen Tabellen. Kgl. Danske Videnskab. – Guildline Instruments.
Selskabs Skrifter, 7 Rackke, Naturvidensk. Og Mathem. http://www.idronaut.it
Afd. 12, pp. 1–151. – Idronaut, S.r.l.
Lewis EL and Perkin RG (1978) Salinity: Its definition and http://www.agu.org
calculation. Journal of Geophysical Research 83: 466--478. – News article ‘Athelstan Spilhaus dies at 86’ (30 March
Miyake M, Emery WJ, and Lovett J (1981) An evaluation 1998), Science and Society (American Geophysical Union).
of expendable salinity–temperature profilers in the http://www.mnc.net
eastern North Pacific. Journal of Physical Oceano- – Nordic Visitors Norway.
graphy 11: 1159--1165. http://www.oceansensors.com
Molcard R and Williams AJ, III (1975) Deep stepped – Ocean Sensors, Inc.
structure in the Tyrrhenian Sea. Memoires Societe http://www.emersonprocess.com
Royale des Sciences de Liege VII: 191--210. – Rosemount Temperature Products, Emerson Process
Ridout P and Higgs N (online) An Overview of the IAPSO Management.
Standard Seawater Service. Havant: OSIL. http:// http://www.seabird.com
www.ptb.de/de/org/3/31/313/230ptbsem/230ptbsem_ – Sea-Bird Electronics, Inc.
osil_ridout.pdf (accessed Mar. 2008). http://www.whoi.edu
UNESCO (1981) International Oceanographic Tables, – Woods Hole Oceanographic Institution.
Vol. 3. UNESCO Technical Papers in Marine Science
39, pp. 1–111. Paris: UNESCO.
 DEEP-SEA DRILLING METHODOLOGY
K. Moran, University of Rhode Island, Narragansett, ship’s motion from the drill pipe; and a pump that
RI, USA flushes sea water through the drill pipe.
& 2009 Elsevier Ltd. All rights reserved. Open hole methods are successfully used in all of
the Earth’s oceans (Figure 2). Scientific ocean-drilling
achievements include drilling in very deep water
(6 km) and to 42 km below the seafloor (Table 1).
Although there have been many achievements using
Introduction these methods, there are also limitations. Although the
exact depth limit of open-hole drilling method is not
The technology developed and used in past scientific yet known, it is likely limited to 2–4 km below the
drilling programs, the Deep Sea Drilling Project and seafloor. This limitation exists because the drill fluid
the Ocean Drilling Program, has now been expanded must be modified to a lower density so that the deep
in the current Integrated Ocean Drilling Program cuttings can be lifted from the bit and flushed out of
(IODP). These technologies include innovative dril- the hole when drilling deep into the seafloor. Another
ling methods, sampling tools and procedures, in situ
measurement tools, and seafloor observatories. This
new IODP technology will be used to drill deeper
into the seafloor than was possible in the previous
scientific programs. The first drilling target of the
IODP using the new technology is the seismogenic
zone offshore Japan, a location deep in the Earth
(7–14 km) where earthquakes are generated.

Drilling Technology
The Deep Sea Drilling Project and the Ocean Drilling
Program used the same basic drilling technology, the Drill pipe
open hole method. Today, the IODP has extended
this capability to include closed hole methods,
known as riser drilling.
Drilling fluid is
pumped down
through drill pipe
Open Hole or Nonriser Drilling
Drilling fluid and
Drilling is the process of establishing a borehole. The cuttings flow
open hole method uses a single drill pipe that hangs Cuttings into ocean
from the drill ship’s derrick, a tall framework posi-
tioned over the drill hole used to support the drill pipe.
Seafloor
The drill pipe is rotated using drilling systems, spe-
cifically a hydraulically powered top drive located
Surface casing
above the drill floor of the ship. Surface sea water is
flushed through the center of the pipe to lubricate the
rotating bit that cuts the rock and then flushes sedi- Drilling fluid
and cuttings flow
ment and rock cuttings away to the seafloor (Figure 1). up between the
Open hole refers to the resulting borehole which re- Second casing drill pipe and the
mains open to the ocean during drilling. This method borehole or casing
is also called a riserless drilling system. Important
parts of the deep-water drilling system are a drilling
Uncased hole
derrick that is large and strong enough to hang a Drill bit
long length of drill pipe reaching deep ocean and
subseafloor depths (up to 8 km); a system that rotates
the drill pipe; a motion compensator that isolates the Figure 1 Diagram of a nonriser drilling system.

35
 36
DEEP-SEA DRILLING METHODOLOGY
Figure 2 Map of all sites drilled by the Deep Sea Drilling Project (DSDP), the Ocean Drilling Program (ODP), and the IODP.
 DEEP-SEA DRILLING METHODOLOGY 37

Table 1 Fact sheet about the JOIDES Resolution, the research vessel used by the Ocean Drilling Program

Total number of days in port 445 days

Total number of days at sea 4751 days
Total distance traveled 507 420 km
Total number of holes drilled 1445 holes
Deepest water level drilled Leg 129 5980 m
Deepest hole drilled Leg 148, Hole 504B 2111 m
(South-eastern
Pacific Ocean,
off coast of Ecuador)
Total amount of core 180 880 m
Most core recovered on single Leg 175 – Benguela 8003 m
(15 Aug.–10 Oct. 1997)
Northern-most site drilled Leg 113, site 911 Latitude 80.47441 N
Longitude 8.22731 E
Southern-most site drilled Leg 151, site 693 Latitude 70.83151 S
Longitude 14.57351 W
Year and place of constitution 1978 Halifax, Nova Scotia,
Canada
Laboratories and other scientific equipment installed 1984 Pascagoula, Mississippi
Gross tonnage 9719 tons
Net tonnage 2915 tons
Engines/generators Seven 16 cyl index Diesel
5@2100 kW (2815 hp)
2@15500 kW (2010 hp)
Length 143 m
Beam 21 m
Derrick 62 m
Speed 11 knots
Crusing range 120 days
Scientific and technical party 50 people
Ship’s Crew 65 people
Laboratory space 1115 m2
Drill string 8838 m

limitation of the open hole method is that drilling technology has been used, in shallow to intermediate
must be restricted to locations where hydrocarbons water depths, by the offshore oil industry to explore
are unlikely to be encountered. In an open hole, there for, and produce oil and gas.
is no way to control the drilling fluid pressure. In lo- Riser drilling uses two pipes: a drill pipe similar to
cations where oil and gas may exist, the formations that used for open hole drilling and a wider diameter
are frequently overpressured (similar to a champagne riser pipe that surrounds the drill pipe and is ce-
bottle). If these formations would be punctured with mented into the seafloor (Figure 3). The system is
an open hole system, the drill pipe would act like a closed because drill fluid (seawater and additives) is
straw that connects this overpressured zone in the pumped down the drillpipe (to lubricate the bit and
rock to the ocean and the ship. This type of puncture flush rock cuttings away from the bit) and then re-
is called a ‘blow-out’ and is a serious drilling hazard. turned to the ship via the riser. With riser drilling, the
The explosion as gases are vented through the straw to drill fluid density can be varied and borehole pressure
the ship’s drill floor could cause serious damage, or can be monitored and controlled, thus overcoming
worse yet, the change in the density of the seawater as the two limitations of open hole drilling. The single
the gas bubbles are released into the overlying ocean limitation of riser drilling is water depth. The current
could cause the ship to sink. Without a system to water depth limit of riser technology is approxi-
control the pressure in the borehole, there is no way to mately 3 km. This water depth limit occurs because
prevent a blow-out. the riser, filled with drill fluid and cuttings, puts a
large amount of pressure on the rock. This pressure is
Closed Hole or Riser Drilling
greater than the strength of the rock; thus, the rock
The new deep-sea drilling technology used in IODP breaks apart under the riser pressure, causing the
is a closed system, also known as riser drilling. This drilling system to fail.
38 DEEP-SEA DRILLING METHODOLOGY

Cuttings unlatch the wireline tool, return it to the ship, and

returned recover the core sample.
to ship
In scientific ocean drilling, three standard wireline
core sampling tools are used: the advanced piston
corer, the extended core barrel, and the rotary core
barrel.
The piston corer is advanced into the sediment
ahead of the drill bit using pressure applied by ship-
board pumps through the drill pipe (Figure 4(a)). The
Riser drill fluid pressure in the pipe is increased until the
Drill pipe corer shoots into the sediment. After the corer is shot
10 m ahead of the bit, it is recovered using the wireline
overshot. The drill pipe and bit are then advanced by
rotary drilling another 10 m, in preparation for taking
Drilling fluid is
pumped down another core sample. The piston corer is designed to
Drilling fluid and
through drill pipe
cuttings flow up recover undisturbed core samples of soft ooze and
between the drill sediments up to 250 m below the seafloor. In soft to
pipe and the riser hard sediments, the piston corer can achieve 100%
recovery. However, because the cores are taken se-
quentially, sediment between consecutive cores may
not be recovered. To ensure that a continuous sedi-
Seafloor
mentary section is recovered, particularly for paleo-
climate studies, the IODP drills a minimum of three
Surface casing
boreholes at one site. The positions of the breaks
between consecutive core samples are staggered in
Drilling fluid
and cuttings flow
each borehole so that if sediment is not sampled in
up between the one borehole at a core break depth, it will be re-
drill pipe and the covered in the second or third borehole.
Second casing borehole or casing
The extended core barrel is a modification of
the oil industry’s rotary corer and is designed to re-
cover core samples of sedimentary rock formation
Uncased hole
Drill bit (Figure 4(b)). Typically, the extended core barrel is
deployed at depths below the seafloor at which the
sediment is too hard for sampling by the piston corer.
Figure 3 Diagram of a riser drilling system. The extended corer uses the rotation of the drill
string to deepen the borehole and cut the core sam-
ple. The cutting action is done with a small bit at-
tached to the core barrel. An innovation of this tool
Deep-Sea Sampling
is an internal spring that allows the core barrel’s
Scientific ocean drilling not only requires a borehole, smaller bit to extend ahead of the drill bit in softer
but – more importantly – the recovery of high- formations. In hard formations, where greater cut-
quality core samples taken as continuously as pos- ting action is needed, the spring is compressed and
sible. Recovering sediment and rock samples from the small core bit rotates with the main drill bit.
below the seafloor in deep water requires the use of The rotary core barrel is a direct descendant of the
wireline tools. Wireline tools are pumped down the rotary coring system used in the oil industry and is
center of the drill pipe to the bottom (called the similar to the extended core barrel in its retracted
bottom hole assembly) where they are mechanically mode. The rotary corer is designed to recover core
latched into place near the drill bit in preparation for samples from medium to very hard formations, in-
sampling. Different types of tools are advanced into cluding igneous rock. The corer uses the rotation of
the geological formation and take a core sample in the drill string and the main drill bit to deepen the
different ways, depending on the type of sediment or hole and cut the core sample (Figure 4(c)).
rock. After the tool samples the rock formation, it is In all drilling operations, there are times when the
unlatched from the bottom hole assembly with a drill pipe must be recovered to the ship, for example,
mechanical device, called an overshot, that is sent to change a worn bit or to install different bottom
down the pipe on a wire. The overshot is used to hole equipment. Before pulling the pipe out of the
 DEEP-SEA DRILLING METHODOLOGY 39

Figure 4 Diagrams of the Ocean Drilling Program’s coring tools: (a) advanced piston corer; (b) extended core barrel; and (c) rotary
core barrel.

borehole, a re-entry cone (Figure 5) is dropped down source. However, when gas hydrate is sampled, it must
the outside of the drill pipe where it free-falls to the be kept at in situ pressure conditions to maintain the
seafloor. The cone is used as a guide to re-enter the integrity of the core. Thus, a pressure core sampler is
borehole with a new bit or equipment. The cone is a used to sample hydrates. The sampler is similar to the
very small target in deep water (1–6 km) and ancil- extended corer in that it has its own bit, but it has an
lary tools are needed to locate it for re-entry into the internal valve that closes before the sampler is re-
borehole. The cone is first located acoustically, using moved from the formation. The closed valve main-
seafloor transponders. To precisely pinpoint the tains the sample at in situ pressure conditions.
cone, a video camera is lowered with the drill pipe to
visually pinpoint the location and drop the drill pipe
back into the borehole.
Special corers are used to sample unusual and dif-
Drilling Measurements
ficult formations. For example, gas hydrates, ice-like Once sampling is completed, logging tools are lowered
material that is stable under high pressure and low into the borehole to measure the in situ geophysical
temperature, commonly occur in deep water below and chemical properties of the formation. In IODP,
the seafloor and require special samplers. Gas hydrates logging tools, developed for use in the oil industry, are
have generated much public interest since they can most commonly leased from Schlumberger. The ori-
contain methane gas trapped within their structure, gins of logging go back to 1911 when the science of
which is thought to be a potential future energy geophysics was new and was just beginning to be used
40 DEEP-SEA DRILLING METHODOLOGY

Drill pipe
connection
Removable
line wiper

Derrick Side-entry port

for logging tools
and wireline

Float valve
Moon pool
23′ 4–3/8′′
Thrusters
O.A.L.

Logging
tool

Drill pipe

Drill pipe
5.7 mi

connection

Figure 6 Diagram of the side-entry sub technology.

associates, they extended the electrical prospecting

technique from the surface to the oil well. Now, the
use of electric prospecting, called logging, is widely
accepted as a standard method in oil exploration.
Logging tools are lowered into the borehole using
Re-entry cone a cable that also transmits the data, in real time, to
Acoustic
beacon the ship. The term logging refers to the type of data
collected. For example, a borehole log is a record or
ledger of the sediment and rock encountered while
drilling. Logs are geophysical and chemical records
Sediment of the borehole. The logging tools typically comprise
transmitters and sensors or a sensor alone encased
in a robust stainless steel tube. Examples of tool
measurements include electrical resistivity, gamma
ray attenuation, natural gamma, acoustic velocity,
and magnetic susceptibility. Log data provide an al-
Hard rock
most continuous record of the sediment and rock
Not to scale
formation along length of the borehole.
Log data are of high quality when collected in the
open hole, outside of the drill pipe. The most com-
mon method for deploying logging tools is to deploy
Figure 5 Diagram of the ship, drill pipe, and re-entry cone used
a device that releases the drill bit from the drill pipe
in the drilling process. once all drilling and sampling operations are com-
pleted. The drill bit falls to the bottom of the bore-
hole and is left there. The drill pipe is retracted and
to explore the internal structure of the Earth. Conrad only 75–100 m of pipe is left at the top of the bore-
and Marcel Schlumberger, the founders of Schlum- hole to keep the upper, loose part of the borehole
berger, conceived the idea that electrical measurements stable. Logging tools are lowered through the drill
could be used to detect ore (precious minerals). pipe to the bottom of the borehole. Log data are
Working at first alone and then with a number of acquired by slowly raising the tool up the borehole at
 DEEP-SEA DRILLING METHODOLOGY 41

a constant speed. When boreholes are unstable and the side-entry sub, positioned well below the ship. In
the walls are collapsing into the open hole, another the way, the logging cable does not interfere with
method is used, unique to scientific ocean drilling. removal of drill pipe.
In unstable conditions, the drill pipe cannot be re- Data in the borehole are also collected using
tracted to within 75 m of the seafloor without bore- wireline-deployed tools. Sediment temperature is
hole walls collapsing and blocking or bridging the measured with a temperature sensor mounted inside
hole. In these situations, after the bit is released, the cutting edge of the piston corer. In addition, other
the logging tools are lowered inside the drill pipe to tools are deployed that do not recover a sample.
the bottom of the hole. The drill pipe is retracted They are pushed into the sediment ahead of the drill
only enough to expose the logging tools to the open bit, left in place for 10–15 min to record tempera-
hole, while protecting the remainder of the borehole ture, and then pulled back to the ship, where the data
walls. Then, drill pipe is retracted at the same speed are downloaded and analyzed. Special wireline tools
as logging tools are pulled up through the borehole. have also been used to measure pressure and fluid
The technology developed that allows for this unique flow properties of sediment and rock.
operation is called the side-entry sub (Figure 6). IODP also uses an oil industry-developed method
When inserted as a part of the drill string, the cable, for logging sediment and igneous rocks – logging-
to which the tools are attached, exits the drill pipe at while-drilling or LWD. In this method, some of the

Data logger

Re-entry cone

Hydraulic
feed
through

Packer

Thermistor string

Zone I Screen

Zone II

Optional osmosampler

Sinker bar

Optional bridge plug screen

Figure 7 Diagram of the advanced CORK system used to seal instruments in the borehole.
42 DEEP-SEA DRILLING METHODOLOGY

same wireline logging sensors are repackaged and Summary

attached to the drill string, immediately behind the
drill bit. The sensors log the geophysical properties Deep-sea scientific drilling applies innovative sam-
of the sediment and rock as drilling proceeds, thus pling, instrument, and observatory technologies to
eliminating problems associated with borehole the study of Earth system science. These range from
collapse. the study of Earth’s past ocean and climate con-
ditions using high-quality sediment cores, to the
study of earthquakes and tectonic processes using
Deep Seafloor Observatories logging tools and seafloor observatories, to exploring
gas hydrates (a potential future energy source) using
The circulation obviation retrofit kit (CORK) is a
specialized sampling tools.
seafloor observatory that measures pressure, tem-
In 2004, the IODP succeeded two earlier scientific
perature, and fluid composition – important param-
programs, Deep Sea Drilling and Ocean Drilling. The
eters for the study of the dynamics of deep-sea
IODP operates two ships: the D/V JOIDES Resolution
hydrologic systems. CORKs are installed by the
and the D/V Chikyu and leases ships for special op-
IODP for measurements over long periods of time
erations in shallow water and ice-covered seas. The
(months to years). Since 1991, observatories have
IODP is supported by the US, Japan, and Europe.
been installed on the deep seafloor in different set-
tings, for example, at mid-ocean ridge hydrothermal
systems and at active margins.
The CORKs are installed by the drill ship. After a See also
borehole is drilled, a CORK is installed to seal in- Deep Submergence: Science of. Manned
struments in the borehole away from the overlying Submersibles, Deep Water. Remotely Operated
ocean (Figure 7). The CORK has two major parts: Vehicles (ROVs).
the CORK body that provides the seal and an in-
strument cable, for measuring fluid pressures, and
temperatures that hangs from the CORK into the Further Reading
borehole. A data recorder is included with the in- DSDP (1969–86) Initial Reports of the Deep Sea Drilling
strument cable. The data recorders have sufficient Project. Washington, DC: US Government Printing
battery power and memory for up to 5 years of op- Office. http://www.deepseadrilling.org/i_reports.htm (ac-
eration. Data are recovered from CORKs using cessed Mar. 2008).
manned submersibles or remotely operated vehicles. Integrated Ocean Drilling Program (2001) Earth, Oceans
The instruments in the CORK measure pressure and and Life: Scientific Investigation of the Earth System
temperature spaced along a cable that extends into Using Multiple Drilling Platforms and New
the sealed borehole. The CORK also includes a valve Technologies, Integrated Ocean Drilling Program,
above the seal where borehole fluids can be sampled. Initial Science Plan, 2003–2013.
Joint Oceanographic Institutions (1996) Understanding
Advanced CORKs are also used to isolate specific
Our Dynamic Earth: Ocean Drilling Program Long
and measure the properties of different sediment or Range Plan. Washington, DC: Joint Oceanographic
rock zones. Institutions.
The IODP installs another type of long-term sea- Oceanography Society (2006) Special Issue: The Impact of
floor observatory for earthquake studies. Seismic the Ocean Drilling Program. Oceanography 19(4).
monitoring instruments are installed in deep bore- ODP (1985–present) Proceedings of the Ocean Drilling
holes located in seismically active regions, for ex- Program, Initial Reports, vols. 101–210. College Station,
ample, off the coast of Japan. These data are used to TX: ODP. http://www-odp.tamu.edu/publications (acces-
help established predictive measures to prevent loss sed Mar. 2008).
of life and damage to cities during large earthquakes. ODP (1985–present) Proceedings of the Ocean Drilling
Deep-sea seismic observatories contain a strain- Program, Scientific Results, vols. 101–210. College
Station, TX: ODP. http://www-odp.tamu.edu/publications
meter, two seismometers, a tilmeter, and a tempera-
(accessed Mar. 2008).
ture sensor. The observatories have replaceable ODP (2004–present) Proceedings of the Integrated Ocean
data-recording devices and batteries like CORKs, Drilling Program, vols. 300–. College Station, TX:
and are serviced by remotely operated vehicles. Integrated Ocean Drilling Program.
Eventually, real-time power supply and data retrieval WHOI (1993–94) Oceanus: 25 Years of Ocean Drilling,
will be possible when some of the observatories are vol. 36, no. 4. Woods Hole, MA: Woods Hole Oceano-
connected to nearby deep-sea fiber-optic cables. graphic Institution.
 EXPENDABLE SENSORS
J. Scott, DERA Winfrith, Dorchester, Dorset, UK and air-dropped expendables can clearly outstrip
Copyright & 2001 Elsevier Ltd.
even this.
The technique also provides standard results, and
the expendable bathythermograph, or XBT, is now
considered a central component of global climate
monitoring programs such as the Global Ocean
Observation System (GOOS). Near-real-time trans-
Introduction fer of these data from ships under way is now also an
Expendable sensors represent an approach to ocean important input to global meteorological forecasting.
measurement in which some degree of measurement The XBT was originally intended to improve on
precision may be sacrificed in the interests of lower the (nonexpendable) mechanical bathythermograph
costs and operational expediency. Two requirements (MBT) which was the principal operational naval
of physical oceanography have driven their devel- device up to the mid-1970s. The MBT was lowered
opment: the problem of achieving adequate spatial into the water from a vessel, and it inscribed a tem-
sampling of the ocean on timescales commensurate perature-depth trace, with a sharp stylus on a small
with temporal variability; and the requirement by coated glass slide. The temperature-sensing element
naval forces for under-way assessments of sonar was a xylene-filled copper tube, whose (temperature-
propagation conditions – the first (and still the dependent) pressure moved the stylus across the slide
dominant) application of operational oceanography. via a Bourdon tube. Stylus movement along the slide
The naval requirement first arose in the area of was determined by a copper bellows, compressed by
physical oceanography, in the need to know the depth the increasing water pressure. Data were read from
variation of water temperature. In practice, of the the trace using an optical projector and scale.
three parameters that determine sound speed – tem- The XBT was a major advance, allowing operation
perature, salinity, and pressure – it is temperature that while under way and dispensing with the intricate
predominates. Pressure is normally deducible with measurement routine of the MBT, with the need for a
adequate precision from depth, and salinity is nor- deployment/recovery winch and with the need for
mally sufficiently constant to be neglected or simply calibration. It uses a pre-calibrated thermistor meas-
‘modeled’ using an archived (T,S) relation. However, urement, read onboard in real time. Inference of its
salinity may be important near ice, in fiords, and es- depth uses knowledge of its rate of fall through the
tuaries, and in regions of freshwater influence (ROFIs). water. There are now several manufacturers, although
The naval requirement is normally for the vertical the originators. Sippican Inc. (Marion, MA, USA), still
sound speed profile, and it is the shape of this profile lead in the number of available probe types.
that is important, rather than its mean value. Current expendable probe capabilities include, in
The expendable measurement facility was quickly addition to temperature, the measurement of sound
taken up by the civilian oceanographic community. It speed, conductivity, ocean current, optical prop-
gives a means of tackling the problem of how to erties, and (recently) seabed properties. This review
make synoptic ocean structure measurements where of expendables summarizes the variety of measurable
features are likely to move significantly during a parameters and then outlines the available deploy-
survey. A survey with spatial scales small enough to ment options. A number of examples are then given
capture interesting features is seriously degraded by of their use in oceanographic research. Expendables
their movement and development. Particularly at specific to naval activities, such as noise-measuring
mid- to high latitudes, a survey using conventional (sonobuoy) systems, will not be covered here, al-
profiling instruments – such as the conductivity- though in some cases they have a limited ocean
temperature-depth (CTD) probe – cannot be carried measurement capability. Air-deployed drifting sys-
out in a time that is small compared with the time- tems are also omitted.
scales of motion and development of features such as
frontal boundaries and eddies. Expendable Sensor Types
Surveys are severely limited by deployments that
Expendable Bathythermograph (XBT)
require a vessel to be regularly stationary for casts
with a profiling speed of B1 m s1. Expendable The purpose of the XBT is to provide a vertical
probes allow use at ship speeds up to 20–30 knots, profile of temperature, from the surface to as great a

43
44 EXPENDABLE SENSORS

depth as required, if possible to the seabed. A ther- connecting wire. Electrical contact with the probe is
mistor measures temperature as the probe descends, achieved when the unit is loaded into the launcher,
and the depth for each measurement is deduced from allowing initialization of the onboard electronics
time of descent using an empirical equation. A before the probe is released by withdrawal of a
number of variants are available with different depth ‘firing pin’ from its tail section.
capabilities, the deepest (the T5) reaching 1830 m, Throughout the operation of the device the probe
which may necessitate data extrapolation in deeper remains connected to its shell, and thence to the
water. Apart from expense, which increases with onboard electronics, by two-strand wire. This wire is
depth capability, operational constraints become arranged in two coils, one within the shell, dispens-
more restrictive for the deeper probes. Whereas the ing wire horizontally as the vessel moves away from
T-7 (760 m) can operate at platform speeds up to 15 the launch location, and one within the probe body,
knots, a T-5 is limited to 6 knots.1,2Other probe which dispenses wire upwards as the probe falls.
types are (or have been) available,3 but T-7 and T-5 Data are collected until wire breakage, either when
types are in most regular use. the probe reaches the seabed or (in deeper water)
The stated accuracy of all probe types is 7 0.151C when the wire has been expended without the bot-
and 72% of indicated depth, with a depth reso- tom being reached. If the deploying vessel travels
lution of 0.65 m. faster than the design speed, the upper coil of wire
may be exhausted first.
The success of XBT is the result of a number of
Operational effects of finite depth Various critical design features. One of these is the small
approaches have been adopted to overcome the compartment in the shell that contains the electrical
limited depth capability of XBTs. The best means of contacts, which is designed to avoid the problem of
doing this is generally accepted to be extrapolation making good instant electrical contacts between a
with the help of relevant (same survey) full-depth probe, which may have spent many months awaiting
CTD casts. This has the added advantage of use, and a launcher normally sited on an exposed
allowing a check on the XBT depth data. In naval ship deck. In the compartment, the probe contacts
operational terms, however,4 this is rarely an are embedded within a thick gelatinous insulator,
option. In conditions where the measured which is penetrated by the pointed launcher contacts
temperature has stabilized at the maximum depth, as the breech closes. The material cleans the launcher
extrapolation to the seabed using the data trend contacts as they penetrate, and maintains their clean
may be reasonable. A second approach uses state by the practice of leaving the spent shell in the
extrapolation using archived data, although if these launcher between probe launches.
are mismatched, this may be a problem. The free-falling probe itself involves three
principal components: the thermistor element, mak-
ing the temperature measurement; the two-strand
Probe design The standard XBT has two main wire that connects the thermistor circuit to the
parts: a protective ‘shell’ which remains on the onboard electronics; and a weighted, hydro-
vessel after launch, and the probe itself, which falls dynamically shaped body. Each of these components
through the water and passes data along the plays a vital part in the remarkable success of the
instrument as a whole. The thermistor, of course, is
indispensable, this small fragment of temperature-
1
The 450m T-4 (460 m, 30 knots), with a rated ship speed of sensitive semiconductor providing the measurement
30 knots, used to be the routine choice for operational use, but this capability of the unit. This is positioned in an aper-
generally gave way to the deeper T-7 at the end of the 1980s.
2
ture at the probe tip.
The specified maximum may be exceeded, but premature-
The connecting wire may represent a technical
breakage of the wire will then limit the depth of data collected.
3 achievement at least as great as any of the other XBT
The T-7, T-5, and T-4 are complemented by the T-6 (460 m, 15
knots), T-10 (200 m, 10 knots), and T-11 (460 m, 6 knots), the components. Two thin strands of copper wire are
latter giving a 0.18 m vertical resolution). A T-7 variation called covered by a thin insulating lacquer which binds
‘Deep Blue’ (760 m, 20 knots) was developed for (faster moving) them securely but which is sufficiently non-sticky to
Volunteer Observing Ships. avoid the problem of self-attachment within the
4
The naval application differs significantly from purely coils, in which hundreds of meters are compactly
oceanographic applications. At great depth the sound speed in-
wound.
creases slowly with depth (pressure), providing weak upward re-
fraction of sound and decreased seabed interaction. Even small The probe body consists of a weighty metal nose
temperature gradients may negate this effect, and small errors in cone attached to a lightweight faired hollow plastic
the extrapolation may have disproportionate effects. tail, equipped with fins to ensure vertical travel,
 EXPENDABLE SENSORS 45

and an even metering of wire from the contained resolution. Depth precision is quoted as 74.6 m or
coil. The metal cone acts as a ‘sea electrode’ to com- 72% of indicated depth, whichever is greater.
plete the measurement circuit with the (effectively
grounded) vessel. The final part of the instrument is a Expendable Conductivity-Temperature-Depth
soft plastic cap, removed before use, which restricts Probe (XCTD)
the movement (and possible damage) of the probe in
The XCTD is one of the newer expendable probes,
its packaged state.
measuring conductivity as well as temperature. This
is not a simple extension of capability, since precision
Probe operation Operation of the XBT involves conductivity measurement is acknowledged to be
five stages (1) removal of the shell of the previous particularly difficult, and first-time operation must
probe from the launcher; (2) unpacking and be assured, even following a lengthy unattended
insertion of the fresh probe into the launcher, shelf-life. A four-electrode configuration is used in a
completing the electrical circuit by closing the resistive measurement of conductivity, with a ther-
breech; (3) initialization of the onboard electronics mistor for the temperature measurement. As with
to recognize the probe and to prepare for data other probes, depth is inferred from the fall rate of
acquisition; (4) launch, by withdrawing the ‘firing the probe. In the XCTD the measuring system con-
pin’; and (5) following data acquisition, completion verts basic C, T data into frequency-modulated sig-
of the data file closure procedure. Data acquisition nals for transmission along the standard two-wire
begins when the probe completes the earth–loop connection. Each unit is calibrated in a three-salt-
circuit on reaching the water, and continues until water-bath procedure following manufacture, with
the wire breaks. Acquisition may be ended before the resulting calibration coefficients stored in non-
this if, for example, the probe is known to have volatile memory in the probe canister.
reached the seabed. This process is common to all
ship-launched probes. Expendable Current Profiler (XCP)
The Sippican XCP is the first expendable technique
Expendable Sound Velocity Probe (XSV) for ocean current measurement. Its measurement of
In the XSV the active sensor, instead of the XBT’s current velocity utilizes measurements of the weak6
thermistor, is a small sound speed sensor using electrical current generated by the motion of con-
the ‘sing-around’ principle. In this, an ultrasonic ducting sea water through the Earth’s magnetic field,
transmitter/receiver pair are arranged with a fixed which is directly proportional to the velocity at any
separation in an electrical circuit with strong feed- given depth. The falling probe interrupts this current
back. The circuit’s ‘resonance’ (‘sing-around’) fre- and measures the electrical potential thus produced.
quency is determined mainly by the time acoustic The probe spins at a prescribed rate, converting the
path, and its measurement allows inference of the potential seen by separated electrodes into a sinus-
sound speed. oidal signal whose orientation is deduced from a co-
XSVs are almost solely limited to military use,5 rotating ‘compass’ coil. This allows resolution of the
principally by operational submarines, and both air- measured current into north and east components;
launched and submarine-launched variants are the third measurement made is that of temperature,
available for this reason. For naval operations, they by the thermistor mounted in the usual probe nose
offer an improvement over the XBT in regions such position. The data are passed along the wire as three
as Arctic, Mediterranean, and coastal waters where frequency-modulated signals.
salinity variations may cause significant sound speed The XCP is available only as a stand-alone in-
changes. strument using a radiofrequency (rf) link, similar to
The probes have a specified precision of 70.25 m that used for air-launched versions of the XBT and
s1, and they are available with two main depth XSV. It may therefore be used from either ship or
options: the XSV-01 (850 m, 15 knots) and the XSV- aircraft, and reception need not be from the de-
02 (2000 m, 8 knots). Both give depth resolution of ploying platform. An aircraft allows greater recep-
0.32 m. A higher resolution, slower-falling XSV-03 tion range, particularly in high sea states. A time
(850 m, 5 knots) is available, giving 0.1 m depth interval is allowed between launch and probe release,
to allow a deploying ship to move away, reducing its

5
An important exception is their potential use in precision
6
hydrographic surveying, to assess the mean sound speed of the In the measurement, a water velocity of 1 m s1 corresponds to
water column. about 5 mV at mid-latitudes.
46 EXPENDABLE SENSORS

electromagnetic disturbance. Operation involves Normal (Surface Ship) Deployment

only the removal of a number of a few items of
The standard deployment of all probes is from a
protective packaging before the unit is dropped into
surface vessel, normally from the stern, so that the
the sea. Energizing of the seawater battery, deploy-
wire emerges freely behind the vessel as it moves
ment of the rf antenna, operation of the probe, and
away from the launch point. Two types of launcher
eventual scuttling all take place without intervention.
are normally available: a robust (heavy) military
The probes have a 1500 m capability, and the 16 Hz
standard deck launcher, usually mounted perman-
(rotation rate determined) sampling rate allows
ently on deck, and a much smaller hand-held unit
0.3 m vertical resolution, with a specified velocity
which may be used even from small craft. Care is
resolution of 710 mm s1 rms and 73% rms hori-
required with either type in strong wind conditions,
zontal shear current accuracy. The specified tem-
as the thin wire may be ‘caught’ by the wind and may
perature resolution is 70.21C.
become entangled with parts of the vessel’s super-
structure. (Through-hull launchers are an option
used by some military vessels, allowing operation in
Expendable Optical Irradiance Probe (XKT)
adverse weather conditions.)
The XKT, another relatively recent innovation, is Since vessel heading is normally set by survey or
used to measure the vertical profile of light pene- operational demands, rather than by wind direction,
trating the upper layer of the ocean, allowing an it is often found that having a pair of launchers,
estimate of the optical diffuse attenuation, K, in mounted on the port and starboard quarters of the
addition to temperature. The upward-looking probe vessel, allows greater flexibility in cross-winds. A
has a cosine spatial response and is sensitive to the hand-held launcher is sometimes used to complement
wavelength band 490710 nm, operating down to a single deck-mounted launcher for this reason. A
200 m with about 0.15 m s1 vertical resolution. Ir- less conventional way of addressing this problem is
radiance is measured over a dynamic range of 105 to apply mild restraint to the emerging wire – such as
within 5% log conformity and 106 within 10%. An loosely guiding the wire through the fingers. This can
air-dropped version is available. reduce the effect of the wind on pulling excess wire
A second variety of optical properties probe, the from the dispenser within the shell, although care
XOTD/AXOTD, measuring suspended particle con- must be taken not to impede its normal flow.
centrations in addition to temperature, has been re- Successful use of expendables may also be limited
ported as ‘under development’. This operates using if the vessel is towing equipment, since the XBT wire
the scattering of light from an included source and is streaming aft can become fouled by tow cables.
intended for the particle concentration range 5 mg l1 The only processing of the data normally needed
to 3 g l1 in the depth range to 500 m. involves the removal of values from the top 3–5 m,
influenced by transient effects as the probe adjusts to
the water temperature, and the removal of values
Expendable Bottom Penetrometer Probe (XBP) obtained after the probe has reached the seabed, an
event frequently denoted by a sharp spike in the data.
The XBP is the most recent addition to the armoury
of expendables, and it is currently still at the devel-
opment stage, available by special arrangement for
evaluation. The requirement which it fulfils is, once Deployment Variations
again, driven by military operations, and relates to The XBT and XSV are available with additional de-
the need to know certain seabed properties, par- ployment options – air-launched and submarine-
ticularly in shallow water (o200 m). Sonar behavior launched, – developed mainly for military use. As
in shallow water is often determined by the geoa- was noted above, the XCP is suited to both ship and
coustic properties of the seabed, and aspects of mine airborne deployment.
counter-measures, particularly relating to the prob-
ability of mine burial, are sensitively dependent on
The Air-Launched XBT and XSV (AXBT, AXSV)
the properties of seafloor sediments.
The sensor carried by the XBP in place of the The measurement parts of the AXBT and AXSV are
XBT’s thermistor is an accelerometer, whose purpose substantially the same as for the standard probes,
is to monitor the deceleration of the probe on impact including the sensor, the wire attaching the falling
with the seabed. Hard or rocky seabeds involve rapid probe to the surface unit, and the weighted probe
deceleration, whilst sediments allow a smoother, body. In this case, however, the only wire coil used is
longer period of retardation. that within the probe, connecting the sensor to a
 EXPENDABLE SENSORS 47

buoyant electronics package which conditions the submarine to assess the sonar propagation charac-
temperature signal and communicates it via an rf teristics of its environment. Without it, a submarine
transmitter link to the launch aircraft. AXBTs and would need either to surface for a conventional de-
AXSVs are packaged in the standard-size canister ployment or to move vertically, making measure-
used for sonobuoys – the ‘A-size’, 914 mm  124 mm ments with onboard sensors.
– and they may be launched at air speeds up to 370 The technique is related to that of the AXBT, in
knots and altitudes up to more than 9000 feet. that the temperature profile is measured by a probe
Descent is controlled by a parachute, deployed when falling from a buoyant package which floats at the
the buoy leaves the aircraft, and operation begins surface. In this case, however, the package rises to the
after a short delay in which the probe reaches tem- surface under its own buoyancy following its sub-
perature stability and the seawater battery (for the rf merged launch from a signal ejector probe, and re-
transmitter) becomes energized. mains connected by wire to the submarine. As might
Each unit has a user-selectable rf frequency, which be imagined from its requirement to pass probes
allows simultaneous monitoring of a number of through the pressure hull of a submarine, this tech-
probes. Although as many as 99 channels may nique involves expensive technology, and is unlikely
be available, the number of probes being deployed is to be used for purely oceanographic, as opposed to
also limited by the number of channels available for operational, purposes.
simultaneous monitoring by the aircraft. Although An equivalent version of the XCTD is understood
transmissions cease when the probe has reached to be imminent.
its maximum depth, and the scuttling mechanism is
initiated, another probe using the same frequency
cannot be launched until this has occurred.
Data Recording and Handling
The receiver used for AXBTs is a standard unit
normally fitted only to military (Maritime Patrol) Current practice for the recording of data normally
aircraft. involves a PC, with a dedicated electronic interface
Two types of AXBT are available, designed for unit which checks the continuity and integrity of the
maximum depths of 302 m and 760 m. Their spatial individual probe electronics before launch and then
resolution is rather better than that of the standard transfers data only after it has detected the probe
XBT, at about 0.15 m. Specified depth accuracy is 2% reaching the water. Data display options exist using
of indicated depth, and temperature accuracy is either dedicated display programs or standard data
70.181C. The standard AXSV operators to 760 m, handling routines. As in many oceanographic appli-
with vertical resolution of about 0.15 m, 2% accuracy, cations, PCs have dramatically simplified the data
and a specified sound speed accuracy of 70.25 m s1. recording process. In the years between the emer-
In practice, although AXBTs allow rapid deploy- gence of the XBT and the eventual prominence of the
ment over substantial horizontal scales, it is difficult PC there was a period in which a dedicated inboard
to simultaneously achieve high spatial sampling, electronic unit was necessary for acquisition and
because of the combined effect of the finite reload data display using a paper trace. Digital recording on
time and the finite number of available receiving magnetic media (long-since obsolete) was also an
channels. To achieve a probe spacing smaller than option, although the practice of digitization from the
about 30 km along a single track it is normally ne- paper trace persisted operationally into the 1980s.
cessary to make at least two passes along the track, This inboard equipment frequently tended to be
the second interleaving dropping probes between the temperamental – a feature that is difficult to forgive
stations covered on the first. Despite the operational when linked to the use of expendable probes.
difficulties, and the relative inaccessibility of such The operational context of XBT data (for me-
activities to nonmilitary agencies, AXBTs are the teorological and military purposes) has led to the
only means of executing large-area surveys (hun- establishment, by the World Meteorological Organ-
dreds of kilometers) of dynamic regions for which ization (WMO) of a standard data exchange format,
the synoptic requirement requires a time-spread of known as the JJYY format (formerly JJXX the for-
o1 day. mat was officially changed from JJXX to JJYY by the
WMO on November 8, 1995), following the alpha-
betical code used to prefix each ‘bathy report’. De-
The Submarine-Launched XBT and ASV (SSXBT
tails such as date, time, location, probe type, and
and SSXSV)
recorder type are included in these reports. This
These variants of the XBT and XSV satisfy the format involves reduction of the XBT data to a small
uniquely military requirement for a submerged number of ‘inflection points’, or ‘break points’,
48 EXPENDABLE SENSORS

which are intended to capture the main features of secondary standard before use. Although this
the temperature profile whilst minimizing the data method can realistically assess only one temperature,
transfer requirement. giving an error to be applied as an offset to the
subsequent launch, it is a reasonably practical means
of improving confidence in the data.
Measurement Precision
Assessment of the precision of an expendable device Use for Ocean Surveying
is necessarily limited by the expected loss of the Although XBT use for purely oceanographic survey
probe following use. However, practical experience purposes is probably still dominated by naval re-
normally indicates performance within specified tol- quirements this is an effective way of enhancing a
erances for the measured variables. Experience is traditional survey using CTD casts. Without impact
considerably greater for the XBT than for the other on survey time it is possible to increase the spatial
probe types, as this is used much more widely for sampling rate by a factor of two or more by inter-
routine purposes. polating XBT launches between CTDs. Another
The growing importance of XBTs in the popu- common use is in the support of surveys undertaken
lation of global databases has led to some detailed with towed undulators or instrument chains, XBTs
consideration of their precision in the scientific lit- allowing a degree of downward extrapolation of the
erature. This has centered principally on the deriv- detailed upper-layer data that these collect. A third
ation of probe depth using the fall-rate equations common use of XBTs is as a ‘fall-back’ option for use
provided by the manufacturers. Direct verification when weather conditions are unsuitable for other
for individual probe types is not possible because of equipment.
the large vertical distances involved, and the in- AXBTs, and their deployment, are considerably
appropriateness of attaching verification sensors. more expensive, and tend to be used only when
The best available means of checking the depth data rapidity is essential. For example, the use of repeated
of expendables, of any kind, is to carry out parallel large-scale AXBT surveys of the highly dynamic
measurements with a temperature-measuring instru- Iceland-Faroes frontal zone has been reported.
ment, such as a CTD, which has a direct pressure- AXBTs have also been used to give near-synoptic
measuring capability. Distinctive individual tem- sections using aircraft underflights of satellite altim-
perature features in the water column may then be eter tracks, to validate the use of residual height data
used to indicate a depth correction for the region of for ocean monitoring.
the survey. The viability of these techniques depends on the
A number of surveys, several of them involving proportion of budget available for expendable items,
dense sampling of XBT and CTD casts, have found and the expense of the probes must be seen in the
significant systematic errors greater than the quoted context of the high basic cost of trials.
tolerances, the manufacturer’s equation always The other main contribution made to ocean sci-
underestimating the fall rate. A number of alter- ence by expendables relates to deductions made from
native fall-rate equations have been proposed, and it data archives. These are frequently dominated by
appears that a consensus is steadily being reached on XBT data and they often allow spatial and temporal
the optimum equation. coverage of regions that would otherwise have in-
The calibration issue is particularly significant for sufficient coverage for reliable deductions to be
assessing trends in climate change, and it is ac- made.
knowledged to be particularly important to follow None of the other probe types described here
an agreed standard procedure for handling the comes close to the XBT in its contribution to the
known depth errors in databases. Use of a variety of science. Perhaps the most detailed and intensive use
fall-rate equations would lead to major confusion in of the other expendables has involved the XCP,
interpretation of ensembles of data. It is now ac- whose data (drawn from a number of different ocean
cepted that only data using the manufacturer’s regions) have been used to draw conclusions about
equation should be used for archived data, and that the horizontal shear environment of the ocean.
it should be left to subsequent analysis to make
whatever adjustments are felt necessary.
It is generally found that temperature errors are
See also
within the bounds indicated by the manufacturer.
However, it has been indicated that performance Acoustics, Shallow Water. Inherent Optical
may be improved by calibrating each probe against a Properties and Irradiance. Ships. Sonar Systems.
 EXPENDABLE SENSORS 49

Further Reading 1980–1981. Journal of Physical Oceanography 14:

185--192.
Black PG, Elsberry RL, and Shay LK (1988) Airborne Smart JH (1988) Comparison of modelled and observed
surveys of ocean current and temperature perturbations dependence of shear on stratification in the upper
induced by hurricanes. Advances in Underwater ocean. Dynamics of Atmospheres and Oceans 12:
Technology, Ocean Science and Offshore Engineering 127--142.
16: 51--58. Stoll RD and Akal T (1999) XBP – Tool for rapid
Budéus G and Krause G (1993) On-cruise calibration of assessment of seabed sediment properties. Sea
XBT probes. Deep-Sea Research 40: 1359--1363. Technology February: 47--51.
Carnes MR and Mitchell JL (1990) Synthetic temperature Thadathil P, Ghosh AK, and Muraleedharan PM (1998) An
profiles derived from Geosat altimetry: Comparison with evaluation of XBT depth equations for the Indian
air-dropped expendable bathythermograph profiles. Ocean. Deep-Sea Research 45: 819--827.
Journal of Geophysical Research 95: 17979--17992.
Smart JH (1984) Spatial variability of major frontal
systems in the North Atlantic Norwegian Sea area:
 GRABS FOR SHELF BENTHIC SAMPLING
P. F. Kingston, Heriot-Watt University, Edinburgh, UK position and lowered to the seabed (Figure 1). On the
& 2009 Elsevier Ltd. All rights reserved.
bottom, the relaxing of the tension on the lowering
warp released the buckets and subsequent hauling
caused them to close before they left the bottom. The
instrument is still used today but is seriously limited
in its range of usefulness, working efficiently only in
Introduction very soft mud.
Petersen’s grab formed the basis for the design of
The sedimentary environment is theoretically one of
many that came after. One enduring example is the
the easiest to sample quantitatively and one of the
van Veen grab, a sampler that is in common use
most convenient ways to secure such samples is by
today (Figure 2). The main improvement over
means of grabs. Grab samplers are used for both
Petersen’s design is the provision of long arms at-
faunal samples, when the grab contents are retained
tached to the buckets to provide additional leverage
in their entirety and then sieved to remove the biota
to the closing action. The arms also provided a
from the sediment, and for chemical/physical sam-
means by which the complex closing mechanism of
ples when a subsample is usually taken from the
the Petersen grab could be simplified with the haul-
surface of the sediment obtained. In both cases, the
ing warp being attached to chains on the ends of the
sampling program is reliant on the grab sampler
arms. The mechanical advantage of the long arms
taking consistent and relatively undisturbed sediment
can be improved further by using an endless warp
samples.
rig; this has the added advantage of helping to pre-
vent the grab being jerked off the bottom if the ship
rolls as the grab is closing. The van Veen grab was
Conventional Grab Samplers
designed in 1933 and is still widely used in benthic
The forerunner of the grab samplers used today is the infaunal studies owing to its simple design, robust-
Petersen grab, designed by C.G.J. Petersen to con- ness, and digging efficiency. The van Veen grab
duct benthic faunal investigations in Danish fiords in
the early part of the twentieth century. It consisted of
two quadrant buckets that were held in an open

Figure 1 Petersen grab. Figure 2 van Veen grab.

50
 GRABS FOR SHELF BENTHIC SAMPLING 51

Figure 3 Diagram of a Hunter grab. Reproduced from Hunter B

and Simpson AE (1976) A benthic grab designed for easy
operation and durability. Journal of the Marine Biological
Association 56: 951–957.

typically covers a surface area of 0.1 m2, although

instruments of twice this size are sometimes used. Figure 4 Smith–McIntyre grab.
A more recent design of frameless grab is the
Hunter grab (Figure 3). This is of a more compact
design than the van Veen. The jaws are closed by
levers attached to the buckets in a parallelogram
configuration, giving the mechanism a good overall
mechanical advantage. The closing action requires
no chains or pulleys and the instrument can be op-
erated by one person. Its disadvantage is that the
bucket design does not encourage good initial pene-
tration of the sediment, which is important in hard-
packed sediments.
A disadvantage of the grab samplers discussed so
far is that there is little latitude for horizontal
movement of the ship while the sample is being se-
cured: the smallest amount of drift and the sampler is
likely to be pulled over. The Smith–McIntyre grab
was designed to reduce this problem by mounting the
grab buckets in a stabilizing frame (Figure 4). Initial
penetration of the leading edge of the buckets is as-
sisted by the use of powerful springs and the buckets Figure 5 Day grab.
are closed by cables pulling on attached short arms in
a similar way to that on the van Veen grab. The
driving springs are released by two trigger plates, one (Figure 5). The Day grab is widely used, particularly
on either side of the supporting frame to ensure that for monitoring work, despite its poor performance in
the sampler is resting flat on the seabed before the hard-packed sandy sediments.
sample is taken. In firm sand the Smith–McIntyre Most of the grabs thus far discussed have
grab penetrates to about the same depth of sediment been designed to take samples with a surface area
as the van Veen grab. Its main disadvantage is the of 0.1 or 0.2 m2. The Baird grab, however, takes
need to cock the spring mechanism on deck before samples of 0.5 m2 by means of two inclined digging
deploying the sampler, a process that can be quite plates that are pulled together by tension on the
hazardous in rough weather. warp (Figure 6). The grab is useful where a relatively
The Day grab is a simplified form of the Smith– large surface area needs to be covered, but has
McIntyre instrument in which the trigger and closing the disadvantage of taking a shallow bite and having
mechanism remains the same, but without spring the surface of the sample exposed while it is being
assistance for initial penetration of the buckets hauled in.
52 GRABS FOR SHELF BENTHIC SAMPLING

Warp Activation on direct measurements of the sampler); however, in

medium sand, the digging performance is reduced to
All the grabs described above use the warp acting a maximum depth of only 8 cm (Figure 7(e)). The
against the weight of the sampler to close the jaws. influence of warp action on the digging efficiency of a
However, direct contact with the vessel on the sur- grab sampler can also depend on the way in which
face during the closure of the grab mechanism poses the sampler is rigged. This is particularly true of the
several problems. van Veen grab. Figure 7(b) shows the bite profile of
the chain-rigged sampler in which the end of each
arm is directly connected to the warp by a chain. The
Warp Heave
vertical sides of the profile represent the initial
As tension is taken up by the warp to close the jaws, penetration of the grab while the central rise shows
there is a tendency for the grab to be pulled up off the the upward movement of the grab as the jaws close.
bottom, resulting in a shallower bite than might be Figure 7(c) shows the bite profile of a van Veen of
expected from the geometry of the sampler. This similar size and weight (30 kg) rigged with an endless
tendency is related to the total weight of the sampler warp in which the arms are closed by a loop of wire
and the speed of hauling and is exacerbated by firm passing through a block at the end of each arm (as in
sediments. For example, the theoretical maximum Figure 2). The vertical profile of the initial pene-
depth of bite of a 120-kg Day grab is 13 cm (based tration is again apparent; however, in this case, the
overall depth of the sampler in the sediment is
maintained as the jaws close. The endless warp rig
increases the mechanical advantage of the pull of the
warp while decreasing the speed at which the jaws
are closed. The result is that the sampler is ‘insulated’
from surface conditions to a greater extent than
when chain-rigged, giving a better digging efficiency.

Grab ‘Bounce’
In calm sea conditions it is relatively easy to control
the rate of warp heave and obtain at least some
consistency in the volume of sediment secured.
However, such conditions are seldom experienced in
the open sea where it is more usual to encounter
Figure 6 Baird grab. wave action. Few ships used in offshore benthic

(a) (b) (c)

2 2 2
4 4 4
6 6 6
8 8 8
10 10 10
Petersen grab (30 kg) Chain-rigged van Veen grab (30 kg) Endless-warp-rigged van Veen grab
(short-armed, 30 kg)

(d) (e) (f)

2 2 2
4 4 4
6 6 6
8
10 8 8
12 10 10
14
Endless-warp-rigged van Veen Day grab (120 kg) Smith−mcintyre grab (120 kg)
grab (long-armed, 70 kg)

Figure 7 Digging profiles of a range of commonly used benthic grab samplers obtained in a test tank using a fine sand substratum.
(a) Peterson grab (30 kg); (b) chain-rigged van Veen grab (30 kg); (c) endless-warp-rigged van Veen grab (short-armed, 30 kg); (d)
endless-warp-rigged van Veen grab (long-armed, 70 kg); (e) Day grab (120 kg); (f) Smith–McIntyre grab (120 kg).
 GRABS FOR SHELF BENTHIC SAMPLING 53

studies are fitted with winches with heave compen- such as the van Veen grab, that do not have stabilizing
sators so that the effect of ship’s roll is to introduce frames. Diver observations have shown, however, that
an erratic motion to the warp. This may result in the at least in shallow water, where the drift effect is at its
grab ‘bouncing’ off the bottom where the ship rises greatest on the bottom, even the framed heavily
just as bottom contact is made, or in the grab being weighted Day and Smith–McIntyre grabs can be
snatched off the bottom where the ship rises just as toppled.
hauling commences. In the former instance, it is
unlikely that any sediment is secured; in the latter, Initial Penetration
the amount of material and its integrity as a sample
will vary considerably, depending on the exact cir- It is clear that the weight of the sampler is an im-
cumstances of its retrieval. portant element in determining the volume of the
The intensity of this effect will depend on the se- sample secured. Much of the improved digging effi-
verity of the weather conditions. Figure 8 shows the ciency of the van Veen grab shown in Figure 7(d) can
relationship between wind speed and grab failure rate, be attributed to the addition of an extra 40 kg of
which is over 60% of hauls at wind force 8. What is weight which increased the initial penetration of the
of more concern to the scientist attempting to obtain sampler on contact with the sediment surface.
quantitative samples is the dramatic increase in vari- Initial penetration is one of the most important
ability with increase in wind speed with a coefficient factors in the sequence of events in grab operation,
of variation between 20 and 30 at force 7. The high determining the final volume of sediment secured.
cost of ship-time places considerable pressure on op- Figure 9 shows the relationship between initial
erators to work in as severe weather conditions as penetration and final sample volume obtained for a
possible and it is not unusual for sampling to continue van Veen grab. Over 70% of the final volume is
in wind force 7 conditions with all its disadvantages. determined by the initial penetration. Subsequent
digging of the sampler is hampered, as already
shown, by the pull of the warp.
Drift
For most benthic faunal studies it is important for
For a warp-activated grab sampler to operate effi- the sampler to penetrate at least 5 cm into the sedi-
ciently it should be hauled with the warp positioned ment (for a 0.1 m2 surface area sample this gives 5 l
vertically above. Where there is a strong wind or of sample). In terms of number of species and indi-
current, these conditions may be difficult to achieve. viduals, over 90% of benthic macrofauna are found
The result is that the grab samplers are pulled on to in the top 4–5 cm of sediment. Figure 10 shows how
their sides. This is a particular problem with samplers, the number of individuals relates to average sample

20
80
18
Number of failures due to motion as the % of

70
16
Initial penetration depth (cm)

60 14
number of attempts

50 12

10
40
8
30
6
20
4
10
2

0 0
0 1 2 3 4 5 6 7 8 0 2 4 6 8 10 12 14 16 18 20 22 24
Sea state (Beaufort scale) Sample volume (l)

Figure 8 Relationship between wind speed and grab failure Figure 9 Relationship between initial penetration of a van Veen
rate. grab sampler and volume of sediment secured.
54 GRABS FOR SHELF BENTHIC SAMPLING

1500

1400
Number of individuals per 0.5 m2

1300

1200

1100

1000

900

800

700

600
Figure 11 Shipek grab.
500
4.5 5 5.5 6 6.5 7 7.5
value for chemical contamination studies), the effect
Average volume of grab sample (l)
on the numbers of benthic fauna is generally very
Figure 10 Relationship between number of benthic fauna small.
individuals captured and sample volume for a boreal offshore
sand substratum. Self-Activated Bottom Samplers
There can be little doubt that one of the most im-
volume for 18 stations in the southern North Sea portant factors responsible for sampler failure or
(each liter recorded represents 1 cm of penetration). sample variability in heavy seas is the reliance of
Although there is a considerable variation in the most presently used instruments on warp-activated
numbers of individuals between stations, there is no closure. The most immediate and obvious answer to
significant trend linking increased abundance with this problem is to make the closing action in-
increased sample volume penetration. No sample dependent of the warp by incorporating a self-
volumes of less than 4.5 l were taken, indicating that powering mechanism.
at that level of penetration most of the fauna were
being captured. Spring-Powered Samplers
Samplers in which the jaws are held rigidly in a
One solution to the problem is to use a spring to
frame have no initial penetration if the edges of the
actuate the sampler buckets. Such instruments are in
jaw buckets, when held in the open position, are on a
existence, possibly the most widely used being the
level with the base of the frame. The lack of any
Shipek grab, a small sampler (0.04 m2) consisting of
initial penetration in such instruments has the added
a spring-loaded scoop (Figure 11). This instrument is
disadvantage in benthic fauna work of under-
widely used where small superficial sediment samples
sampling at the edges of the bite profile (see Figures
are required for physical or chemical analysis. The
7(e) and 7(f)), although the addition of weight will
use of a pretensioned spring unfortunately sets
usually increase the sample volume obtained.
practical limits on the size of the sampler, since to
cock a spring in order to operate a sampler capable
Pressure Wave Effect of taking a 0.1 m2 sample would require a force that
would be impracticable to apply routinely on deck.
The descent of the grab necessarily creates a bow
In addition, in rough weather conditions, a loaded
wave. Under field conditions, it is usually im-
sampler of this size would be very hazardous to
practicable to lower the grab at a rate that will
deploy.
eliminate a preceding bow wave, even if the sea were
flat calm. There have been several investigations of
Compressed-Air-Powered Samplers
the effects of ‘downwash’ both theoretical, using
artificially placed surface objects, and in situ. The Another approach has been to use compressed air
effects of downwash can be reduced by replacing the power. In the 1960s, Flury fitted a compressed air
upper surface of the buckets with an open mesh. ram to a modified Petersen grab with success.
Although there is still a considerable effect on the However, the restricted depth range of the instru-
surface flock layer (rendering the samples of dubious ment and the inconvenience of having to recharge the
 GRABS FOR SHELF BENTHIC SAMPLING 55

air reservoir for each haul limited its potential for Water inlet valve
routine offshore work. with release delay
mechanism held shut
Piston by weight or sampler
Hydraulically Powered Samplers
Hydraulically powered grabs are commonly used for
Ram operated by
large-scale sediment shifting operations such as sea- hydrostatic pressure
bed dredging. The Bedford Institute of Ocean-
ography, Nova Scotia, successfully scaled down this Connecting rod
technology to that of a practical benthic sampler.
Their instrument is relatively large, standing 2.5 m
high and weighing some 1136 kg. It covers a surface
area of 0.5 m2 and samples to a maximum sediment
depth of 25 cm. At full penetration, the sediment
volume taken is about 100 l. The buckets are driven
closed by hydraulic rams powered from the surface.
The grab is also fitted with an underwater television
camera which allows the operator to visually select
the precise sampling area on the seabed, close and
open the bucket remotely, and verify that the bucket
closed properly prior to recovery. The top of the Figure 12 Diagram of a grab sampler using hydrostatic
buckets remain open during descent to minimize the pressure to close the jaws.
effect of downwash and close on retrieval to reduce
washout of the sample on ascent. The current oper- and, on release, are drawn open by reversal of the
ating depth of the instrument is 500 m. The instru- piston motion from air pressure built up on the
ment has been successfully used on several major underside of the piston during its initial power
offshore studies, but does require the use of a sub- stroke. Since the powering of the grab jaws is in-
stantial vessel for its deployment. dependent of the warp, the sampler may be used
successfully in a much wider range of surface wea-
Hydrostatically Powered Samplers ther conditions than conventional grabs.
Hydrostatically powered samplers use the potential
energy of the difference in hydrostatic pressure at the
sea surface and the seabed. The idea of using this Alternatives to Grab Samplers
power source is not new. In the early part of the Ideally a benthic sediment sample for faunal studies
twentieth century, a ‘hydraulic engine’ was in use by should be straight-sided to the maximum depth of its
marine geologists that harnessed hydrostatic pressure excavation and should retain the original stratifi-
to drive a rock drill. Hydrostatic power has also been cation of the sediment. Grab samplers by the very
used to drive corers largely for geological studies. nature of their action will never achieve this end.
However, these instruments were principally con-
cerned with deep sediment corers and were not de-
Suction Samplers
signed to collect macrofauna or material at the
sediment–water interface. One answer to this problem is to employ some sort
A more recent development has been that of a grab of corer designed to take samples of sufficient surface
built by Heriot-Watt University, Edinburgh. The area to satisfy the present approaches to benthic
sampler uses water pressure difference to operate a studies. The Knudsen sampler is such a device and is
hydraulic ram that is activated when the grab reaches theoretically capable of taking the perfect benthic
the seabed. Figure 12 shows the general layout of the sample. It uses a suction technique to drive a core
instrument. Water enters the upper chamber of the tube of 0.1 m2 cross-sectional area 30 cm into the
cylinder when the sampler is on the seabed, forcing sediment. Water is pumped out of the core tube on
down a piston that is connected to a system of levers the seabed by a pump that is powered by unwinding
that close the jaws. The actuating valve is held shut a cable from a drum. The sample is retrieved by
by the weight of the sampler and there is a delay pulling the core out sideways using a wishbone ar-
mechanism to prevent premature closure of the jaws rangement and returning it to the surface bottom-
resulting from ‘bounce’. Back on the ship, the jaws side up (Figures 13 and 14). Under ideal conditions,
are held shut by an overcenter locking mechanism the device will take a straight-sided sample to a depth
56 GRABS FOR SHELF BENTHIC SAMPLING

Figure 13 Knudsen sampler in descent position.

of 30 cm. However, conditions have to be flat calm in

order to allow time for the pump to operate on the
seabed and evacuate the water from the core. This
limits the use of the Knudsen sampler and it is gen-
erally not suitable for sampling in unsheltered con-
ditions offshore. Mounting the sampler in a
stabilizing frame can improve its success rate and it is
used regularly for inshore monitoring work where it
is necessary to capture deep burrowing species.

Spade Box Samplers

Another approach to the problem is to drive an
open-ended box into the sediment, using the weight
of the sampler, and arrange for a shutter to close off
the bottom end. The most widespread design of such Figure 14 Knudsen sampler in ascent position.
an instrument is that of the spade box sampler, first
described by Reineck in the 1950s and later sub-
jected to various modifications. The sampler consists
of a removable steel box open at both ends and
driven into the sediment by its own weight. The
lower end of the box is closed by a shutter supported
on an arm pivoted in such a way as to cause it to
slide through the sediment and across the mouth of
the box (Figure 15). As with the grab samplers pre-
viously described, the shutter is driven by the act of
hauling on the warp with all the attendant dis-
advantages. Nevertheless, box corers are very suc-
cessful and are used widely for obtaining relatively Lead weights
undisturbed samples of up to 0.25 m2 surface area
(Figure 16). One big advantage of the box sampler is Box closure
that the box can usually be removed with the sample (‘spade’)
and its overlying water left intact, allowing detailed
studies of the sediment surface. Furthermore, it is Box
possible to subsample using small-diameter corers holder
for studies of chemical and physical characteristics. Box
Despite their potential of securing the ‘ideal’ sedi-
ment sample, box corers are rarely used for routine
benthic monitoring work. This is largely because of
their size (a box corer capable of taking a 0.1 m2 Figure 15 Diagram of a Reineck spade box sampler.
 GRABS FOR SHELF BENTHIC SAMPLING 57

Cable drum

Scoops

Figure 17 Diagram of a Holme scoop. Reproduced from Holme

NA (1953) The biomass of the bottom fauna in the English
Channel off Plymouth. Journal of the Marine Biological
Association 32: 1–49.

macrobenthos sampling. At present, there is no in-

strument that fulfils the requirements for a quick
turnaround precision multiple corer for offshore
sampling.

Figure 16 A 0.25 m2 spade box sampler. Sampling Difficult Sediments

Most of the samplers so far discussed operate rea-
sample weighs over 750 kg and stands 2 m high) and sonably well in mud or sand substrata. Few operate
the difficulty in deployment and recovery in heavy satisfactorily in gravel or stony mixed ground either
seas. because the bottom is too hard for the sampler to
penetrate the substratum or because of the increased
likelihood of a stone holding the jaws open when
Precision Corers
they are drawn together. To get around this problem
For chemical monitoring, it is important that the various types of scoops have been devised. The
sediment–water interface is maintained intact, for it Holme grab has a double scoop action with two
is the surface flock layer that will contain the most buckets rotating in opposite directions to minimize
recently deposited material. Unfortunately, such un- any lateral movement during digging. The scoops are
disturbed samples are rarely obtained using grab closed by means of a cable and pulley arrangement
samplers or box corers. Precision corers are capable (Figure 17) and simultaneously take two samples of
of securing undisturbed surface sediment cores; 0.05 m2 surface area.
however, they are unsuitable for routine offshore The Hamon grab, which has proved to be very
work because of the time taken to secure a sample on effective in coarse, loose sediments, takes a single
the seabed and dependence on warp-activated clo- rectangular scoop of the substratum covering a sur-
sure. Additionally, the cross-sectional area of the face area of about 0.29 m2. The scoop is forced into
core (0.002–0.004 m2) would necessitate the taking the sediment by a long lever driven by pulleys that
of large numbers of replicate samples in order to are powered by the pull of the warp (Figure 18).
capture sufficient numbers of benthic macrofauna to Although the samples may not always be as con-
be useful. This would be impracticable given the time sistent as those from a more conventional grab
taken to take a single sample. Large multiple pre- sampler, the Hamon grab has found widespread use
cision corers have been constructed; these are usually where regular sampling on rough ground is im-
too large and difficult to deploy for routine possible by any other means.
58 GRABS FOR SHELF BENTHIC SAMPLING

A generic disadvantage of most samplers presently

in use is that they rely on slackening of the warp to
trigger the action and the heave of the warp to drive
the closing mechanism. In calm conditions, this
presents no great problem, but with increasing sea
state the vertical movement of the warp decreases
reliability dramatically until the variability and fi-
nally failure rate of hauls make further sampling ef-
fort fruitless.

Specific Problems, Requirements,

and Future Developments
Scoop Chemical Studies
Studies involving sediment chemistry require pre-
cision sampling if undisturbed samples at the sedi-
ment/water interface are to be obtained. This can be
critical, particularly when the results of recent sedi-
mentation are of interest. The impracticality of using
Release
hook
Stop plate existing hydraulically damped corers such as the
Craib corer for offshore work has led to the wide-
Figure 18 Diagram of a Hamon scoop. spread use of less ‘weather sensitive’ devices such as
spade box corers and grabs for routine monitoring
purposes. However, studies carried out have shown
Present State of Technology that these samplers produce a considerable ‘down-
wash effect’, blowing the surface flocculent layer
It is perhaps surprising that given the high state away before the sample is secured. This can have
of technology of survey vessels, position fixing, and serious consequences if any meaningful estimation of
analytical equipment, the most commonly used the surface chemistry of the sediment is desired.
samplers are relatively primitive (being designed Repetitive and accurate sampling is also a pre-
some 40 or more years ago). Yet the quality of requisite for determining spatial and temporal
the sample is of fundamental importance to any change in sediment chemistry.
research or monitoring work. Currently the
most popular instruments are grab samplers,
probably because of their wide operational weather
window and apparent reliability. Samplers such
Meiofauna Studies
as the Day, van Veen, and Smith–McIntyre grab
samplers are still routinely used for sampling sedi- Meiofauna has increasingly been shown to have
ment for chemical and biological analysis despite potential as an important tool in benthic monitoring.
their well-documented shortcomings. As discussed One of the major factors limiting its wider adoption
earlier, the most important of these are the substan- is the lack of a suitable sampler. Although instru-
tial downwash that precedes the sampler as it des- ments such as the Craib corer and its multicorer
cends and the disturbance of the trapped sediment derivatives are capable of sampling the critically
layers by the closing action of the jaws. Both chem- important superficial sediment layer, these designs
ical and meiofaunal studies are particularly vulner- provide a poor level of success on harder sediments
able to these. and in anything but near-perfect weather conditions.
Although box corers go some way to reducing They also have a slow turnaround time. Box corers
disturbance of the sediment strata, the all-important are widely used as an alternative; however, they are
surface flocculent layer is invariably washed away. A known to be unreliable in their sampling of meio-
big disadvantage of the box corer for routine off- fauna. As with the macrobenthos, meiobenthic
shore work is that it is sensitive to weather con- patchiness results in low levels of precision of
ditions; in addition, the larger instruments do not abundance estimates unless large numbers of samples
perform well on sand substrata. are taken.
 GRABS FOR SHELF BENTHIC SAMPLING 59

Macrobenthic Studies The problem is to be able to secure the 50 core

The measurement and prediction of spatial and samples per site that would be needed for the mac-
temporal variation in natural populations are of rofaunal monitoring in a timescale that would be
great importance to population biologists, both for realistic offshore. At present, there is no instrument
fundamental research into population dynamics and capable of supporting such a sampling demand and
productivity and in the characterization of benthic operating in the range of sediment types that wide-
communities for determining change induced by en- scale monitoring studies demand. Clearly a single
vironmental impact. Though always an important core sampler would be impracticable, and one must
consideration, cost-effectiveness of sampling and look to the future development of a multiple corer
sample processing is not so crucial in fundamental that is capable of flexibility in its operation, which
research, since time and funding may be tailored to allows a quick on-deck turnaround between hauls.
fit objectives. This is rarely the case in routine
monitoring work where often resolution and
timescales have to fit the resources available. Further Reading
Benthic fauna are contagiously distributed and to
sample such communities with a precision that will Ankar S (1977) Digging profile and penetration of the van
Veen grab in different sediment types. Contributions
enable distinction between temporal variation and
from the Askö Laboratory, University of Stockholm,
incipient change resulting from pollution effects, it is Sweden 16: 1--12.
generally accepted that five replicate 0.1 m2 hauls Beukema JJ (1974) The efficiency of the van Veen grab
from each station are necessary (giving a precision at compared with the Reineck box sampler. Journal du
which the standard error is no more than 20% of the Conseil Permanent International pour l’Exploration de
mean). This frequency of sampling requires ap- la Mer 35: 319--327.
proximately 10–15 man-days of sediment faunal Eleftheriou A and McIntyre AD (eds.) (2005) Methods for
analysis per sample station. While this may be ac- the Study of the Marine Benthos. Oxford, UK:
ceptable in community structure studies in which Blackwell.
time and manpower (and thus cost) are not a pri- Flury JA (1967) Modified Petersen grab. Journal of the
mary consideration, this high cost of analyzing Fisheries Research Board of Canada 20: 1549--1550.
Holme NA (1953) The biomass of the bottom fauna in the
samples is of importance in routine monitoring
English Channel off Plymouth. Journal of the Marine
studies and has led monitoring agencies and offshore
Biological Association 32: 1--49.
operators to reduce sampling frequency on cost Hunter B and Simpson AE (1976) A benthic grab designed
grounds to as few as two replicates per station. This for easy operation and durability. Journal of the Marine
reduces the precision with which faunal abundance Biological Association 56: 951--957.
can be estimated to a level at which only gross Riddle MJ (1988) Bite profiles of some benthic grab
change can be demonstrated. However, sampling to samplers. Estuarine, Coastal and Shelf Science 29(3):
an acceptable precision may be achieved from an 285--292.
area equivalent to 0.1 m2 if a smaller sampling unit is Thorsen G (1957) Sampling the benthos. In: Hedgepeth JW
used. For example, 50 5-cm core samples (with a (ed.) Treatise on Marine Ecology and Paleoecology, Vol.
similar total surface area) have been shown to give a 1: Ecology. Washington, DC: The Geological Society of
America.
similar precision to that of 5 to 12 0.1-m2 grab
samples. Thus a similar degree of precision may be
obtained for around one-fifth to one-twelfth the
analytical costs using a conventional approach.
 TOWED VEHICLES
I. Helmond, CSIRO Marine Research, controlling computer on board the vessel. The
Tasmania, Australia electromechanical cable allows the data to be
Copyright & 2001 Elsevier Ltd. transmitted to the ship and displayed and pro-
cessed in real time. This can be an advantage
when following a feature such as an ocean front;
the ship’s course can be adjusted to optimize the
data collection. It also has the advantage of en-
Introduction abling fast, real-time response, an important
consideration for bottom avoidance when oper-
As a platform for marine instruments the towed ve- ating in shallow water. As with most towed ve-
hicle (often termed a ‘towfish’ or ‘towed body’) hicles, cable drag dominates performance, so
combines the advantages of a ship-mounted instru- cable fairing is commonly used to reduce drag.
ment that gathers surface data while under way, and Examples of this type of towed vehicle are the
an instrument lowered from a stationary ship to Batfish (Guildline Instruments, Canada), SeaSoar
gather data at depth. This article discusses the types (Chelsea Instruments, UK) and the Scanfish
of vehicles, the significance of the hydrodynamic (MacArtney A/S, Denmark).
drag of the tow cable and methods to reduce it. It • A vehicle with controllable wings that is totally
also outlines the basic hydrodynamics of towed ve- self-contained. It is preprogrammed for maximum
hicles and presents the results of a model of the ve- and minimum depths and records the data in-
hicle/cable system, indicating depths and cable ternally. As such a vehicle can be towed on a
tensions for a typical system. simple wire rope, it is convenient to use on ships
not equipped for research. The lack of real-time
control and data can be a disadvantage. An ex-
Types of Towed Vehicles ample of this type is the Aquashuttle (Chelsea
A towed vehicle system has three main components: Instruments, UK).
the vehicle, the tow cable and a winch. The vehicles • A passive vehicle, often with fixed wings, where
fall into two broad categories: those with active changes in depth are made by winching the tow
depth control and those without. cable in and out. This necessitates a high-speed,
computer-controlled winch to produce the depth
Vehicles With Active Depth Control variation, but the vehicle can be simple. A recent
development of this type is the Moving Vessel
Depth-controlled towed vehicles (or ‘undulators’) Profiler (Brooke Ocean Technology Ltd, Canada).
can move vertically in the water column while being This system has a winch that spools out the cable
towed horizontally by the ship. The main advantage fast enough to allow the vehicle to free fall while
they have over the lowered instrument is that they the ship is under way, and then retrieves it after it
can quickly and conveniently measure vertical pro- has reached its maximum depth, which may be as
files of ocean properties with high horizontal spatial deep as 800 m. Because the profiler free-falls on a
resolution. The main disadvantage is that it is dif- slack cable, the usual cable drag constraints are
ficult to reach depths greater than 1000 m while not as relevant. This allows good depths to be
being towed at useful speeds; most available systems achieved without the complication of cable fairing.
are limited to 500 m at best. The reason for the depth
limitation is that the hydrodynamic drag of the tow Vehicles Without Active Depth Control
cable must be overcome by a downward force on the
The vehicle without active depth control produces
vehicle, produced either by weight or a downward
the depressing force by means of its weight, fixed
hydrodynamic force from hydrofoils or wings.
wings or both. It maintains a constant depth for a
However the cable’s strength limits the allowable size
given tow-cable length and tow speed. The vehicle
of these forces.
can be towed with either an electromechanical cable
The following are the common types of depth-
to provide real-time data to the ship (as used for
controlled vehicles.
underwater survey instruments such as the side-scan
• A vehicle with controllable wings towed with an sonar) or with a wire rope (as often used for plank-
electromechanical cable that connects it to a ton recorders).

60
 TOWED VEHICLES 61

Tow Cable Drag 10

5
Cable CDN
Tow cables are either wire rope or double-armored 2

Drag coefficient (CD)

electromechanical cables. In the latter, two layers of 1 Smooth cylinder
0.5
armor provide mechanical strength, and because the
0.2
layers are wound in opposite directions they are 0.1
approximately torque balanced, i.e., the cable has 0.05
CDT
little tendency to rotate when loaded. The electrical 0.02
0.01
conductors handle data and power and optical fibers 0.005
are used for high data rates. Except for systems with 0.002
short cables used for shallow operations the tow 0.001
2 3 4 5 6
cable is usually the dominant part of a towed vehicle 1 10 10 10 10 10 10
system. Even a modest cable of 8 mm diameter and Reynolds number (Re)
500 m length has a mass of about 150 kg and a
Figure 1 Normal and tangential drag coefficients for a typical
longitudinal cross-sectional area of 4 m2. This large
cable.
cross-sectional area means that the cable drag dom-
inates the performance of the system.
cable diameter d. The Strouhal number Sn is defined
as Sn ¼ fd=uN , and Sn E 0:2 over the range of Rey-
Drag Caused by Flow Normal to the Cable nolds numbers 102 oRe o105 . If this frequency is
The normal drag on a moving body in a fluid is close to a natural frequency of the cable, an amplified
given by oscillation occurs. A tow cable has many natural
frequencies and there are many modes excited by
vortex shedding; the result is a continuous oscillation
DN ¼ CDN 12ru2N A
of the cable. These vortex-induced oscillations, which
For a cable, the cross-sectional area A is the can have an amplitude of up to three cable diameters,
product of the cable’s diameter (d) and length (l) (see increase cable fatigue and drag. The increase in drag
list of symbols at end of article). This drag is the sum can be as much as 200%, resulting in drag co-
of drag due to the shape of the body (the form or efficients as high as 3. Values around 2 are common.
pressure drag) and drag due to surface friction The values of the drag coefficient for towed cables
(additionally a shape that produces lift also generates cited in the literature differ widely because each case
induced drag). The value of the drag coefficient CDN has its own set of conditions: cable curvature, ten-
depends on the Reynolds number Re (the ratio of sion, Reynolds number and surface roughness vary
inertial to viscous forces), which is defined as from case to case. A good starting point for esti-
Re ¼ ud=v. For a long smooth cylinder with normal mating the normal cable drag is CDN E 1:5 for cables
flow at Reynolds numbers less than about 3  105 the that are not strumming and CDN E 2 for strumming
flow is laminar and CDN E 1:2. At higher Reynolds cables.
numbers, the flow becomes turbulent and CDN drops For the towed vehicle to be able to dive, the nor-
to about 0.35. This change in drag coefficient is due mal cable drag force must be overcome by down-
to the large area of separated flow in the wake of the ward or depressing forces: the cable weight, the
cylinder in laminar flow decreasing when the flow towed vehicle weight and the hydrodynamic forces
becomes turbulent. For most cables used for towed produced by the vehicle. The normal drag does not
vehicles, the value of the Reynolds number is less contribute directly to cable tension but by influ-
than 105. To exceed this value a 10 mm-diameter encing the angle of the cable in the water it deter-
cable moving through water requires a normal vel- mines the component of cable weight that
ocity greater than 10 m s1. For cables with a rough contributes to tension.
surface, such as the usual stranded cable, CDN E 1:5
when 103oReo105 (Figure 1). Drag Caused by Flow Tangential to the Cable
There is a mechanism that increases the drag co-
efficient of a cable above the value of a rigid cylinder. The tangential drag on a long cylinder is due to
This is vortex-induced oscillation, commonly referred surface friction. It is given by:
to as ‘strumming’. Vortices shed from the region of
flow separation alter the local pressure distribution DT ¼ CDT 12ru2T pA
and the cable experiences a time-varying force at the
frequency of the vortex shedding. This frequency, f , is For tow cables with Reynolds numbers greater
a function of the normal flow velocity uN and the than about 103, a typical value for the tangential
62 TOWED VEHICLES

drag coefficient (CDT) E 0.005 (Figure 1). With a Another consequence of the high tangential drag is
towed vehicle system, the tangential drag contributes that this loading must be transferred from the fairing
significantly to cable tension but has little influence sections to the cable. Every 2–3 m, a ‘stop ring’ is
on the depth achieved. swaged or clamped to the cable to take the load. If
this force accumulates over too great a length the
fairing sections will not rotate freely, and in the ex-
treme they can break under the high compressive
Reduction of Cable Drag
load.
Normal cable drag can be reduced by an attachment Another form of rigid fairing is the ‘clip-on’ fairing
that gives the cable a streamlined or ‘fair’ shape. (see Figure 2). Unlike the ‘wrap-round’ type this
Alternatively the attachment can be designed to split does not totally enclose the cable, but is essentially
the wake, so that the shed vortices cannot become an after-body attached to the cable with clips. Be-
correlated over a significant length of cable and cause of the gap between the cable and the fairing the
strumming is prevented. These attachments are usu- drag coefficient is higher. Typical values are
ally called ‘fairing’. CDN E 0.4.
A problem that can occur with rigid fairing is the
phenomenon of ‘tow-off’. If the fairing sections are
Rigid Airfoil-shaped Fairing not free to align accurately with the flow, they can
‘Wrap-round’ fairing is the most effective method generate a considerable lift force, which can cause
of reducing normal drag. An example is shown in the cable to tow off to the side and decrease the
Figure 2. A good airfoil shape can have a normal drag depth it achieves.
coefficient of 0.05, but the practicalities of having a Although the rigid airfoil fairing is the most
rigid moulded plastic shape that can wrap around a effective method of decreasing drag, it comes at a
cable, be passed over sheaves and spooled onto a high cost. Not only is it expensive but it also requires
winch often results in a drag coefficient of about 0.2. special winches and handling gear. However, if a
The greater drag is primarily due to the circular nose system is set up well it gives reliable performance.
of the fairing (to accommodate the cable) and gaps
between fairing sections. Because of the large surface
Flexible Ribbon and ‘Hair’ Fairing
area of the fairing, the tangential drag coefficient
(based on cable surface area) is about 0.05. This Ribbon fairing is made of a flexible material in the
means that, although the normal drag coefficient of a form of trailing ribbons (Figure 2). Fibers or ‘hairs’
faired cable is only a tenth of a bare cable, the tan- attached to the cable are also effective. These fairing
gential drag coefficient is about ten times greater. A devices do not usually produce a fair shape, but
consequence of this is large cable tensions. For a achieve their effect by splitting the wake. Their main
typical system with 500 m of faired cable, at least effect is, therefore, to reduce strumming and reduce
50% of the cable tension can be due to the fairing. the normal drag coefficient to the bare cable value of

Cable Fixed to cable

Fairing

Stop ring Stop ring

(A) (B) (C) (D)

Figure 2 (A) ‘Wrap-round’ fairing; (B) ‘clip-on’ fairing; (C) ‘ribbon’ fairing; (D) ‘hair’ fairing.
 TOWED VEHICLES 63

about 1.5. In some cases ribbon fairing can stream- L

Chord line
line the cable, reducing the normal drag coefficient to
about 0.7 (due to the reduction of form drag). Hair
fairing reduces strumming but can increase the nor- Angle
mal drag coefficient so that CDN E2. of 
attack D
These devices increase the surface area over a bare AC
cable so the tangential drag is increased, resulting in M
u
greater cable tensions. Flexible fairing does not need Mean camber line
special handling gear and can be wound onto a
Figure 4 Forces and moment on an airfoil.
regular winch. However, the fairing deteriorates ra-
ther quickly requiring regular repair and
replacement.
M ¼ CM 12ru2 Sc
Basic Aerodynamics of the Towed The lift coefficient CL is proportional to the angle of
Vehicle attack. The theoretical relationship for a thin, sym-
Vehicles that Generate Lift metrical airfoil gives the slope of the curve of lift
coefficient against angle of attack
Most towed vehicles use wings (hydrofoils) to gen-
erate the force required to pull the tow cable down.
dCL =da ¼ a0 ¼ 2p=radian ¼ 0:11=degree
Figure 3 shows an example of a winged vehicle. As
shown in Figure 4, a wing moving through a fluid
The theoretical aerodynamic center is at the quarter
experiences a force perpendicular to the direction of
chord point ðc=4Þ and CM ¼ 0. If the angle of attack
flow (the lift), a force directly opposing the motion
is increased beyond a certain value the flow separates
(the drag), and a moment tending to rotate the wing
from the low-pressure side of the wing rapidly
(the pitching moment). The pitching moment is
causing the lift to decrease and the drag to increase.
usually referenced about a point termed the aero-
This is termed ‘stall’ (Figure 5).
dynamic center, chosen so that the moment co-
An asymmetrical or cambered airfoil, where
efficient is constant with angle of attack.
the camber line (the line drawn halfway between the
The lift force is given by:
upper and lower surfaces) deviates from the chord
line (Figure 4), has the same theoretical slope of
L ¼ CL 12ru2 S the lift curve, 2p, but has a nonzero value of the
lift coefficient when a ¼ 0. The aerodynamic center
The drag force is given by:
is also at the quarter chord point but CMa0
(Figure 6).
D ¼ CD 12ru2 S

The pitching moment is given by:

CL
1.5
Moment coefficient (C M)

1.0
Lift coefficient (CL)

0.5 _ 0.1

_ 0.5 _ 0.1
CM
_ 1.0

_ 1.5

_ 15 _ 10 _ 5 5 10 15

Angle of attack
Figure 3 CSIRO (Australia) modified SeaSoar with a Sea-Bird
Electronics Inc. conductivity, temperature and depth (CTD) Figure 5 Typical section characteristics for symmetrical airfoil
instrument. type NACA0012.
64 TOWED VEHICLES

perturbations such as those experienced when tow-

ing in rough seas.
CL The delta wing (a wing with a triangular plan-
1.5
form) is a popular form for vehicles without active

Moment coefficient (C M)
1.0
depth control. Flow over a delta wing is dominated
Lift coefficient (CL)

0.5 by large leading-edge vortices and cross-flow that

enable the wing to operate at large angles of attack
without stalling. A delta wing has a typical lift curve
_ 0.5 _ 0.1
CM slope of about 0.05/degree (about half that of a
_ 1.0 _ 0.2 conventional wing), but can operate with an angle of
_ 1.5 _ 0.3 attack up to 301 before stalling. Delta wings make
robust depressors and are often given large dihedral
_ 15 _ 10 _ 5
angles to increase roll stability. The dihedral is the
5 10 15
angle of inclination of the wings in relation to the
Angle of attack lateral axis.
To control the depth of a towed vehicle, the
Figure 6 Typical section characteristics for asymmetrical airfoil magnitude and direction of the wing’s lift force are
type NACA4412. usually varied by:

These properties describe airfoil sections that are • the use of control surfaces (flaps) on the trailing
two-dimensional. In a real wing, the span is finite edge;
and there is spanwise flow. The effect of this is a
‘leakage’ around the wing tips from the high-pressure • an independent control surface, usually at the tail;
side to the low-pressure side. This generates wing-tip • rotating the entire wing about its aerodynamic
center to vary its angle of attack.
vortices, which in turn produce a downward flow
around the wing – the downwash. The angle of this
local flow relative to the wing subtracts from the The first two methods cause the whole vehicle to
angle of attack so that the wing actually experiences adopt an angle of attack and so the body of the ve-
a smaller effective angle of attack. Since the lift hicle also generates lift. Some towed vehicles, such as
vector is normal to the local relative flow it becomes the Aquashuttle (Chelsea Instruments, UK), operate
inclined behind the vertical and so has a rearward without wings, generating all the lift from the body.
component – the induced drag. This drag can be the Others, such as the Scanfish (MacArtney A/S, Den-
dominant drag on a towed vehicle. The induced drag mark) and the V-Fin (Endeco Inc, USA), are effect-
coefficient is given by: ively flying wings. The third method, used by the
Batfish (Guildline Instruments, Canada) and SeaSoar
(Chelsea Instruments, UK), maintains the body
CDi ¼ C2L =ðpARÞ aligned with the flow, which is an advantage for
some types of sensors that need to be aligned to
The reduction in the effective angle of attack of the flow. Both these vehicles use a highly cambered
a wing with finite span also reduces the slope of the wing section (NACA6412) that has a large moment
lift curve, a. coefficient ðCM E  0:13Þ. Thus a large torque is
needed to rotate the wing. If a small operating torque
is required the wings are typically controlled by an
dCL =da ¼ a ¼ a0 =ð1 þ a0 =pARÞ
electric servomotor. If large forces are needed a
hydraulic system is used. A symmetrical wing section
The reduced slope of the lift curve means that the pivoted at the quarter chord point requires only
wing has a higher angle of attack at stall. a small torque, but the control system needs to be
To achieve the necessary mechanical strength, the robust enough to survive rough handling on the
aspect ratio of wings used on towed vehicles is usu- ship’s deck.
ally very low. This results in high induced drag and To gain the maximum benefit of any lift force, the
low values of the lift curve slope. It will be shown vehicle must fly so that the force is directed near to
later that the higher induced drag is not significant. vertical, that is it should fly without a significant roll
The lower sensitivity to changes in the angle of angle. A towed vehicle often needs to be able to
attack (and the higher angle at stall) can be an ad- direct its lift force both down and up, to maximize its
vantage: it makes the vehicle more tolerant of flight depth range. A consequence of this is that dihedral, a
 TOWED VEHICLES 65

T servomotor systems. The other basic requirement is

for pitch and yaw stability. This is achieved by en-
suring that the tow point is ahead of the aero-
dynamic center (AC) of the vehicle (Figure 7). The
position of the vehicle’s aerodynamic center is con-
1
trolled by providing a suitably sized tailfin.

D
AC Center of
gravity
Vehicles That Do Not Generate Lift
These passive towed vehicles use their weight to
produce the required downward force. The drag of
W
L the cable is overcome by the combined weight of
cable and vehicle. Depth is controlled by varying the
Figure 7 Basic forces on a towed vehicle. cable length. The depth is also very dependent on the
tow speed. This is because the cable drag is pro-
portional to the square of the tow speed and the
common method of providing roll stability, cannot depressing force is fixed by the cable and vehicle
be used. What would be stable for lift in one dir- weight. This contrasts with the vehicle that generates
ection would be unstable in the other. By having the lift; its depth is less speed-dependent because both lift
tow cable attachment point above the center of and drag vary with speed in the same manner. A
gravity of the vehicle, the vehicle’s weight contributes heavy passive vehicle can have good pitch and roll
to roll stability (Figure 7). But to stabilize the large stability if the position of the tow point, center of
lift forces needed for good depth performance, add- mass, center of gravity and the aerodynamic center
itional aerodynamic control by means of ailerons or are carefully chosen. The lack of flow separation
similar devices is needed. These can be simple sys- over lifting surfaces also makes them acoustically
tems driven directly with gravity by using a heavy quiet. This stability and quietness make them useful
pendulum device or more sophisticated electric vehicles for underwater acoustic work (Figure 8).

Figure 8 The CSIRO (Australia) multifrequency towed vehicle used for fish stock measurements. Note the ribbon-faired cable.
66 TOWED VEHICLES

shown in Figure 7 the angle of the cable at the ve-

0 hicle, f1 is determined by the weight ðWÞ, lift ðLÞ
and drag ðDÞ of the vehicle.
DN
tan f1 ¼ ðW þ LÞ=D

The cable profile starts at an angle of f1 at the

D vehicle and gradually approaches f0 up the cable.
u The vehicle drag is the sum of the form drag, the
DT friction drag and the induced drag. With vehicles that
generate lift the induced drag is the main component.
Even with a poorly streamlined vehicle it dominates,
w providing perhaps 75% of the drag. The rather poor
performance of the typical low aspect ratio wing used
Figure 9 Forces on the tow cable.
on towed vehicles gives the vehicle a lift to drag ratio
of about 3. This makes the cable angle f1 E 701.
Performance of the Vehicle/Cable Further improvement in the lift to drag ratio does not
System gain much in cable angle or depth. Table 1 compares
the equilibrium depths for bare and faired cables;
When a cable is towed through water, it assumes an Table 2 compares the performance of bare and faired
equilibrium angle where the drag force (D) is bal- cable when towing a vehicle with controllable wings.
anced by the cable weight (w). If the cable properties These data were produced by a computer model of
are uniform, this angle is constant along the cable the vehicle/cable system.
length
By referring to Figure 9
Effects of Wave-Induced Ship Motion
tan f0 ¼ w=D on the Towed Vehicle
Or expressed in terms of the normal drag coefficient A problem with towing in rough seas is that the
wave-induced motion of the towing vessel is propa-
 1=2 gated down the tow cable to the vehicle. Motion
cos f0 ¼ 17 1 þ 4B2 =2B
normal to the cable is rapidly damped, but there is
where B ¼ ru2 dCDN =ð2wÞ surprisingly little attenuation of the tangential mo-
The equilibrium depth is ðl sin f0 Þ. When a vehicle tion of the cable. The amplitude of the perturbation
is attached to the end of the cable it perturbs this of the vehicle is approximately the same as the cable
equilibrium depth by the extent that its weight and at the ship. This causes changes in the vehicle’s pitch
lift force either drag the cable deeper when diving, or angle and depth which can be very significant for
lift the cable when climbing. This defines the depth towed acoustic systems and vehicles such as camera
range of the vehicle/cable system (Figure 10). As units operating very close to the seafloor.

Equilibrium depth 0

Depth range

1

Figure 10 Towed cable profile.

 TOWED VEHICLES 67

Table 1 Comparison of equilibrium depths for 500 m of 8.2 mm diameter cable with a
weight of 2.5 N m1 towed at 4 m s1 for faired and bare cable

Equilibrium depth (m) Tension (kN) f0 (degrees)

Faired CDN ¼ 0.2 142 3.6 17

Bare CDN ¼ 2.0 61 0.7 7

The following methods are used to minimize the Speeds of 3–5 ms1 are common for oceanographic
effects of ship motion. surveys and depths of up to 1000 m can be achieved.
Depth-controlled vehicles operate with vertical vel-
• The vehicle can be tuned to minimize the pitching
ocities up to about 1 m s1.
effect by carefully adjusting the position of the
Depth-controlled vehicles are commonly operated
vehicle’s aerodynamic center and center of mass in
in an undulating mode with maximum and minimum
relation to the tow point.
depths set to specific values to give a triangular flight
• A constant-tension winch reduces the cable dis-
path. The depth is measured by the water pressure at
placement at the ship by spooling cable in and out
the vehicle and the wings or control surfaces are
as the ship surges.
adjusted to make the vehicle follow the defined path
• A device called an accumulator produces an effect
by a servo system. The servo-loop parameters are
similar to the constant-tension winch. The cable
usually controlled by the shipboard computer;
runs over a pair of sheaves that are mounted on a
however the actual servo-loop system may reside in
sprung or pneumatic arrangement that allows
the towed vehicle or combine ship- and vehicle-based
them to take up and pay out cable as needed.
components. The control algorithm needs to be
• In the ‘two-body system’ the instrumented vehicle
carefully tuned to achieve smooth flight.
is passive and near-neutrally buoyant. It is towed
behind the depressor or depth-controlled vehicle
on a cable that is approximately horizontal. This
geometry decouples ship’s motion from the second Sensors
vehicle.
Some of the earliest towed vehicles were used
• A system that has a cable angle close to horizontal
to collect plankton (in fact a trawl net is a form of
at the ship is insensitive to ship pitch and heave as
towed vehicle). These plankton collectors are
these displacements are almost normal to the
often separate nets and depressors but can also
cable. A system that uses a long cable, a cable
be single units. An early system, the Hardy Con-
without fairing or a high tow speed has this
tinuous Plankton Recorder, dates from the 1930s.
characteristic.
Several commercially available vehicles are a devel-
opment of this type, for example the Aquashuttle
(Chelsea Instruments, UK) and the U-Tow (Valeport
Flight Control
Ltd, UK).
Tow speeds vary from as low as 1 ms1 for seafloor Depth-controlled vehicles are commonly equipped
survey instruments to 10 ms1 for high-speed sys- with a conductivity, temperature and depth (CTD)
tems. The fast systems are limited to shallow depths. instrument. They are also suitable platforms for

Table 2 Comparison of depths, cable tensions and cable angles (f2 cable angle at ship, f1 cable angle
at vehicle) for the same cable as Table 1 towing a typical vehicle with the following characteristics: weight
in water, 1500 N; cross–sectional area, 0.2 m2; wing area, 0.5 m2; wing aspect ratio, 1; tow speed, 4 m s1.
Positive lift coefficients indicate lift force upwards

Wing CL Depth (m) Tension (kN) f2 (degrees) f1 (degrees)

Faired CDN ¼ 0.2  1.0 360 11.2 36 72

þ 0.5 0 4.3 7  36
Bare CDN ¼ 2.0  1.0 140 6.5 8 72
þ 0.5 31 1.6 7  36
þ 1.0 0 3.2 6  55
68 TOWED VEHICLES

many other types of sensors, such as fluorometers, can also be blocked by marine organisms especially
radiometers, nutrient analysers and transmiss- when towing near the surface. Dual cells allow rec-
ometers. In the case of a CTD it is recommended that ognition of this problem. Figure 11 shows the data
the sensors be duplicated. When a CTD is lowered from a CTD section across an ocean front demon-
from a stationary ship in the usual manner the cali- strating the high spatial resolution realized with a
bration of the conductivity sensor is checked by towed system.
collecting water samples for laboratory analysis. Passive towed vehicles are often used for acoustic
This option is not usually available on a towed in- survey work. Examples are side-scan sonars, towed
strument so a check of sensor stability can be ob- multibeam systems for seafloor mapping, and ve-
tained by using dual sensors. The conductivity cells hicles for estimating fish stocks. The towed vehicle

100 12
Depth (m)

14
13 11
12
11
200
10

300

Salinity

35.0 34.8
35.5 35.1

100 35.6
Depth (m)

35.4
35.0 34.8
34.9
34.8
200

35.5
300

Density (sigma-t)

26.55 26.60

100
Depth (m)

26.50 26.70
26.55 26.70
200 26.75

26.80

300

0 50 100 150 200 250 300

Distance (km)

Figure 11 An example of a conductivity, temperature and depth (CTD) section across the Sub-Tropical Front south of Australia
using a SeaSoar towed vehicle equipped with a Seabird CTD instrument. (From Tomczak M and Pender L (1999) Density
compensation in the Sub-Tropical Front in the Indian Ocean South of Australia.http://www.es.flinders.edu.au/ Bmattom/STF/
fr1098.html.
 TOWED VEHICLES 69

D Drag force
DN Normal drag force
DT Tangential drag force
L Lift force
M Pitching moment
Re Reynolds number ¼ ud/v
S Wing planform area
Sn Strouhal number ¼ fd/u
T Cable tension
W Tow vehicle weight
a Slope of the lift curve for a wing ¼ dCL/da
a0 Slope of the lift curve for an airfoil sec-
tion ¼ dCL/da
Figure 12 A faired cable winch and SeaSoar towed vehicle. b Wing span
This winch holds 5000 m of cable, 400 m faired and 4600 m of c Chord length
bare cable. A combination of faired and bare cable can be a cost- cg Center of gravity
effective method of achieving greater depths. d Cable diameter
f Frequency
offers advantages over hull-mounted transducers by l Cable length
deploying the acoustic transducer away from the u Velocity
high noise environment and bubble layer near the uN Normal velocity
ship and closer to the object of interest. A well-de- uT Tangential velocity
signed system can also be a more stable platform for w Cable weight per unit length
the acoustic transducers than a ship in rough seas. a Angle of attack
f Angle of the cable to the horizontal
f0 Cable equilibrium angle
The Winch f1 Cable angle at the towed vehicle
f2 Cable angle at the ship
Towed-vehicle systems using electromechanical
m Dynamic viscosity (E1  10  3 kg m  1 s for
cables usually require a special, purpose-built winch
water)
with accurate spooling gear and slip-rings to make
n Kinematic viscosity ¼ m/r (E1  10  6m2s  1
the electrical connection to the rotating drum. Cable
for water)
lengths can vary from 100 m to 5000 m. If a cable
r Density (E1000 kg m  3 for water)
with rigid fairing is used special blocks and fairing
guiding devices are needed to handle the cable. The
faired cable has a large bending radius and can only
be wound onto the winch drum in a single layer. As See also
illustrated in Figure 12, this type of winch is quite
large. If the towed vehicle system uses wire rope for Acoustic Scattering by Marine Organisms. Ships.
the tow cable, then the winch can be a standard type. Sonar Systems.

Nomenclature
A Area Further Reading
AC Aerodynamic center
Abbott IH and von Doenhoff AE (1959) Theory of Wing
AR Wing aspect ratio ¼ b2/S
Sections. New York: Dover Publications.
CDi Induced drag coefficient Anderson JD (1991) Fundamentals of Aerodynamics, 2nd
CDN Normal drag coefficient edn. New York: McGraw-Hill.
CDT Tangential drag coefficient Wingham PJ (1983) Comparative steady state deep towing
CL Lift coefficient performance of bare and faired cable systems. Ocean
CM Moment coefficient Engineering 10(1): 1--32.
 ZOOPLANKTON SAMPLING WITH NETS AND
TRAWLS
P. H. Wiebe, Woods Hole Oceanographic equipment normally used to sample zooplankton
Institution, Woods Hole, MA, USA until recent technological developments enabled the
M. C. Benfield, Louisiana State University, use of high-frequency acoustics and optical systems
Baton Rouge, LA, USA as well.
Copyright & 2001 Elsevier Ltd.

Net Systems
A variety of net systems have been developed over
the past 100 þ years and versions of all of these de-
Introduction vices are still in use today. They can be categorized
In the late 1800s and early 1900s, quantitative ocean into eight groups: non-opening/closing nets, simple
plankton sampling began with non-opening/closing opening/closing nets, high-speed samplers, neuston
nets, opening/closing nets (mostly messenger-based), samplers, planktobenthos plankton nets, closing
high-speed samplers, and planktobenthos net sys- cod-end samplers, multiple-net systems, and moored
tems. Technology gains inelectrical/electronic sys- plankton collection systems.
tems enabled investigators to advance beyondsimple
vertically or obliquely towed nets to multiple cod-
Non-opening/Closing Nets
end systems and multiple net systems in the 1950s
and 1960s. Recent technological innovation has en- Numerous variants of the simple non-opening/
abled net systems to be complemented or replaced by closing plankton net have been developed, which are
optical and acoustics-based systems. Multi-sensor principally hauled vertically. Most are simple ring-
zooplankton collection systems are now the norm nets with mouth openings ranging from 25 to 113 cm
and in the future, we can anticipate seeing the de- in diameter and conical or cylinder-cone nets 300–
velopment of real-time four-dimensional plankton 500 cm in length. Among the ring-nets that have
sampling and concurrent environmental measure- been widely used are the Juday net (Figure 1B),
ments systems, and ocean-basin scale sampling with International Standard Net, the British N-series nets,
autonomous vehicles. the Norpac net, the Indian Ocean Standard net
From the beginning of modern biological (Figure 1C), the ICITA net, the WP2 net, the Cal-
oceanography in the late 1800s, remotely operated COFI net, and the MARMAP Bongo net (Figure 1D).
instruments have been fundamental to observing Early nets were made from silk, but today nets are
and collecting zooplankton. For most of the past made from a square mesh nylon netting. Typical
century, biological sampling of the deep ocean has meshes used on zooplankton nets range from 150 mm
depended upon winches and steel cables to deploy a to 505 mm, although larger and smaller mesh sizes
variety of instruments. The development of quan- are available. Most of these nets are designed to be
titative zooplankton collecting systems began with hauled vertically. They are lowered to depth cod-end
Victor Hensen in the 1880s (Figure 1A). His first and then pulled back to the surface with animals
methods covered the whole scope of plankton being caught on the way up. Others, such as the
sampling from the building and handling of CalCOFI net and the Bongo net are designed to be
nets to the final counting of organisms in the towed obliquely from the surface down to a max-
laboratory. imum depth of tow and then back to the surface. The
Three kinds of samplers developed in parallel: Reeve net was a simple ring-net with a very large
waterbottle samplers that take discrete samples of cod-end bucket designed to capture zooplankton
a small volume of water (a few liters), pumping alive. The Isaacs-Kidd midwater trawl (IKMT) has
systems that sample intermediate volumes of water been used to collect samples of the larger macro-
(tens of liters to tens of cubic meters), and nets of zooplankton and micronekton. It has a pentagonal
many different shapes and sizes that are towed ver- mouth opening and a dihedral depressor vane as part
tically, horizontally, or obliquely and sample much of the mouth opening. Four sizes of IKMTs, 3 foot
larger volumes of water (tens to thousands of cubic (91 cm), 6 foot (183 cm), 10 foot (304 cm), and 15
meters) (Table 1). Net systems dominated the foot (457 cm) are often cited.

70
 ZOOPLANKTON SAMPLING WITH NETS AND TRAWLS 71

(B)

(A)
(C)

(E)

(D)

Figure 1 Some commonly used non-opening/closed nets. (A) The Hensen net. (Reproduced with permission from Winpenny, 1937.)
(B) The Juday net; note the use of messenger release on this version of the net. (Reproduced with permission from Juday, 1916.)
(C) The Indian Ocean Standard net. (Reproduced with permission from Currie, 1963.) (D) The Bongo net with CTD (c. 1999).
(Photograph courtesy of P. Wiebe.) (E) The Tucker trawl. (Reproduced with permission from Tucker, 1951.)

Non-opening/closing nets with rectangular mouth Simple Opening/Closing Nets

openings were not widely used until the Tucker trawl
was first described in 1951 (Figure 1E). This simple The development of nets that could obtain depth-
trawl design with a 180 cm  180 cm mouth opening specific samples evolved from those of very simple
gave rise to a substantial number of opening/closing design (a simple ring net) at an early stage. In the late
net systems described below. 1800s and early 1900s, there was considerable effort
72 ZOOPLANKTON SAMPLING WITH NETS AND TRAWLS

Table 1 Summary of zooplankton sampling gear types

Sampling gear Type of sampling Size fraction Resolving scale Typical operating
range

Vertical Horizontal Vertical Horizontal

Conventional methods
Waterbottles Discrete samples Micro/meso 0.1–1 m — 4000 m —
Small nets Vertically integrating Micro/meso 5–100 m — 500 m —
Large nets Vertical, obliquely Meso/macro 5–1000 m 50–5000 m 1000 m 10 km
Horizontally integrating
High-speed samplers Obliquely, horizontally Meso/macro 5–200 m 500–5000 m 200 m 10 km
integrating
Pumps Discrete samples Micro/meso 0.1–100 m — 200 m —
Multiple net systems
Continuous plankton Horizontally integrating Meso 10–100 m 10–100 m 100 m 1000 km
recorder
Longhurst-Hardy Obliquely, horizontally Meso 5–20 m 15–100 m 1000 m 10 km
plankton recorder integrating
MOCNESS Obliquely, horizontally Meso/macro 1–200 m 100–2000 m 5000 m 20 km
integrating
BIONESS Obliquely, horizontally Meso/macro 1–200 m 100–2000 m 5000 m 20 km
integrating
RMT Obliquely, horizontally Meso/macro 1–200 m 100–2000 m 5000 m 20 km
integrating
Multinet Vertically Obliquely, Meso/macro 2–1000 m 100–2000 m 5000 m 5 km
horizontally

Electronic optical or acoustical systems

Electronic High resolution in the Meso 0.5–1 m 5–1000 m 300 m 100s of km
plankton-counter horizontal/vertical plane
In situ silhouette High resolution in the Meso 0.5–1 m 5–1000 m 1000 m 10 km
camera net system horizontal/vertical plane
Optical plankton counter High resolution in the Meso 0.5–1 m 5–1000 m 300 m 100s of km
horizontal/vertical plane
Video plankton recorder High resolution in the Meso 0.01–1 m 5–1000 m 200 m 100s of km
horizontal/vertical plane
Ichthyoplankton recorder High resolution in the Meso 0.1–1 m 5–1000 m 200 m 10 km
horizontal/vertical plane
Multifrequency acoustic High resolution in the Meso/macro 0.5–1 m 5–1000 m 100 m 10 km
profiler system horizontal/vertical plane
Dual-beam acoustic High resolution in the Meso/macro 0.5–1 m 1–1000 m 800 m 100s of km
profiler horizontal/vertical plane
Split-beam acoustic High resolution in the Meso/macro 0.5–1 m 1–1000 m 1000 m 100s of km
profiler horizontal/vertical plane
ADCP High resolution in the Meso/macro 10 m 5–500 m 500 m 100s of km
horizontal/vertical plane

Most vertical nets are hauled at a speed of 0.5–1 m s1. Normal speed for horizontal tows are B2 knots (1 m s1) and for high-speed
samplers B5 knots (2.6 m s1). For further categorization of pumping systems which are used by a number of investigators, reference
is made to the review paper by Miller and Judkins (1981). (Reproduced with permission from Sameoto D, Wiebe P, Runge S, et al.
(2000) Collecting zooplankton. In: Harris R, Wiebe P, Lenz J, Skjoldal HR, and Huntley M (eds.) ICES Zooplankton Methodology
Manual, pp. 55–81. New York: Academic Press.)

to develop devices that closed or opened and closed quickly followed. In the mid-1930s, the Leavitt net
nets at depth. Most employed mechanical release system became popular and variants of this system
devices which were attached to the towing wire and are still being used today (Figure 2B). Another
activated by messengers traveling down the towing popular system still in use today is the Clarke and
wire. The single-messenger Nansen closing mech- Bumpus sampler, a two-messenger zooplankton col-
anism and its variants were very popular during most lection system that can be deployed as multiple units
of early to mid-twentieth century (Figure 2). Double- on the wire and has a positive means of opening and
messenger systems that opened and then closed a net closing the mouth of the net (Figure 2C).
 ZOOPLANKTON SAMPLING WITH NETS AND TRAWLS 73

r
(A) m (B)
(A) c g
p t1 a
A f Cod end
d e
r l h t2
l b w
a a
e
a
r g
B t1 d (B) e
g t2
g b a
c
w

r
C
r
t1
(C) g
t2 a c
d
b e
w

(D)
b b
s s s s

L L

Fig. 1. Closing- Fig. 2. The net

net open, ready closed, as it is
to sent down. hauled up.

G
(C)
Q2
R

Q1 MN 3

F
J

Y
E T
1
P
D
A 8
O
H C I
4

K
L

Figure 2 Some commonly used simple opening/closing nets. (A) The single-messenger Nansen closing net. (Reproduced with
permission from Nansen, 1915.) (B) The two-messenger Leavitt net. (Reproduced with permission from Leavitt, 1935.) (C) The two-
messenger Clarke-Bumpus net. (Reproduced with permission from Clarke and Bumpus, 1939.) The plankton purse seine (D)
represents an unusual way to collect plankton from a specific region. (Reproduced with permission from Murphy and Clutter, 1972.)

Mechanical tripping mechanisms activated by fall of the net rather than the reverse – so-called
pressure, by combinations of messengers and flow- pop-down nets; to sample under sea ice using the
meter revolutions, or clocks have also been English umbrella net; to sample plankton from sev-
devised. eral depths simultaneously, using a combination of
Nontraditional approaches to collecting plankton nets and a pumping system; to sample plankton from
include designs to catch plankton on the downward the nuclear submarine, SSN Seadragon; to open and
74 ZOOPLANKTON SAMPLING WITH NETS AND TRAWLS

close a Tucker-style trawl using two towing cables, and made opened bodied; hence it became a modified
one for the top spreader bar and one for the bottom, Gulf V. The Ministry of Agriculture, Fisheries and
with each cable going to a separate winch; and to Food MAFF/Guildline high-speed samplers, de-
capture plankton and fish larvae with a plankton veloped in the 1980s, were also modified Lowestoft
purse seine (Figure 2D). samplers. These systems have a Guildline CTD sen-
sor unit with oxygen, pH, and digital flowmeter as
additional probes with telemetry through a con-
High-speed Samplers
ducting cable. Recently in the 1990s, the Gulf VII/
Most of the net systems described above were towed Pro net and MAFF/Guildline high-speed samplers
at speeds o3 knots (150 cm s1). High-speed sam- were developed that are routinely towed at 5–7
plers typically towed at speeds of 3–8 knots (150– knots.
400 cm s1) were also developed in the late 1800s and Other high-speed samplers were developed during
early 1900s to sample in bad weather, for underway the 1950s and 1960s, including a high-speed plank-
sampling between stations, or to reduce the effects of ton sampler which could collect a series of samples
net avoidance by the larger zooplankton. The Hardy during a tow; the ‘Bary Catcher’ that had an opening/
plankton indicator, developed in the 1920s, was the closing mechanism in the mouth of the sampler
first widely used device. The original version was (Figure 3D); a vertical high-speed sampler with a
17.8 cm in diameter and 91.4 cm in length with a rectangular mouth opening that could be closed
circular filtering disk on which plankton were col- using the Juday method; an automatic high-speed
lected. It was subsequently modified (and renamed plankton sampler with 21 small nets that were se-
the standard plankton indicator) to make it smaller, quentially closed by means of a cam/screw assembly
more streamlined, and equipped with a depressor and driven by a ships log (propellor); and the Clarke Jet
stabilizing fins (Figure 3). An even smaller version, net that was an encased high-speed sampler with an
the Small Plankton Sampler, was developed. In the elaborate internal passageway designed to reduce the
1950s, it was further modified and named the Small flow speed of water within the sampler to that nor-
Plankton Indicator, and in the 1960s, it was modified mally experienced by a slowly towed net.
again so that multiple units could be used on the The continuous plankton recorder (CPR) is in a
towing wire at speeds of 7–8 knots with a multiplane class by itself when it comes to high-speed plankton
kit otter depressor at the end of the wire. Until the samplers, because it can take many samples and can
1950s, only one high-speed collector was designed be towed from commercial ships (Figure 3E). Ori-
with a double-messenger system that enabled the ginally built in the 1920s, it has evolved over the
mouth to be opened and closed; most could not make years to become the mainstay in a plankton survey
depth-specific collections. program in the North Atlantic. This encased sampler
The ‘Gulf’ series of high-speed samplers developed weight 87 kg and is about 50 cm wide by 50 cm tall
in the 1950s and early 1960s gave rise to a number of by 100 cm long. The 1.27 cm  1.27 cm rectangular
high-speed samplers still in use today. The first was aperture expands into a larger tunnel opening. The
the Gulf I-A which looked similar to earlier high- tunnel passes through the lower portion of the
speed samplers. The Gulf III was a much larger high- sampler and out of the back. Below the tunnel is one
speed sampler that was enclosed in a metal case. The spool of silk gauze which threads across the tunnel
Gulf V was an unencased and scaled-down version of and captures the plankton. A second spool of silk
the Gulf III (Figure 3B). The Gulf III and Gulf V gauze lies above the tunnel and is threaded to meet
samplers have been very popular, and have been the first gauze strip as it leaves the tunnel, sand-
modified numerous times. In the early 1960s, a five- wiching the plankton between the two strips. The
bucket cod-end sampling device was added to the gauze strips are wound up on a take-up spool which
Gulf III that was electrically activated from a deck resides in a formalin-filled tank above the flow-
unit through two-conductor cable. HAI (shark) was through tunnel, preserving the plankton. The take-up
the German version of the Gulf III built in the mid- spool is driven by a propellor on the back of the
1960s. A hemispherical nose cone and an opening/ sampler behind the tail fins. This sampler is usually
closing lid were added to the HAI. This German towed at 20 knots from commercial transport vessels
system evolved further when ‘Nackthai’ (naked at a fixed depth of about 10 m below the surface,
shark), a modified Gulf V sampler, was developed in thus it only samples the surface layer of the ocean.
the late 1960s. Also in the 1960s, the British modi- The undulating oceanographic recorder (UOR) was
fied the Gulf III sampler, which was subsequently developed in the 1970s to extend the vertical sam-
called the Lowestoft sampler (Figure 3C). Sub- pling capability of high-speed plankton collection
sequently, the Lowestoft sampler was scaled down systems. The UOR carries sensors to measure
 ZOOPLANKTON SAMPLING WITH NETS AND TRAWLS 75

(B)

(A)

(C)

(D)

5 6 7

(E)

Figure 3 Some examples of high-speed plankton samplers. (A) The standard plankton indicator. (Reproduced with permission from
Hardy, 1936.) (B) The encased Gulf III sampler. (Reproduced with permission from Gehringer, 1952.) (C) The open-bodied Lowestoft
sampler (Gulf V type). (Reproduced with permission from Lockwood, 1974.) (D) The Bary catcher. (Reproduced with permission from
Bary, 1958.) (E) The continuous plankton recorder (CPR). (Reproduced with permission from Hardy, 1936.)
76 ZOOPLANKTON SAMPLING WITH NETS AND TRAWLS

temperature, salinity, and pressure; data are logged

internally at 30 observations per minute. A propellor
drives the rollers winding up the gauze and provides
the power for the electronics.

Neuston Samplers
Nets to collect neuston, the zooplankton that live
within a few centimeters of the sea surface, by-and-
large are non-opening/closing. The first net specific-
ally designed to sample zooplankton neuston was
built in about 1960. A rectangular mouth opening
(A)
design is typical of most of the systems. Neuston nets
come either with a single net which collects animals p
right at the water surface or vertically stacked sets of e
two to six nets extending from the surface to about
100 cm depth (Figure 4). Normally they are towed
from a vessel, but a ‘push-net’ was developed in the af
h
20
1970s with a pair of rectangular nets positioned side- 10
by-side in a framework and mounted in front of a bb
10
small catamaran boat that pushed the frame through
the water at B2.6 knots. 20
b c2
Planktobenthos Plankton Nets 20
n
The ocean bottom is also special habitat structure for c1
ac 20
zooplankton, and gear to sample zooplankton living
here (‘planktobenthos’) was developed early. The e
first nets were designed in the 1890s specifically to
sample plankton living very near the bottom. Non- (B) 25
opening/closing systems were succeeded by samplers
with mechanically operated opening/closing doors or
with a self-closing device (Figure 5A).
An entirely different strategy has been to employ
manned submersibles or deep-towed vehicles to col-
lect deep-sea planktobenthos. A pair of nets mounted
on the front of DSRV Alvin was used for making net Detail A
collections at depths 41000 m in the 1970s; the
pilot opened and closed the net (Figure 5B). A mul-
tiple net system was used on the Deep-Tow towed
body. This system was attached to the bottom of the
Deep-Tow and used for sampling within a few tens of Detail B
meters above the deep-sea floor in the 1980s
(Figure 5C). This net system was later adapted for
use on DSRV Alvin for near-bottom studies of
plankton in the vicinity of hydrothermal vent sites in
the 1990s.
On other benthic habitats, such as coral reefs,
fixed or stationary net systems which orient to the
current’s flow and filter out zooplankton drifting by, (C)
nets pushed by divers, and traps have been used to
capture plankton close to the bottom. The Hori- Figure 4 Neuston net samplers collect plankton living at the
zontal Plankton Sampler (HOPLASA) creates its own sea surface. (A) A single net system. (Reproduced with
permission from David, 1965.) (B) A multinet system.
current to collect zooplankton on or near the bottom
(Reproduced with permission from Ellertsen, 1977.) (C) A push
in coral reef areas with variable or little current flow net. (Reproduced with permission from Miller, 1973.)
(Figure 5D).
 ZOOPLANKTON SAMPLING WITH NETS AND TRAWLS 77

(A)

(B)

(C) (D)

Figure 5 Some planktobenthos samplers. (A) Early system with opening/closing doors. (Reproduced with permission from
Wickstead, 1953.) (B) DSR Alvin opening/closing system. (Reproduced with permission from Grice, 1972.) (C) The Deep-Tow multiple
net system. (Reproduced with permission from Wishner, 1980.) (D) A system for coral reef sampling (HOPLASA). (Reproduced with
permission from Rutzler, 1980.)

Closing Cod-end Systems

systems began to do more than collect animals at
In the late 1950s and 1960s, conducting cables and specific depth intervals. Single nets equipped with
transistorized electronics were beginning to be closing cod-end devices preceded multiple net sys-
adapted for oceanographic use and sophisticated net tems by only a few years. One of the first systems
78 ZOOPLANKTON SAMPLING WITH NETS AND TRAWLS

used a 1950s version of a serial device in the high- and later modified by adding environmental sensors
speed sampler that was mechanically driven by a and an electronically controlled opening/closing
propellor. Another had a pressure-actuated catch- mechanism.
dividing bucket (CDB) attached to the back of an The Longhurst-Hardy plankton recorder (LHPR),
IKMT (Figure 6A). The Mark III Discrete Depth a modification of the CPR, was developed in the
Plankton Sampler (DDPS) also developed for use 1960s (Figure 6D). The recorder box was attached to
with an IKMT or a 1 m diameter net, had four catch the back end of a net and gauze strips in the box were
chambers separated by solenoid-activated damper advanced in discrete steps (15 s to 60 s) by an elec-
doors (Figure 6B). This latter system was one of the tronics package on the tow frame; data on pressure,
first to carry underwater electronics to sample depth temperature, and flow were logged on an internal
and temperature, and to telemeter the data up a recorder; power was supplied by a NICAD battery
single conductor cable for display at the surface. The pack. The LHPR was redesigned in the 1970s to
multiple plankton sampler (MPS, described below) reduce problems with hang-ups and stalling of
was turned into a cod-end sampler for an IKMT animals in the net which caused smearing of the

142 (147)cm
(A) B
(B) (Overall length)

10 (15) cm Stabilizing fin

(Outside DIA) 19 (26) cm DIA
Q Power line cover

R
Solenoid 9 cm
Z K 30 cm
D Power connection
F
T1
19 cm
C1
V Screen area = 190 (342) sq cm
G
P1
W M A2
A1
S2

S1
(D)

(C)

Figure 6 Some discrete-depth samplers using a closing cod-end. (A) The catch dividing cod-end. (Reproduced with permission from
Foxton, 1963.) (B) The Mark III multiple cod-end bucket. (Reproduced with permission from Aron et al. 1964.) (C) ARIES.
(Reproduced with permission from Dunn et al. 1993.) (D) A version of the LHPR. (Photograph courtesy of J. Smith, 1966.)
 ZOOPLANKTON SAMPLING WITH NETS AND TRAWLS 79

distributions of animals and losses of animals from System (BIONESS), with 10 nets, developed in the
the recorder box. The modified LHPR was used 1980s (Figure 7D). A modified version of the MPS
without a net on the conning-tower of the US Navy was developed in Germany at about the same time
research submarine Dolphin in the 1980s. Another and named the Multinet; it carried five nets, which
modification of the LHPR was made by the British in were opened and closed electronically via conducting
1980s. They used an unenclosed Lowestoft sampler cable (Figure 7C). A scaled-up version of BIONESS
to mount a pair of recorder boxes to collect meso- built in the 1990s was the Large Opening Closing
and micro-zooplankton. The system acoustically High Speed Net and Environmental Sampling System
telemetered depth, flow, and temperature. It also (LOCHNESS). Another variant of the MPS was the
carried a chlorophyll sensor with a recorder system. Ocean Research Institute’s (Japan) vertical multiple
The LHPR was further modified for use in catching plankton sampler developed in the 1990s in which
Antarctic krill. A descendant of the LHPR developed the nets are opened/closed by surface commands
in the 1990s is the Autosampling and Recording transmitted via conducting cable to an underwater
Instrumental Environmental Sampler (ARIES) unit.
(Figure 6C). This cod-end plankton sampling device
is a stretched version of the Lowestoft-modified Gull Moored Plankton Collection Systems
III frame. It has a multiple cod-end system, water
Only a few instrument systems have been developed
sampler, data logger, and an acoustical telemetry
that autonomously collect time-series samples of
system.
plankton from moorings. Most were patterned after
the CPR or LHPR (e.g. the O’Hara automatic
Multiple Net Systems plankton sampler built in the 1980s; a modified
version of the O’Hara system built in the 1990s; the
The development of multiple net systems began with
moored, automated, serial zooplankton pump
the simple non-opening/closing Tucker trawl system.
(MASZP) built in the late 1980s) (Figure 8). The lack
In the mid-1960s, timing clocks were used to open
of such systems may be due to the difficulty of
and close the Tucker trawl mouth. Then late in the
powering them for long periods underwater.
1960s, the British rectangular mouth opening trawl
(RMT), which was opened and closed acoustically,
was developed. The RMT was expanded into the Optical Systems
NIO Combination Net (RMT 1 þ 8), which carries
nets with 1 m2 and 8 m2 mouth openings (Figure 7A). Optical survey instruments can be divided into two
This was expanded into a multiple net system with categories, based on whether the systems produce an
three sets of 1 m and 8 m nets controlled acoustically. image of their zooplankton targets (e.g. video,
The acoustic command and telemetry system for the photographic, and digital camera systems) or use the
RMT 1 þ 8 was replaced in the 1990s by a micro- interruption of a light source to detect and estimate
computer-controlled unit connected by conducting the size of particles (e.g. the optical plankton coun-
cable to an underwater electronics unit. ter). The first attempts to quantify plankton optically
In a parallel development in the 1970s, a five-net appear to have been made in the 1950s using a beam
and a nine-net Tucker Multiple Net Trawl was de- of light projected into the chamber from a 300 W
veloped on the West Coast of the USA. The system mercury vapor lamp and a Focabell camera (Orion
was powered electrically through conducting wire Camera, Tokyo).
and controlled from the surface. A modified Tucker
Image-forming Systems Mounted on Non-opening/
trawl system, the Multiple Opening/Closing Net and
Closing Nets
Environmental Sensing System (MOCNESS), with
nine nets and a rigid mouth opening was built soon In the 1980s, a 35 mm still camera with a high-
after on the US east coast (Figure 7E). The current capacity film magazine in front of the cod-end of a
versions of the MOCNESS are computer-controlled plankton net attached to a rigid frame was used to
(Table 2). Sensors include pressure, temperature, take in situ silhouette photographs of zooplankton as
conductivity, fluorometer, transmissometer, oxygen, they passed into the cod-end. This was a field ap-
and light. plication of the laboratory-based silhouette photog-
The design of the Bé multiple plankton sampler raphy system developed in the late 1970s. The
(MPS) (Figure 7B), initially messenger operated in camera provided a series of photographic images at
the late 1950s and then pressure-actuated in the points along the trajectory of the net separated by
1960s, was the basis for the Bedford Institute of o1 m. In the development of the ichthyoplankton
Oceanography Net and Environmental Sensing recorder, the still camera was replaced with a video
80 ZOOPLANKTON SAMPLING WITH NETS AND TRAWLS

(A) (B) (C)

Net
Monitor

Bar 1

Closing Bar (6)

RMT 1
Opening Bar (7)

Bar 2

Closing Bar (3)

RMT 2
Opening Bar (4)

Bar 5

(E)

(D)

Figure 7 Some examples of multiple net plankton sampling systems. (A) The RMT 1 þ 8. (Reproduced with permission from Baker,
1973.) (B) The Bé net. (Reproduced with permission from Bé, 1959.) (C) The Multinet. (Photograph courtesy of B. Niehof.) (D) The
BIONESS. (Photograph courtesy of P. Wiebe, 1993.) (E) The 1 m2 MOCNESS. (Photograph courtesy of Wiebe, 1998.)

camera, which was located in front of the cod-end of imaging zooplankton within a defined volume of
a high-speed Gulf V-type net (Nackthai). It had an water (Figure 9A). The original VPR had four video
estimated horizontal spatial resolution of 3 cm. One cameras; each camera imaged concentrically located
consequence of going from camera film to video tape volumes of water ranging from 1 ml to 1000 ml,
was a loss of image resolution. but it has been modified to a one- or two-camera
system. It has been possible to image undisturbed
animals in their natural orientations. The current
Stand-alone Image-forming Systems
VPR image processing system is capable of digitizing
The video plankton recorder (VPR) was developed in each video field in real time and scanning the fields
the early 1990s as a towed instrument capable of for targets using user-defined search criteria for
 ZOOPLANKTON SAMPLING WITH NETS AND TRAWLS 81

Table 2 MOCNESS system dimensions and weights

System Number of Width of Height of Net width Mouth area Length of Approx. Rec. wire
nets frame (m) frame (m) (m) at 451 net (m) weight in diameter
towing air (kg) (mm)
angle (m)

MOCNESS-1/4 9 0.838 1.430 0.50 0.5 6.00 70 6.4

MOCNESS-1/4- 18/20 1.430 1.430 0.50 0.5 6.00 155 7.4
Double
MOCNESS-1 9 1.240 2.870 1.00 1.0 6.00 150 7.4
MOCNESS-1-Double 18/20 2.560 2.870 1.00 1.0 6.00 320 12.1
MOCNESS-2 9 1.650 3.150 1.41 2.0 6.00 210 11.8
MOCNESS-4 6 2.140 4.080 2.00 4.0 8.44 460 11.8
MOCNESS-10 6 3.410 4.690 3.17 10.0 18.25 640 11.8
MOCNESS-20 6 5.500 7.300 4.47 20.0 14.50 940 17.3

The MOCNESS systems are denoted by the mouth area when being towed. Thus a MOCNESS-1/4 has a 0.25 m2 mouth opening.
The ‘Double’ systems have two sets of nets side-by-side in a single rigid framework. Nets can be opened and closed on one side and
then opened and closed on the other.

brightness, focus, and size. The targets are identified and foraminiferans; and one has been mounted on
using a zooplankton identification program to pro- a 1 m2 MOCNESS net system to map the fine-scale
vide near-real-time maps of the zooplankton distributions of the larval cod prey items. A moored
distributions. system called the Autonomous Vertically Profiling
A number of VPR-based systems are currently in Plankton Observatory (AVPPO) utilizes an internally
operation or under development: a single-camera recording, two-camera VPR, and has been deployed
system is mounted on the BIOMAPER II vehicle in coastal waters off New England.
(described below); an internally recording VPR has Image resolution constraints inherent in the use of
been constructed and used to quantify radiolarians standard video formats have driven the development

Sampler unit Flow generation & measurement unit

Time- or event-triggered
Rubber hose connection
automated, serial,
Flow meter Outboard motor plankton pump

Intake

0 20 40
Control unit & pressure case Centimetres Supply
spool
Net storage reel
Net storage Chafing rail
Take-up
chamber Hall effect spool
mechanism
Fixable bath
Inlet Collection net

Wind motor Filling tube

housing
Preservative Pump Outlet
Takeup reel
chamber
Drain plug

Motor
0 10 20
Carriage chassis
Centimetres
Front view Sampler unit Side view (A) (B)

Figure 8 Two examples of moored plankton collecting systems. (A) A modified version of the O’Hara sampler. (Reproduced with
permission from Lewis and Heckl, 1991.) (B) MASZP. (Reproduced with permission from Doherty et al. 1993.)
82 ZOOPLANKTON SAMPLING WITH NETS AND TRAWLS

(A)

(B)

(C)

Figure 9 Examples of optical or electrical systems for collecting zooplankton data. (A) The VPR. (Photograph courtesy of P. Alatalo,
1999.) (B) The in-situ zooplankton detecting device. (Photograph courtesy of P. Wiebe, c. 1972.) (C) The optical plankton counter
(OPC). (Photograph courtesy of M. Zhou, 2000.)
 ZOOPLANKTON SAMPLING WITH NETS AND TRAWLS 83

of optical systems that utilizes higher-resolution counting and sizing system appeared in the late
formats. A modification of the continuous underway 1960s and was referred to as the in situ zooplankton
fish egg sampler (CUFES, described below) utilizes detecting device (Figure 9B). A shipboard version of
a line-scanning digital camera to quantify the abun- the device was connected to a continuously pumped
dances of fish eggs. The shadowed image particle stream of water and employed to analyze spatial
profiling and evaluation recorder (SIPPER) utilizes heterogeneity of zooplankton in surface waters in
high-resolution digital line-scanning cameras to relation to chlorophyll fluorescence and temperature.
quantify zooplankton passing through a laser light A version of this conductive zooplankton counter
sheet. The SIPPER has been mounted either on a was deployed aboard a Batfish towed vehicle in the
towed vehicle called the high-resolution sampler 1980s.
(HRS) or an AUV. A second group of particle detectors utilized pho-
The need for systems to quantify the abundance of todetectors rather than changes in voltage. The
‘marine snow’ prompted development of profiling Opto-Electronic Plankton Sizer was a laboratory-
systems based on both still and video cameras. In the based system designed in the 1970s to automate
1980s, a profiling system called the large amorphous the measurement of preserved plankton samples.
aggregates (LAA) camera was constructed which The HIAC particle size analyzer was modified at the
employed a photographic camera and a pair of Lowestoft Laboratory during the late 1970s for
strobes to photograph marine aggregates. A video plankton counting. The optical plankton counter
profiling instrument called the underwater video (OPC) was developed during the mid-1980s
profiler (UVP) has been used to quantify the vertical (Figure 9C). This instrument measures changes in the
distribution and size frequency of marine snow, and intensity of a light beam that occur when a particle
to examine the distributions of macrozooplankton. crosses the beam. The OPC has been mounted on a
The UVP consists of a Hi-8 video camera imaging a variety of towed platforms or in shore-based or
collimated light sheet coupled with a CTD, data shipboard applications. The OPC has also been in-
logger, and batteries. A profiling system called corporated into a shipboard device called the con-
ZOOVIS recently has been developed around a high tinuous underway fish egg sampling system (CUFES)
resolution (2048  2048 pixel) digital camera and which enumerates the distribution and abundance of
CTD linked to a surface workstation via a fiber-optic fish eggs in surface waters. In spite of the prevalence
cable. A color video camera has been mounted on the of OPC systems in current use, interpretation of OPC
front of a Sea Owl II remotely operated vehicle data remains a subject of some controversy.
(ROV) and used to quantify the vertical distribution
of gelatinous zooplankton off the west coast of
Optical Instruments for Nonquantitative Studies
Sweden.
Still holographic imaging of plankton in a la- The ecoSCOPE is an optical video-endoscope that
boratory was first reported in 1966. It was refined in enables direct observation of predator–prey inter-
the 1970s to record movies of live plankton in the actions between juvenile fish and zooplankton. The
laboratory. In the 1990s, a submersible internally ecoSCOPE has been operated from an ROV, from
recording in-line holographic camera that records up the keel of a sailing vessel, and in towed and moored
to 300 holograms on a film emulsion was developed. modes, but the best recordings of predator/prey
Many zooplankton produce or induce the pro- interactions have come from free-drifting deploy-
duction of bioluminescent light that can be detected ments, when the instrument was hovering within
with sensitive CCD cameras. One system is mounted schools of feeding juvenile herring. A software
on the Johnson SeaLink manned submersible and package called dynIMAGE animates sequential im-
consists of an intensified silicon-intensified target ages keeping the fish and its prey in the middle of the
(ISIT) video camera mounted on and aimed forward viewing field.
at a 1 m diameter transect screen to quantify the Optical sensors can provide valuable ground-
distribution, abundance, and identities of biolumin- truthing for acoustical sensors. In the 1990s, a
escent zooplankton. megapixel digital still camera was mounted on a
FishTV sonar array and the resulting system was
named the Optical-Acoustical Submersible Imaging
Particle Detection Systems
System (OASIS). In this system, high acoustic returns
Particle detection systems refer to non-image- are used to trigger the camera taking a picture of the
forming devices that utilize interruption of an elec- acoustical target. An analog video camera aimed at
trical current or a light beam to detect and estimate the focal point of an acoustic array mounted on the
the size of a passing particle. The first in situ particle front of a MAXRover ROV has been used to take
84 ZOOPLANKTON SAMPLING WITH NETS AND TRAWLS

pictures of individual zooplankton passing through (Figure 10A). An EG&G Edgerton model 205 cam-
the acoustic beam. era and aflash light were mounted on the top of a
modified MOCNESS and on the top of BIONESS to
take black and white photographs about 2 m in front
High-frequency Acoustics of the net mouth. The BIONESS has also been
High-frequency acoustics (Z38–1000 kHz) provide equipped with an OPC and video lighting system,
the foundation for another class of tools to study and used in conjunction with an echosounder.
zooplankton. The utility of the acoustic systems de- The BIo-Optical Multi-frequency Acoustical and
rives from their ability to operate with high ping Physical Environmental Recorder – BIOMAPER II –
rates and precision range-gating. Mapping plank- was developed to conduct high-speed, large-area
tonic distributions on a wide range of space and surveys of zooplankton and environmental property
timescales is becoming possible because of the con- distributions to depths of 500 m (Figure 10B).
tinued development of acoustics systems and ap- Mounted inside are a multi-frequency sonar (up-
propriate ground-truthing methods. There are two wards-looking and downwards-looking pairs of
fundamental measurements: volume backscattering transducers operating at five frequencies: 43, 120,
(integration of the energy return from all individuals 200, 420, and 1000 kHz), an environmental sensor
in a given ensonified volume, i.e. echo integration) package (CTD, fluorometer, transmissometer), and
and target strength (echo strength from an indi- several other bio-optical sensors (down- and up-
vidual). Statistical procedures have been developed welling spectral radiometers, spectrally matched at-
to estimate animal assemblage size distribution using tenuation, and absorption meters). A single-camera
the data from single-beam transducers. In some video plankton recorder (VPR) system is mounted
cases, it is possible to extract estimates of animal above and just forward of the nose piece. The lower
target strength distribution in addition to volume four acoustical frequencies involve split-beam tech-
backscattering from a series of single-beam trans- nology and are able to make target strength and echo
ducers operating at different frequencies. Multi- integration measurements.
beam acoustical systems provide a direct means of A variety of vehicles have been built that actively
determining individual target strength (TS). The two change their vertical position without changing the
current designs, dual-beam and split-beam, both towing wire length. Examples for surveying zoo-
provide a hardware solution to the problem of TS plankton include the undulating oceanographic re-
determination. corder and SeaSoar equipped with optical (VPR and
OPC) and/or acoustical (the Tracor Acoustical Pro-
filing System, TAPS). Remotely operated vehicles
The Current State of Plankton (ROVs) have also been equipped with acoustical and
video systems to study zooplankton. A SeaRover
Sampling Systems
ROV was equipped with the same dual-beam
The diversity of zooplankton samplers in use today acoustic system and environmental sensors. A VPR
reflects the fact that no single collection system ad- rigged to provide 3-D images of plankton and an
equately samples all zooplankton. Non-opening/ environmental sensor package (temperature, con-
closing nets, such as the WP2, the modified Juday ductivity, pressure, fluorescence) were mounted on
net, and the Bongo net, are used in large ocean sur- the front of the ROV JASON and on the SeaRover
veys. Simple, double-messenger opening/closing nets ROV (Figure 10C). FishTV (FTV) has been used on a
similar to those developed in the first half of the last Phantom IV ROV and a combination of acoustics
century are still manufactured and used. The Multi- and video has been used on the front of a
net, RMT 1 þ 8, BIONESS, and MOCNESS are MAXRover ROV. Dual-beam acoustics (420 and
widely used multiple-net systems that also carry 1000 kHz) have also been deployed on the DSRV
additional sensors to measure other water properties. Johnson SeaLink.
Plankton pumps are also being used, especially to
collect micro-zooplankton.
The advent of high-speed computers and towing
Future Developments
cables with optical fibers and electrical conductors
have enabled development of multi-sensor towed The future promises vastly increased application of
systems which provide real-time data while the in- remote sensing techniques and sensor development,
strument package is deployed. The MOCNESS and real-time data telemetry, processing, and display.
has been equipped with a high-frequency acoustic Three-dimensional (space) and four-dimensional
system for forward or sideways range-gated viewing visualization (space and time) of biological and
 ZOOPLANKTON SAMPLING WITH NETS AND TRAWLS 85

(A) (B)

(C)

Figure 10 Examples of multi-sensor plankton sampling systems. (A) MOCNESS with a dual-beam acoustic system. (Photograph
courtesy of P. Wiebe, 1994.) (B) BIOMAPER-II. (Photograph courtesy of P. Wiebe, 1999.) (C) The JASON-ROV with 3-D VPR system.
(Photograph courtesy of P.Alatalo, 1995.)

acoustic data are also an increasingly important as- for biological studies is the Autosub-1 that carries a
pect of data processing. For a number of research gyrocompass, ADCP, an echosounder, and acoustic
programs today, the development of an image of the telemetry and surface radio electronics. It can be
spatial arrangement of organisms is but the first step programmed to run a geographically based course
in efforts to study and understand their relationships using GPS surface positions and dead reckoning.
to each other and to their environment. Thus, there is The autonomous Lagrangian circulation explorer
need for real-time 3-D and 4-D images. (ALACE) and the more recently developed profiling
Autonomous self-propelled vehicles (AUVs) have version (PALACE) floats that carry temperature and
only recently begun to be used widely to gather- conductivity probes are vertically migrating neutrally
oceanographic data. The remote environmental buoyant drifters. They track the movements of water
measuring units (REMUS) are a new class of small at depths between the surface and 1000–2000 m
AUVs which can carry an impressive array of en- depth. Hundreds to thousands of the PALACE floats
vironmental sensors including a VPR. Another class will be deployed over the next few years and it is
of autonomous vehicles is epitomized by the au- expected that they will become a mainstay in the
tonomous benthic explorer (ABE), which is equipped Global Ocean Observing System (GOOS). The next
with precise navigation and control systems that generation of neutrally buoyant floats is an autono-
enable it to descend to a worksite, navigate preset mous glider named SPRAY. SPRAY will be able
tracklines or terrain-follow, and find a docking sta- to sail along specific preprogrammed tracklines. A
tion. A much larger AUV which has been employed further step in their development is to provide
86 ZOOPLANKTON SAMPLING WITH NETS AND TRAWLS

biological instrumentation to complement the phys- Further Reading

ical sensors.
High-resolution optical systems, suchas the VPR, Harris RP, Wiebe PH, Lenz J, Skjoldal HR, and Huntley M
combined with computer-based identification pro- (eds.) (2000) ICES Zooplankton Methodology Manual.
grams can now provide higher level taxa identifi- New York: Academic Press.
Kofoid CA (1991) On a self-closing plankton net for
cations in near-real time. Classification of species
horizontal towing. University of California Publications
using acoustic signatures is less well developed and it in Zoology 8: 312--340.
now seems unlikely that the technology to develop Miller CB and Judkins DC (1981) Design of pumping
species-specific acoustic signatures will be developed systems for sampling zooplankton with descriptions of
soon. Molecularly based species identification is two high-capacity samplers for coastal studies. Biol
likely to make significant strides in the next decade. It Oceanogr 1: 29--56.
is now conceivable that this information will enable Omori M and Ikeda Y (1976) Methods in Marine
simultaneous analysis, identification, and quantifi- Zooplankton Ecology. New York: John Wiley & Sons.
cation of all species occurring in a zooplankton Schulze PC, Strickler JR, Bergström BI, et al. (1992) Video
sample. systems for in situ studies of zooplankton. Arch
Hydrobiol Beih Ergebn Limnol 36: 1--21.
Sprules WG, Bergström B, Cyr H, et al. (1992) Non-video
optical instruments for studying zooplankton
See also distribution and abundance. Arch Hydrobiol Beih
Ergebn Limnol 36: 45--58.
Acoustic Scattering by Marine Organisms.
Tranter DJ (ed.) (1968) Part I. Reviews on Zooplankton
Continuous Plankton Recorders. Grabs for Shelf
Sampling Methods. Monographs on Oceanographic
Benthic Sampling. Platforms: Autonomous Under-
Methodology, Zooplankton Sampling. UNESCO
water Vehicles; Satellite Remote Sensing SAR.
MOORINGS, FLOATS, AND CURRENT
METERS
 DRIFTERS AND FLOATS
P. L. Richardson, Woods Hole Oceanographic values of these statistics can be plotted and contoured
Institution, Woods Hole, MA, USA to reveal, for example, patterns of ocean circulation
& 2009 Elsevier Ltd. All rights reserved. and the sources and sinks of eddy energy. Maps of
velocity fields can be combined with measurements
of hydrography to give the three-dimensional velocity
field of the ocean. Oceanographers are using the newly
acquired drifter data in these ways and also incor-
Introduction porating them into models of ocean circulation.
Starting in the 1970s, surface drifters and subsurface Care must be used in interpreting drifter mea-
neutrally buoyant floats have been developed, im- surements because they are often imperfect current
proved, and tracked in large numbers in the ocean. followers. For example, surface drifters have a small
For the first time we have now obtained worldwide downwind slip relative to the surrounding water.
maps of the surface and subsurface velocity at a few They also tend to be concentrated by currents into
depths. New profiling floats are measuring and re- near-surface converge regions. The surface water
porting in real time the evolving temperature and that converges can descend below the surface, but
salinity structure of the upper 2 km of the ocean in drifters are constrained to remain at the sea surface
ways that were impossible a decade ago. These in the convergence region. Surface currents some-
measurements are documenting variations of the times converge drifters into swift ocean jets like the
world ocean’s temperature and salinity structure. Gulf Stream. This can result in oversampling these
The new data are revealing insights about ocean features and in gridded mean velocities that are dif-
circulation and its time variability that were not ferent from averages of moored current meter
available without drifters and floats. measurements, which are called Eulerian measure-
Surface drifters and subsurface floats measure ments of the flow. Bin averages of drifter velocities
ocean trajectories that show where water parcels go, can give misleading results if the drifters are very
how fast they go there, and how vigorously they are unevenly distributed in space. For example, drifters
mixed by eddies. Ocean trajectories, which are called launched in a cluster in a region of zero mean velo-
a Lagrangian description of the flow, are useful both city tend to diffuse away from the cluster center by
for visualizing ocean motion and for determining its eddy motions, implying a divergent flow regime. The
velocity characteristics. The superposition of nu- dispersal of such a cluster gives important infor-
merous trajectories reveals that very different kinds mation about how tracers or pollutants might also
of circulation patterns occur in different regions. disperse in the ocean. Drifters tend to diffuse faster
Time variability is illustrated by the tangle of cross- toward regions of higher eddy energy, resulting in a
ing trajectories. Trajectories often show the compli- mean velocity toward the direction of higher energy.
cated relationship between currents and nearby This is because drifters located in a region of high
seafloor topography and coastlines. Drifters and eddy kinetic energy drift faster than those in a region
floats have been used to follow discrete eddies like of low kinetic energy. Errors concerning array biases
Gulf Stream rings and ‘meddies’ (Mediterranean need to be estimated and considered along with
water eddies) continuously for years. When a drifter gridded maps of velocity.
becomes trapped in the rotating swirl flow around an
eddy’s center, the path of the eddy and its swirl velo-
Surface Drifters
city can be inferred from the drifter trajectory. Thus
the movement of a single drifter represents the huge Surface drifter measurements of currents have been
mass of water being advected by the eddy. Trajec- made for as long as people have been going to sea.
tories of drifters launched in clusters have provided The earliest measurements were visual sightings of
important information about dispersion, eddy dif- natural and man-made floating objects within sight
fusivity, and stirring in the ocean. of land or from an anchored ship that served as a
When a sufficient number of drifters are in a region, reference. Starting at least 400 years ago, mariners
velocity measurements along trajectories can be reported using subsurface drogues of different shapes
grouped into variously sized geographical bins and and sizes tethered to surface floats to measure cur-
calculations made of velocity statistics such as mean rents (Figure 1). The drogues were designed to have a
velocity, seasonal variations, and eddy energy. Gridded large area of drag relative to the surface float so that

89
90 DRIFTERS AND FLOATS

compilations of historical ship drift measurements.

Pilot charts used by most mariners today are still
based on historical ship drifts. Problems with the
ship drift technique are the fairly large random errors
of each velocity measurement (B20 cm s  1) and the
suspected systematic downwind leeway or slip of a
ship through the water due to wind and wave forces.
New velocity maps based on satellite-tracked drifters
are providing a much more accurate and higher-
resolution replacement of ship drift maps.
Drifting derelict ships gave an early measurement
of ocean trajectories during the nineteenth century.
Wooden vessels that had been damaged in storms
were often abandoned at sea and left to drift for
months to years. Repeated sightings of individual
vessels reported in the US pilot charts provided
trajectories.

Other Drifters
Bottles with notes and other floating objects have
been a popular form of surface drifter over the years.
The vectors between launch and recovery on some
distant shore provided some interesting maps but
ones that were difficult to interpret. The technique
was improved and exploited in the North Atlantic by
Prince Albert I of Monaco during the late 1800s.
More recently, 61 000 Nike shoes and 29 000 plastic
Figure 1 Schematic of an early drifter and drogue from the
Challenger expedition (1872–76). Adapted from Niiler PP, Davis
toy animals were accidentally released from dam-
RE, and White HJ (1987) Water following characteristics of a aged containers lost overboard from ships in storms
mixed layer drifter. Deep-Sea Research 24: 1867–1881. in the North Pacific. The recovery of thousands of
these drifters along the west coast of North America
has given some interesting results about mean cur-
the drifter would be advected primarily with the
rents and dispersion.
water at the drogue depth and not be strongly biased
Bottom drifters are very slightly negatively buoy-
by wind, waves, and the vertical shear of near-
ant and drift along the seafloor until they come
surface currents. Over the years many kinds of dro-
ashore and are recovered. The vectors between
gues, tethers, and surface floats have been tried,
launch and recovery show long-term mean currents
including drogues in the form of crossed vanes, fishing
near the seafloor.
nets, parachutes, window shades, and cylinders.
Tracking
Ship Drifts
Early measurements of drifters were visual sightings
Probably the most successful historical drifter is a using telescopes, compasses, and sextants to measure
ship; the drift of ships underway as they crossed bearings and locations. Later during the 1950s, radio
oceans provided millions of ocean current measure- direction finding and radar were used to track
ments. A ship drift measurement is obtained by drifters over longer ranges and times from shore,
subtracting the velocity between two measured pos- ship, and airplane. Some drifter trajectories in the
ition fixes from the estimated dead reckoning velo- 1960s were obtained by Fritz Fuglister and Charlie
city of the ship through the water over the same time Parker in the Gulf Stream and its rings using radar.
interval. The difference in velocity is considered to be These early experiments did not obtain very
a measure of the surface current. This technique many detailed and long trajectories but did reveal
depends on good navigation, which became common interesting features of the circulation. It was clearly
by the end of the nineteenth century. Most of apparent that a remote, accurate, relatively in-
what we have learned about the large-scale patterns expensive, long-term tracking system was needed.
of ocean currents until very recently came from This was soon provided in the 1970s by satellite
 DRIFTERS AND FLOATS 91

tracking, which revolutionized the tracking of drift- long drogues remained attached. Over the years
ers in the ocean. many people tried various approaches to solve these
The first satellite tracking of drifters occurred in problems, but it was mainly due to the impetus of
1970 using the Interrogation Recording and Loca- two large experiments, Tropical Ocean and Global
tion System (IRLS) system flown on Nimbus 3 and 4 Atmosphere (TOGA) and World Ocean Circulation
satellites. This system measured the slant range and Experiment (WOCE), and with the persistent efforts
bearing of a radio transmitter on a drifter. The IRLS of Peter Niiler and colleagues that a good surface
drifters were very expensive, too expensive at drifter was finally developed, standardized, and
$50 000 to be used in large numbers (but cheap deployed in large numbers. The so-called WOCE
compared to the cost of the satellite). During drifters have good water-following characteristics
1972–73, several drifters were tracked with the and the slip of the drogue has been calibrated in
(Corporative Application Satellite) EOLE system, different conditions. The WOCE drifter works
which incorporated Doppler measurements of the fairly reliably and often survives longer than a year
drifter radio transmissions to determine position. In at sea. As of March 2007 there were around 1300
the mid-1970s, NASA developed the Random Access drifters being tracked in the oceans as part of the
Measurement System (RAMS), which used Doppler Global Drifter Program. Data assembly and quality
measurements and was flown on the Nimbus 6 sat- control is performed by the Drifter Data Assembly
ellite. The radio transmitters were relatively in- Center at the National Oceanic and Atmospheric
expensive at $1300, and tracking was provided free Administration (NOAA) in Miami, Florida. Recent
by NASA (as proof of concept), which enabled many analyses include mapping surface velocity over
oceanographers to begin satellite tracking of surface broad regions and the generation of maps of mean
drifters. The modern Doppler-based satellite tracking sea level pressure based on drifter measurements.
used today, the French Argos system flown on polar The WOCE drifter consists of a spherical sur-
orbiting satellites, is similar to the early RAMS but face float 35 cm in diameter, a 0.56-cm diameter
provides improved position performance. The mod- plastic-impregnated wire tether, with a 20-cm
ern drifter transmitter emits a 0.5-W signal at diameter subsurface float located at 275 cm below
402 MHz approximately every minute. Positions are the surface and a drogue in the shape of a 644-cm-
obtained by Service Argos about six times per day in long cloth cylinder 92 cm in diameter with circular
the equatorial region increasing to 15 times per day holes in its sides (Figure 2). The fiberglass surface
near 601 N. Position errors are around 300 m. The float contains a radio transmitter, batteries, antenna,
cost per day of satellite tracking is around $10, and sensors including a thermometer and a sub-
which becomes quite expensive for continuous year- mergence sensor that indicates if the drogue is at-
long trajectories. To reduce costs, some drifters have tached. Additional sensors can be added to measure
been programmed to transmit only one day out of conductivity, atmospheric pressure, light, sound, etc.
three or for one-third of each day. This causes gaps in The basic WOCE drifter costs around $2500, ready
the trajectories that need to be interpolated. for deployment.
Recently, drifters with Global Positioning System The WOCE drifter’s drogue is centered at a depth
(GPS) receivers have been deployed to obtain more of 15 m below the sea surface. The ratio of the drag
accurate (B10 m) and virtually continuous fixes. GPS area of the drogue to the drag area of tether and float
position and sensor data need to be transmitted to is around 41:1, which results in the drogue’s slip
shore via the Argos system or another satellite that through the water being less than 1 cm s  1 in winds
can relay data. Experiments are underway using new of 10 m s  1. The slip was measured to be pro-
satellite systems to relay information both ways – to portional to wind speed and inversely proportional
the drifters and to the shore – in order to increase to the drag area ratio. From this information, the slip
bandwidth, to decrease costs, and to modify can be estimated and subtracted from the drifter
sampling. velocity.

WOCE drifter The development of satellite track-

Drogue Depth
ing in the 1970s quickly revealed the weakness of
available drifters – most performed poorly and most Drogues have been placed at many different depths
did not survive long at sea. Many problems needed to suit particular experiments. The drogues of
to be overcome: fishbite, chafe, shockloading, Coastal Ocean Dynamics Experiment (CODE)
biofouling, corrosion, etc. Early drogues tended to drifters developed by Russ Davis were located in the
fall off fairly quickly and, since drogue sensors were upper meter of the water column to measure the
not used or did not work well, no one knew how surface velocity. WOCE drogues are placed at 15 m
92 DRIFTERS AND FLOATS

often to track subsurface coherent eddies such as

meddies.
∅ 35 cm 307 cm

Subsurface Floats
Many kinds of freely drifting subsurface floats are
∅ 20 cm being used to measure ocean currents, although most
floats are usually either autonomous or acoustic. An
autonomous float measures a series of subsurface
displacements and velocities between periodic surface
satellite fixes. An acoustic float measures continuous
subsurface trajectories and velocities using acoustic
1500 cm tracking. Acoustic floats provide high-resolution
ocean trajectories but require an acoustic tracking
array and the effort to calculate subsurface positions,
both of which add cost. Thousands of autonomous
and acoustic floats have been deployed to measure the
general circulation in the world ocean at various
∅ 46 cm depths but concentrated near 800 m. Historical and
WOCE era float data can be seen and obtained on the
WOCE float website along with references to a series
of detailed scientific papers, and newer float data on
the Argo website.

WOCE Autonomous Float

644 cm
The autonomous WOCE float was developed in the
1990s by Russ Davis and Doug Webb. The float
typically drifts submerged for a few weeks at a time
and periodically rises to the sea surface where it
transmits data and is positioned by the Argos satellite
Drag area ratio = 41.3 system. After around a day drifting on the surface,
92 cm the float resubmerges to its mission depth, typically
somewhere in the upper kilometer of the ocean, and
Figure 2 Schematic of WOCE surface drifter. Adapted from continues to drift for another few weeks. Around
Sybrandy AL and Niiler PP (1991) WOCE/TOGA Lagrangian
Drifter Construction Manual, SIO Reference 91/6, WOCE Report
100 round trips are possible over a lifetime up to
No. 63. La Jolla, CA: Scripps Institution of Oceanography. 6 years.
The float consists of an aluminum pressure hull
1 m in length and 0.17 m in diameter (Figure 3).
to measure a representative velocity in the Ekman A hydraulic pump moves oil between internal and
layer but below the fastest surface currents. Many external bladders, forcing changes of volume and
scientists have deployed drogues at around 100 m to buoyancy and enabling the float to ascend and de-
measure the geostrophic velocity below the Ekman scend. An antenna transmits to Argos and a damping
layer. An argument for the 100-m depth is that it is plate keeps the float from submerging while it is
better to place the drogue below the complicated floating in waves on the sea surface. Some floats are
velocity structures in the Ekman layer, Langmuir drogued to follow a pressure surface; others can be
circulations, and near-surface convergence regions. programmed with active ballasting to follow a par-
An argument against the 100-m depth is that the ticular temperature or density surface; more com-
drag of the long tether and surface float in the rela- plicated sampling schemes are possible.
tively fast Ekman layer could create excessive slip of Autonomous floats have been equipped with
the drogue and bias the drifter measurement of temperature and conductivity sensors to measure
geostrophic velocity. The controversy continues. The vertical profiles as the floats rise to the surface.
15-m depth is widely used today, but many earlier Electric potential sensors have been added to some
drifters had deeper drogues at around 100 m. Some floats by Tom Sanford and Doug Webb in order to
drogues have been placed as deep as 500–1000 m measure vertical profiles of horizontal velocity. These
 DRIFTERS AND FLOATS 93

Antenna port

Evacuation port

Damping disk
Argos antenna

Internal reservior
107 cm
Argos transmitter

Controller and
circuit boards
70 cm

Microprocessor battery packs

Pump battery packs

Motor

Filter

Hydraulic pump

Latching valve

Pressure case

External bladder
17 cm

Figure 3 Schematic of Autonomous Lagrangian Circulation Explorer (ALACE) float. For ascent, the hydraulic pump moves oil down
from an internal reservior to an external bladder. For descent, the latching valve is opened, allowing oil to flow back into the internal
reservior. The antenna shown at the right is mounted on the top hemispherical end cap. Adapted from Davis RE, Webb DC, Regier LA,
and Dufour J (1992) The Autonomous Lagrangian Circulation Explorer (ALACE). Journal of Atmospheric and Oceanic Technology 9:
264–285.

floats were recently used to measure the ocean re- The basic drift data from an autonomous float are
sponse of hurricanes. subsurface displacements or velocity vectors between
Starting in 2000, an array of profiling floats began surface satellite fixes or between extrapolated pos-
to be launched as part of an international program itions at the times of descent and ascent. Errors in
called Argo. Plans are to build up the float array position are estimated to be around 3 km. The surface
reaching 3000 profiling floats by 2007 and to replace drifts cause gaps in the series of subsurface displace-
them as they are lost. As of March 2007, there are ments, so the displacements cannot be connected into
around 2800 Argo floats operational. The floats a continuous subsurface trajectory. Subsurface dis-
profile temperature and salinity to a depth of 2000 m placements are typically measured over several
and measure velocity at the drift depth near 1000 m. weeks, which attenuates the higher-frequency mo-
Profiles of temperature and salinity are being used to tions of ocean eddies. The main benefit of these floats
map large areas of the ocean including velocity at the is that they can be used to map the low-frequency
drift depth and are being incorporated into predictive ocean circulation worldwide relatively inexpensively.
numerical models. The profiles are being combined The cost of a WOCE profiling autonomous float is
with earlier and sparser hydrographic profiles to around $16 000 (cheaper than a day of an ocean-
document oceanic climate changes. going ship).
94 DRIFTERS AND FLOATS

A recent development is the addition of small of drifting slowly in a generally northward direction
wings plus streamlining that transforms the float into as had been predicted, the floats drifted fast and er-
a simple autonomous glider as it ascends and des- ratically, providing convincing evidence of eddy
cends. These gliders are self-propelled through the motions that were much swifter than long-term
ocean with typical horizontal speeds of 30 cm s  1 mean circulation. Previously, the deep interior flow
while moving vertically. Movable internal ballast is was considered too sluggish to be measured with
used to bank a glider, forcing it to turn. Gliders can moored current meters. The discovery of mesoscale
be programmed to return to a specific location to variability or ocean eddies by Swallow and James
hold position, to execute surveys, and to transit Crease using floats radically changed the perception
ocean basins along lines. Doug Webb at Webb Re- of deep currents and spurred the further develop-
search Corporation is equipping some with thermal ments of both floats and current meters.
engines that extract energy from the ocean’s thermal Swallow floats had a short acoustic range and re-
stratification in temperate regions in order to con- quired a nearby ship to track them, which was dif-
tinuously power the glider. Phase changes of a fluid ficult and expensive. It was quickly realized that
are used to force buoyancy changes. Some gliders much longer trajectories were needed in order to
incorporate suitable navigation and measure vertical measure the ocean variability and the lower-
profiles of velocity. Recently, fleets of gliders have frequency circulation. Accomplishing this required a
been directed from shore to survey the evolving neutrally buoyant float capable of transmitting sig-
structure of coastal regions. nificantly more acoustic energy and operating un-
attended for long times at great pressures. Second,
access was required to military undersea listening
Acoustic Floats
stations, so that the acoustic signals could be rou-
In the mid-1950s, Henry Stommel and John Swallow tinely recorded and used to track the floats. In the
pioneered the concept and development of freely late 1960s, Tom Rossby and Doug Webb successfully
drifting neutrally buoyant acoustic floats to measure developed and tested the sound fixing and ranging
subsurface currents. The method uses acoustics be- (SOFAR) float, named after the SOFAR acoustic
cause the ocean is relatively transparent to sound channel.
propagation. The deep sound channel centered at a SOFAR floats transmit a low-frequency (250 Hz)
depth around 1000 m enables long-range acoustic signal that sounds in air somewhat like a faint boat
propagation. The compressibility of hollow alu- whistle. The acoustic signal spreads horizontally
minum and glass pressure vessels is less than that of through the SOFAR channel and can be heard at
water, so that a float can be ballasted to equilibrate ranges of roughly 2500 km. The acoustic arrival
and remain near a particular depth or density. For times measured at fixed listening stations are used to
example, if the float is displaced too deep, it com- calculate distances to the float and to triangulate its
presses less than water and becomes relatively position. The first success with a SOFAR float drift of
buoyant, rising back to its equilibrium level, which is four months in 1969 led to further developments and
consequently stable. Once neutrally buoyant, a float the first large deployment of floats in 1973 as part of
can drift with the currents at that depth for long Mid-Ocean Dynamics Experiment (MODE). Very
times. interesting scientific results using the float data led to
In 1955, Swallow built the first successful floats many more experiments and wider use of floats.
(since called Swallow floats) and tracked them for a Later improvements included swept-frequency co-
few days by means of hydrophones lowered from a herent signaling in 1974, active depth control in
ship. A moored buoy provided a reference point for 1976, higher power for longer range in 1980, and
the ship positioning. The first pressure hulls were microprocessors and better electronics in 1983.
made out of surplus aluminum scaffolding tubes; Moored autonomous undersea listening stations
Swallow thinned the walls with caustic soda to ad- were developed in 1980, freeing experiments from
just compressibility and buoyancy. Although several military stations and enabling floats to be tracked in
floats failed, two worked successfully, which led to the Gulf Stream and other regions for the first time.
further experiments. In 1957, Swallow tracked deep SOFAR floats are large (B5-m long) and heavy
floats as they drifted rapidly southward offshore of (B430 kg), which makes them difficult to use in large
South Carolina, providing the first convincing proof numbers. In 1984, Rossby developed the RAFOS
of a swift, narrow southward flowing deep western (SOFAR spelled backward) float, a much smaller,
boundary current previously predicted by Stommel. cheaper float that listens to moored sound sources
A second experiment in 1959 tracked deep Swallow and at the end of its mission surfaces and reports
floats in the Sargasso Sea west of Bermuda. Instead back data via satellite. This float made it much easier
 DRIFTERS AND FLOATS 95

and cheaper to conduct larger experiments; this style from an array of moored undersea sound sources.
of float was improved and tracked in large numbers At the end of the mission, a few months to a few years
in the North and South Atlantic as part of WOCE. in length, the float drops an external weight, rises to
Various float groups have collaborated in tracking the sea surface, and transmits recorded times of ar-
floats at different depths and in maintaining moored rival, temperatures, and pressures to the Argos sys-
tracking arrays. tem. The float remains drifting on the surface for
roughly a month before all the data are received and
WOCE RAFOS Float relayed ashore by satellite. A typical float costs
$4000–5000 and is considered expendable because it
The modern acoustic RAFOS float consists of a glass is difficult and expensive to retrieve. The times of
hull 8.5 cm in diameter and 150–200-cm long, en- arrival are used with the known transmit times of
closing an electronic package, Argos beacon, and
sources and the estimated speed of sound to trian-
temperature and pressure sensors (Figure 4). An gulate the float’s position.
acoustic transducer and external drop weight are at- A drifting RAFOS float closely follows a pressure
tached to an aluminum end cap on the bottom. surface. A compressee consisting of a spring and
RAFOS floats are capable of operating at depths from piston in a cylinder is sometimes suspended below a
just below the sea surface to around 4000 m. Usually RAFOS float, so that it matches the compressibility
several times per day they listen and record the times of seawater. If the compressibilities are the same, the
of arrival of 80 s 250-Hz acoustic signals transmitted float will remain on or close to a constant density
surface and more closely follow water parcels. Some
8.5 floats have active ballasting and can track a column
cm of water by cycling between two density surfaces.
To ballast a RAFOS float, it is weighed in air and
Copper
water, which gives its volume. Its compressibility is
foil
antenna measured by weighing the float at different pressures
in a water-filled tank. The amount of weight to be
Glass added to make the float neutrally buoyant at the
pressure target depth (or density) is calculated using the
housing compressibility and thermal expansion of the float
and the temperature and density of the water in the
Displacement tank and at the target depth. Floats usually equili-
measuring brate within 50 m of their target depths or density.
scale
Some floats combine acoustic tracking with the
190_200
cm Radio active buoyancy of the autonomous float, so that the
transmitter float can periodically surface and relay data to shore
at intervals of a few months. This avoids the long
wait for multiyear RAFOS floats to surface and
avoids the loss of all data should a float fail during its
Electronics mission. French MARVOR floats developed by
Michel Ollitrault report data back every 3 months
and typically survive for 5 years.
Battery
Drogues have been added to neutrally buoyant
pack floats by Eric d’Asaro to enable them to better
Pressure measure three-dimensional trajectories. Vertical velo-
transducer cities from these floats are especially interesting in
Acoustic
the upper ocean and in the deep convective regions
hydrophone like the Labrador Sea in winter. Another technique
used to measure vertical water velocity is the addi-
Ballast tion of tilted vanes attached to the outside of a float.
weight
Water moving vertically past the vanes forces the
Compressee float to spin and this is measured and recorded.
(isopycnal model)
At least two moored sound sources are required to
Figure 4 Schematic of RAFOS acoustic float. Adapted from position a RAFOS float. Often three or more are used
Rossby HT, Dorson D, and Fontaine J (1986) The RAFOS system. to improve accuracy. The sources transmit an 80-s
Journal of Atmospheric and Oceanic Technology 3: 672–679. swept-frequency 250-Hz signal a few times per day
96 DRIFTERS AND FLOATS

for up to 5 years. Sources are similar to the old Griffa A, Kirwan AD, Mariano AJ, Rossby T, and
SOFAR floats and cost around $33 000. Mooring Ozgokmen TM (eds.) (2007) Lagrangian Analysis and
costs of wire rope, flotation, acoustic release, and Prediction of Coastal and Ocean Dynamics.
other recovery aids can double this figure. Recently, Cambridge, MA: Cambridge University Press.
louder, more efficient, and more expensive sound Niiler PP, Davis RE, and White HJ (1987) Water following
characteristics of a mixed layer drifter. Deep-Sea
sources have increased tracking ranges up to 4000 km.
Research 24: 1867--1881.
Errors of acoustic positioning are difficult to esti- Pazan SE and Niiler P (2004) New global drifter data set
mate and vary depending on the size and shape of the available. EOS, Transactions of the American Geo-
tracking array, the accuracy of float and source physical Union 85(2): 17.
clocks, how well the speed of sound is known, etc. Richardson PL (1997) Drifting in the wind: Leeway error
Estimates of absolute position errors range from a in shipdrift data. Deep-Sea Research I 44: 1877--1903.
few kilometers up to 10 km (or more). Fix-to-fix Rossby HT, Dorson D, and Fontaine J (1986) The RAFOS
relative errors are usually less than this because some system. Journal of Atmospheric and Oceanic Tech-
errors cancel and others such as clock errors vary nology 3: 672--679.
slowly in time. Corrections are made for the Doppler Sanford TB, Price JF, Webb DC, and Girton JB (2007)
shift caused by a float’s movement toward or away Highly resolved observations and simulations of the
ocean response to a hurricane. Geophysical Research
from a source. The typical correction amounts to
Letters 34: L13604.
around 1.3 km for a speed of 10 cm s  1. Tides and Siedler G, Church J, and Gould J (eds.) (2001) Ocean
inertial oscillations add high-frequency noise to Circulation and Climate, Observing and Modelling the
positions and velocities, but since a float integrates Global Ocean. London: Harcourt.
these motions, it provides an accurate measure of Swallow JC (1955) A neutrally-buoyant float for mea-
lower-frequency motions. suring deep currents. Deep-Sea Research 3: 74--81.
Sybrandy AL and Niiler PP (1991) WOCE/TOGA
Further Reading Lagrangian Drifter Construction Manual, SIO Ref-
erence 91/6, WOCE Report No. 63. La Jolla, CA:
Burns LG (2007) Tracking Trash: Flotsam, Jetsam, and the Scripps Institution of Oceanography.
Science of Ocean Motion, 56pp. Boston, MA:
Houghton Mifflin. Relevant Websites
D’Asaro EA, Farmer DM, Osse JT, and Dairiki GT (2000)
A Lagrangian float. Journal of Atmospheric and http://www.meds-sdmm.dfo-mpo.gc.ca
Oceanic Technology 13: 1230--1246. – Archived Drifter Data, Integrated Science Data
Davis RE (2005) Intermediate-depth circulation of the Management, Fisheries and Oceans Canada.
Indian and South Pacific Oceans measured by http://www.argo.ucsd.edu
autonomous floats. Journal of Physical Oceanography – Argo Floats, ARGO home page.
35: 683--707. http://www.argo.net
Davis RE, Sherman JT, and Dufour J (2001) Profiling – Argo Floats, ARGO.NET, The International Argo
ALACEs and other advances in autonomous subsurface Project Home Page.
floats. Journal of Atmospheric and Oceanic Technology http://www.aoml.noaa.gov
18: 982--993. – Global Drifter Program, Atlantic Oceanographic and
Davis RE, Webb DC, Regier LA, and Dufour J (1992) The Meteorological Laboratory, NOAA.
Autonomous Lagrangian Circulation Explorer (ALACE). http://wfdac.whoi.edu
Journal of Atmospheric and Oceanic Technology 9: – WOCE Float Data, WOCE Subsurface Float Data
264--285. Assembly Center.
Gould WJ (2005) From swallow floats to Argo – the
development of neutrally buoyant floats. Deep-Sea
Research 52: 529--543.
 MOORINGS
R. P. Trask and R. A. Weller, Woods Hole Early attempts at mooring work in the 1960s began
Oceanographic Institution, Woods Hole, MA, USA with surface moorings. Problems with the mooring
& 2009 Elsevier Ltd. All rights reserved. materials and the dynamic conditions encountered
at the ocean surface resulted in poor performance
by these early designs, and attention turned to de-
veloping subsurface moorings. The introduction of
wire rope as a material for fabricating mooring lines
Introduction and the advent of a remotely triggered mechanism
The need to measure ocean currents throughout the to release the mooring’s anchor were significant
water column for extended periods in order to better milestones that helped make the subsurface mooring a
understand ocean dynamics was a driving force that viable option. It has since proved to be a very suc-
led to the development of oceanographic moorings. cessful oceanographic tool. Interest in the upper ocean
Today’s moorings are used as ‘platforms’ from which and the air–sea interface prompted a reexamination
a variety of measurements can be made. These in- of the surface mooring design. The evolution of
clude not only the speed and direction of currents, the subsurface mooring as a standard platform for
but also other physical parameters, such as con- oceanographic observations and the more recent de-
ductivity (salinity), temperature, and sea state, as velopment of reliable surface moorings are summar-
well as surface meteorology, bio-optical parameters, ized here.
sedimentation rates, and chemical properties.
Moorings typically have three basic components:
Subsurface Mooring Evolution
an anchor, some type of chain or line to which in-
strumentation can be attached, and flotation devices Early moorings consisted of a surface float, surplus
that keep the line and instrumentation from falling to railroad car wheels for an anchor, and lightweight
the seafloor. Shackles and links are typically used to synthetic line, such as polypropylene or nylon, to
connect mooring components and to secure instru- connect the surface float to the anchor. Several
ments in line. The choice of hardware, line, and kilometers of line are required for a full-depth ocean
flotation for a particular application, as well as the mooring, and weight of the line itself, even in water,
size and design of the anchor, depends on the type is not negligible. Instrumentation was connected to
of mooring and the environment in which it is the synthetic line along its length. The anchor was
deployed. connected to the mooring line by means of a corro-
Most moorings fall into two broad categories – sible weak link. The initial method of recovering the
surface and subsurface. The main difference between moorings was to connect to the surface float and pull
the two is that the surface mooring has a buoy it with the hope that the tension would break the
floating on the ocean surface, whereas the subsurface weak link leaving the anchor behind. Unfortunately,
mooring does not. Although the two mooring types at the time of recovery, the mooring line was often
have similar components, the capabilities of the two weaker than the weak link and would break,
are very different. With a surface buoy, it is possible allowing the line and instrumentation below the
to measure surface meteorology, telemeter data, and break to fall to the seafloor.
make very near-surface measurements in the upper Studies showed that the synthetic ropes were being
ocean. The surface mooring, however, is exposed to damaged by fish. Analysis of many failed lines re-
ocean storms with high wind and wave conditions vealed tooth fragments and bite patterns that were
and therefore must be constructed to withstand the used to identify the type of fish responsible for the
forces associated with those environmental con- damage. Statistics concerning the number of fish-
ditions. In addition, the wave action may transmit bites, their depths, and their locations were collected,
some unwanted motion to subsurface instruments if and it was found that the majority of fishbites oc-
care is not taken. The subsurface mooring, on the curred in the upper 1500–2000 m of the water col-
other hand, is away from the surface forcing and can umn. Prevention of mooring failure due to fish attack
be fabricated from smaller, lighter components, required lines that could resist fishbite.
which are less expensive and easier to handle. Ropes made of high-strength carbon steel wires
However, it is difficult to make near-surface meas- were an obvious candidate. Wire rope would not
urements from a subsurface mooring. only provide protection from fish attack, but also

97
98 MOORINGS

would have minimal stretch, unlike the synthetic against corrosion, some wire ropes have a plastic
ropes, and would provide high strength with rela- jacket extruded over the wire. Types of plastics used
tively low drag. Many types of wire rope con- for jacketing materials include polyvinyl chloride,
struction and sizes were tested, in addition to polypropylene, and high-density polyethylene.
methods for terminating the wire rope; terminations In the early years, mooring recoveries that were
are the fittings attached to the ends of wire sections. initiated by pulling on deteriorated mooring lines
In constructing a mooring whose components can be often resulted in line breakage and instrument loss.
shipped separately and handled safely on the deck of A preferable approach was to detach the mooring
a ship at sea, the practice is to cut the wire into from its anchor prior to hauling on the mooring line.
sections of specific lengths (shots) that allow con- This would reduce the load on the mooring line since
nection to other wire shots or to instrumentation in the line would never ‘feel’ the weight of the anchor
series (end to end). A desirable termination is one nor the tension required to pull the anchor out of the
that is as strong as the wire rope itself. If the tech- bottom sediments. This approach became possible
nique used to terminate a rope imposes stress con- with the development of an acoustically commanded
centrations, which significantly reduce the strength anchor release. The acoustic release is deployed in-
of the wire rope, then the whole system is weakened. line on the mooring and is typically positioned below
Methods of terminating wire include the formation all instrumentation and close to the anchor. To ac-
of eyes into which shackles can be attached either tivate the release mechanism, a coded acoustic signal
from swaged fittings or from zinc- or resin-poured is sent from the recovery vessel. The acoustic release
sockets. Swaged terminations utilize a fitting that is detects the signal and disconnects from the anchor.
slid onto the end of the wire and pressed or swaged When mooring work was in its infancy, the surface
onto the wire with a hydraulic press. In the case of a buoy was a vital, visible link to the mooring below.
poured-socket termination, the wire is inserted into Without it, the exact location of the mooring was
the socket and the individual wires are splayed out- unknown. The introduction of acoustic releases not
ward or ‘broomed out’. Once the wires are properly only provided a way to disconnect the line and in-
cleaned and positioned, a filler material (molten zinc struments from the anchor, but also provided a way
or uncured epoxy resin) is poured into the socket and to locate the exact position of the mooring by
allowed to harden. A strain relief boot is often used acoustic direction finding. This eliminated the need
in conjunction with a swaged fitting termination, as for a surface float, which at that time, before the
well as with the poured sockets. The boots are often development of meteorological sensors for buoys,
an injection-molded urethane material designed to was used solely for recovery purposes. Instead of
extend from the fitting out over a short section of the having a mooring that stretched all the way from the
wire to minimize the bending fatigue that can occur ocean bottom to the surface, the mooring was
between the flexible wire and rigid fitting. shortened so that the top of the mooring was pos-
At present, galvanized 3  19 wire rope is widely itioned below the surface of the water. Sufficient
used for oceanographic applications. The desig- buoyancy was placed at the top of the mooring to
nation ‘3  19’ denotes three strands or groups, each keep all of the mooring components as vertical
with 19 individual wires: the 19 wires are twisted as possible throughout the water column. With this
together to form a strand. Three strands are then design, which became known as the subsurface
wound together to form the rope. The rotation mooring, the mooring would ascend to the surface
characteristics of wire rope are critically important in once it was acoustically released from its anchor. At
certain oceanographic applications. If the rope has the surface, it could be pulled out of the water by the
the tendency to spin or rotate excessively when waiting recovery vessel. A great advantage of the
placed under tension, there is a tendency for that subsurface mooring is having the hardware below
wire to develop loops when the tension is reduced the ocean surface, which is the most dynamic part of
quickly. If the load is quickly applied again to the the water column. As a result, there is a considerable
line, the loops are pulled tight into kinks, which reduction in component fatigue due to surface-wave
can severely weaken the wire rope. Wire rope with action. In addition, the mooring is no longer visible
minimal rotation characteristics is preferred for to surface vessels and is less vulnerable to vandalism.
mooring applications, particularly surface mooring The buoyancy used on subsurface moorings is
work. Wire ropes are available with varying degrees usually in the shape of a sphere, because of its low
of rotation resistance. Swivels are sometimes placed drag coefficient. Other shapes have been used de-
in series with the wire to minimize the chances of pending on the specific application. Various materials
kink formation. In addition to using galvanized wires are used, including steel spheres, glass spheres
when fabricating the rope to provide protection with protective plastic covers, ceramic spheres, and
 MOORINGS 99

syntactic foam spheres. Syntactic foams consist of

small pressure-resistant glass microspheres (2–300 mm
in diameter), as well as larger glass fiber-reinforced
spheres (0.15–10 cm in diameter) embedded in a
thermosetting plastic binder. An advantage of the
syntactic foam is that it can be molded to form cus- 1.52-m syntactic sphere with radio,
95 m
tom shapes. Unlike the steel spheres, whose use is light, satellite beacon

depth-limited (maximum working depth is approxi- 3 m 1.27 cm chain

100 m
mately 1000 m), the syntactic foam can be engineered Vector averaging current meter (VACM)
to withstand full ocean-depth pressures. In addition, it 292 m 0.48 cm wire
can be designed to provide the same buoyancy as a Glass float transponder
(3) Glass flotation balls
string of glass balls (commonly 43 cm in diameter) in plastic hardhats
with considerably less drag. This makes syntactic
401 m VACM
foam spheres attractive for use in high-current re-
gimes, since the drag on the mooring will be less; 393 m 0.48 cm wire
consequently, less buoyancy will be needed to keep (3) Glass flotation balls
the mooring near-vertical. in plastic hardhats
Subsurface moorings with a single element of 800 m VACM
buoyancy at the top are still at some risk. Should the
591 m 0.48 cm wire
buoyant element be lost or damaged, the mooring
would fall to the bottom, leaving no secondary (3) Glass flotation balls
means of bringing it back to the surface when the in plastic hardhats
acoustic release mechanism is activated. To provide a 1399 m
VACM
higher degree of reliability, buoyancy is often pro-
591 m 0.48 cm wire
vided in the form of glass balls attached along the
length of the mooring. In addition, buoyancy is often (3) Glass flotation balls
added to the bottom of the mooring, just above the in plastic hardhats

acoustic release, to provide what is sometimes re- 2000 m

VACM
ferred to as the ‘backup recovery’. With this design 491 m 0.48 cm wire
feature (Figure 1), should a mooring component fail
(4) Glass flotation balls
and the upper part of the mooring be lost, no matter in plastic hardhats
where the failure occurs, there should be sufficient 2500 m
buoyancy below that point to bring the remaining VACM

section of the mooring back to the surface. Instru- 281 m 0.48 cm wire
mentation that would otherwise have been lost if (14) Glass flotation balls
deployed with a single buoyant element is recover- in plastic hardhats
able with this configuration. Equally important, the 2799 m
VACM
recovery provides the opportunity to identify the 2 m 0.95 cm chain
failed component and correct the problem. As a re- 2804 m Acoustic release
5 m 0.95 cm chain
covery aid, pressure-activated submersible satellite 156 m 0.64 cm wire
20 m 1.9 cm nylon
transmitters are frequently installed on the upper 5 m 0.64 cm wire
buoyancy sphere. In the event of a mooring com- 5 m 1.27 cm chain

ponent failure that causes the top of the mooring to Anchor 3000 m deep

surface, it can be tracked via satellite. This allows for

possible recovery of whatever instrumentation hangs Figure 1 A typical subsurface mooring design. Design by S.
Worrilow.
below.
The number of measurements made from a sub-
surface mooring depends on several factors including instruments possibly above and below the failed unit
the load the mooring is designed to support and the that continue to collect data. Such a design requires
resources available for instrumenting the mooring. numerous lengths of terminated wire, which must be
Depending on the resources, a mooring may have deployed in a particular order so the individual in-
on the order of 10 instruments located at discrete struments end up at the desired depths.
depths. If one or more of the instruments mal- An alternative approach is to have a single in-
function during their deployment, the data from that strument that uses the mooring wire as a guide along
particular depth are missing, but there may be other which an instrument moves up and down profiling
100 MOORINGS

the water column on a predetermined schedule. In is a chance that the mooring may not collect any data.
this configuration (Figure 2), the bulk of the mooring For that reason, it is not uncommon for a mooring to
consists of a single continuous length of wire, elim- have a combination of instrument types, which in-
inating the need for multiple lengths used between clude profiling as well as several discrete instruments.
discrete instruments. An advantage of such a system Duplicate measurements by different instruments
is that the data are collected from the entire sampling can improve the chances of collecting a full data set.
area and not from specific depths, which, due to their Since redundant moorings are seldom an option, a
deployed location, may not be in an optimal position mooring failure can have a catastrophic impact on
for observing an interesting phenomenon. One the total data return. With both instruments and
profiling instrument has the potential to replace mooring components, attention to detail is critical.
many instruments deployed along the mooring. If, The care taken in preparing and testing instru-
however, the mooring only carries one profiling in- mentation, in the selection of quality mooring com-
strument and that instrument malfunctions, there ponents, and in the fabrication techniques utilized is
often a deciding factor. Other factors that can impact
a mooring’s success include the quality of the infor-
mation that went into the design of the mooring, that
is, how similar were the predicted environmental
conditions to the actual conditions encountered? Is
the design unique or is it a variation of a design that
has historically worked well? Attention to detail
during the actual deployment of the mooring, as well
as uncontrollable outside influences such as fishing
activities in the area, can also be contributing factors.
Despite all the variables, it is possible to routinely
Subsurface buoy achieve success rates that are greater than 90%.
Top stop

Wire rope Advances in Surface Mooring

Technology
Growing interest in understanding interactions be-
tween the ocean and the atmosphere has rekindled
Profiler
interest in using surface moorings. The surface
mooring is a unique structure. It extends from above
the surface to the ocean bottom, providing a platform
from which both meteorological and oceanographic
measurements can be made in waters that range from
shallow to 5 km in depth. Surface-mooring designs
must consider the effects of surface waves, ocean
currents, biofouling, and other factors that can vary
with the time of year, location, and regional climate
Bottom stop
and weather patterns. The success of a surface-
mooring deployment often depends on the abilities
Backup glass ball flotation both to accurately estimate the range of conditions
that the mooring may encounter while deployed and
to design a structure that will survive those con-
Acoustic releases
ditions. The primary goal of any mooring deployment
is to keep the mooring on location and making ac-
curate measurements. Adverse environmental con-
Anchor
ditions not only influence the longevity itself but also
impact the instruments that the mooring supports. It
is often very difficult to keep the instruments working
Figure 2 A subsurface mooring schematic with a profiling
instrument that runs up and down the mooring line on a under such conditions for long periods.
predetermined schedule. Illustration by J. Doucette/WHOI Surface moorings are used to support submerged
Graphics. oceanographic instrumentation from very close to
 MOORINGS 101

the surface (sometimes floating at the surface) to near Component sizes are usually increased to compen-
the bottom, which is typically 5 km in depth. Meas- sate for the larger forces and the increased wear.
urements of physical properties, such as temperature, Materials include chain, plastic-jacketed wire rope,
velocity, and conductivity (salinity), as well as and synthetic line. Chain is used directly beneath the
of biological parameters, such as photosynthetically buoy for strength, ease of handling, and, because of
available radiation (PAR), beam transmission, its additional mass, stabilization of the buoy during
chlorophyll fluorescence, and dissolved oxygen, are its deployment. If the water is sufficiently deep and
routinely made from surface moorings. The sur- the design permits, the wire rope is usually extended
face buoy also provides a platform from which to a depth of at least 1500 m and often as deep as
meteorological measurements can be made and a 2000 m for fishbite protection.
structure from which both surface- and subsurface- The surface mooring needs some form of built-in
collected data can be telemetered via satellite. Me- ‘compliance’ (ability to stretch) to compensate for
teorological sensors typically deployed on a surface large vertical excursions that the buoy may experi-
buoy measure wind speed, wind direction, air tem- ence during the change of tides and with passing
perature, relative humidity, barometric pressure, waves and swell. The compliance also compensates
precipitation, and long-wave and short-wave radi- for the buoy being displaced laterally on the surface
ation. The meteorological data are stored in memory by the drag forces associated with ocean currents and
and telemetered via satellite to a receiving station prevents the buoy from being pulled under when
ashore. The telemetered data often play an important such forces are applied. In deep-water applications,
part in real-time analysis and reaction to conditions compliance is provided through the use of synthetic
on site. The data can also be passed to weather materials, such as nylon. The synthetic line acts like a
centers for forecasting purposes. large rubber band that stretches as necessary to
There are a number of different types of surface maintain the connections between the surface-
buoys. Some shapes have been in use since the early following buoy and the anchor on the bottom.
days of mooring work, and others are relatively new. A challenge in the design process, particularly in
Buoy shapes include the toroid or ‘donut’, the discus, shallow water, is to achieve an appropriate mix of
and the hemispherical hull. The toroid hull in various compliant materials and fishbite-resistant materials,
configurations is widely used throughout the scien- which tend to be unstretchable. The ‘scope’ of the
tific community. Where a significant amount of in- mooring – the ratio of the total unstretched length of
strumentation must be supported, a discus-shaped the mooring components to the water depth – can be
buoy with as much as 6800 kg of buoyancy may be one of the sensitive design factors. A mooring with a
used for both deep- and shallow-water applications. scope of less than 1.0 relies on the stretch of the
Some buoys are designed with modular buoyancy nylon for the anchor to reach the bottom. Such a taut
elements that can be added to maximize the available mooring remains fairly vertical with a relatively
buoyancy. The discus buoy design is widely used by small watch circle (the diameter of the area on the
the US National Data Buoy Center in Mississippi in ocean surface where the buoy can move about while
coastal waters, at the Great Lakes stations, and for still anchored to the ocean bottom), but it carries a
directional wave measurements. The 3-m discus- penalty: such a vertical mooring is under consider-
shaped hull was also adopted by the Atmospheric able tension, or ‘preloaded’, at the time of deploy-
Environment Service in Canada for its coastal buoys. ment. Currents and waves impose additional loads
Smaller discus-shaped hulls are used for shallow- beyond the initial preloaded condition. Moorings
water applications. with scopes between 1.0 and c. 1.1 are generally
Buoy hulls are made of aluminum, steel, fiberglass referred to as ‘semi-taut’ designs. The mooring
over foam, and various closed-cell foams. Several shown in Figure 3 is typical of a semi-taut design.
closed-cell foams are extremely resistant to wear and Early surface moorings were designed using only a
have low maintenance. Ionomer foam and poly- static analysis program, which used steady-state
ethylene foam are common materials for buoy and current profiles as input to predict mooring per-
fender applications. Depending on the material, formance. However, experience has shown that it is
various outer skin treatments are used to increase the necessary to consider the combined effects of strong
hull’s resilience to wear. These include the appli- currents and surface waves. An investigation of the
cation of heat and pressure, as well as bonding a dynamic effects of surface forcing on the perform-
different material, such as urethane, to the exterior ance of surface moorings found that semi-taut
of the hull. moorings could have a resonant response to forcing
The mooring materials used on surface moorings in the range of surface wave periods, causing
resemble those used on the subsurface moorings. high dynamic loads. These high tensions limit the
102 MOORINGS

3-m discus buoy with meteorological

Discus buoy with meteorological sensors
sensors, tension recorder, and satellite transmitter

Bridle with temperature sensor, SEACAT, tensiometer

,
3.5 m dissolved oxygen and backup Argos transmitter
3 m 1.9 cm chain 0.4 m 1.9 cm Syst. 3 chain
4.5 m Temperature logger
3.9 m 1.27 cm chain 0.4 m 1.9 cm Syst. 3 chain
10 m Vector measuring current meter (VMCM) 5m Vector measuring current meter (VMCM)
and temperature pod (Tpod) 2 m 1.9 cm Syst. 3 chain
5.5 m 0.95 cm wire rope
Multivariable moored system (MVMS):
10 m current, temperature, dissolved oxygen,
20 m Vector averaging current meter (VACM) light, and other sensors
2.5 m 1.9 Syst. 3 chain
9.5 m 0.95 cm wire rope 15 m VMCM

30 m VMCM 20 m TPod 7.25 m 1.27 cm wire

5.5 m 0.95 cm wire rope 25 m VMCM

30 m TPod 7.25 m 1.27 cm wire

40 m VACM
35 m MVMS
39.5 m 0.95 cm wire rope
2.5 m 1.9 cm Syst. 3 chain
80 m VMCM
40 m Acoustic instrument
with TPod
2.5 m 1.9 cm Syst. 3 chain
45 m VMCM
37.6 m 0.95 cm wire rope
50 m TPod 7.25 m 1.27 cm wire

55 m VMCM

120 m VMCM 60 m TPod 7.25 m 1.27 cm wire

37.4 m 0.95 cm wire rope 65 m MVMS

VMCM 72.5 m TPod 12.1 m 1.27 cm wire

160 m
80 m MVMS
537.2 m 0.95 cm wire rope
90 m TPod 17.5 m 1.11 cm wire
VMCM 100 m SEACAT conductivity/temperature sensor
700 m 125 m TPod 49 m 1.1 cm wire
150 m SEACAT
175 m TPod 49 m 1.1 cm wire
1300 m 0.79 cm wire rope 200 m SEACAT
2000 m 1000 m 2.06 cm nylon 225 m TPod 49 m 1.1 cm wire
250 m SEACAT
1500 m 1.9 cm nylon 300 m TPod 100 m 1.1 cm wire

(2) Glass flotation balls 1400 m 0.95 cm wire

in plastic hardhats 100 m 0.95 cm wire
one piece
Special wire/nylon 500 m 2.22 cm nylon
termination
732 m 2.22 cm nylon
1500 m 1.9 cm nylon 500 m 2.22 cm nylon
100 m 2.54 cm nylon one piece
500 m 2.86 cm polypro
600 m 2.86 cm polypro
1.27 cm trawler chain

(57) Glass flotation balls

in plastic hardhats (82) Glass flotation balls
in plastic hardhats

2 m 1.27 cm chain
Acoustic release
5 m 1.27 cm chain 5 m 1.27 cm trawler chain
20 m 2.54 cm nylon Acoustic release 5 m 1.27 cm trawler chain
20 m 2.54 cm Samson nystron
5 m 1.27 cm chain 5 m 1.27 cm trawler chain
Anchor 5400-m deep Anchor 4032-m deep

Figure 3 A semi-taut surface mooring design. Design by Figure 4 An inverse catenary mooring design. Design by
P. Clay. G. Tupper.

instrument-carrying capacity of the mooring and can offers larger scope (typically 1.2) for high-current
lead to failure of mooring components. periods, yet performs well in lesser currents. In low
An alternative design fashioned after the US currents, the positively buoyant synthetic line keeps
National Data Buoy Center ‘inverse catenary’ mooring the slightly negatively buoyant nylon from tangling
has evolved in response to difficulties encountered with the rest of the mooring below it. Thus, the inverse
using taut surface mooring designs. With wire rope in catenary design can tolerate a wider range of en-
the upper part of the mooring and with nylon line vironmental conditions. The inverse catenary design
spliced to a buoyant synthetic line such as poly- lowers the static mooring tension, as shown in
propylene below, the inverse catenary design (Figure 4) Table 1. The dynamic tension contribution to the total
 MOORINGS 103

Table 1 A comparison of semi-taut surface mooring and an 0

inverse catenary design (subjected to the same ocean current
Semi-taut design
forcing)
Tension at buoy : 1565 kg
Tension at anchor : 1777 kg
Semi- Inverse Difference
taut catenary 500

Mooring scope 1.109 1.285

Depth (m)
Tension at the buoy 2065 1602 463
(kg)
1000
Anchor tension (kg) 2292 1783 509
Horizontal excursion 1208 1735 527
Inverse catenary design
(m)
Tension at buoy : 1241 kg
Tension at anchor : 1400 kg
1500

tension, however, is unchanged, and care must still be

taken in the design process to prevent the mooring
from having a resonant response to forcing in the 2000
range of surface wave periods. 0 500 1000 1500 2000
In some regions of the world’s oceans, the dynamic Horizontal excursion (m)
loading due to high wind and sea state conditions Figure 5 A comparison of the shape of a semi-taut design with
may be so severe that ultimate strength consider- that of an inverse catenary design when subjected to the same
ations are superseded by the fatigue properties of the ocean current forcing. Note the differences in buoy and anchor
standard hardware components. In these cases, in tensions as well as the horizontal excursions at the surface.
addition to appropriate mooring design, attention
must be paid to the choice and preparation of
mooring hardware. Cyclic fatigue tests revealed that, deep-current measurements because the mooring line
in certain applications, mooring hardware that had at these depths is sometimes inclined more than 151
been used reliably in the past lost a significant part of from vertical. This is a problem for two reasons: first,
its service life owing to fatigue and either failed or some instruments fitted with compasses do not work
showed evidence of cracks. Where possible, different well if the compass is inclined more than 151; and
hardware components that are less susceptible to second, some velocity sensors require the instrument
fatigue failure in the range of expected tensions are to be nearly vertical. An inverse catenary mooring,
now substituted. with its greater scope, has inclination problems at
In situations where there is no replacement hard- shallower depths as compared to the semi-taut de-
ware available, the fatigue performance is improved sign. Figure 5 compares the mooring shape of a semi-
by shot peening. Shot peening is a process whereby a taut design with that of an inverse catenary mooring
component is blasted with small spherical media, subjected to the same environment conditions.
called shot, in a manner similar to the process of In addition to the inclination problem, there is also
sand blasting. The medium used in shot peening is a depth-variability problem. Compliant members on
more rounded rather than angular and sharp, as in a surface mooring are usually synthetics, which must
sandblasting. Each piece of shot acts like a small ball- be placed below the fishbite zone (nominally 2000-m
peen hammer and tends to dimple the surface that it depth). The deep instruments are, therefore, in line in
strikes. At each dimple site, the surface structure of the synthetics; their depth can vary by several hun-
the material is placed in tension. Immediately below dred meters depending on the stretch of the material.
the surface of each dimple, the material is highly A pressure sensor on the instrument can be used to
stressed in compression so as to counteract the tensile record the instrument depth, but if a particular depth
stress at the surface. A shot-peened part with its is desired, it is not possible with the conventional
many overlapping dimples, therefore, has a surface design. Hence, the trade-off for being able to with-
layer with residual compressive stress. Cracks do not stand a wider range of environmental conditions is a
tend to initiate or propagate in a compressive stress reduction in the depth range for making certain kinds
zone. Since cracks usually start at the surface, a of measurements.
shot-peened component will take longer to develop a A partial solution to the problem of deep
crack, thereby increasing the fatigue life of the part. measurements on a surface mooring is illustrated in
With both the semi-taut and the inverse catenary the mooring design shown in Figure 6, which com-
surface mooring designs, it is difficult to make bines features of both the subsurface- and inverse
104 MOORINGS

0
3-m discus buoy with meteorological
sensors and satellite telemetry
Combination inverse catenary
1000
and subsurface design
Bridle with T-pod, IMET and VAWR Buoy tension = 1438 kg
temperature sensors, and tensiometer 3500 m VMCM angle = 8°
6.7 m 1.9 cm chain 2000
10 m Vector measuring current meter (VMCM)

Depth (m)
17 m 0.95 cm wire
30 m VMCM 3000
17 m 0.95 cm wire
50 m VMCM
Inverse catenary design
4000 Buoy tension = 1471 kg
17 m 0.95 cm wire
3500 m VMCM angle = 25°
70 m VMCM
17 m 0.95 cm wire
5000
90 m VMCM
17 m 0.95 cm wire
110 m VMCM
6000
37.3 m 0.95 cm wire 0 1000 2000 3000 4000 5000 6000
150 m VMCM Horizontal excursion (m)
47.3 m 0.95 cm wire
200 m VMCM
Figure 7 A comparison of the shapes of two mooring designs
subjected to the same ocean current forcing, showing the
97 m 0.95 cm wire
differences in mooring inclination at 3500-m depth.
300 m VMCM
7.3 m 0.95 cm wire
310 m VMCM release near the anchor and another immediately
200 m 0.95 cm wire
237 m 0.79 cm wire above the 3500 m instrument. The compliance of
750 m VMCM
the mooring consists of 1500 m of nylon and poly-
500 m 0.79 cm wire
252 m 0.79 cm wire propylene inserted between the 3500 m instrument
1500 m VMCM 400 m 0.79 cm wire
100 m 0.79 cm wire one piece
and the base of the wire at 2000 m. The combination
50 m 2.06 cm nylon
Special wire/nylon termination
of nylon and polypropylene gives the mooring
450 m nylon one piece
500 m 2.54 cm polypro enough stretch (from the nylon) and built-in buoy-
1.27 cm trawler chain
(47) Glass flotation balls
ancy (from the polypropylene) to handle the range of
Tensiometer with safety chain expected current conditions. The polypropylene ac-
10 m 0.95 cm wire
tually performs a double duty in that during low
3500 m VMCM
current periods the buoyant polypropylene keeps the
excess nylon from tangling with the lower part of the
2078 m 0.95 cm wire mooring; when the currents increase, that buoyant
member becomes available in the form of extra
1.27 trawler chain
(52) Glass flotation balls
scope. The shape of the combination mooring design
1.27 cm trawler chain is compared with the shape of an inverse catenary
Acoustic release
5 m 1.27 cm trawler chain
20 m 2.54 cm nylon
design in Figure 7.
5 m 1.27 cm trawler chain Surface moorings can also be used as a com-
Anchor 5670-m deep
munications link between the ocean surface and
points along the mooring all the way to the seafloor.
Figure 6 A mooring design that combines the features of an With the appropriate mooring components, infor-
inverse catenary mooring with those of a semi-taut mooring in
mation can be passed in both directions. Data col-
order to improve the quality of deep (3500 m) current
measurements made from a surface mooring. Design by lected by instrumentation deployed on the mooring
G. Tupper. line or in close proximity to the mooring can be sent
to the buoy, where it is transmitted via satellite to a
receiving station. Two-way communications between
catenary-type moorings. The upper 2000 m of the a shore station and the buoy via satellite permit buoy
mooring is similar to any surface mooring, with the systems to be reprogrammed so as to modify in-
instrumentation at the appropriate depths and wire strument sampling schemes, as well as to repair
rope in between. The lower part of the mooring from system malfunctions. Being able to diagnose and re-
the bottom up to the 3500 m instrument is all wire pair a buoy data collection system without having to
with a cluster of glass ball flotation just above the dispatch a vessel to the site is of great value.
 MOORINGS 105

Telemetry of data from subsurface instruments on located on the mooring. Those signals are then
surface moorings is possible through various tech- transmitted up the mooring line to the surface buoy
niques. One approach is to utilize electromechanical and then to a receiving station via satellite. Data from
(EM) cable for the transmission of an electrical sig- ocean-bottom seismometers that record undersea
nal. EM cables typically have two elements, a earthquakes have been monitored in this manner
strength member and conductors. One type of EM as part of a prototype tsunami-warning network.
cable has electrical conductors in the center with an Figure 8 depicts ocean-bottom seismographs com-
outer armor of steel wire that provides strength, as municating acoustically with a surface mooring that is
well as fishbite protection of the conductors. Another linked with a satellite network.
EM cable design that has been used successfully The interface where a mooring transitions from just
utilizes 3  19 oceanographic cable with three con- below the ocean surface to a buoy on the surface is a
ductors laid in the valleys that are formed by the challenging area in the design of a mooring because
three strands. The plastic jacket is then extruded over the components used in that section are subject to
the wire and conductors. A disadvantage of this de- excessive wear and fatigue failure from the constant
sign is that the conductors are on the outside of the motion of the surface buoy. To reduce the wear at this
strength member and are more susceptible to fishbite juncture, a universal joint is employed between the
damage than in a cable with the outer armor. buoy and the mooring. It is extremely challenging to
Mooring cables are also being designed with optical pass cables with electrical conductors through this
fibers to take advantage of their capability to trans- interface. The universal joint is configured with a
mit over long distances with minimal losses, and its central hole, which provides an unbending pathway
inherently high data-carrying capacity. for conductors that must pass through the universal.
Not all communications rely solely on special pur- Getting the vulnerable conductors through this dy-
pose cables. Through the use of high-speed acoustic namic near-surface region and into the buoy is not a
modems, signals can be sent from instrumentation trivial part of the design. Several approaches have
deployed on the ocean floor to an acoustic transducer been taken with reasonable success.

Link to satellite network

Acoustically linked surface buoy

Acoustic transducer

Hydrothermal
vent sensors, Glass floats Ocean-bottom
ADCP, and seismograph
modem Acoustic release and modem

Figure 8 A moored buoy system configured to receive acoustically transmitted data from seafloor instruments in near-real time
and able to communicate data and commands between shore-based labs and the observatory via a satellite network. Illustration by
J. Doucette/WHOI Graphics.
106 MOORINGS

One technique is to utilize a special chain assembly capabilities as does the mooring hardware necessary
directly below the buoy consisting of chain that is to keep the structure on station. Other systems,
wrapped with a spiral shaped multiconductor cable which do not require high-speed satellite communi-
and completely encapsulated with urethane. This cations or have only a moderate number of sensors,
component has the required strength and flexibility do not have the same power requirements as a larger
from the chain while the spiraled cable configuration system and can make use of smaller moored plat-
and urethane protects the conductors from bending forms with smaller hardware.
strain produced by the buoy motion.
Another approach that has been used, particularly
in shallow-water applications, is an ultrastretchy
Mooring Deployments
rubber hose that is reinforced with nylon and is
capable of stretching to twice its unstretched length. Deep-ocean surface and subsurface moorings are
Electrical conductors are embedded in the wall of the typically deployed using an anchor-last technique. As
hose. The angle that the conductors make with re- the name implies, the anchor is the last component to
spect to the axis of the hose is a critical part of the be deployed. The entire mooring, starting at the top,
hose design in order to allow the hose to stretch and is put over the side and strung out behind the
not damage the conductors. The system also allows deployment vessel and towed into position. At the
power generated by solar panels on the surface buoy appropriate location, the anchor is dropped.
to flow to the bottom instruments. If the current and the wind are from the
It is possible to use conventional plastic-jacketed same direction, the deployment begins by positioning
wire rope, without copper electrical conductors, to the ship down-current of the desired anchor-drop
send a signal up the mooring cable through the use of position. By doing this, the ship can maintain steer-
inductively coupled modems. In such a system, the age as it slowly steams against the current while the
signal is applied to the primary winding of a toroidal mooring components are deployed and are carried
transformer. The mooring wire only has to pass away from and behind the ship by the wind and
through the toroid, to form a single-turn secondary current. When the wind and current are opposing
that conveys the data along the mooring cable. The each other, it becomes necessary to alter the de-
ends of the mooring cable are grounded to the sea- ployment plan. In such cases, the important factor is
water, which permits a current to flow through the the relative speed of the ship, which is typically
mooring wire and seawater. So as not to have to 50–100 cm s  1 through the water. Depending on the
thread the mooring cable through the toroid, it is length of the mooring, its complexity, and the
split and clamped around the cable. It is not neces- wind and current conditions, the start position could
sary to break the cable at the instrument position or be as much as 10 km from the anchor-drop position.
to provide any electrical connection between the The goal is to put the mooring line over the side
sensor and the cable. Another advantage of this at a rate that is slightly less than the ship’s
system is the flexibility it offers for sensor placement. speed through the water and, thus, have the entire
Since it does not require discrete cable lengths, sen- mooring stretched out without kinks and loops
sors can be clamped along the wire at any location behind the ship by the time it arrives at the anchor-
and easily repositioned if necessary. drop site.
Surface moorings have capabilities which make With the ship at the position for the start of the
them good candidates for use with ocean obser- deployment sequence, the upper buoyancy of the
vatories. The moorings have the flexibility to be de- mooring is lowered into the water. Figures 9(a)–9(g)
ployed nearly anywhere in the world’s oceans and illustrate the deployment sequence of a deep-water
offer near-real-time access due to the satellite com- surface mooring. The mooring components are at-
munications capability from the surface buoy. tached in series and paid out with the assistance of a
Moored observatories can vary in complexity de- winch. Instruments are attached to the mooring at
pending on their intended application. If data must the appropriate locations between premeasured
be collected from high-bandwidth sensors, such as mooring line shots. The last component put in line is
those used in acoustic arrays or for continuous seis- the anchor. The ship tows the mooring into position
mic monitoring, then high-speed satellite links may with the anchor still on deck and actually steams past
be required in conjunction with fiber optic cables the desired anchor position by a distance equal to
connected to junction boxes on the seafloor. Such approximately 7–10% of the water depth. The an-
systems typically require large quantities of power to chor is deployed by either sliding it off the deck of
operate and could require active power generation. the ship and into the water by means of a steel tip
The buoy size increases to accommodate such plate or it is placed into the water with a crane and
 MOORINGS 107

(a) (b) (c)

From trawl Start of

Deploying additional winch buoy deployment
Initial instrument deployment upper instruments Quick-release
hook
Stopped off on deck Stopped off on deck

From From From

mooring winch mooring winch mooring winch
on deck on deck on deck

(d) (e)
Buoy outboard of ship
From trawl ready to be lowered
winch Buoy released and in tow
Quick-release hook From mooring winch
on deck

From
mooring winch
on deck

(f) (g)

Glass ball buoyancy attached

Anchor away

Figure 9 Surface mooring deployment sequence. (a) The first instrument is lowered into the water. (b) Instrumentation in the upper
part of the mooring is lowered into the water before deploying the buoy. (c) The upper part of the mooring is attached to the surface
buoy. (d) The surface buoy is placed into the water. (e) The ship steams forward slowly as additional mooring line and instrumentation
are deployed. (f) The entire mooring is in tow behind the ship as the glass ball buoyancy is deployed. (g) The anchor free-falls to the
ocean bottom, pulling the buoy along the surface.

mechanically released once it is just below the sur- mooring, the anchor can fall at a rate of approxi-
face. As the anchor falls to the bottom, the mooring mately 100 m min  1.
is pulled under with it. The mooring line takes the Some situations make anchor-last deployments
path of least resistance, following the anchor as it difficult or near impossible. An application requiring
descends, resulting in the top of the mooring moving the deployment of moorings through the ice in high
toward the anchor-drop position as the anchor falls latitudes is one example where an anchor-first de-
to the bottom. The normal drag on the mooring line ployment could be used. In these cases, there is not
is greater than the tangential drag; therefore, a water- enough open water to completely stretch the moor-
sheave effect takes place as the anchor falls to the ing out behind the vessel. Instead, the vessel breaks
bottom. The anchor does not, however, fall straight the ice where the mooring is to be deployed, creating
down but rather falls back a distance equal to a small a small pool. Starting with the anchor, the mooring is
percentage of the water depth, hence the reason for deployed vertically through the opening in the ice.
steaming past the desired anchor position before The upper buoyancy is the last component to go
deploying the anchor. Depending on the design of the into the water and the mooring is allowed to drop
108 MOORINGS

straight down to the bottom. Such moorings have to may require multiple legs to provide stability and
be designed so that each component can support the to minimize mooring motion. A horizontal mooring
total weight of all the components below it, in- (Figure 11) may be needed to investigate spatial
cluding the anchor. Larger wire sizes with greater variability.
breaking strengths are typically needed for anchor- Moorings provide one means of collecting tem-
first deployments. poral and spatial data. Oceanographic gliders, which
Another application requiring an anchor-first de- are able to freely move both vertically and horizon-
ployment is a mooring that must be deployed in a tally through adjustments in buoyancy and attitude
specific depth or at a precise location. With the control, offer another approach to collecting spatial
mooring hanging below the deployment ship and the data. Gliders can be configured to travel long dis-
anchor close to the bottom, the vessel can be man- tances (B3000 km) for extended periods (B200
euvered to the desired location, at which time the days) while making approximately 600 vertical ex-
mooring is allowed to free-fall the short distance to cursions. With the proper configuration, gliders
the bottom. could make the vertical and horizontal measure-
ments typical of a moored array. Each approach has
its limitations (power constraints, deployment dur-
ation, and spatial variability) and, depending on the
Discussion and Summary
application, the most appropriate technique should
All moorings have similar components, but each de- be chosen.
sign is unique. Factors such as the mooring’s intended The ability to model mooring performance both
use, the environment in which it will be deployed, the statically and dynamically now permits extensive
water depth, the payload it must support, and the design studies before the mooring is taken into the
deployment period greatly affect the design. Although field. As a result, it is possible to explore new designs
we have discussed vertical arrays, moorings can as- and have greater confidence in how they will perform
sume a variety of orientations. For some applications, prior to cutting any wire or splicing any line. It is
a U-shaped array (Figure 10) is required, or a mooring important to point out, though, that regardless of the

Radar reflector/marker Radar reflector light/Argos

Surlyn telemetry buoy

Urethane chain
0.63 cm polypropelene

25.3 m 0.95-m steel sphere S-tether EM cable

2 m 0.95 cm chain/swivel
WHOI navigator
30.5 m

5 m 0.95 cm chain
in Tygon tubing
16-element hydrophone array
Panther Plast floats
150 m 1.9 cm nylon

150 m 1.9 cm nylon

Yale grip and flounder plate

Acoustic release/pinger Yale grip and flounder plate
Anchor sled Acoustic release
84 m 2 Glass flotation balls
Anchor

Figure 10 A U-shaped moored hydrophone array. Design by J. Kemp.

 MOORINGS 109

160 m
Glass 158 m 0.95 cm Panther 20 m
flotation wire rope Plast
MicroCAT TCP balls floats

1.22 m steel sphere

Ax
pack
2 m 1.27 cm 2 m 1.27 cm
trawler chain 3D ACM trawler chain

44 m 0.95 cm SBE 39 62 m 0.95 cm

wire rope wire rope 64 m

Brancker
TP recorder
15 m 1.27 cm Weight
trawler chain MicroCAT TC
BACS acoustic Swivel
2000 lb. release
1000 lb. 15 m 1.27 cm trawler chain
909 kg 5 m 1.27 cm trawler chain 10 m
454 kg
Dor-Mor 1.9 cm chain
Dor-Mor 15.8 m steamer
anchor
anchor chain

Figure 11 A two-dimensional moored array. TCP, temperature, conductivity, and pressure; BACS, binary acoustic command
system; SBE, Sea-Bird Electronics; ACM, acoustic current meter; Ax pack, acceleration package. Design by R. Trask.

amount of time spent designing, modeling, and fab- Frye D, Ware J, Grund M, et al. (2005) An acoustically-
ricating a mooring, the success of a deployment will linked deep-ocean observatory. Proceedings of Oceans
often come down to the ability of trained personnel 2005 – Europe, vol. 2, pp. 969--974. Woods Hole, MA:
to pay close attention to all the details and to get the WHOI.
mooring safely in and out of the water while working Morrison AT, III, Billings JD, and Doherty KW (2000) The
McLane Moored Profiler: An autonomous platform for
under extremely adverse conditions at sea.
oceanographic measurements. In: OCEANS 2000 MTS/
Further Reading IEEE Conference and Exhibition, Providence, RI, 11
Sep.–14 Sep. 2000, vol. 1, pp. 353–358. East Falmouth,
Berteaux HO (1991) Coastal and Oceanic Buoy MA: McLane Research Laboratories Inc.
Engineering. Woods Hole, MA: Henri Berteaux. Warren BA and Wunsch C (eds.) (1981) Evolution of
Berteaux HO and Prindle B (1987) Deep sea moorings: Physical Oceanography, Scientific Surveys in Honor of
Fishbite handbook. Woods Hole Oceanographic Henry Stommel. Cambridge, MA: Massachusetts
Institution Technical Report WHOI 87-8. Woods Hole, Institute of Technology Press.
MA: WHOI.
Dobson F, Hasse L, and Davis R (eds.) (1980) Air–Sea Inter-
action: Instruments and Methods. New York: Plenum.
 PROFILING CURRENT METERS
A. J. Plueddemann, Woods Hole Oceanographic vertically cycling an instrument package along a
Institution, Woods Hole, MA, USA subsurface mooring line or a taut section of surface
Copyright & 2001 Elsevier Ltd. mooring line that is dynamically decoupled from the
surface buoy. A limitation introduced by these
schemes is that the upper 20–30 m of the water col-
umn are inaccessible to the profiler. The benefit is
Introduction that the relative stability of the mooring line allows
the motion of the instrument relative to the Earth to
Surface moorings, which may be deployed in water be ignored, and the problem reduces to one of pro-
depths from tens of meters to thousands of meters, pelling the single-point velocity sensor or choice
provide a suitable platform for the deployment of along the line. Solutions to this problem fall into four
single-point current meters. Placing multiple sensors classes based on the nature of the propulsion: trac-
along the mooring line gives a discretized velocity tion, buoyancy, waves, and currents.
profile and thus single-point sensors can be used to Traction profilers use an electric motor to drive a
resolve the vertical structure of ocean currents. traction wheel that propels the sensor package along
However, in deep water the number of sensors ne- the mooring line. The package is designed to be
cessary to obtain sufficient vertical resolution neutrally buoyant at mid-depth of the profile.
throughout the water column quickly becomes pro- Buoyancy-driven profilers achieve propulsion by
hibitive. Thus, a variety of techniques have been varying the displacement of the package without
developed to obtain velocity profiles from a single changing its mass. This may be done by alternately
sensor. Ideally, such profiles resolve the oceanic vel- filling and emptying a bladder external to the pres-
ocity fine structure (vertical scales of 1–10 m) and are sure housing with a fluid or gas. Alternatively, a
synoptic in the sense that they are obtained over a piston may be driven in and out of a flooded cham-
time short compared to the characteristic timescale ber. Cycling is typically controlled by a micro-
of the phenomena of interest. processor that turns a pump or motor on and off at
There are three basic approaches to velocity pro- preset times and/or pressures. Mechanical stops at-
filing: vertically cycling single-point sensors, free-fall tached to the mooring line limit travel at the upper
probes, and acoustic (Doppler) profilers. A fourth and lower extremes of the profiling range. Traction
category involves the combination of these ap- and buoyancy-driven systems are similar in that they
proaches. Single-point sensors may be cycled be- carry their energy source with them in the form of
tween the surface and some depth by use of a winch batteries, and they require roughly the same amount
or may be cycled within the water column along a of energy to propel a given instrument package.
mooring line. The simplest free-fall probes are Traction systems are more easily configured to profile
streamlined objects that are dropped through the to great depths. Both techniques are capable of
water column and tracked acoustically. Acoustic providing a total vertical travel distance of order
Doppler techniques rely on a measurement of Dop- 1 Mm per deployment. In a typical configuration a
pler shift from a range-gated transmission that traction system may be configured for 100 round
ensonifies the water column. trips to 5000 m, while a buoyancy-driven system may
be configured for a 1500 round trips to 300 m.
Wave-driven profilers make use of the vertical
Vertically Cycling Sensors
motion of a surface mooring to travel downward.
It is a relatively simple matter to vertically cycle a During the down cycle the clamp attaching the sen-
single-point sensor from a ship using a winch, but sor to the mooring also acts as a ratchet that allows
interpretation of the resulting velocity is complicated the mooring line to pass through only when it is
by motion of the ship relative to the Earth and mo- moving upward relative to the sensor. The sensor
tion of the sensor relative to the ship. Cycling along a package is buoyant, but its inertia tends to keep it in
mooring line mitigates these problems, but the en- place while the mooring line moves upward during
ergetic vertical motion and potentially large inclin- the passage of a wave crest. With successive waves
ations of a surface mooring line are undesirable the sensor ‘crawls’ down the line until it reaches a
(wave-driven profilers are an exception, see below). mechanical stop. The ratchet is then released and the
Thus, the most common implementation involves package rises to the starting point, where the ratchet

110
 PROFILING CURRENT METERS 111

is reset. In a typical configuration such a system may Temperature and salinity sensors can easily be added
profile to 200 m depth every two hours. A current- to the probe, providing a complete hydrographic
driven profiler uses a similar technique, but the profile. Since return to the surface is not essential
relative motion between the mooring line and when tracking is used, an alternative to the recover-
the sensor is due to the generation of lift from the able probe is to use inexpensive, expendable probes,
passage of ambient currents over a wing that is and reverse the roles of the beacons and the
incorporated into the sensor package. Wave- and transponder.
lift-driven devices are distinguished by the ability to A conventional velocity sensor attached to a free-
draw on the surrounding environment for power, so fall probe measures the relative velocity between the
that deployment durations are in principle limited probe and the water (eqn [2]).
only by the power and storage requirements of the
sensors. However, these systems are likely to have a Urel ¼ Ū þ uðzÞ  Uprobe ½2
more limited vertical range than traction or buoy-
ancy-driven systems and may be ineffective where Acoustic tracking can be used to determine Uprobe
waves and currents are small. and recover the total velocity profile. The advantages
over tracking alone are higher precision and higher
vertical resolution. Alternatively, the measured vel-
Free-fall Probes ocity can be used as a basis for a total velocity esti-
A probe that follows the horizontal motion of the mate. In this approach, Urel is used as the input to a
surrounding water during a descent to the ocean predetermined transfer function, which accounts for
bottom and return to the surface can be used to de- the probe’s response to lateral forces and produces an
termine the vertically averaged current. In a typical estimate of the acceleration of the probe’s center of
implementation the probe itself is buoyant, but car- mass. This information is used to estimate Uprobe,
ries a weight that is dropped upon encountering the which is then added to Urel to produce a total vel-
bottom. The depth-average (vector) current Ū is ocity profile. The complication of tracking is elim-
computed from the horizontal distance between the inated, but uncertainties in the transfer function may
drop and recovery points divided by the travel time. result in larger errors.
In principle, velocity precision is limited only by the Without a means of determining Uprobe, only the
precision of the position fixes, but in practice the fact velocity shear (or the baroclinic velocity relative to an
that the probe may not faithfully follow the water unknown constant) can be determined. However, this
must be considered. Vertically averaged currents over approach has significant utility because the probe
variable depth ranges can be obtained by program- motion acts as a high-pass filter in vertical wave-
ming weights to drop before the bottom is reached. number space, i.e., for vertical wavelengths larger
A velocity profile is obtainable if the motion of the than the probe length (typically 2–3 m) the sensor
probe can be tracked during its descent. Tracking is follows the water, while for smaller wavelengths it
done acoustically, often using inexpensive, expend- does not. Thus, an appropriate (fast response) single-
able beacons. Two or more beacons are configured to point sensor can be used to measure the velocity
‘ping’ synchronously relative to a known timing profile at high vertical wavenumber. This technique is
reference, deployed from a ship, and then surveyed routinely used to detect velocity microstructure.
to accurately determine their relative positions. The necessity of tracking a free-fall probe or using
A transponder in the probe is used to detect the ar- a transfer function for the probe’s large-wavelength
rival times of signals from the beacons relative to the response could be eliminated if the sensor was cap-
same time base, allowing the probe’s horizontal able of directly detecting the total water velocity
position relative to the beacons to be determined. during its descent. This can be accomplished by ex-
Vertical position is determined by a pressure sensor. ploiting electromagnetic induction. The motion of
The result is a profile of the total (vector) velocity sea water (a conductor) in the presence of the Earth’s
(eqn [1], where u(z) is the baroclinic component). vertical magnetic field Bz induces horizontal electric
currents Eh in the water column that can be meas-
UðzÞ ¼ Ū þ uðzÞ ½1 ured. For a probe that perfectly follows horizontal
water motion the relationship is given by eqn [3].
Velocity precision depends on the accuracy of the
beacon positions and the precision of the timing; Eh ðzÞ ¼ Bz k  ðUðzÞ  Ū Þ ½3
errors are typically near 1 cm s1. Vertical resolution
is limited by the product of the fall rate and the time Here k is the vertical unit vector and Ū is a weighted
between pings, which is typically several meters. vertical integral of the horizontal velocity. In general,
112 PROFILING CURRENT METERS

the depth-average contribution Ū is not known, and the beam from vertical. Instrument tilt and heading
only the baroclinic component u(z) can be deter- are used to compensate for vertical misalignment of
mined. The electromagnetic induction technique range bins between coplanar beam pairs and convert
yields baroclinic velocity profiles with vertical reso- the measured velocities into geographic components.
lution of 5–10 m and precision of about 1 cm s1. The standard deviation of horizontal velocity es-
The technique has been implemented both on small, timates from an ADCP with transmitter frequency ft
expendable probes (similar in operation to XBT, has the general form of eqn [5].
expendable bathythermographs) and on larger, more
 1
complex recoverable probes. By adding a self-con- su Bc ft Tp siny ½5
tained acoustic velocity profiler to the latter, it is
possible to estimate Ū and determine the total vel- The precision [5], based on a single transmission, is
ocity profile (see Combined Approaches below). often too large for practical applications and is re-
duced by averaging over many transmissions. In the
field, precision may be adversely affected by motion
Acoustic Doppler Sensors of the scatterers and motion of the measurement
An acoustic Doppler sensor estimates fluid velocity platform (the most significant difficulty in computing
by detecting the Doppler frequency shift of the geographic velocity components is compass error). It
acoustic reverberation (or ‘backscatter’) from objects is evident from [5] that velocity precision can be in-
in the water column. For an acoustic Doppler current creased for a given vertical resolution by using a
profiler (ADCP) operating in the 50–500 kHz fre- higher transmission frequency. However, profiling
quency range, the primary scatterers are biological range R decreases with increasing frequency. It is also
(zooplankton and micronekton). The scatterers are possible to increase precision for fixed ft and Tp by
assumed to be drifting passively with the sur- using wideband transmission or coded pulses, and
rounding water, although this may not always be the this technique is now routinely implemented. Com-
case. During operation, an acoustic transducer monly used configurations include high-precision
emits a pulse of acoustic energy along a narrow beam ADCPs (sE2 cm s1; RE20 m) operating near
(3–41 half-power beam width) ensonifying a volume 1 MHz and long-range ADCPs (sE10 cm s1;
of fluid determined by the beam width, the pulse RE500 m) operating near 75 kHz. Very high-reso-
duration, and the distance from the transducers. As lution systems (sE0.1 cm s1; Ro10 m), which
the time after transmission increases, the returned utilize a different approach to data processing, are
signal comes from successively more distant sample also available.
volumes known as range bins. Backscattered energy By selecting an operating frequency matched to the
from each range bin arrives at the transducer with a desired profiling range, it would appear that ADCPs
Doppler shift proportional to the average speed of could provide full water column velocity profiles in
the scatterers within the volume. For a transmission depths up to 500 m. However, difficulties arise in
at frequency ft through a medium with sound speed using ADCPs near surface and bottom boundaries.
c, the velocity in the direction of the beam is related The first one or two range bins (nearest the trans-
to the mean Doppler shift Df eqn [4]. ducer) may be corrupted by transient signals from
the pulse transmission that saturate the system elec-
Ubeam ¼ cDf =2ft ½4 tronics. Range bins near the surface (or the bottom if
the instrument is down-looking) are corrupted by
Estimation of Ubeam for successive range bins results reflections from the side-lobes of the acoustic beam.
in a profile of water velocity as a function of distance As a result, many applications use single-point sen-
along the beam. sors near the surface and bottom boundaries com-
The typical ADCP configuration consists of four bined with ADCPs that profile the central water
downward-slanting beams separated by 901 in azi- column.
muth and inclined at 201 or 301 from vertical. The ADCPs mounted in a frame or housing that sits
four beams form two coplanar pairs that can be directly on the seafloor generally provide the highest-
combined to estimate horizontal and vertical velocity quality current profiles, because platform motion is
components as long as the horizontal scale of eliminated and compass error is reduced to a con-
the motion is greater than the beam separation. The stant (rather than a function of direction). On sub-
along-beam extent of a range bin is related to surface moorings, ADCPs are often deployed as the
the transmitted pulse duration Tp by Drp ¼ cTp/2. uppermost element, facing upward into undisturbed
The vertical resolution is set by the vertical extent of water. This is an attractive option when bottom
the range bin, Dzp ¼ Drp cos y, where y is the angle of mounting is impractical owing to deep water and
 PROFILING CURRENT METERS 113

surface conditions prohibit use of a surface element. ADCP (e.g., ft ¼ 150 kHz) is mounted on a CTD/
Platform motion is increased relative to a bottom rosette frame with the transducers facing down-
mount, but remains significantly less than on a sur- ward. As the frame is lowered during occupation of
face mooring. On surface moorings, ADCPs may be a hydrographic station, the ADCP transmits rapidly
mounted within the buoy bridle facing downward, or (E1 Hz), collecting a series of overlapping profiles.
attached to the mooring line facing either upward or Each profile is relative to an unknown (but assumed
downward. Deployment in a buoy bridle is attractive constant) velocity resulting from horizontal motion
because near-surface currents can be detected, but of the frame during the cast. By separating the
performance may be degraded as a result of wave instrument motion into a portion Uship due to ship
motion. It has been shown that small sensors drift and a portion Uframe due to the motion of
mounted in-line along a mooring do not affect ADCP the frame relative to the ship, the relative velocity
performance, and thus the deployment of relatively observed by the LADCP can be written as eqn [6].
inexpensive point sensors measuring temperature (or
temperature and conductivity) along with one or  
more ADCPs has become standard practice on both Urel ðzÞ ¼ Ū þ uðzÞ  Uship þ Uframe ½6
surface and subsurface moorings.
Practical issues important to shipboard ADCP In post-processing these profiles are differentiated
operation include installation method, transducer with depth to eliminate the unknowns Ū, Uship, and
alignment, platform motion, and navigation. The Uframe (all assumed to be constant for a given profile)
installation must account for the presence of bubbles, and the depth bins are indexed to pressure using the
which are generated under the hull in rough seas and pressure record from the CTD. The overlapping
swept past the transducers, interfering with acoustic shear profiles are then averaged together in common
transmissions. This problem may be alleviated by pressure bins and integrated to obtain the baroclinic
extending the transducers below the bubble layer, or velocity profile u(z). If the profiles are continuous
by using a faired housing, or both. Separating the during the cast (i.e., not affected by drop-outs
ADCP from the ship and deploying it in a towed due to low scattering strength or acoustic inter-
body eliminates interference from bubbles and re- ference near the bottom boundary) the vertical
duces platform motion, but the complexity of oper- integral of Uframe will be zero. The depth-average
ation is increased. After installation, the transducers component can then be estimated from eqn [7],
must be ‘calibrated’ to account for imperfect mech- where /Urel(z)S indicates a vertical average (taken
anical alignment, which may result in both magni- to be equivalent to a time average over the cast) and
tude and direction errors in the observed current. /UshipS is estimated from position fixes at the start
The accuracy of the calibration depends principally and end of the cast.
on the quality of the shipboard navigation and
compass. Of course, the accuracy of navigation also   
determines the accuracy of the absolute velocity ob- Ū ¼ Urel ðzÞi þ Uship ½7
tained from a shipboard ADCP. A distinct advantage
of shallow water applications is that the speed of the In practice a small correction due to /uðzÞSa0 may
ship over the earth can be estimated directly from the be included.
ADCP in ‘bottom track’ mode, alleviating the de- Another combined approach merges a free-fall
pendence on independent navigation. However, the probe fitted with an electromagnetic velocity sensor
availability of more accurate navigation (via the with an ADCP. In this application, a relatively long-
Differential Global Positioning System) has allowed range ADCP is mounted at the bottom of the probe
absolute velocity profiling from ships in the absence with transducers facing downward. Rather than
of bottom tracking to be done routinely. producing a profile of the water column velocity, the
ADCP is configured to detect the motion of the in-
strument relative to the bottom. Of course, this is
only effective when the probe is within acoustic
Combined Approaches
range of the bottom (typically a few hundred
A combined approach that has become relatively meters). Within this depth interval, the ADCP pro-
common on deep-ocean hydrographic cruises is the vides an independent measure of U(z), the motion of
use of an ADCP in conjunction with a vertical cyc- the instrument relative to the Earth, and the ver-
ling CTD (conductivity- temperature- depth profiler), tically averaged flowŪ can be determined. The
often called a lowered ADCP or LADCP. In the velocity profile U(z) can then be estimated over the
most common application, a relatively long-range entire water column from eqn [3].
114 PROFILING CURRENT METERS

See also Collar PG (1993) A Review of Observational Techniques

and Instruments for Current Measurement in the Open
Sonar Systems. Sea. Institute of Ocean Science, Deacon Laboratory,
Report No. 304.
Emery WJ and Thomson RE (1998) Data Analysis
Further Reading Methods in Physical Oceanography. New York:
Elsevier.
Anderson SP, Terray EA, Rizoli-White JA, and Williams AJ Pinkel R (1980) Acoustic Doppler techniques. In: Dobson
(1999) Proceedings of the IEEE Sixth Working EF, Hasse L, and Davis R (eds.) Air–Sea Interaction,
Conference on Current Measurement, 333pp. pp. 171--199. New York: Plenum Press.
Piscataway, NJ: IEEE Press.
Baker JD (1981) Ocean instruments and experiment
design. In: Warren BA and Wunsch C (eds.) Evolution
of Physical Oceanography. Cambridge, MA: MIT Press.
 SINGLE POINT CURRENT METERS
P. Collar and G. Griffiths, Southampton can be used, for example in shallow seas, or at the
Oceanography Centre, Southampton, UK deep ocean floor. Care then needs to be taken to
Copyright & 2001 Elsevier Ltd. avoid, as far as possible, disturbance to the flow by
the sensor itself, and by any supporting structure. In
the case of moored current meters the design of the
mooring must minimize vibration, which can lead to
the sensor sampling in its own turbulent wake,
thereby generating significant errors. With proper
Introduction attention to design of the mooring or platform, and
A current meter estimates the speed and direction of selection of an appropriate current meter, it should
water moving relative to the instrument. The single be possible to make most deep-sea measurements to
point current meter is, therefore, only part of a within about 1 cm s1 in speed and 2–51 in direction,
measurement system that includes the mooring or and with rather better precision in the case of bot-
mounting hardware or technique. This article begins tom-mounted instruments. Many of the issues of
with a discussion of the interaction between the current meter data quality have been reviewed in the
current meter, the method of mounting and the literature.
characteristics of the currents within the environ- Particular problems arise in the case of near-sur-
ment being studied. This is followed by an intro- face measurements. Wave orbital motion decays ex-
duction to the principles of current meter design, ponentially with depth but may be considered
which are largely independent of the chosen imple- significant, if somewhat arbitrarily, to a depth
mentation technology. Some examples of commonly equivalent to half the wavelength of the dominant
used instruments follow, with an assessment of the surface waves. In the open ocean the influence of
strengths and weaknesses of the different types of surface waves can thus easily extend to a depth of
sensor. The importance of direction measurement several tens of meters. Within this region the dif-
and calibration are discussed as prerequisites to ficulty presented by the lack of a fixed Eulerian frame
making accurate observations. (For typical current of reference for current measurement is compounded
meter moorings see Moorings.) This article con- by the presence of three-dimensional wave orbital
cludes with a note on the future for current meas- velocities. These can be large compared with the
urement systems. horizontal mean flow, making it difficult to avoid
The first self-recording current meters were in- flow obstruction by the sensor itself and necessitating
genious mechanical devices such as the Pillsbury in- a linear response over a large dynamic range. If in-
strument (first used in 1884) and the Ekman current struments are suspended some way beneath a surface
meter, available in 1904. However, the slowness of buoy large errors can result from vertical motion
progress during the first half of the twentieth century induced by the surface buoy relative to the local
is reflected in the view of the German hydrographer water mass.
Bohnecke in 1954 that ‘The subject of current However, a more fundamental problem arises in
measurements has kept the oceanographers busy the surface wave zone. This may be illustrated by
for more than a hundred years without having reference to a water particle undergoing progressive
found – this must honestly be admitted – an entirely wave motion in a simple small amplitude wave.
satisfactory solution.’ In the 1960s and 1970s the Neglecting any underlying current, a particle at
growing need for current measurements in the deep depth z experiences a net Lagrangian displacement,
ocean provided a stimulus for the development of or Stokes drift, in the direction of wave travel of
robust, self-contained recording instruments capable O½a2 skexpð2kzÞ, where a is the wave amplitude,
of deployment over periods of months. k is the wavenumber and s is its angular frequency.
Measurement of current in the open sea is usually In 10 s waves of amplitude 2 m this amounts to about
achieved by mounting the instrument on a mooring. 4.5 cm s1 at a depth of 10 m. However, a fixed in-
Movement of the mooring makes true fixed-point, or strument, even if perfect in all respects will not be
Eulerian, measurement impossible, although careful able to detect the Stokes drift. Nevertheless an in-
attention to mooring design can generally provide an strument that is moving in a closed path in response
acceptably good approximation to a fixed point. In to wave action, but unable to follow drifting par-
some circumstances a fixed measurement platform ticles, will record some value of current speed related

115
116 SINGLE POINT CURRENT METERS

in a complex but generally unknown fashion to the in the open sea. The reader is referred to the bibli-
drift. Close to the surface, where the path of a cur- ography for discussion of these points.
rent sensor over a wave cycle can be more easily
arranged to approximate the path of a water particle,
the value recorded by the instrument should more
Current Meter Design
closely resemble the surface value of the local Stokes Fluid motion can be sensed in a number of ways:
drift. This has been verified in laboratory measure- techniques most frequently employed nowadays
ments involving simple waves, but is not easily tested include the rotation of a mechanical rotor,

Two-axis
acoustic
transducer

Rotor

Large vane

(A) (B)

Orthogonal
Vane Annular EM
sensors

Dual
Savonius
Rotor

(C) (D)

Three
orthogonal
acoustic
transducers
Four
electrodes
spaced at
Vane
90˚on the
equator

(E) (F)

Figure 1 Current meters based on different sensors. (A) Aanderaa RCM4 deep ocean rotor-vane instrument; (B) Aanderaa RCM9
single cell Doppler current meter; (C) EGG Vector averaging current meter with dual Savonius rotor (at the base) and small vane
(immediately above); (D) Vector averaging electromagnetic current meter based on an annular sensor; (E) Interocean S4
electromagnetic current meter; (F) Nortek Aquadopp high precision single cell Doppler instrument, capable of measuring horizontal
and vertical currents.
Table 1 Main characteristics of some contemporary current meters and a vector averaging current meter (VACM) from the 1970s
SINGLE POINT CURRENT METERS
117
118 SINGLE POINT CURRENT METERS

electromagnetic sensing, acoustic travel time meas- Vector Averaging

urement, and measurement of the Doppler frequency
Apart from the study of turbulence, the existence of
of back-scattered acoustic energy. Figure 1 shows
significant wave energy and instrument motion down
examples of practical current meters based on these
to periods of 1 s means that a sampling rate ðfs Þ of
techniques, and Table 1 shows the main character-
Z2 Hz is often used. At this frequency substantial
istics of some commonly used instruments. The
amounts of data are generated and, unless the high
evolution in design of experimental and commercial
frequency content is specifically of interest, it is usual
instruments from 1970 to 2000 can be traced by
to average before storing data. If done correctly this
comparing the descriptions of the current sensors
involves the summation of orthogonal Cartesian
and in situ processing to the 3-D current mapping
components individually prior to computation of the
discussed in contemporary literature. Acoustic Dop-
magnitude. Any other form of averaging can produce
pler and correlation back-scatter techniques can also
erroneous results.
measure current profiles, as discussed elsewhere in
If the instrument makes a polar measurement, for
this volume (see Profiling Current Meters).
example if it measures flow by determining instant-
In quasi-steady flow, relatively unsophisticated
aneous rotor speed Vi and the instrument is aligned
instruments often produce acceptable results. How-
with the current using a vane whose measured angle
ever, in circumstances in which an instrument may
relative to north is yi the averages are formed:
need to cope with a broad frequency band of fluid
motions, as in the wave zone or when subject to 1 Pn
appreciable mooring motion, there are implications Ē ¼ Vi sin yi
n i¼1
for the design of the sampling system.
If the sensor is to determine horizontal current it
should be completely insensitive to any vertical 1 Pn
N̄ ¼ Vi cos yi
component, while responding linearly to horizontal n i¼1
components across a frequency band which includes
If on the other hand the instrument measures orth-
the wave spectrum.
ogonal velocity components Xi , Yi directly, as for
Then if, as is usual, the sensor output is sampled in
example in electromagnetic or acoustic sensors, it
a discrete manner, the provisions of the Nyquist
forms:
sampling theorem must be observed, i.e., the sampling
rate must be at least twice the highest frequency 1 Pn
component of interest, whereas negligible spectral Ē ¼ ðXi cos yi þ Yi sin yi Þ
n i¼1
content should exist at frequencies above the highest
frequency of interest. The highest frequencies that
need to be measured are encountered in velocity 1 Pn
N̄ ¼ ðXi sin yi þ Yi cos yi Þ
fluctuations in small-scale turbulence, for example in n i¼1
measurements of Reynolds stress from the time-aver-
where yi is the instantaneous angle between the Y
aged product of a horizontal velocity component with
axis and north; n is chosen so as to reduce noisy
the vertical velocity. A frequency response to at least
contributions from, for example, the wave spectrum;
50 Hz is generally required, perhaps even higher
a value of nfs1 > 50s is usual.
frequencies if the measurements are being made
The averaged magnitude and direction are then
from a moving platform. Satisfying spatial sampling
given by:
criteria is as important as satisfying temporal sam-
pling requirements. Hence, the sampling path length, 
2 2 0:5
or sampling volume of the sensor must be less than Ū ¼ ðĒÞ þðN̄Þ
the spatial scale corresponding to the highest fre-
quencies of interest. In this case, specialist turbulence
dissipation probes that employ miniature sensors Ē
ȳ ¼ tan1
measuring velocity shear are used (see Turbulence N̄
Sensors).
Experiments involving the use of laser back-scatter
instruments have been carried out at sea, for example
Mechanical Current Meters
to measure fine-scale turbulence near the ocean floor,
but their characteristics are generally better matched At first, mechanical current meters were relatively
to high resolution studies in fluid dynamics in the simple. For example, the early, mechanically encoded
laboratory. Aanderaa current meter of the 1960s combined a
 SINGLE POINT CURRENT METERS 119

scalar average of speed with a spot measurement of B

the direction (Figure 1A). Speed was measured by a
rotor consisting of six impellers of cylindrical shape
mounted between circular end plates. The rotor shaft V
ran in ball-race bearings at each end, and at the
lower end two magnets communicated the rotation
X
to an internal recording device. The large plastic
vane, with a counterweight at the rear end, aligned
the instrument with the current.
As experience in a range of deployment conditions
and types of mooring widened, such sampling
schemes were found to be unsuitable when the sensor E
experienced accelerating flow as a result of wave
motion or mooring movement. The introduction of
vector averaging schemes followed, initially in the XX
vector averaging current meter (VACM) (Figure 1C),
C
and provided a substantial improvement in accuracy
in such conditions. Improved sampling regimes were Figure 2 Sketch showing the Faraday effect, which forms the
facilitated in later instruments by low power micro- basis of the electomagnetic current meter. The effect results in a
processor technology. It was also realized that it is potential difference E ¼ BVL induced between two electrodes (X
necessary to understand fully the behavior of speed/ and XX) with a separation L when a conductor (sea water) moves
at a resolved velocity V perpendicular to the line X–XX and
velocity and direction sensors in unsteady flow perpendicular to a magnetic field with a flux density of B induced
conditions. by coil C.
By the time the dual orthogonal propeller vector
measuring current meter (VMCM) was developed in
the late 1970s sufficient was understood about the and flow separation, are of critical importance in
pitfalls of near-surface current measurement to real- determining the degree of sensor linearity as well as
ize that rotor design required a combination of the directional response. Modeling techniques can
modeling and experimental testing in order to ensure help to evaluate specific cases.
a linear response. For example, the propellor in the Forms of sensor head that have been considered or
VMCM was designed to avoid nonlinearity due used include various solids of revolution, such as
to the different response times to accelerating and spheres, cylinders, and ellipsoids. Although hydro-
decelerating flows that had been found in the dynamic performance weighs heavily in choice of
‘S’-shaped Savonius rotor of the VACM. Today, shape, this may be balanced by consideration of ease
mechanical current meter development might be re- of fabrication and robustness. One neat solution in-
garded as mature. corporates the entire instrument within a spherical
housing that can be inserted directly into a mooring
line (Figure 1E). For a smooth sphere, the resulting
Electromagnetic Current Meters instrument dimensions would normally give rise to a
In electromagnetic current meters an alternating transition from a laminar to a turbulent boundary
current (a.c.) or switched direct current (d.c.) mag- layer over the instrument at some point within its
netic field is imposed on the surrounding sea water working velocity range, at a Reynolds number of
using a coil buried in the sensing head, and meas- B105, but this is forestalled by use of a ribbed sur-
urements of the potential gradients arising from the face so as to introduce a fully turbulent boundary
Faraday effect are made using orthogonally mounted layer at all measurable current speeds. Good linearity
pairs of electrodes, as illustrated in Figure 2. Some is thereby achieved.
electromagnetic techniques make use of the Earth’s Lower flow disturbance can be achieved using an
field, but in self-contained instruments simple d.c. open form of head construction (Figure 1D) which
excitation is avoided. This is because unwanted po- has been shown to provide excellent linearity and
tential differences arising, for example, from elec- off-axis response, the only disadvantage relative to
trochemical effects can exceed flow-induced solid heads being greater complexity in construction
potential differences, which are typically between 20 and perhaps some reduction in robustness.
and 100 mV m1 s1, by two orders of magnitude. Unlike mechanical current meters electromagnetic
Flow-field characteristics around the sensor head, instruments have no zero velocity threshold. In the
including hydrodynamic boundary layer thickness past zero stability has presented a problem, but with
120 SINGLE POINT CURRENT METERS

modern electronics, and care in head design and applications as a miniature profiling sensor for tur-
fabrication, stability to within a few mm s1 over bulence measurement, and a buoy-mounted instru-
many months of immersion should be achieved. ment with 3 m path length providing surface current
measurements.
ATT current meters offer well-defined spatial
Acoustic Travel Time (ATT) Current averaging, high resolution of currents (better than 1
Meters mm s1), potentially good linearity and high fre-
quency response. The main disadvantage, tackled
ATT systems are based on the valid assumption that with varying degrees of success in individual types of
the resultant velocity of an acoustic pressure wave instrument, is associated with disturbance of flow in
propagating at any point in a moving fluid is the the acoustic path by transducers, support struts, and
vector sum of the fluid velocity at that point and the the instrument housing.
sound velocity in the fluid at rest. The method in-
volves the measurement of the difference in propa-
gation time of an acoustic pulse along reciprocal Remote Sensing Single-Point Current
paths of known length in the moving fluid, although
Meters
the principle can be realized equally in terms of
measurement of phase or of frequency difference. One current measurement technique that avoids flow
Using reciprocal paths removes the need to know the obstruction altogether is that of acoustic back-scat-
precise speed of sound. The three techniques present ter, using either Doppler shift or spatial or temporal
differing design constraints. Typically an acoustic cross-correlation. In the past, these computationally
path length l may be of order 10 cm. For resolution intensive techniques were restricted to use in current
of currents Dv to 1 cm s1, the required time dis- profilers, where the relatively expensive instrument
crimination of acoustic pulse arrivals can be calcu- could nevertheless substitute for an array of less-ex-
lated from: pensive single point current meters. Nowadays, the
availability of low cost, low-power yet high-per-
Dt ¼ Dv  l=c2 formance digital signal processing circuits has made
it possible and economic to produce single-point
or about 4  1010 s (since the sound speed, c, is acoustic back-scatter current meters (Figure 1B & F).
about 1500 m s1) requiring stable, wide-band de- Such instruments provide a combination of several
tection in the electronic circuitry. In contrast, phase desirable specifications, including: rapid data output
measurement, made on continuous wave signals, is rate, with 25 Hz being common; a dynamic range
effected within a narrow bandwidth, thereby relax- extending from 1 mm s1 to several m s1; an ac-
curacy of 71% or 7o5 mm s1; a typical sampling
ing the front-end design in the receiver. Phase meas-
volume of a few cubic centimeters and the capability
urement provides good zero stability and low power
of operating within a few millimeters of a boundary.
consumption, but the pathlength may be constrained
by the need to avoid phase ambiguity. These characteristics make this class of instrument
almost ideal for current measurement within
Whichever method is chosen, hydrodynamic con-
boundary layers, e.g., in the surf zone, while also
siderations are important in achieving accuracy: rigid
enabling the collection of concurrent velocity and
mounting arrangements which do not disturb the
directional wave spectrum information through
flow significantly are required for the transducers at
sensing the wave orbital velocity components.
each end of the acoustic path. Techniques for min-
imizing flow obstruction have included the use of
mirrors to route sound paths away from wakes and,
Directional Measurement
with the development of substantial in situ process-
ing, the use of redundant acoustic paths. For a given The directional reference for measurement of current
instrument orientation, the least-disturbed paths can is invariably supplied by a magnetic compass, two
be selected for processing. main types of which are in common use. The first
ATT techniques have been implemented in various type is the traditional bar magnet, often mounted on
forms for a range of applications, including mini- an optically read encoded disk. The entire assembly
ature probes for laboratory tanks, profiling instru- is mounted on jeweled bearings, with arrangements
ments and self-recording current meters. Of the three for damping and gimballing. In the fluxgate compass,
basic methods, the measurement of frequency dif- the second type of sensor, a soft magnetic core is
ference seems to have been the least exploited, al- driven into saturation by an a.c. signal. Orthogonal
though it has been successfully used in such diverse secondary windings detect the out-of-balance
 SINGLE POINT CURRENT METERS 121

harmonic signals caused by the polarizing effect of for an accurate heading reference. Modern instru-
the Earth’s field and, from an appropriately summed ments can correct for heading-dependent errors in
output, the orientation of the sensor relative to the real time as well as correcting for a user-supplied
Earth’s field can be determined. In current meters a magnetic variation. However, older instruments
gimballed two-component system may be used, but usually require the corrections to be applied at the
as in the case of the magnet compass, this does re- post-processing stage.
quire that the system will respond correctly to any There is a variety of practices relating to routine
rotational and translational motions arising from calibration, ranging from checks before and after
mooring or platform motion. every deployment to almost complete lack of checks.
It has been argued that sensitivities of acoustic and
Calibration, Evaluation and electromagnetic sensors are determined by invariant
physical dimensions and stable electronic gains,
Intercomparison whereas mechanical instruments require only a sim-
The calibration, evaluation, and intercomparison of ple in-air test to ensure free revolution of the rotor.
current measuring instruments are closely related However, good practice is represented by regular
and are central to the issue of data quality assurance. calibration checks in water.
Basic velocity calibration can be carried out in a tank Current meters generally behave well in steady
of nominally still water by moving the instrument, flows but, as remarked above, in the near surface
usually suspended from a moving carriage, at a zone, or in the presence of appreciable mooring or
constant, independently measured velocity. Compass platform motion, substantial differences can occur in
calibration is done, typically to a precision of B11, in data recorded by different instruments at the same
an area free from stray magnetic fields either using a nominal place and time. The fact is that no amount
precisely orientated compass table equipped with a of simple rectilinear calibration in steady flow con-
vernier scale or by invoking a self-calibration pro- ditions can reveal the instrument response to the
gram built into the instrument that obviates the need complex broadband fluid motions experienced in the

Figure 3 Acoustic travel time current meter as one instrument among many on a package capable of crawling up and down a wire
mooring to obtain profiles of properties in water depths of up to 5000 m. (Illustration courtesy of McLane Research Inc.)
122 SINGLE POINT CURRENT METERS

sea and as yet there are no standard instruments or electromagnetic current meter of similar perform-
procedures for more comprehensive calibration. ance (Figure 1 and Table 1).
Some efforts have, however, been made to model the Another trend brought about by the growth of
errors incurred in some specific instruments, with a processing capability in situ is towards the in-
view to the prediction of performance at sea from corporation of current measurement within a com-
dynamic simulation data acquired in the laboratory plete measurement system embracing a range of
test tank. physical, chemical, and biological parameters
Laboratory tests in controlled conditions thus (Figure 3). Operational requirements for current data
provide a necessary, though insufficient basis for may also in time result in the routine deployment of
judging performance, and when a new instrument, or telemetering systems. At present, satellite telemetry
technique, is first used at sea considerable effort is of surface and near-surface measurements is well
put into intercomparisons with other, longer estab- established, but telemetry of midwater measurements
lished instruments or techniques. Not surprisingly, is not yet common practice.
most of the impetus for testing and intercomparison
has come from the scientific community; the costs See also
of providing anything other than basic performance
data in controlled flow conditions is, with some Moorings. Profiling Current Meters. Sonar Systems.
justification, considered prohibitive by manu- Turbulence Sensors.
facturers. Extensive information on the performance
at sea of instruments of many types is, therefore, Further Reading
to be found in the scientific literature, although
cheaper instruments are generally less well Appell GF and Curtin TB (1991) Special issue on current
represented. measurement. IEEE Journal of Oceanic Engineering
16(4): 305--414.
Collar PG, Carson RM, and Griffiths G (1983)
Measurement of near-surface current from a moored
Evolutionary Trends wave-slope follower. Deep-Sea Research 30A(1):
63--75.
As a result of the advances in electronics and battery Dobson F, Hasse L, and Davis R (1980) Air–Sea
technology in recent years, and the painstaking Interaction: Instruments and Methods. New York:
evaluation work accompanying the introduction of Plenum Press.
new instrument types, sufficient is now known about Hine A (1968) Magnetic Compasses and Magnetometers.
current measurement that it can in this sense at least London: Adam Hilger.
be regarded as a relatively mature technology. Yet Howarth MJ (1989) Current Meter Data Quality. Co-
clear evolutionary trends are in evidence, driven by operative Research Report No. 165. Copenhagen:
an increasing operational need for data in support of International Council for the Exploration of the Sea.
large-scale monitoring programmes. A further factor Myers JJ, Holm CH, and McAllister RF (1969) Handbook
is the growing commercial involvement in data of Ocean and Underwater Engineering. New York:
McGraw-Hill.
gathering. The tendency is towards cheaper, lighter
Pinkel R and Smith JA (1999) Into the Third Dimension:
instruments which are more easily handled at sea, Development of Phased Array Doppler Sonar.
and which can be deployed in larger numbers. An Proceedings of the IEEE Sixth Working Conference on
example of changes in size, recording capacity and Current Measurement. San Diego, March 1999.
weight that have taken place over the past 25 years is Shercliff JA (1962) The Theory of Electromagnetic
shown by comparing the Vector Averaging Current Flow Measurement. Cambridge: Cambridge University
Meter from the 1970 s with a modern acoustic or Press.
UNDERWATER VEHICLES
 DEEP SUBMERGENCE, SCIENCE OF
D. J. Fornari, Woods Hole Oceanographic Institution, Enabling Deep Submergence
Woods Hole, USA
Technologies
Copyright & 2001 Elsevier Ltd.
The events that enabled these breakthroughs was the
intensive exploration that typified oceanographic
expeditions in the 1950s to 1970s, and focused de-
velopment of oceanographic technology and instru-
Introduction mentation that facilitated discoveries on many
disciplinary levels. Significant among the enabling
The past half-century of oceanographic research has
technologies were satellite communication and global
demonstrated that the oceans and seafloor hold the
positioning, microchip technology and the widespread
keys to understanding many of the processes re-
sponsible for shaping our planet. The Earth’s ocean development of computers that could be taken into the
floor contains the most accurate (and complete) re- field, and increasingly sophisticated geophysical and
cord of geologic and tectonic history for the past acoustic modeling and imaging techniques. The other
200 million years that is available for a planet in our key enabling technologies which supplanted tradi-
solar system. For the past 30 years, the exploration tional mid-twentieth century methods for imaging and
and study of seafloor terrain throughout the world’s sampling the seafloor from the beach to the abyss were
oceans using ship-based survey systems and deep submersible vehicles of various types, remote-sensing
submergence platforms has resulted in unraveling instruments, and sophisticated acoustic systems de-
plate boundary processes within the paradigm of sea signed to resolve a wide spatial and temporal range of
floor spreading; this research has revolutionized the ocean floor and oceanographic processes.
Earth and Oceanographic sciences. This new view of Oceanographic science is by nature multi-
how the Earth works has provided a quantitative disciplinary. Science carried out using deep sub-
context for mineral exploration, land utilization, and mergence vehicles of all types has traditionally
earthquake hazard assessment, and provided con- involved a wide range of research components be-
ceptual models which planetary scientists have used cause of the time and expense involved with con-
to understand the structure and morphology of other ducting field research on the seafloor using human
planets in our solar system. occupied submersibles or remotely operated vehicles
Much of this new knowledge stems from studying (ROVs), and most recently autonomous underwater
the seafloor – its morphology, geophysical structure, vehicles (AUVs) (Figures 2–6). Through the use of all
and characteristics, and the chemical composition of these vehicle systems, and most recently with the
rocks collected from the ocean floor. Similarly, the advent of seafloor observatories, deep submergence
discoveries in the late 1970s of deep sea ‘black science is poised to enter a new millennium where
smoker’ hydrothermal vents at the midocean ridge scientists will gain a more detailed understanding of
(MOR) crest (Figure 1) and the chemosynthetic- the complex linkages between physical, chemical,
based animal communities that inhabit the vents biological, and geological processes occurring at and
have changed the biological sciences, provided a beneath the seafloor in various tectonic settings.
quantitative context for understanding global ocean Understanding the temporal dimension of seafloor
chemical balances, and suggest modern analogs for and sub-seafloor processes will require continued use
the origin of life on Earth and extraterrestrial life of deep ocean submersibles and utilization of newly
processes. Intimately tied to these research themes is developed ROVs and AUVs for conducting time-
the study of the physical oceanography of the global series and observatory-based research in the deep
ocean water masses and their chemistry and dy- ocean and at the seafloor. These approaches will
namics, which has resulted in unprecedented per- provide new insights into intriguing problems con-
spectives on the processes which drive climate and cerning the interrelated processes of crustal gener-
climate change on our planet. These are but a few of ation, evolution, and transport of geochemical fluids
the many examples of how deep submergence re- in the crust and into the oceans, and origins and
search has revolutionized our understanding of our proliferation of life both on Earth and beyond.
Earth and ocean history, and provide a glimpse at the Since the early twentieth century, people have been
diversity of scientific frontiers that await exploration venturing into the ocean in a wide range of diving
in the years to come. vehicles from bathyscapes to deep diving submersibles.

125
126 DEEP SUBMERGENCE, SCIENCE OF

(A) (B)

(C) (D)

Figure 1 Hydrothermal vents on the southern East Pacific Rise axis at depths of 2500–2800 m. (A) Titanium fluid sampling bottles
aligned along the front rail of Alvin’s basket in preparation for fluid sampling. (B) Alvin’s manipulator claw preparing to sample the
hydrothermal sulfide chimney. (C) View from inside Alvin’s forward-looking view port of the temperature probe being inserted into a
vent orifice to measure the fluid temperature. (D) Hydrothermal vent after a small chimney was sampled which opened up the orifice
through which the fluids are exiting the seafloor. Photos courtesy of Woods Hole Oceanographic Institution – Alvin Group, D. J. Fornari
and K. Von Damm and M. Lilley.

Even in ancient times, there was written and graphic observational access to the deep ocean and seafloor
evidence of the human spirit seeking the mysteries of down to 6000 m depth using the ROV and tethered
the ocean and seafloor. There is unquestionably the vehicles of the NDSF (Figure 7). These vehicle systems
continuing need to take the unique human visual and of the NDSF include the ROV Jason, and the tethered
cognitive abilities into the ocean and to the seafloor to optical/acoustic mapping systems Argo II and DSL-
make observations and facilitate measurements. For 120 sonar (a 120 kHz split-beam sonar system capable
about the past 40 years submersible vehicles of various of providing 1–2 m pixel resolution back-scatter im-
types have been developed largely to support strategic agery of the seafloor and phase-bathymetric maps with
naval operations of various countries. As a result of B4 m pixel resolution (Figure 7). These fiberoptic-
that effort, the US deep-diving submersible Alvin was based ROV and mapping systems can work at depths
constructed. Alvin is part of the National Deep Sub- as great as 6000 m.
mergence Facility (NDSF) of the University National Alvin has completed over 3600 dives (more than
Oceanographic Laboratories System (UNOLS) oper- any other submersible of its type), and has partici-
ated by the Woods Hole Oceanographic Institution pated in making key discoveries such as: imaging,
(Figure 7). Alvin provides routine scientific and en- mapping, and sampling the volcanic seafloor on the
gineering access to depths as great as 4500 m. The MOR crest; structural, petrological, and geo-
US academic research community also has routine, chemical studies of transforms faults; structural
 DEEP SUBMERGENCE, SCIENCE OF 127

(A) (A)

(B)

Figure 3 (A) The ROV Jason being lifted off the stern of a
research vessel at the start of a dive. (B) ROV Jason recovering
amphora on the floor of the Mediterranean Sea. Photos courtesy
of Woods Hole Oceanographic Institution – Alvin Group and R.
Ballard.
(B)

Figure 2 (A) The submersible Alvin being lifted onto the sternof which provided the first hint that a vast subsurface
R/V Atlantis, its support ship. (B) Alvin descending to the
seafloor. Photos courtesy of Woods Hole Oceanographic
biosphere exists in the crust of the Earth on the ocean
Institution– Alvin Group and R. Catanach. floor (Figure 7).

studies of portions of deep-sea trenches off Central

America; petrological and geochemical studies of Deep Submergence Science Topics
volcanoes in back-arc basins in the western Pacific
Some of the recent achievements in various fields of
Ocean; sedimentary and structural studies of sub-
deep submergence science include the following.
marine canyons, discovering MOR hydrothermal
vents; and collecting samples and making time series 1. Discoveries of deep ocean hydrothermal com-
measurements of biological communities at hydro- munities and hot (43501C) metal-rich vents on
thermal vents in many MOR settings in the Atlantic many segments of the global mid-ocean ridge
and Pacific Oceans. In 1991, scientists in Alvin were (MOR);
also the first to witness the vast biological reper- 2. Documentation of the immediate after-effects of
cussions of submarine eruptions at the MOR axis, submarine eruptions on the northern East Pacific
128 DEEP SUBMERGENCE, SCIENCE OF

(A) (D)

(B) (E)

(C) (F)

Figure 4 Photographs taken from the ROV Jason at hydrothermal vents on the Mid-Atlantic Ridge near 371N on the summit ofLucky
Strike Seamount at a depth of 1700 m. All photographs are of a vent named ‘Marker d4.’ (A) Overall view of vent lookingNorth. White
areas are anhydrite and barite deposits, yellowish areas are covered with clumps of vent mussels. (B) Close-up of theside of the vent,
Jason’s sampling basket is in the foreground. (C) Close-up view of mussels on the side of the hydrothermalchimney; the hydrothermal
vent where hot fluids are exiting the mound is at the upper right, where the image is blurry because ofthe shimmering effect of the hot
water. Nozzles of a titanium sampling bottle are at the middle-right edge. (D) Insertinga self-recording temperature probe into a
beehive chimney. (E) Titanium fluid sampling bottles being held by Jason’s manipulatorduring sampling. (F) Close-up of nozzles of
sampling bottle inserted into vent orifice during sampling. Photos (D)}(F) are framegrabs of Jason video data. Photos courtesy of
Woods Hole Oceanographic Institution – ROV Group, D. J. Fornari, S. Humphris, andT. Shank.
 DEEP SUBMERGENCE, SCIENCE OF 129

(A)

(B)

Figure 5 (A) ROV Tiburon of the Monterey Bay Aquarium

Research Institute (MBARI) in the hanger of its support ship R/V Figure 6 The autonomous underwater vehicle (AUV) ABE
Western Flyer. This electric-powered ROV can dive to 4000 m. A (Autonomous Benthic Explorer) of the Woods Hole
steel-armored, electro-optical cable connects Tiburon to the R/V Oceanographic Institution’s Deep Submergence Laboratory,
Western Flyer and delivers power to the vehicle. Electric which can survey the seafloor completely autonomously to
thrusters allow fine maneuvering while minimizing underwater depths up to 4000 m and is especially well suited to working in
noise and vehicle disturbance. A variable buoyancy control rugged terrain such as is found on the Mid Ocean Ridge. Photo
system, together with the syntactic foam pack, enables Tiburon to courtesy of Woods Hole Oceanographic Institution.
hover inches above the seafloor without creating turbulence, to
pick up a rock sample, or maneuver quickly to follow an animal.
(B) ROV Ventana of the MBARI being launched from its support
ship R/V Pt. Lobos. This ROV gives researchers the opportunity 4. Discoveries of extensive fluid flow and vent-based
to make remote observations of the seafloor to depths of 1850 m. biological communities along continental margins
The vehicle has two manipulator arms } a seven-function arm and subduction zones;
with five spatially correspondent joints and another seven-
function robot arm with six spatially correspondent joints. Both
5. Initial deployment of ocean floor observatories of
arms can use a variety of end effectors to suit the type of work various types which enable the monitoring and
being done. Ventana is also equipped with a conductivity, sampling of geological, physical, biological and
temperature and density (CTD) package including a chemical processes at and beneath the seafloor
dissolvedoxygen sensor and a transmissometer. Photos (Figures 10 and 11).
courtesy of MBARI.
These studies have revolutionized our concepts
Rise, Axial Seamount, Gorda Ridge and CoAxial of deep ocean processes and highlighted the need
Segment of the Juan de Fuca Ridge; for more detailed, time-series, multidisciplinary
3. Utilization of Ocean Drilling Program bore holes research.
and specialized vehicle systems (e.g. Scripps Within the field of biological oceanography,
Institution’s Re-Entry Vehicle) (Figure 8) and in- major recent advances have come from the study
strument suites (e.g. CORKs) (Figure 9) for a wide of the new life forms and chemoautotrophic
range of physical properties, fluid flow and seis- processes discovered at hydrothermal vents (Fig-
mological experiments; ure 10). These advances have fundamentally altered
130 DEEP SUBMERGENCE, SCIENCE OF

Figure 7 (A) Summary of the US National Deep Submergence Facility (NDSF) vehicles operated for the University
NationalOceanographic Laboratories System (UNOLS) by the Woods Hole Oceanographic Institution (WHOI). (B) Montage of the
NDSFvehicles showing examples of the various types of data they collect. The figure also shows the nested quality of the
surveysconducted by the various vehicles which allows scientists to explore and map features with dimensions of tens of kilometers
(topright multibeam sonar map), to detailed sonar back scatter and bathymetry swaths which have pixel resolution of 1–2 m (DSL-
120sonar), which are then further explored with the Argo II imaging system, or sampled using Alvin or ROV Jason. Graphics by
P.Oberlander, WHOI; photos courtesy of WHOI – Alvin and ROV Group.
 DEEP SUBMERGENCE, SCIENCE OF 131

_ 1000
37˚ 25′ N _ 1500

Depth (m)
_ 2000
_ 2500
_ 3000
37˚ 05′ N _ 3500
32˚ 10′ W
32˚ 20′ W 10 m

DSL 120 Sonar

Multibeam Sonar

Argo II
Imaging
System

1 km
DSV Alvin

3m

ROV Jason

5m 0.75 m
(B)
Figure 7 (Continued)

biological classification schemes, extended the Recent marine biological studies show that: (1)
known thermal and chemical limits of life, and the biodiversity of every marine community is
have pushed the search for origins of life on Earth vastly greater than previously recognized; (2) both
as well as for new life forms on other planetary sampling statistics and molecular tools indicate that
bodies. the large majority of marine species have not been
132 DEEP SUBMERGENCE, SCIENCE OF

described; (3) the complexity of biological com-

munities is far greater than previously realized; and
(4) the response of various communities to both
natural and anthropogenic forcing is far more
complex than had been understood even a decade
ago. Focused studies over long time periods will
be required to characterize these communities
fully.
The field of marine biology is heading toward a
more global time-series approach as a function of
recent discoveries largely in the photic zone of the
oceans and in the deep ocean at MOR hydrothermal
vents. The marine biological, chemical, and physical
oceanographic research which will be carried out in
the next decade and beyond will certainly have a
profound impact on our understanding of the com-
plex food webs in the ocean which control prod-
uctivity at every level and have direct implications
for commercial harvesting of a wide range of re-
sources from the ocean. Meeting the challenge of
deciphering the various chemical, biological, and
physical influences on these phenomena will require
a better understanding and resolution of the causes
and consequences of change on scales from hours to
millennia. Understanding ocean ecosystems and their
constituents will improve dramatically in response to
emerging molecular, chemical, optical, and acous-
tical technologies. Given the relative paucity of in-
formation on deep-sea fauna in general, and
(A)
especially the relatively recent discovery of chemo-
Figure 8 (A) The Scripps Institution of Oceanography’s Control synthetic ecosystems at MOR hydrothermal vents,
Vehicle is a specialized ROV that can place instrument strings this will continue to be a focus for deep ocean bio-
inside deep-sea boreholes, using a conventional oceanographic logical research in the coming decade and beyond.
vessel capable of dynamic positioning and equipped with a winch
carrying 17.3 mm (0.68 in) electromechanical cable with a single
Time-series and observatory-based research and
coax (RG8-type). The control vehicle is 3.5 m tall and weighs sampling techniques will be required to answer the
about 500 kg in water (1000 kg in air). It consists of a stainless- myriad of questions regarding the evolution and
steel frame that contains two orthogonal horizontal hydraulic physiology of these unique biological systems (Fig-
thrusters, a compass, a Paroscientific pressure gauge, four ures 10 and 11).
250 W lights, a video camera, sonar systems, electronic
interfaces to electrical releases and to a logging probe, and
Present and future foci for deep submergence sci-
electronics to control all these sensors and handle data telemetry ence is MOR crests, hydrothermal systems, and the
to and from the ship. Telemetry on the tow cable’s single coax is volcano–tectonic processes that create the architecture
achieved by analog frequency division multiplexing over a of the Earth’s crust. Geochemists from the marine
frequency band extending from 20 kHz to about 800 kHz. The geology and geophysics community have emphasized
sonars include a 325 kHz sector-scanning sonar, a 23.5 kHz
narrow beam acoustic altimeter, and a 12 kHz sonar for long
the need for studies of: (1) the flux-frequency distri-
baseline acoustic navigation. This system was used successfully bution for ridge-crest hydrothermal activity (heat,
in the 1998 Ocean Seismic Network Pilot Experiment at ODP fluid, chemistry); (2) the role played by fluid flow in
Hole 843B, 225 km south-west of the island of Oahu, Hawaii. In gas hydrate accumulation and determination of how
1999, the control vehicle’s analog telemetry module was important hydrates are to climate change; (3) slope
converted into a digital system using fiberoptic technology thus
providing bandwidth capabilities in excess of 100 Mbaud. (B)
stability; and (4) determination of how much of a role
Cartoon showing the configuration of the control vehicle deployed the microbial community plays in subsurface chemical
from a research ship as it enters an Ocean Drilling Program bore and physical transformations. Another focus involves
hole to insert an instrument string. Photo and drawing courtesy of subduction zone processes, including: an assessment
Scripps Institution of Oceanography – Marine Physical of the fluxes of fluids and solids through the seismo-
Laboratory, F. Spiess, and C. de Moustier.
genic zone; long-term monitoring of changes in seis-
micity, strain, and fluid flux in the seismogenic zone;
 DEEP SUBMERGENCE, SCIENCE OF 133

Soft tether
and
Multiconductor cable
0.68″
Control
vehicle Tow cable

300 m

Probe

Acoustic
Cone
transponder

(B)

Figure 8b (Continued)

and determining the nature of materials in the seis- and biological activity will be necessary. Whereas
mogenic zone. traditional geological and geophysical tools will
To answer these types of questions, the marine continue to provide some means to address aspects of
geology and geophysics community requires sys- these problems, it is clear that an array of deep
tematic studies of temporal evolution of diverse areas submergence vehicles, in situ sensors, and ocean floor
on the ocean floor through research that includes observatory systems will be required to address these
mapping, dating, sampling, geophysical investi- topics and unravel the variations in the processes that
gations, and drilling arrays of crustal holes (Figures occur over short (seconds/minutes) to decadal time-
8, 9 and 11). The researchers in this field stress that scales. The infrastructural requirements, facility, and
the creation of true seafloor observatories at sites development needs required to support the research
with different tectonic variables, with continuous questions to be asked include: a capability for long-
monitoring of geological, hydrothermal, chemical, term seafloor monitoring; effective detection and
134 DEEP SUBMERGENCE, SCIENCE OF

Figure 9 (A) Diagram of the upper portion of an OceanDrilling Program (ODP) borehole with a Circulation ObviationRetrofit Kit
(CORK) assembly. These units serve the samepurpose as a ‘cork’ which seals a bottle; in the deep-seacase, the bottle is the seafloor
which contains fluids that arecirculating in the ocean crust. The CORK allows scientists toaccess the circulating fluids and make
controlled hydrologicmeasurements of the pressures and physical properties of thefluids. (B) A CORK observatory on the ocean floor
in ODP hole858G off the Pacific north-west coast. (C) A CORK with instrumentsinstalled to measure sub-seafloor fluid circulation
processes.Diagram courtesy of Woods Hole Oceanographic Institutionand J. Doucette; photo courtesy of K. Becker andE. Davis.

bathyurid crab (center). (F) A time-series temperature probe (with black and yellow tape), deployed at a hydrothermal vent (Tube worm
pillar) on the East Pacific Rise crest near 9149.60 N at a depth of 2495 m. The vent is surrounded by a large community of tube worms.
(G) Seafloor markers along the Bio-Geo Transect, a series of 210 markers placed on the seafloor in 1992 to monitor the changes in
hydrothermal vent biology and seafloor geology over a 1.4 km long section of the East Pacific Rise axis that have occurred after the
1991 volcanic eruption at this site. Photographs courtesy of T. Shank, Woods Hole Oceanographic Institution and R. Lutz, Rutgers
University, D. J. Fornari and Woods Hole Oceanographic Institution – Alvin Group.
 DEEP SUBMERGENCE, SCIENCE OF 135

(A)

(B)

(F)

(C)

(D)
(G)

(E)

Figure 10 Time-series sequence of photographs taken of the same area of seafloor from the submersible Alvin of a hydrothermal
vent site on the East Pacific Rise axis near 9149.80 N at a depth of 2500 m. (A) ‘Snow blower’ vent spewing white bacterial by-product
during the 1991 eruption. (B) Same field of view as (A) about 9 months later. Diffuse venting is still occurring as is bacterial production.
White areas in the crevices of the lava flow are juvenile tube worms. (C) Patches of Riftia tube worms colonizing the vent area B18
months after the March 1991 eruption. (D) Tube worm community has continued to develop, and the venting continues over 5 years
after the eruption. (E) Close-up photograph of zoarcid vent fish (middle-left), tube worms, mussels (yellowish oblong individuals) and
136 DEEP SUBMERGENCE, SCIENCE OF

(A)

Figure 11 (A) Diagram of the deployment of the Hawaii-2 Observatory (H2O) one of the first long-term, deep seafloor observatories
deployed in the past few years. Scientists used the ROV Jason to splice an abandoned submarine telephone cable into a termination
frame which acts as an undersea telephone jack. Attached by an umbilical is a junction box, which serves as an electrical outlet for up
to six scientific instruments. An ocean bottom seismometer and hydrophone are now functioning at this observatory. (B) The H2O
junction box as deployed on the seafloor and photographed by ROV Jason. Drawing courtesy of Jayne Doucette, Woods Hole
Oceanographic Institution (WHOI), A. Chave, and WHOI–ROV group.

response capability for a variety of seafloor events kilometers long) at depths that range from 1500 m to
(volcanic, seismic, chemical); adequately supported, 6000 m and greater in the deepest trenches. Clearly,
state-of-the-art seafloor sampling and observational the spectrum of scientific problems and environments
facilities (e.g. submersible, ROVs, and AUVs), and where they must be investigated require access to the
accurate navigation systems, software, and support deep ocean floor with a range of safe, reliable,
for shipboard integration of data from mutiscalar multifaceted, high-resolution vehicles, sensors, and
and nested surveys (Figures 7B, 11 and 12). samplers, operated from support ships that have
As discussed above, the disciplines involved in global reach and good station-keeping capabilities in
deep-submergence science are varied and the scales rough weather. Providing the right complement of
of investigation range many orders of magnitude deep-submergence vehicles and versatile support
from molecules and micrometer-sized bacteria to ships from which they can operate, and the funding
segment-scales of the MOR system (10s to 100s of to operate those facilities cost-effectively, is both a
 DEEP SUBMERGENCE, SCIENCE OF 137

water column near the bottom without any guidance

from human operators. It follows programmed
track-lines precisely and follows the bottom at
heights from 5 to 30 m, depending on the type of
survey conducted. ABE’s unusual shape allows it to
maintain control over a wide range of speeds. Al-
though ABE spends most of its survey time driving
forward at constant speed, ABE can slow down or
even stop to avoid hitting the seafloor. In practice,
ABE has surveyed areas in and around steep scarps
and cliffs, and not only survived encounters with the
extreme terrain, but also obtained good sensor data
throughout the mission.
Recently ABE has been used for several geological
(B) and geophysical research programs on the MOR in
the north-east and south Pacific which have further
Figure 11 (Continued).
proved its reliability as a seafloor survey vehicle, and
pointed to its unique characteristics to collect de-
requirement and a challenge for satisfying the ob- tailed, near-bottom geological and geophysical data,
jectives of deep-sea research in the coming decade and to ground-truth a wide range of seafloor terrains
and into the twenty-first century. (Figure 12). These new perspectives on seafloor
To meet present and future research and engin- geology and insights into the geophysical properties
eering objectives, particularly with a multidisciplinary of the ocean crust have greatly improved our ability
approach, deep submergence science will require a to image the deep ocean and seafloor and have al-
mix of vehicle systems and infrastructures. As deep ready fostered a paradigm shift in field techniques
submergence science investigations extend into previ- and measurements which will surely result in new
ously unexplored portions of the global seafloor, it is perspectives for Earth and oceanographic processes
critical that scientists have access to sufficient vehicles in the coming decades.
with the capability to sample, observe and make time-
series measurements in these environments. Sub-
mersibles, which provide the cognitive presence of
humans and heavy payload capabilities will be critical
to future observational, time-series and observatory-
Conclusions
based research in the coming decades. Fiberoptic- One of the most outstanding scientific revelations of
based ROVs and tethered systems, especially when the twentieth century is the realization that ocean
used in closely timed, nested investigations offer un- processes and the creation of the Earth’s crust within
paralleled maneuverability, mapping and sampling the oceans may determine the livability of our planet
capabilities with long bottom times and without the in terms of climate, resources, and hazards. Our dis-
limitation of human/vehicle endurance. AUVs of coveries may even enable us to determine how life
various designs will provide unprecedented access to itself began on Earth and whether it exists on other
the global ocean, deep ocean and seafloor without worlds. The next step is toward discovering the
dedicated support from a surface ship. linkages between various phenomena and processes in
AUVs represent vanguard technology that will the oceans and in exploring the interdependencies of
revolutionize seafloor and oceanographic measure- these through time. Marine scientists recognize that
ments and observations in the decades to come. Over technological advances in oceanographic sensors and
approximately the past 5 years scientists at several vehicle capabilities are escalating at a increasingly
universities and private laboratories have made rapid pace, and have created enormous opportunities
enormous advances in the capabilities and field- to achieve a scope of understanding unprecedented
readiness of AUV systems. One such system is the even a decade ago. This new knowledge will build on
Autonomous Benthic Explorer (ABE) developed by the discoveries in marine sciences over the last several
engineers and scientists at the Woods Hole Oceano- decades, many of which have been made possible only
graphic Institution (Figure 6). ABE can survey the through advances in vehicle and sensor technology.
seafloor completely autonomously and is especially With the rapidly escalating advances in technology,
well suited to working in rugged terrain such as is marine scientists agree that the time is ripe to focus
found on the MOR. ABE maps the seafloor and the efforts on understanding the connections in terms of
138 DEEP SUBMERGENCE, SCIENCE OF

Magnetic
Field(nT)
59500
59000
58500
58000
57500
46˚31 ′30″N 57000 46˚31 ′30″N
56500
56000
24 5 0
_ 2 46

55500
40 0
_ 2
_

55000
4

54500
_ 54000
24 53500
30 53000
52500
52000
51500
46˚31 ′00″N 51000 46˚31 ′00″N
50500
50000
49500
49000
48500
48000
90
24
_
0
247
_

46˚30 ′30″N 46˚30 ′30″N

(A) (B)

Thickness
(meters)
45
24
22
46˚31 ′30″N 20 46˚31 ′30″N
18
16
14
12
10
8
6
4
46˚31 ′00″N 2 46˚31 ′00″N
1
0

400m 400m

46˚30 ′30″N 46˚30 ′30″N

129˚35 ′30″N 129˚35 ′00"N 129˚34 ′30″N 129˚35 ′30″N 129˚35 ′00″N 129˚34 ′30″N
(C) (D)

Figure 12 (A) Bathymetry map (contour interval 10 m) showing location of 1993 CoAxial lava eruption (gray) on the Juan de Fuca
Ridge off the coast of Washington, and ABE tracklines (each color is a separate dive). (B) Magnetic field map based on ABE tracklines
showing strong magnetic field over new lava flow. (C) Computed lava flow thickness assuming an average lava magnetization of 60 A/
m compared with (D) lava flow thickness determined from differential swath bathymetry. Figure courtesy of M. Tivey, Woods Hole
Oceanographic Institution.
 DEEP SUBMERGENCE, SCIENCE OF 139

interdependency of phenomena at work in the world Haymon RH, Fornari DJ, Von Damm KL, et al. (1993)
oceans and their variability through time. Direct submersible observation of a volcanic eruption
on the Mid-Ocean Ridge: 1991 eruption of the East
Pacific Rise crest at 91450 –520 N. Earth and Planetary
Science Letters 119: 85--101.
See also Humphris SE, Zierenberg RA, Mullineaux L, and
Manned Submersibles, Deep Water. Remotely Thomson R (1995) Seafloor Hydrothermal systems:
Operated Vehicles (ROVs). Physical, Chemical, Biological, and Geological
Interactions. American Geophysical Union Monograph,
vol. 91, 466 pp.
Ryan WBF (chair) et al., (Committee on seafloor
Further Reading observatories: challenges and opportunities) (2000)
Illuminating the Hidden Planet, the Future of Seafloor
Becker K and Davis EE (2000) Plugging the seafloor with Observatory Science. Washington, DC: Ocean Studies
CORKs. Oceanus 42(1): 14--16. Board, National Research Council, National Academy
Chadwick WW, Embley RW, and Fox C (1995) Seabeam Press.
depth changes associated with recent lava flows, coaxial Shank TM, Fornari DJ, Von Damm KL, et al. (1998)
segment, Juan de Fuca Ridge: evidence for multiple Temporal and spatial patterns of biological community
eruptions between 1981–1993. Geophysical Research development at nascent deep-sea hydrothermal vents
Letters 22: 167--170. along the East Pacific Rise, 9149.60 N–9150.40 N. Deep
Chave AD, Duennebier F, and Butler R (2000) Putting Sea Research, II 45: 465--515.
H2O in the ocean. Oceanus 42(1): 6--9. Tivey MA, Johnson HP, Bradley A, and Yoerger D (1998)
de Moustier C, Spiess FN, Jabson D, et al. (2000) Deep-sea Thickness measurements of submarine lava flows
borehole re-entry with fiber optic wire line technology. determined from near-bottom magnetic field mapping
Proceedings 2000 of the International Symposium on by autonomous underwater vehicle. Geophysical
Underwater Technology. pp. 23–26. Research Letters 25: 805--808.
Embley RW and Baker E (1999) Interdisciplinary group UNOLS (University National Laboratory System) (1994)
explores seafloor eruption with remotely operated The Global Abyss: An Assessment of Deep
vehicle. Eos Transactions of the American Geophysical Submergence Science in the United States.
Union 80(19): 213--219 222. Narragansett, RI: UNOLS Office, University of Rhode
Fornari DJ, Shank T, Von Damm KL, et al. (1998) Time- Island.
series temperature measurements at high-temperature Von Damm KL (2000) Chemistry of hydrothermal vent
hydrothermal vents, East Pacific Rise 91490 –510 N: fluids from 9–101N, East Pacific Rise: ‘Time zero’ the
monitoring a crustal cracking event. Earth and immediate post-eruptive period. Journal of Geophysical
Planetary Science Letters 160: 419--431. Research 105: 11203--11222.
 GLIDERS
C. C. Eriksen, University of Washington, Seattle, The tasks gliders do best are those for which ships
WA, USA are least suited: intensive, regular, and sustained ob-
& 2009 Elsevier Ltd. All rights reserved. servations of oceanic properties that are readily mea-
sured by electronic means. Observations that require
large or power-hungry instruments, physical collection
of samples, or specialized labor are unsuitable for
gliders and continue the need for ship support. Gliders
Introduction excel at measuring standard properties typically
Underwater gliders are a recently developed class of collected by a conductivity–temperature–depth (CTD)
autonomous underwater vehicle (AUV) driven by package. Among their many advantages is their ability
buoyancy changes to fly along saw-tooth trajectories to function well through the most severe seas the ocean
through the ocean. Gliders quickly have become the has to offer, day and night, around the globe. They are
AUVs with the highest endurance and longest range. a means to erasing the fair-weather bias of ship-based
They are able to sample the ocean interior at com- observations.
paratively low cost because they can operate in- One may sample the ocean at deliberately chosen
dependently of ships, some for the better part of a locations and times with gliders, examine the data
year under global remote control. very shortly after it is collected, and alter the sam-
Sampling the ocean on space and timescales as fine pling plan as often as a glider reaches the sea surface
as those dominating its variability and over long and communicates. The density of observations can
ranges and durations has been a challenge through- readily be scaled to the phenomena of interest by
out the history of ocean science. Traditionally, most adjusting the number and distribution of platforms
knowledge of the ocean interior has been collected to address local, regional, or global variability on
by ships. Starting in the late twentieth century, diurnal, fortnightly, or seasonal scales, for example.
moored observations have complemented those from Of course, gliders are not without limitation: long
ship-based surveys, and more recently satellite re- range and high endurance are achieved at the cost of
mote sensing has provided images of sea surface traveling slowly through the ocean. When they cease
conditions globally on ever-finer space and time- to communicate, they are lost.
scales. Were these techniques inexpensive, there
would be no particular need for autonomous data
A Short History
collection by moving platforms. Unfortunately, ships
stay at sea only for a month or two and rarely are John Swallow began the era of exploring the ocean
directed to sample the same region for multiple interior autonomously in the 1950s with acoustically
cruises. Moorings sampling the open ocean number tracked floats ballasted to follow water motions
in the hundreds globally, typically in the dozens in at depth, famously deployed in the various North
the deep sea. This means variability of the ocean Atlantic locales. In following decades, acoustic float-
interior is generally undersampled. Undersampling is tracking technology extended to ocean basin scale.
arguably the principal impediment to description and Russ Davis and Doug Webb turned to satellite
understanding of the ocean. tracking to follow the motion of Autonomous
Gliders have been developed to address the Lagrangian Circulation Explorer (ALACE) floats
shortcomings of conventional observing means so that periodically adjusted their buoyancy to reach
that such dominant sources of variability as meso- the sea surface for communication, and then re-
scale eddies, fronts, and boundary currents can be turned to a ‘parking’ depth to drift. ALACE floats
resolved simultaneously in space and time and over were the precursor to Argo floats, platforms used in
sufficiently long periods and wide domains to allow an ambitious international program to seed the
them to be understood. They are able to make pro- ocean with 3000 profiling floats. While Argo floats
gress in solving the sampling problem simply by the continue to track horizontal water movements, the
numbers their economy affords. The salient charac- focus of the Argo program is largely on CTD profiles
teristic of gliders is that they cost roughly the collected periodically to describe large-scale features
equivalent of a few days of research vessel operation of ocean circulation.
to build and can be operated for a few months for the Gliders are the descendants of profiling floats.
cost of a day of ship time. They can be thought of as profiling floats with wings.

140
 GLIDERS 141

Their essential difference is that their sampling lo- several months (Seaglider and Spray) duration while
cation is controlled, subject to a limited ability to traveling at B0.5 kt (B0.25 m s  1). All are battery-
navigate through currents. Floats randomly sample powered, navigate by dead-reckoning underwater
where currents carry them. While gliders do not offer between GPS navigational fixes obtained at the sea
a Lagrangian view of the ocean, they are commanded surface, and send data and receive commands via a
to describe oceanic conditions at deliberate times and constellation of low Earth-orbit satellites on long-
locations, adding to the Eulerian description of the range missions.
ocean. Derivatives of these three gliders are currently
Henry Stommel recognized the potential of under development, including a version powered in
underwater gliders (the idea was shared with him by part by extracting thermal energy from ocean
his friend, neighbor, and colleague Doug Webb), and stratification, one that operates under ice, and one
published a science fiction article describing them in capable of making open ocean full-depth dives. In
1989. Written as the retrospective of a scientist addition, parallel development of a vehicle con-
working at a glider network control station in 2021, siderably larger, faster, and capable of flying along a
the article has proved remarkably prescient in de- shallower glider path is underway.
scribing the gliders that were developed over the next
decade and a half. Stommel envisioned a fleet of
about 1000 gliders plying the oceans, continuously
Design Considerations
collecting profiles and communicating them ashore
via satellite, while being controlled to variously oc- The three operational gliders were designed to ad-
cupy hydrographic sections or chase features within dress the sampling deficiencies of infrequent ship-
the ocean. The fundamental vision was of a distrib- based surveys and widely separated moorings in
uted network of relatively inexpensive platforms. His describing oceanic internal structure. Mesoscale ed-
optimism on the pace of development can under- dies with characteristic space and timescales of
standably be blamed on his enthusiasm, for he pre- O(100 km) and O(1 month) provide the principal
dicted the maiden voyage at sea of a glider to have noise to ocean general circulation and climate vari-
taken place in 1994 and last 198 days. In reality, the ability. Fronts, eddies, internal waves, diurnal cycles,
first ocean deployment longer than a week took place and tidal fluctuations contribute to variability on
in 1999 and the longest mission as of this writing smaller scales in problems of regional focus. Mea-
lasted 217 days, accomplished in 2005. surement platforms capable of resolving these features
The Autonomous Oceanographic Sampling Net- require sampling on scales O(10 km) and O(10 h)
work program set up by the US Office of Naval while the need to observe many realizations of ran-
Research (ONR) in 1995 provided 5 years of support dom processes calls for sampling to be carried out for
that resulted in gliders becoming a reality. The three several weeks or months over tens or hundreds of
operational gliders, Slocum (Webb Research Corp.), kilometers. In order to observe at much lower cost
Spray (Scripps Institution of Oceanography and than with ships or moorings, the three operational
Woods Hole Oceanographic Institution), and Sea- gliders were designed with endurance and range
glider (University of Washington), were developed capability to meet these demands.
under this program and have subsequently been Gliding is the conversion of a buoyancy force into
supported both by ONR and the US National Science a body’s forward as well as vertical motion. While
Foundation. Four key technical elements enabled aeronautical gliders are heavier than the air around
the successful development of underwater gliders: them, hence always fall through it, underwater gli-
small reliable buoyancy engines, low-power com- ders are designed to alternately be heavier and lighter
puter microprocessors, Global Positioning System than the surrounding ocean so that they may glide
(GPS) navigation, and low-power duplex satellite both when diving and climbing. On average, under-
communication. water gliders displace an amount of water equal to
Seaglider, Slocum, and Spray were designed to their mass. They accomplish gliding by adjusting
fulfill similar missions, hence understandably share their volume displacement downward to dive and
many elements (Figure 1). All three are of similar upward to climb. Gliders pitch themselves down-
size, roughly 50 kg in mass, in order that both ward and are supported by wings to descend along a
manufacturing and operating costs would be rela- slanting path (conversely to ascend). Wings are glider
tively small. The vehicles can be launched and re- propellers. They convert the vertical force of buoy-
covered from small vessels by two people without ancy into forward motion.
power-assisted equipment. They are designed to The desire to obtain vertical as well as horizontal
carry out missions from a few weeks (Slocum) to structure of the ocean is conveniently and efficiently
 142
GLIDERS

Figure 1 The operational gliders (clockwise from left): Seaglider pitched down at the sea surface with its trailing antenna raised (photo by Keith Van Thiel, University of Washington), top
view of Spray at the sea surface (photo by Robert Todd, Scripps Institution of Oceanography), and rear view of Slocum climbing (photo by David Fratantoni, Woods Hole Oceanographic
Institution).
 GLIDERS 143

provided by gliding, provided glide slopes are steep make up for downstream drift within them by glid-
compared to slopes of oceanic property surfaces. The ing upstream in weak or opposing flows on their
operational gliders variously operate along glide flanks. Navigating the swirling currents of a strong
slopes in the range 0.25 to unity (e.g., B14–451), eddy is also possible if glider position within an eddy
similar to the glide slope of a NASA space shuttle. By is recognized.
contrast, high-performance sail planes attain slopes The operational gliders all carry a fixed energy
as gentle as 0.02, not far from the higher slopes of supply in the form of primary batteries. Packs con-
property surfaces within the ocean. sisting of lithium cells provide the highest energy
Sinking through the ocean is gratis, while climbing density, but are classified as hazardous material.
through it requires energy input. The force of am- Gliders using these are subject to various shipping
bient pressure can be used to partially collapse ve- restrictions, particularly as air freight. Nevertheless,
hicle volume, raise its mass density over that of lithium battery packs enable glider missions of half a
seawater surrounding it, and make it sink. Diving year or more. Up to about one quarter of the total
requires only the energy to open a valve to allow a mass of these gliders can be devoted to carrying
swim bladder to drain into a reservoir within the batteries, so packs of the equivalent of B100 lithium
hull. To stop diving and begin ascent, displacement D-cells can be used, providing roughly 10 MJ of en-
volume must be increased, requiring work against ergy, enough for B1000 dive cycles using B10 kJ per
pressure to expand a swim bladder, for example. cycle. The total average power consumption may be
Work is required not only to provide sufficient po- as little as 0.5 W, resulting in mission endurance of
tential energy to bring a glider to the sea surface, but several months. Recent improvements suggest that
also to provide kinetic energy to fly through the missions longer than a year are feasible for some
ocean. Once made buoyant, gliding eventually brings battery-powered gliders. A buoyancy engine pow-
a vehicle to the sea surface where it can raise an ered by ocean thermal stratification, as envisioned
antenna to navigate and communicate. by Stommel and Webb, is under development for
Kinetic energy provided through buoyancy forcing Slocum, but will be restricted to regions where
is ultimately dissipated by hydrodynamic drag in sufficient ocean temperature stratification is found.
gliding. Since drag is generally quadratic with speed, In principle, longer endurance and range can be
halving speed quadruples endurance and doubles attained by larger vehicles, since drag is determined
range. Underwater gliders achieve endurance and by area while energy storage scales with volume.
range 2 orders of magnitude higher than con- Instrument payload also scales with volume, and also
ventional propeller-driven AUVs by virtue of em- expense. Practically, AUVs larger than B60 kg re-
ploying speeds an order of magnitude smaller. While quire power assistance for handling at sea. Both the
a slow propeller-driven AUV might attain consider- significant economy of manual handling by a pair of
able endurance and range traveling horizontally, it operators and the smaller unit manufacture and
too must work against the ocean’s potential energy service costs associated with small platforms are
gradient (gravity) to reach the sea surface to navigate consistent with the vision expressed by Stommel’s
or communicate electromagnetically. 1989 article and also that guiding the Argo project:
Slocum, Spray, and Seaglider were each designed to that of a network of a large number of relatively
operate at speeds barely faster than typical ocean small, inexpensive platforms. By keeping gliders
currents. Since currents are typically surface-intensified small, networks of them are tolerant to occasional
due to baroclinicity, deeper dives tend to reduce vehicle loss.
average current encountered by a glider, hence en-
hance the ability to navigate effectively. Gliders can
navigate more effectively in regions where currents
Glider Dive Cycles
reverse with depth, despite being swift, than in re-
gions of strong barotropic flow. For example, glider Gliders are fully autonomous from the time they
navigation in a swift equatorial current system or leave the sea surface until they return. At the surface,
within a strait with strongly baroclinic flow is ex- they use a GPS fix to compute a desired heading to a
pected to be more effective than in weakly stratified target (or use a preset or communicated heading),
coastal jets over the continental shelf. Gliders have then reduce volume displacement to start diving.
operated across the Kuroshio and Gulf Stream by Retracting a piston accomplishes buoyancy re-
gliding with an upstream component to ferry across duction in the 200-m-depth rated Slocum, while in
them, much as a canoeist paddles both upstream and Spray and Seaglider an external bladder is deflated to
across stream in order to cross a swift river. The finite reduce buoyancy by opening a valve, since sub-
spatial extent of ocean currents can be exploited to barometric pressure is maintained inside the pressure
144 GLIDERS

hull. Glider pitch attitude is adjusted by redistrib- maximum lift to drag ratio. As an example, buoyancy,
uting mass within the vehicles, both by changing power, and attack angle dependence on vertical and
buoyancy and moving fixed masses such as battery horizontal speed are shown in Figure 2 for a typical
packs. Gliders turn to attain the desired heading Seaglider. Contours of power (solid curves) reflect
through the use of an active rudder or by redistrib- the nearly quadratic nature of the drag law for this
uting mass to roll the wings. Typical turn radii of the vehicle: operating at speeds of B0.2 m s  1 costs
operational gliders are tens of meters. B0.1 W, while operating at B0.4 m s  1 costs B0.4 W
Under water, navigation is by dead reckoning: (of delivered power – the buoyancy engines are at
simply following the chosen compass course. While most about 50% efficient). The maximum horizontal
course through the water varies due to environ- speed component attainable for a given power use is at
mental effects such as water-column shear of hori- a glide slope only slightly steeper than the stall glide
zontal current as well as to vehicle control, the slope, as can be seen from the departure of the power
overall horizontal displacement due to gliding at curve shape from near-circular about the origin to
various headings and speeds can be measured or in- steeply curved to form a cone of exclusion at the stall
ferred. The difference between displacement between slope. The extreme marks the most efficient glide slope
surface GPS fixes and gliding displacement gives for range, B0.28 for speeds of B0.25 m s  1 to B0.20
an estimate of depth-averaged current. The depth- for speeds of B1 m s  1 for the example given in
averaged current for each dive cycle can be used to Figure 2.
guide the choice of vehicle heading to approach a A given buoyancy force leads to a wide range of
target through a prediction scheme, either on board speed, depending on glide slope, as illustrated by the
or communicated from afar. dashed curves in Figure 2. For example, a buoyancy
of 150 g can provide a horizontal speed of 0.22 m s  1
near the glide slope of maximum efficiency, or
0.33 m s  1 at a glide slope of 0.8 (using about
Hydrodynamics
4 times as much power). The maximum total speed is
The existing operational underwater gliders attain highest for a given buoyancy if a glider were to be
effectively steady flight in some tens of seconds due able to rise perfectly vertically (requiring no wing lift),
to relatively slow acceleration relative to hydro- but the maximum horizontal speed is attained at
dynamic drag. The balance of buoyancy by lift slopes of 0.78–0.84 (glide angles of 30–401 above
and drag forces determines the glide slope. Glider horizontal) for the particular combination of lift,
design requires sufficiently small drag that the ratio drag, and induced drag coefficients in the Seaglider
of drag to lift, equal in steady flight to glide slope, example. The buoyancy range plotted in Figure 2
can provide the desired performance. The wings of exceeds what the Seaglider buoyancy engine is cap-
Seaglider, Slocum, and Spray are somewhat small able of providing in both positive and negative
relative to body vehicle size in comparison to buoyancy (the end-to-end range is B850 g), but div-
those of typical aeronautical sail planes. Both con- ing through a light near surface layer may produce
siderations of how steeply to dive through the instantaneously buoyancy as high as is plotted in the
ocean’s structure and ease of handling on launch and figure.
recovery guided wing sizing. Vehicle drag depends on Attack angles for gliders (the difference between
both shape and surface texture, form, and skin pitch and glide angle, where the latter is always
drag, respectively. A significant component of drag is steeper) range from less than 11 for steep dive slopes
that induced by lift, typically parametrized as pro- to several degrees for glide slopes near stall. For an
portional to the square of attack angle. Consider- underwater flying wing, for example, glider pitch can
ations of drag led Spray to approximate a slender be in the opposite sense to glide angle to produce
ellipsoidal form with a long cylinder capped by el- large lift. For the Seaglider example, typical attack
lipsoidal ends and Seaglider to use a long smooth fore- angles are a few degrees for the most efficient glide
body (to maintain laminar boundary layer flow) and slope. Attack angle affects flow around the hull,
short after-section (where turbulence is tolerated). amplifying it as a function of position.
Tow tank, wind tunnel, and open-ocean experience
have demonstrated that small appendages can ac-
count for disproportionately large contributions to
Instrumentation
total vehicle drag.
A glider with a given set of lift, drag, and induced Size, power, hydrodynamic smoothness, and mass
drag coefficients can be flown over a range of speed invariance are the principal constraints on instru-
and glide slope, the latter limited by stall at the mentation carried on gliders. For the operational
 GLIDERS 145

Seaglider performance model

1 40
3 350 0 4 5 7

600
50

750
650
55
0

700
0.5

0
− 6

45
0.9 3
300

0
2250
0.8
4
2
3

5
200

35
0.7 25
.5

0
0
−0

400
−1
Vertical speed (m s−1)

300
0.6 150

500

600
0.8 1

550
0.9 2

20

650

750
0
450

80
0.7

700
−1
0

0.5
0.5 0.6
100

4
0.4

3
0.4 .5

150
−0
0.
0.3 8
−1 −1.5 −2

250
0.

1 9
40 350
0.
7
0.30.2
0.
−2
5
50 −2.5

0
0
00.0.1 1.5 0. 0 −3
9 4 00500 75800

0.6
.08

0
0 0 6
−2.5 5055 3670

30
0.2 0
100

2
0
0.05.06 7 . 0 2 0
45 −4 −4.5
0
0 3.5
20

0.0.0341 4 −5.5
0.02 −2. 5 3 5
4. −5
3.5 −78.5
0.3

5.5 −6 −6.57.5
0.1 0.0 9 5−5.5 10
9.5
4
−0. 5 0

−4 4.5 5 6.5 7 8 10
9.516350 3
98387
6.5 72.0 0.02
0.6
45 0.9
6001
700
650
0 550
0500
1
0.2
0.07
0.06
0.050.03
0.01
0.04
9
.5
00.1
.3
.0
15 8.5 50 510 0.8
0.40.7
0.5
400
.5250 02
34.5
3.5 8310700
80

4
12. 7
5.5550 6 0
3 0500087.5
7 5
9.5
.5800010
00.4
.2858
0 93049
.7
.5
0 .57
03.57 10
800
0.3
200
0.08 450
300 1
0.
0.6
750 5
600
0.09
0.1
100 6
7
0.5
0.0
0.0
0 55
0
002900
.5
.0
5
1.0
.0 .2
6347 24150
1 09.5
88.5
7.5
50 0535
0.00
2.5
5 0 0
5200 8.
4 05 0
5
0.205 6
10
0067 9
2
58
8
7
500
.5.5
.5
00 6350
1.5
53 6
4
5 38
35.56.5 3994.5
1 .5 23.5 010
730
0.0 05
40.1 5.5
01
95.0
0. 1 520.0
63
07
204
.0 410
36 79.
52810
.59785.5
4200 8 610
5.50.505
1.535683 87 80.0
100 0 .0
0
.5
0 0.
Buoyancy (g) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Power (W) Horizontal speed (m s−1)
Attack angle (°) a = 0.002 758 9 b = 0.009 257 5 c = 9.228 2 e −06 Λ0.25 = 211.863 4

Figure 2 Model Seaglider performance based on steady flight with lift, drag, and induced drag coefficients a, b, and c where lift and
drag forces follow L ¼ ql 2aa and D ¼ ql 2(bq  1/4 þ c a2), where l ¼ 1.8 m is the Seaglider body length, a is attack angle, and dynamic
pressure is given by q ¼ r(u2 þ w2)/2 for horizontal and vertical speed components u and w for water density r. Solid, dashed, and
dotted curves contour power, buoyancy, and attack angle, respectively, as functions of horizontal and vertical speed. Contours bunch
at the stall limit. Glide slope is the ratio w/u and attack angles are labeled as the difference between vehicle pitch and glide angle,
referenced to the horizon (e.g., negative attack for upward glide).

gliders, sensors and their packaging are practically interest, together with measurements of vehicle
limited to perhaps B500-cc volume and power pitch, volume, and heading, these can be used to es-
consumption a fraction of that used for glider timate glider speed through the water. Other instru-
propulsion. On long missions, as much as 85% mentation routinely installed on gliders include
of glider battery energy might be devoted to run- dissolved oxygen sensors, bio-optical sensors for
ning the buoyancy engine, limiting the average chlorophyll fluorescence, optical backscatter, and
power consumption available to instrumentation transmissivity, and acoustic sensors for both active
and operating the onboard microprocessor to less and passive measurements. Specially adapted acoustic
than B0.1 W. Average power of this level is achieved Doppler current profilers have also been carried
by lowering the duty cycle of typical instruments on gliders. All of these instruments involve trade-
considerably; that is, most of the time sensors are offs between capability, sampling rate, and mission
turned off and only briefly are energized for sam- endurance.
pling. The low-power sleep current of the micro-
processor is a significant constraint on mission
endurance.
Oceanic fields that are readily measured elec-
Missions
tronically are well suited to gliders. Temperature, The track of a Seaglider in the Labrador Sea
salinity, and pressure are readily measured elec- (Figure 3) illustrates the ability of a glider to make
tronically at little energy cost. In addition to scientific transects in regions where depth-averaged flow is
146 GLIDERS

65° W 60° W 55° W 50° W 45° W

70
0

GPS fix and iridium call

60
400 40
05

0
00

20 cm s−1 current

60 5
0 00
upper 1-km average

200
10
0

300
0
20
0
20

100
100

65° N Greenland 65° N

600

400
100
300 0
70
800

200
300 0 300
90 Nuuk
20
0
1000

2000
0
40

6080
04 0
00

00
5

5070
00
30
0

90
0
30
0

10

100
00
00
30

600 0
20
20
300

00

60° N 200

60° N
450

00
60
00080

700
30
91
000
0000

100
200
0

30
00
20
30

0
0

0
40

50
0
20
55° N 0 55° N
0
30

65° W 60° W 55° W 50° W 45° W

Figure 3 Track of a Seaglider in the Labrador Sea from 24 Sep. 2004 to 29 Apr. 2005, the longest AUV mission to date. This vehicle
was launched from R/V Knorr in Davis Strait and recovered more than 7 months later offshore Nuuk, Greenland. Pink symbols mark
the location of GPS fixes and Iridium calls between dive cycles. Blue arrows indicate depth-averaged currents (to the shallower of
bottom depth and 1 km), where the scale length is typical glider speed through the water. Depth contour intervals are 100–1000 m,
500 m thereafter with heavy contours every 1 km.
 GLIDERS 147

sufficiently weak, but be hampered in its ability to courier for data recorded by these devices, and to
navigate within strong currents. Launched from a control them.
research vessel offshore west Greenland, this vehicle
made two transects across Davis Strait before being
sent south to Hamilton Bank offshore southeast Further Reading
Labrador. It was swept across part of the continental
shelf region before escaping the Labrador Current to Davis RE, Eriksen CC, and Jones CP (2003) Autonomous
continue northeastward on its return to Greenland buoyancy-driven underwater gliders. In: Griffiths G (ed.)
near Kap Desolation. On its way north toward re- Technology and Applications of Autonomous Underwater
covery, it was caught in an anticyclonic eddy that Vehicles, pp. 37--58. New York: Taylor and Francis.
Rudnick DL, Davis RE, Eriksen CC, Fratantoni DM, and
drifted offshore. This glider was directed mainly
Perry MJ (2004) Underwater gliders for ocean research.
northward as it cycled around the eddy core four
Journal of the Marine Technology Society 38: 73--84.
times before it finally escaped by being guided radi- Stommel H (1989) The Slocum mission. Oceanography 2:
ally outward across the eddy. It continued north- 22--25.
eastward to Fyllas Bank, offshore Nuuk, Greenland,
where it was recovered from a fishing vessel, having Relevant Websites
completed the longest AUV mission to date as of this
writing, lasting 217 days, traveling 3750 km through http://iop.apl.washington.edu
the water, and making 663 dives, the majority to – Operations Summary: Custom View, Seaglider Status,
1000-m depth. Integrative Observational Platforms, Ocean Physics
Several different types of glider missions have been Department, Applied Physics Laboratory, University of
Washington.
undertaken, of which the example in Figure 3 is just
http://spray.ucsd.edu
one: a solitary glider survey. In other applications, – SIO IDG Spray Home.
gliders have been used to make repeat surveys along http://www.webbresearch.com
the same track, controlled to maintain position – Slocum Glider, Webb Research Corporation.
(a ‘virtual mooring’), used in numbers to survey a http://seaglider.washington.edu
region intensively, used to interact with one another, – The Seaglider Fabrication Center of the University of
to communicate with moored sensors, act as a Washington.
 MANNED SUBMERSIBLES, DEEP WATER
H. Hotta, H. Momma, and S. Takagawa, History of Deep Submersibles
Japan Marine Science & Technology Center, Japan
The first modern deep diving by humans, to a depth
Copyright & 2001 Elsevier Ltd.
of 923 m, was achieved in 1934 by William Beebe, an
American zoologist, and Otis Burton using the
bathysphere, which means ‘deep sphere’. The
bathysphere was a small spherical shell made of cast
iron, 135 cm in inside diameter designed for two
observers. The bathysphere had an entrance hatch
Introduction and a small glass view port. As the sphere was low-
ered by a cable and lacked thrusters, it was im-
Deep-ocean underwater investigations are much
possible to maneuver.
more difficult to carry out than investigations on
The next advance, using a free-swimming vehicle,
land or in outer space. This is because electro-
occurred after World War II, in 1947. The bath-
magnetic waves, such as light and radio waves, do
yscaph FNRS II was invented by Auguste Piccard,
not penetrate deep into sea water, and they cannot be
who had been studying cosmic rays using a manned
used for remote sensing and data transmission.
balloon in Switzerland. The principle of the bath-
Moreover, deep-sea underwater environments are
yscaph was the same as that of a balloon. Instead of
physically and physiologically too severe for humans
hydrogen or helium gas, gasoline was used as the
to endure the high pressures and low temperatures.
buoyant material. During descent, air ballast tanks
First of all, pressure increases by 1 atmosphere for
were filled with sea water, and for ascent, iron shot
every 10 meters depth because the density of water is
ballast was released. The pressure hull was made of
1000 times greater than that of air. Furthermore, as
we have no gills we can not breathe under water. drop-forged iron hemispheres, 2 m in inside diameter
Water temperature decreases to 11C or less in the and 90 mm in thickness, allowing for two crew
deep sea and there is almost no ambient light at members. It was able to maneuver around the sea-
depth because sunlight can not penetrate through floor by thrusters driven by electric motors. Later, the
more than a few hundred meters of sea water. These second bathyscaph, Trieste, was sold to the US Navy,
are several of the reasons why we need either and independently at the same time, the French Navy
manned or unmanned submersibles to work in the developed the bathyscaph FNRS III, and later
deep sea. Archimede. In 1960, the Trieste made a dive into the
A typical manned submersible consists of four Challenger Deep in the Mariana Trench, to a depth
major components: a pressure hull, propellers of 10 918 m. This historic dive was conducted by
(thrusters), buoyant materials, and observational Jacques Piccard, son of Auguste Piccard, and Don
instruments. The pressure hull is a spherical shell Walsh from the US Navy. The bathyscaph was the
made of high-strength steel or titanium. The typical first generation of deep-diving manned submersibles.
internal diameter of the hull is approximately 2 m, It was very big and slow as it needed more than 100
which allows up to three people to stay at one at- kiloliter capacity gasoline tanks to provide flotation
mosphere for 8–12 h during underwater operations. for the 2 m diameter pressure hull.
In case of emergency, a life-support system enables a In 1964, the second generation of deep sub-
stay of three to five days. Several thrusters are usually mersibles began. Alvin was funded by the US Navy
installed on the body of the submersible to give under the guidance of the Woods Hole Oceano-
maneuverability. The buoyant material is syntactic graphic Institution (WHOI). At first, its depth cap-
foam, which is made of glass microballoons and an ability was only 1800 m. It was small enough to be
adhesive matrix. Its specific gravity is approximately able to put on board the R/V Lulu, which became its
0.5 gf ml1. Observational instruments such as cam- support ship. Instead of gasoline flotation, syntactic
eras, lights, sonar, CTD (conductivity, temperature, foam was used. Alvin had horizontal and vertical
and depth sensors) etc., are also very important for thrusters to maneuver freely in three dimensions.
gathering information on the deep-sea environment. Scientific instruments, including manipulators, cam-
It should be mentioned that the power consumption eras, sonar and a navigation system, were installed.
of the lights can reach as much as 15% of the total Three observation windows were available for the
power consumption of the submersible. three crew members. In France, the two-person 3000

148
 Vent air and
fill sea water in ballast tank to
descend

Variable ballast tank

(Air and sea water)
Drop weights

Sea water
Blow out sea water from
ballast tank
Descent Ascent

Jettison drop weights partially Jettison all the drop weight to ascend
for neutral buoyancy

Sea water

Weight control by variable ballast tank

(Air and sea water)

Figure 1 Principle of descent and ascent for a modern deep submersible.

MANNED SUBMERSIBLES, DEEP WATER
149
150 MANNED SUBMERSIBLES, DEEP WATER

m-class submersible Cyana was built. These two by the increasing pressure. Thus the submersible
vehicles typified submersibles during the 1960s. At slows slightly as it dives deeper (Figure 2).
present, the depth capability of the Alvin has been When the submersible approaches the seafloor
increased to 4500 m by replacing the high-strength (50–100 m in altitude, i.e., height above the bottom),
steel pressure hull with a titanium alloy sphere in a portion of its ballast (usually lead or some other
1973. In the 1980s, 6000 m-class submersibles, such heavy material) is jettisoned to achieve neutral
as the Nautile from France, the Sea Cliff of the US buoyancy. Perfect neutral buoyancy occurs when the
Navy, the Mir I and Mir II from Russia and the positively buoyant materials (things which tend to
Shinkai 6500 from Japan, were built. They were float) on the submersible balance the negatively
theoretically able to cover more than 98% of the buoyant materials (things which tend to sink). This
world’s ocean floor. allows the vehicle to hover weightless in position and
What will the third generation of deep sub- move freely about. As perfect neutral buoyancy is
mersibles be like? Manned submersibles of the third difficult to maintain, most submersibles have aux-
generation, which would be capable of exceeding iliary weight-adjusting (trim and ballast) systems.
10 000 m depth, have not yet been developed at the This consists of a sea-water pumping system to draw
time of this report. One possibility is a small and in or expel water, thus adjusting the buoyancy of the
highly maneuverable one- or two-person submersible submersible.
with a transparent acrylic or ceramic pressure hull. Upon completing its mission, the remaining ballast
Another possibility is a deep submergence labora- is jettisoned and the submersible now with positive
tory, which would be able to carry several scientists buoyancy begins ascending. When resurfaced, air
and crew long distances and long durations without from a high-pressure bottle is blown into the air
the assistance of a mother ship. This would be the ballast tank to give enough draft to the submersible
realization of the dream like ‘Nautilus’ in 20 000 for the recovery operation.
Leagues Under The Sea by French novelist Jules
Verne. Strong scientific and/or social goals would be
Water Pressure
needed for such a submersible design to be pursued.
And there is a third possibility that the next gener- Water pressure increases by 0.1 MPa per 10 m depth.
ation will be evolutionary upgrades of existing Thus every component sensitive to pressure must be
second-generation submersibles. isolated from intense pressure changes. First and
foremost are the passengers which are protected
against great ambient pressure by a pressure hull or
Principles of Modern Submersibles pressure vessel, maintained at surface pressure. The
Descent and Ascent
ambient pressure exerts strong compressional force
on the pressure hull which is therefore designed to
There are several methods to submerge vehicles into avoid any tensile stress. The strongest geometric
the deep sea. The simplest way is to suspend a sphere
by a cable, known as a bathysphere. Mobility,
)

1.08
_1

however, is greatly limited. A second method relies

Water density (gf ml

on powerful thrusters to adjust vertical position in

relatively shallow water. The submersible Deep 1.05

Flight is a high-speed design which uses thrust power

coupled with fins for motion control like the wings of 1.02 Water pressure (MPa)
a jet fighter. It descends and ascends obliquely in the 50 100
water column at speeds up to 10 knots. Most sub- 0

mersibles employ a third method that, while using Water pressure is affected by
weak thrusters to control attitude and horizontal compression of sea water
Water depth (m)

movement, relies principally on an adjustable buoy-

ancy system for descent and ascent (Figure 1). 5000

When on the surface, the submersible’s air ballast Simple calculation

of water pressure
tank is filled with air creating positive buoyancy, (proportional to water depth)
hence it floats. When the dive begins, air is vented
10 000
from the ballast tank and filled with sea water, thus
creating negative buoyancy and sinking the vehicle.
As the submersible dives deeper, buoyancy increases Figure 2 Relations between water depth, pressure and water
modestly due to the increasing water density created density at a water temperature of 01C and salinity of 34.5%.
 MANNED SUBMERSIBLES, DEEP WATER 151

shape against outside pressure relative to volume and pressure compensation systems. Technology is being
hence weight is a sphere, followed by a cylinder developed to apply ambient pressure to electronic
(capped at both ends). However, it is not easy to devices such as integrated circuits (ICs) and large
arrange instruments inside a sphere effectively. scale ICs (LSIs).
In order to increase mobility, it is important to
make submersibles small and light. The pressure hull Buoyancy
is one of the largest and heaviest components of the
submersible. The hull must be as small (and light) as With the exception of some shallow-water sub-
possible, while affording appropriate strength mersibles, the total weight of the essential systems is
against external pressure. Thus for deep-diving sub- larger than the total buoyancy. This means that extra
mersibles, a spherical pressure hull is employed buoyancy is needed to balance the excess weight.
whereas shallower vehicles can use a cylindrical Wood or foam-rubber cannot be used for this pur-
shape if so desired. pose because they shrink under increasing water
The material used for the pressure hull is critical. pressure. The material providing buoyancy must
In earlier vehicles, steel was used. Later, titanium have a relatively small specific gravity while re-
alloy was the material of choice. Titanium alloy has maining strong under high-pressure conditions.
very high tensile strength, and is resistant to cor- Historically, gasoline was used to provide buoy-
rosion and relatively light (specific gravity B60% ancy in bathyspheres as it did not lose buoyancy
that of steel). Recently, the trend in submersible under pressure. However, its specific gravity was too
construction is to use nonmetallic materials, such as large for practical use – huge volumes are needed to
fiber- or graphite-reinforced plastics (FRP or GRP), offset the weight. With the invention of syntactic
or ceramics. foam, a superior material for deep-diving sub-
Components not sensitive to pressure or saline mersibles became available. Syntactic foam consists
conditions need no special consideration. Though of tiny microscopic spheres of glass embedded in
those devices which require electrical insulation resin. These microballoons are 40–200 mm in diam-
need to be housed in oil-filled compartments called eter, and are closely packed with resin filling in the
oil-filled pressure compensation systems (Figure 3). surrounding spaces. Proper selection of the balloons
These systems do not require heavy pressure hulls and resin allows the proper pressure tolerance and
and thus reduce the weight of the submersible specific gravity to be created. For example, the syn-
overall. Electric motors, hydraulic systems, batteries, tactic foam used by the Shinkai 6500 is tolerant up to
wiring, and power transistors are all housed in 130 MPa with a specific gravity of 0.54 gf ml1 and
the foam used by the ROV Kaiko is tolerant up to
160 MPa with a specific gravity of 0.63 gf ml1.
Pressure resistant housing (thick wall)

Rotary seal Life Support

(high friction)
The pressure hull is a very small space where crew
members must stay for up to 20 h, depending on their
Electric motor
. . . . mission. Since the pressure hull is maintained at
ambient pressure, no decompression of the occu-
pants is needed. High-pressure oxygen bottles pro-
vide oxygen within the pressure hull, while carbon
scrubbers absorb carbon-dioxide (Figure 4). Extra
Pressure compensated housing (thin wall) life-support is required with varying standards de-
Rotary seal
(low friction) pending on the country (from 32 h to 120 h)

Energy
Electric motor
. . . Energy for deep-diving submersibles is supplied by
rechargeable (secondary) batteries. There are several
types including: lead acid, nickel cadmium, nickel
hydrogen, oxidized silver zinc. Batteries which con-
tain a higher density of energy are preferable to reduce
Oil bladder (for compensation of pressure and the oil volume)
the weight and volume of the submersible. However,
Figure 3 Pressure resistant and pressure compensated such batteries are very expensive. These batteries are
housings for electric motors. housed in oil-filled pressure-compensated systems to
152 MANNED SUBMERSIBLES, DEEP WATER

white light must be considered. Thus, for example,

engineers must consider both the compactness of
Valve video cameras and their color sensitivity.

Oxygen bottle Manipulator

Pressure hulls, thrusters, batteries, and buoyancy
materials are all essential parts of the modern sub-
Fan
mersible. One of the most important tools is the
manipulator arm. Manipulators extend the arms and
hands of the pilot, allowing sample collection and
deployment of experimental equipment. Most ma-
Desiccant CO2
nipulators are driven by hydraulic pressure. The
absorbent
most advanced manipulators operate in a master-
slave system. The operator handles a master arm
Figure 4 Life support system for a deep submersible. (controller) which imitates human arm and hand
movements, and the motions are translated to the
slave-unit (manipulator) which follows precisely the
motions of the master arm. There are usually one or
two manipulators on a research submersible.
Video camera
Navigation
Underwater navigation is one of the most crucial
elements of deep-sea submersible researches. Usually,
long base line (LBL) or super short base line (SSBL)
acoustic navigation systems are used depending on
the accuracy required. The positioning error of LBL
systems is 5–15 m, and it is approximately 2–5% of
slant range for SSBL systems. An advantage of SSBL
systems is that seafloor transponders are not neces-
Video camera sary, whereas at least two seafloor transponders are
necessary for LBL systems. In both systems, absolute
or geodetic position is determined by surface navi-
gation systems, such as Differential Global Pos-
itioning System (DGPS). Although the position of the
Figure 5 Examples of inside alignments of the pressure case. submersible is usually reported from the mother ship
by voice, the Shinkai 6500 has an automatic LBL
navigation system.
reduce weight. Recent developments in fuel cell
technology offer the promise of higher density energy Surface Support (the Mother Ship)
coupled with higher efficiency. This has great poten-
tial for submersible applications. The R/V Lulu, the original support ship of the sub-
mersible Alvin, was retired and replaced by the R/V
Atlantis II in 1983. In 1997 Atlantis II was replaced
Instrumentation
by the newly commissioned Research Vessel Atlantis.
There are many scientific and observational instru- The support ship not only supports the diving oper-
ments employed on research submersibles. Due to ation, including launch, recovery, communication,
the limited payload, these instruments must be as and positioning, but also provides a place to conduct
light and small as possible. For example, bulky on-board research during a cruise. Accordingly, re-
camera bodies must be streamlined and aligned with search laboratories, computers, instruments for on-
lenses in a compact manner, thus reducing the size board data analysis, multi-narrow beam echo
and weight of their pressure housing (see Figure 5). sounders, etc., are necessary. Also, remotely operated
Furthermore, physical conditions such as extremely vehicles (ROVs) or autonomous underwater vehicles
cold temperatures or the differential absorption of (AUVs), can be operated during the nighttime, or in
 MANNED SUBMERSIBLES, DEEP WATER 153

case the manned submersible cannot be operated transponders. Still video images are transmitted to
because of bad weather. the support ship through the acoustic link.

Sea Cliff (USA)

Deep Submersibles in the World Sea Cliff was originally built in 1968 as a 3000 m-
Alvin (USA) class submersible by General Dynamics Corp. for the
US Navy as a sister submersible for the Turtle. In
Alvin (Figure 6) was built in 1964 by Litton Indus- 1985, the Sea Cliff was converted into a 6000 m-
tries with funds of the US Navy and operated by class deep submersible. It is 7.9 m long, 3.7 m wide,
WHOI. Its original depth capability was 1800 m 3.7 m high and weighs 23 tf in air. (Sea Cliff and
with a steel pressure hull. Later, the pressure hull was Turtle are currently out of commission.)
replaced by titanium alloy to increase depth cap-
ability up to 4500 m. Alvin’s size and weight are: Mir I and Mir II (Russia)
length 7.1 m; width 2.6 m; height 3.7 m and weight
17 tf in air. The outside diameter of the pressure hull The 6000 m-class submersibles Mir I and Mir II
is 2.08 m with a wall thickness of 49 mm. It is (Figure 8) were built in 1987 by Rauma Repola in
equipped with three view ports, 120 mm in inside Finland for the P.P. Shirshov Institute of Oceanology
diameter, two manipulators with seven degrees of in Russia. They are 7.8 m long, 3.8 m wide, 3.65 m
freedom. The original catamaran support ship, R/V high and weigh 18.7 tf in air. Inside diameter of the
Lulu, was replaced by the R/V Atlantis II. Launch pressure hull, which is made of high-strength steel, is
and recovery take place at the stern A-frame. It has 2.1 m with a wall thickness of 40 mm. Launch and
been the leading deep submersible in the world. recovery take place by an articulated crane over the
side of the support ship, the R/V Academik Mistilav
Keldysh. If necessary, both Mir I and Mir II are
Nautile (France) launched simultaneously to carry out cooperative or
independent research. Another characteristic feature
Nautile (Figure 7) was built in 1985 by IFREMER
of the Mir is a powerful secondary battery, 100 kWh
(French Institution for Marine Research and Devel-
of total energy, which allows it to stay more than 20
opment) in France. Depth capability of 6000 m was
hours underwater, or to carry out more than 14 h of
aimed to cover 98% of the world’s ocean floor. It is
continuous operation on the bottom.
8 m long, 2.7 m wide, 3.81 m in height and weighs
19.3 tf in air. It is equipped with three view ports, a
Shinkai 6500 (Japan)
manipulator, a grabber and a small companion ROV
Robin. The position of the submersible is directly The Shinkai 6500 (Figure 9) was built in 1989 by
calculated by interrogating the seafloor Mitsubishi Heavy Industries and operated by the

Figure 6 US submersible Alvin.

154 MANNED SUBMERSIBLES, DEEP WATER

Figure 8 Russian 6000 m-class submersible Mir I or Mir II.

Figure 7 French 6000 m-class submersible Nautile.

Japan Marine Science Technology Center (JAM- Cape Cod in New England. After an extensive search
STEC). It is 9.5 m long, 2.71 m wide, 3.21 m high for the submarine, the bathyscaph Trieste made dives
and weighs 25.8 tf in air. The pressure hull is made of to inspect the wreck in detail and recover small ob-
titanium alloy, 73.5 mm in thickness, and has an jects. The operation demonstrated the importance of
inside diameter of 2 m. It is equipped with three view using deep submersibles and advanced deep ocean
ports, two manipulators with seven degrees of free- technology to increase knowledge of the deep ocean.
dom. Position of the submersible is calculated and In 1966, hydrogen bombs were lost with a downed
displayed in real time by directly interrogating the US B-52 bomber off Palomares, Spain. The Alvin
seafloor transponders. Still color video images are showed the great utility of deep submersibles by lo-
transmitted automatically at 10 s intervals to the cating and assisting in the recovery of lost objects
support ship, the R/V Yokosuka, through the from the sea.
acoustic link during the diving operation. Launch Between 1973 and 1974, project FAMOUS
and recovery take place at the stern A-frame of the (French–American Mid-Ocean Undersea Study) was
R/V Yokosuka. conducted in the Mid-Atlantic Ridge off the Azores
using the French bathyscaph Archimede and the US
submersible Alvin. The project was the first system-
Major Contributions of Deep atic and successful use of deep submersibles for sci-
ence. They discovered and sampled fresh pillow lavas
Submersibles and lava flows at 3000 m deep in the rift valley,
The dive to the Challenger Deep in the Mariana where the oceanic crusts were being created, pro-
Trench by the bathyscaph Trieste in 1960 was one of viding visual evidence of Plate Tectonics. In 1977,
the most spectacular achievements of the twentieth Alvin discovered a hydrothermal vent and vent ani-
century. However, the dive was mainly for adventure mals in the East Pacific Rise off the Galapagos Is-
rather than for science. In 1963, the US nuclear lands at a depth of 2450 m. Discovery of these
submarine Thresher sank in 2500 m of water off chemosynthetic animals, which were not dependent
 MANNED SUBMERSIBLES, DEEP WATER 155

Figure 9 Japanese 6000 m-class submersible Shinkai 6500.

on photosynthesis, had a profound impact on biol- Further Reading

ogy in the twentieth century.
Manned submersibles now compete with un- Beebe W (1934) Half Miles Down. New York: Harcourt
manned submersibles, such as ROVs and AUVs. Brace.
Because of the expense of operation and mainten- Busby RF (1990) Undersea Vehicles Directory – 1990–91,
4th edn. Arlington, VA: Busby Associates.
ance, national funding is necessary for manned sub-
Funnel C (ed.) (1999) Jane’s Underwater Technology, 2nd
mersibles. However, ROVs and AUVs can be edn. UK: Jane’s Information Group Limited.
operated by private companies or institutions. In Kaharl VA (1990) Water Baby – The Story of Alvin. New
spite of the costs, the ability of the human observer York: Oxford University Press.
to rapidly process information to make decisions Piccard A (1956) Earth Sky and Sea. New York: Oxford
provides an advantage and justifies continued use of University Press.
manned submersibles. Piccard J and Dietz RS (1961) Seven Miles Down. New
York: G.P. Putnam.
See also
Deep Submergence: Science of. Manned Sub-
mersibles, Shallow Water. Moorings. Remotely
Operated Vehicles (ROVs).
 MANNED SUBMERSIBLES, SHALLOW WATER
T. Askew, Harbor Branch Oceanographic permitted divers to walk around outside. In 1776,
Institute, Ft Pierce, FL, USA Dr David Bushnell built and navigated the first sub-
Copyright & 2001 Elsevier Ltd. marine employed in war-like operations. Bushnell’s
Turtle was built of wood, egg-shaped with a conning
tower on top and propelled horizontally and ver-
tically by a primitive form of screw propeller after
flooding a tank which allowed it to submerge.
Introduction In the early 1800s, Robert Fulton, inventor of
the steamship, built two iron-formed copper-clad
Early man’s insatiable curiosity to look beneath the submarines, Nautilus and Mute. Both vehicles carried
surface of the sea in search of natural treasures that out successful tests, but were never used operationally.
were useful in a primitive, comfortless mode of living The first ‘modern submersible’ was Simon Lake’s
were a true test of his limits of endurance. The fragile Argonaut I, a small vehicle with wheels and a bottom
vehicle of the human body quickly discovered that hatch that could be opened after the interior was
most of the sea’s depths were unapproachable with- pressurized to ambient. While there are numerous
out some form of protection against the destructive other early submarines, the manned submersible did
hostilities of the ocean. not emerge as a useful and functional means of ac-
Modern technology has paved the way for man to complishing underwater work until the early 1960s.
conquer the hostile marine environment by creating It was during this same period of time that the
a host of manned undersea vehicles. Called sub- French-built Soucoupe, sometimes referred to as the
mersibles, these small engineering marvels carry out ‘diving saucer’, came into being. Made famous by
missions of science, exploration, and engineering. Jacques Cousteau on his weekly television series, the
The ability to conduct science and other operations Soucoupe is credited with introducing the general
under the sea rather than from the surface has population to underwater science. Launched in
stimulated the submersible builder/operator to fur- 1959, the diving saucer was able to dive to 350 m.
ther develop the specialized tools and instruments The USS Thresher tragedy in 1963 appears to have
which provide humans with the opportunity to be spurred a movement among several large corpor-
present and perform tasks in relative comfort in ations such as General Motors (Deep Ocean Work
ocean bottom locations that would otherwise be Boat, DOWB), General Dynamics (Star I, II, III, Sea
destructive to human life. Cliff, and Turtle), Westinghouse (Deepstar 2000,
Over the past fifty years a depth of 1000 m has 4000), and General Mills (Alvin), along with nu-
surfaced as the transition point for shallow vs deep merous other start-up companies formed solely to
water manned submersibles. During the prior 100 manufacture submersibles. Perry Submarine Build-
years any device that enabled man to explore the ers, a Florida-based company, started manufacturing
ocean depths beyond breath-holding capabilities small shallow-water, three-person submersibles in
would have been considered deep. 1962, and continued until 1980 (Figures 1 and 2).
International Hydrodynamics Ltd (based in Van-
couver, BC, Canada) commenced building the Pisces
History series of submersibles in 1962. The pressure hull
While it is difficult to pinpoint the advent of the first material in the 1960s was for the most part steel with
submersible it is thought that in 1620 Cornelius van one or more view ports. The operating depth ranged
Drebel constructed a vehicle under contract to King from 30 m to 600 m, which was considered very deep
James I of England. It was operated by 12 rowers for a free-swimming, untethered vehicle.
with leather sleeves, waterproofing the oar ports. It is The US Navy began design work on Deep Jeep in
said that the craft navigated the Thames River for 1960 and after 4 years of trials and tribulations it
several hours at a depth of 4 m and carried a secret was commissioned with a design depth of 609 m.
substance that purified the air, perhaps soda lime? A two-person vehicle, Deep Jeep included many
In 1707, Dr Edmund Halley built a diving bell features incorporated in today’s submersibles, such
with a ‘lock-out’ capability. It had glass ports above as a dropable battery pod, electric propulsion motors
to provide light, provisions for replenishing its air, that operate in silicone oil-filled housings, and
and crude umbilical-supplied diving helmets which shaped resin blocks filled with glass micro-balloons

156
 MANNED SUBMERSIBLES, SHALLOW WATER 157

Figure 1 Perry-Link Deep Diver, 1967. Owned by Harbor Branch Oceanographic Institution. Length 6.7 m, beam 1.5 m, height
2.6 m, weight 7485 kg, crew 1, observers 2, duration 3–5 hours.

used to create buoyancy. Deep Jeep was eventually unveiled and is still operating today with well over
transferred to the Scripps Institution of Ocean- 3000 dives logged.
ography in 1966 after a stint searching for a lost ‘H’ A few of the submersibles were designed with a
bomb off Palomares, Spain. Unfortunately, Deep diver ‘lock-out’ capability. The first modern vehicle
Jeep was never placed into service as a scientific was the Perry-Link ‘Deep Diver’ built in 1967 and
submersible due to a lack of funding. The missing able to dive to 366 m. This feature enabled a separate
bomb was actually found by another vehicle, Alvin. compartment carrying divers to be pressurized in-
Alvin did get funding and proved to be useful as a ternally to the same depth as outside, thus allowing
scientific tool. the occupants to open a hatch and exit where they
The Nekton series of small two-person sub- could perform various tasks while under the super-
mersibles appeared in 1968, 1970, and 1971. The vision of the pilots. Once the work was completed, the
Alpha, Beta, and Gamma were the brainchildren of divers would re-enter the diving compartment, closing
Doug Privitt, who started building small sub- the outer and inner hatches; thereby maintaining the
mersibles for recreation in the 1950s. The Nektons bottom depth until reaching the surface, where they
had a depth capability of 304 m. The tiny sub- could decompress either by remaining in the com-
mersibles conducted hundreds of dives for scientific partment or transferring into a larger, more comfort-
purposes as well as for military and oilfield cus- able decompression chamber via a transfer trunk.
tomers. In 1982, the Nekton Delta, a slightly larger Acrylic plastic was tested for the first time as a new
submersible with a depth rating of 365 m was material for pressure hulls in 1966 by the US Navy.
158 MANNED SUBMERSIBLES, SHALLOW WATER

Figure 2 Perry PC 1204 Clelia, 1976. Owned and operated by Harbor Branch Oceanographic Institution. Length 6.7 m, beam 2.4 m,
height 2.4 m, weight 8160 kg, crew 1, observers 2, duration 3–5 hours.

The Hikino, a unique submersible that incorporated impractical. The deeper diving vehicles utilize various
a 142 cm diameter and a 0.635 cm thick hull, was drop weight methods; most use two sets of weights
only able to dive to 6 m. This experimental vehicle (usually scrap steel cut into uniform blocks). One set
was used to gain experience with plastic hulls, which of weights is released upon reaching the bottom,
eventually led to the development of Kumukahi, allowing the vehicle to maneuver, travel, and perform
Nemo (Naval Experimental Manned Observatory), tasks in a neutral condition. The other set of weights
Makakai, and Johnson-Sea-Link. is dropped to make the vehicle buoyant, which carries
The Kumukahi, launched in 1969, incorporated a it back to the surface once the dive is complete.
unique 135 cm acrylic plastic sphere formed in four The shallow vehicles use their thrusters and/or
sections. It was 3.175 cm thick and could dive to water ballast to descend to the bottom and some of
92 m. the more sophisticated submersibles have variable
The Nemo, launched in 1970, and Makakai, ballast systems which allow the pilot to achieve a
launched in 1971, both utilized spheres made of 12 neutral condition by varying the water level in a
curved pentagons formed from a 6.35 cm flat sheet of pressure tank.
PlexiglasTM. The pentagons were bonded together to One advantage of the shallow vehicles is that the
make one large sphere capable of diving to 183 m. view ports (commonly called windows) can be much
The Johnson-Sea-Link, designed by Edwin Link larger both in size and numbers, and where an acrylic
and built by Aluminum Company of America plastic sphere is used the entire hull becomes a
(ALCOA), utilized a Nemo-style PlexiglasTM sphere, window.
167.64 cm in diameter, 10.16 cm thick, and made of Since the late 1960s and early 1970s acrylic plastic
12 curved pentagons formed from flat sheet and pressure hulls have emerged as an ideal engineering
bonded together. This new thicker hull had an op- solution to create a strong, transparent, corrosion-
erational depth of 304 m. resistant, nonmagnetic pressure hull. The limiting
factor of the acrylic sphere is its ability to resist im-
plosion from external pressure at great depths. Its
Present Day Submersibles
strength comes from the shape and wall thickness.
The submersibles currently in use today are for the Therefore, the greater the depth the operator aims to
most part classified as either shallow-water or deep- reach, the thicker the sphere must be, which results
water vehicles, the discriminating depth being ap- in a hull that is much too heavy to be practical for
proximately 1000 m. This is where the practicality use on a small submersible designed to go deeper
of using compressed gases for ballasting becomes than 1000 m.
 MANNED SUBMERSIBLES, SHALLOW WATER 159

These shallow-water submersibles, once quite nu- shaped submarine. Attached to the frame are the
merous because of their usefulness in the offshore vessel’s ballast tanks, thrusters, compressed air,
oilfield industry, are now limited to a few operators mixed gas flasks, and battery pod.
and mostly used for scientific investigations. The aluminum alloy parts of the submersible,
The Johnson-Sea-Links (J-S-Ls) stand out as two lightweight and strong, along with the acrylic cap-
of the most advanced manned submersibles (Figure 3). sule which was patterned after the prototype used by
J-S-L I, commissioned by the Smithsonian Institution the US Navy on the Nemo, had extraordinary ad-
in January 1971, was named for designer and donors vantages over traditional materials like steel. They
Edwin A. Link and J. Seward Johnson. were most of all immune to the corrosive effects of
Edwin Link, responsible for the submersible’s sea water.
unique design and noted for his inventions in the The emphasis in engineering of the submersible
aviation field, turned his energies to solving the was on safety. Switches, connectors, and all
problems of undersea diving, a technology then still operating gear were especially designed to avoid
in its infancy. One of his objectives was to carry out possible safety hazards. The rear diving compart-
scientific work under water for lengthy periods. ment allows one diver to exit for scientific collections
The Johnson-Sea-Link was the most sophisticated while tethered for communications and breathing air
diving craft he had created for this purpose, and it supply, while the other diver/tender remains inside
promised to be one of the most effective of the new as a safety backup. Once the dive is completed and
generation of small submersible vehicles that were the submersible is recovered by a special deck-
being built to penetrate the shallow depths of the mounted crane on the support ship, the divers can
continental shelf (183 m or 100 fathoms). transfer into a larger decompression chamber via a
Originally designed for a depth of 304 m, the transfer trunk which is bolted to the lock-out/lock-in
vessel’s unique features include a two-person trans- compartment.
parent acrylic sphere, 1.82 m in diameter and Now, 30 years later, the Johnson-Sea-Links with a
10.16 cm thick, that provides panoramic underwater 904 m depth rating, remain state of the art under-
visibility to a pilot and a scientist/observer. Behind water vehicles. Sophisticated hydraulic manipulators
the sphere, there is a separate 2.4 m long cylindrical, work in conjunction with a rotating bin collection
welded, aluminum alloy lock-out/lock-in compart- platform which allow 12 separate locations to be
ment that will enable scientists to exit from its bot- sampled and simultaneously documented by digital
tom and collect specimens of undersea flora and color video cameras mounted on electric pan and tilt
fauna. The acrylic sphere and the aluminum cylinder mechanisms and aimed with lasers. Illumination is
are enclosed within a simple jointed aluminum tub- provided by a variety of underwater lighting systems
ular frame, a configuration that makes the vessel utilizing zenon arc lamps, metal halide, and halogen
resemble a helicopter rather than a conventionally bulbs. Acoustic beacons provide real time position

Figure 3 Johnson-Sea-Link I and II, 1971 and 1976. Owned and operated by Harbor Branch Oceanographic Institution. Length
7.2 m, beam 2.5 m, height 3.1 m, weight 10400 kg, crew 2, observers 2, duration 3–5 hours.
160 MANNED SUBMERSIBLES, SHALLOW WATER

and depth information to shipboard computer including mapping a large area of the right solid
tracking systems that not only show the sub- rocket booster debris at a depth of 365 m. The ve-
mersible’s position on the bottom, but also its rela- hicles proved their worth throughout the operation
tionship to the ship in latitude and longitude via the by consistently performing beyond expectations.
satellite-based global positioning system (GPS). The They were launched and recovered easily and
lock-out/lock-in compartment is now utilized as an quickly. They could work on several contacts per
observation and instrumentation compartment, day, taking NASA engineers to the wreckage for first-
which remains at one atmosphere. hand detailed examination of debris while video
Today’s shallow-water submersibles (average dive cameras recorded what was being seen and said.
3–5 h) require a support vessel to provide the ne- Significant pieces were rigged with lifting bridles for
cessities that are not available due to their relatively recovery. The autonomous operation of the J-S-Ls, a
small size. The batteries must be charged, com- dedicated support vessel, and highly trained oper-
pressed air and oxygen flasks must be replenished. ations personnel made for a successful conclusion to
Carbon dioxide removal material, usually soda lime an operation that had a significant impact on the
or lithium hydroxide, is also replenished so as to future of the US Space Program.
provide maximum life support in case of trouble.
Most submersibles today carry 5 days of life support,
which allows time to effect a rescue should it become
Summary
necessary.
The support vessel also must have a launch/re- There is no question that the manned submersible
covery system capable of safely handling the sub- has earned its place in history. Much of what are
mersible in all sea conditions. Over the last 30 years, now cataloged as new species were discovered in the
the highly trained crews that operate the ship’s last 30 years with the aid of submersibles. The ability
handling systems and the submersibles, have virtu- to conduct marine science experiments in situ led to
ally made the shallow-water submersibles an every- the development of intricate precision instruments,
day scientific tool where the laboratory becomes the sampling devices for delicate invertebrates and gel-
ocean bottom. atinous organisms that previously were only seen in
blobs or pieces due to the primitive methods used to
collect them.
While some suggest that remotely operated ve-
Operations
hicles (ROVs) could, and have replaced the manned
The two Johnson-Sea-Links have accumulated over submersible, in reality they are complementary.
8000 dives for science, engineering, archaeology, and There is no substitute for the autonomous, highly
training purposes since 1971. They have developed maneuverable submersible that can approach and
into highly sophisticated science tools. Literally collect without contact delicate zooplankton, while
thousands of new species of marine life have been observing behavior and measuring the levels of
photographed, documented by video camera and bioluminescence, or probing brine pools and cold
collected without disturbing the surrounding habitat. seep regions in the Gulf of Mexico for special-
Behavioral studies of fish, marine mammals and in- ized collections of biological, geological, and geo-
vertebrates as well as sampling of the water column chemical samples. Tubeworms are routinely marked
and bottom areas for chemical analysis and geological for growth rate studies and collected individually,
studies are everyday tasks for the submersibles. In along with other biological species that thrive in
addition, numerous historical shipwrecks from gal- these chemosynthetic communities. Sediments and
leons to warships like the USS Monitor have been methane ice (gas hydrates) are also selectively re-
explored and documented, preserving their legacy for trieved for later analysis.
future generations. Some new vehicles are still being produced, but
Johnson-Sea-Links I and II (J-S-Ls) were pressed have limited payloads, which restricts them to spe-
into service to assist in locating, identifying, and ul- cific tasks such as underwater camera platforms or
timately recovering many key pieces of the ill-fated observation. Some are easily transportable but
Space Shuttle Challenger. This disaster, viewed by the are small and restricted to one occupant; they can be
world via television, added a new dimension to the carried by smaller support vessels and are more
J-S-Ls’ capabilities. Previously only known for their economical to operate. Man’s desire to explore
pioneering efforts in marine science, they proved to the lakes, oceans, and seas has not diminished. New
be valuable assets in the search and recovery oper- technology will only enable, not reduce, the need for
ation. The J-S-Ls completed a total of 109 dives, man’s presence in these hostile environments.
 MANNED SUBMERSIBLES, SHALLOW WATER 161

See also Forman WR (1968) KUMUKAHI Design and Operations

Manual. Makapuu, HI: Oceanic Institute.
Manned Submersibles, Deep Water. Remotely Forman W (1999) The History of American Deep
Operated Vehicles (ROVs). Submersible Operations. Flagstaff, AZ: Best Publishing
Co.
Link MC (1973) Windows in the Sea. Washington, DC:
Further Reading Smithsonian Institute Press.
Stachiw JD (1986) The Origins of Acrylic Plastic
Askew TM (1980) JOHNSON-SEA-LINK Operations Submersibles. American Society of Mechanical
Manual. Fort Pierce: Harbor Branch Foundation. Engineers, Asme Paper 86-WA/HH-5.
Busby F (1976, 1981) Undersea Vechicles Directory. Van Hoek S and Link MC (1993) From Sky to Sea; A Story
Arlington: Busby & Associates. of Ed Link. Flagstaff, AZ: Best Publishing Co.
 PLATFORMS: AUTONOMOUS UNDERWATER
VEHICLES
J. G. Bellingham, Monterey Bay Aquarium Research number of companies develop subsystems and sen-
Institute, Moss Landing, CA, USA sors for AUVs.
& 2009 Elsevier Ltd. All rights reserved. The most common class of AUV in use today is a
torpedo-like vehicle with a propeller at its stern, and
steerable control surfaces to control turns and ver-
tical motion (see Figure 1). These vehicles are used
when speed or efficiency of motion is an important
consideration. Such torpedo-like vehicles range in
Introduction weight from a few tens of kilograms to thousands of
Autonomous underwater vehicles (AUVs) are un- kilograms. Most typical are vehicles weighing a few
tethered mobile platforms used for survey operations hundred kilograms, with an endurance of about a
by ocean scientists, marine industry, and the military. day at a speed of c. 1.5 m s1. Often they have par-
AUVs are computer-controlled, and may have little allel mid-bodies, which allow the vehicle length to be
or no interaction with a human operator while extended without large hydrodynamic consequences.
carrying out a mission. Being untethered, they must This is useful when it is necessary to add new
also store energy onboard, typically relying on bat- sensors or batteries to a vehicle. A disadvantage of
teries. Motivations for using AUVs include such torpedo-like vehicles is that, like an aircraft, they
factors as ability to access otherwise-inaccessible must maintain forward motion to generate lift over
regions, lower cost of operations, improved data its control surfaces, and thus are not controllable at
quality, and the ability to acquire nearly synoptic very low speed through the water.
observations of processes in the water column. An Gliders are a class of vehicles that use changes in
example of the first is operations under Arctic and buoyancy rather than a propeller for propulsion.
Antarctic ice, an environment in which operations of Gliders use their ability to control buoyancy to gen-
human-occupied vehicles and tethered platforms are erate vertical motion. Vertical motion is translated
either difficult or impossible. Illustrating the next into horizontal motion with lifting surfaces, usually
two points, AUVs are becoming the platform of wings mounted in about the middle of the vehicle (see
choice for deep-water bathymetric surveying in the Figure 1). Several types of gliders weighing about
offshore oil industry because they are less expensive 50 kg are in use today. These comparatively small
than towed platforms as well as produce higher- vehicles are designed to move slowly, about
quality data (because they are decoupled from mo- 0.25 m s1, and operate sensors consuming a watt or
tion of the sea surface). Finally, the use of fleets of less. By minimizing power consumption, these gliders
AUVs enables the rapid acquisition of distributed can operate for periods of months using high-energy-
data sets over regions as large as 10 000 km2. density primary batteries. Disadvantages of this class
AUVs are a new class of platform for the ocean of system are that they are limited to vertical profiling
sciences, and consequently are evolving rapidly. The flight tracks, and can be overwhelmed by ocean cur-
Self Propelled Underwater Research Vehicle (SPURV) rents, especially in the coastal environment or within
AUV, built at the University of Washington Applied boundary currents. However, larger gliders in devel-
Physics Laboratory, was first operated in 1967. opment and testing will operate at higher speeds, and
However, adoption by the ocean sciences community thus not suffer from this limitation.
lagged until the late 1990s. Adoption was spurred on A final class of AUVs uses multiple thrusters to
by two developments: AUV development teams provide capabilities similar to that of a helicopter or a
started supporting science field programs with AUV ship with dynamic positioning (see Figure 1). The
capabilities, and AUVs that nondevelopers could additional thrusters enable maneuvers such as hover-
purchase and operate became available. The first ing, translating sideways, and moving vertically. These
served the purpose of building a user base and vehicles are used when maneuverability is needed, for
demonstrating AUV capabilities. The second enabled example, when operation near a very rough bottom is
scientists to obtain and operate their own vehicles. a necessity. The disadvantage is that the additional
Today, a wide variety of AUVs are available thrusters reduce efficiency for moving large distances
from commercial manufacturers. An even larger or at high speeds.

162
 PLATFORMS: AUTONOMOUS UNDERWATER VEHICLES 163

Figure 1 From top-left corner clockwise: the Hugin AUV, the Spray glider (courtesy, MBARI), the ABE AUV (courtesy, Dana
Yoerger, WHOI), and the Dorado AUV (courtesy, MBARI). Hugin and Dorado are examples of propeller-driven vehicles optimized for
moving through the water efficiently, and have a torpedo-like configuration. Spray is an example of a glider, which is a buoyancy-driven
vehicle that has no propeller. ABE is a highly maneuverable vehicle capable of hovering or pivoting in place, and of moving straight up
or down. Also illustrated are different handling strategies. The large Hugin vehicle is launched and recovered from a ship using a stern
ramp. Spray is hand-launched and recovered. ABE is launched and recovered with a crane. Dorado is shown being launched and
recovered with a capture mechanism suspended from a J-frame.

While the vehicles described above are represen- energy from thermal differences in the ocean for
tative of the most commonly used systems, a wide propulsion have also been tested. A hybrid vehicle is
range of other vehicles are in development or are in being developed for reaching the deepest portion of
limited use. AUVs that come to the surface and use the ocean. The hybrid vehicle operates as a tethered
solar panels to recharge batteries have been demon- platform via a disposable fiber optic link for tasks
strated in seagoing operations. Gliders which extract requiring human perception, in other words as a
164 PLATFORMS: AUTONOMOUS UNDERWATER VEHICLES

remotely operated vehicle (ROV), but operates as an gliders optimized for low speed and long endurance
AUV when that link is severed. These are just a few can operate with hotel loads on the order of a watt or
of the diverse AUVs being developed to answer the less, but at the cost of operating a few very simple
needs of ocean science. sensors.
The power relationship provides insight into a
Terminology variety of AUV design considerations. For example,
the speed at which the vehicle will consume the least
The military term for an AUV is unmanned under-
energy per unit distance traveled (the energetically
water vehicle, or UUV. This phrase is ambiguous in
optimum speed) can be computed from observing
that it can also refer to ROVs. While ROVs are
that power divided by vehicle speed equals energy
unmanned, they are tethered and designed to be
per unit distance. Finding the minimum of P/v with
operated by human, and thus are not considered
respect to v yields the optimum speed from an energy
autonomous. However, common usage is to employ
conservation perspective:
UUV as a synonym for AUV. Military terminology
is relevant as many AUVs in use by the scientific

ZH 1=3
community were developed under navy funding, vopt ¼ ½2
CD Ar
and the military continues to be the largest single
investor in AUV technology. Consequently, the For the example vehicle values given above, the op-
technical literature on AUVs also employs military timum vehicle speed is approximately 1 m s1.
terminology. What can we say about vehicle performance at the
energetically optimum speed? Substituting eqn [2]
into eqn [1] we find that the power consumed at the
Basics of AUV Performance optimum speed is (3/2)H which for our example
Energy is a fundamental limitation for underwater vehicle is 45 W. If the total energy capacity of the
vehicles. Thus, energy efficiency is a fundamental battery system is Ecap, then the maximum range of
driver for vehicle design and operations. This section the vehicle will be:
outlines the relationship between vehicle speed and
1=3
endurance, and its dependence on factors such as 2Ecap Z
dmax ¼ ½3
power consumed by onboard systems. 3 CD ArH 2
A simple but useful model for power consumption
P of an AUV is as follows: If our example vehicle carries 10 kg of high-energy-
density primary batteries providing a total of
1 CD Arv3 1.3  107 J, it will have an endurance of 80 h, and a
P ¼ Pprop þ H where Pprop ¼ ½1 range of 280 km. The same vehicle with 10 kg of
2 Z
rechargeable batteries, with one-third the energy
Here the total electrical power consumed by the ve- capacity of the high-energy-density batteries, will have
hicle, P, is equal to the sum of propulsion power, its endurance and range reduced proportionately.
Pprop and hotel load, H. Hotel load is simply the There are caveats to the above discussion. For
power consumed by all subsystems other than pro- example, propulsion efficiency, Z, is typically a strong
pulsion. Propulsion power is a function of the drag function of speed as the electrical motors used tend
coefficient of the vehicle, CD, the area of the vehicle, to have comparatively narrow ranges of efficiency. In
A, the density of water, r, the speed of the vehicle, v, practice, a vehicle’s propulsion system is optimized
and the efficiency of the propulsion system, Z. for a particular speed and power. Also, vehicles are
What are the typical values for the coefficients often operated at higher speeds than the energetically
in eqn [1]? Consider an ‘example vehicle’ which is a optimum speed given by eqn [2]. For example, op-
12 3/400 (0.32 m) diameter torpedo-like AUV. Note erators of an AUV attended by a ship will be more
that this is a standard for a mid-size class AUV. sensitive to minimizing ship costs than to optimizing
For such a vehicle, parameter values might be energy efficiency of the AUV.
CD ¼ 0.2 (based on frontal area), A ¼ 0.082 m2, and
Z ¼ 0.5. Hotel load would depend on sensors, but an
AUV Systems and Technology
overall value of 30 W might be representative,
although mapping sonars would consume much AUVs are highly integrated devices, containing a
more power. We use r ¼ 1027 kg m–3. Note that these variety of mechanical, electrical, and software sub-
numbers, except seawater density of course, can systems. Figure 2 shows internal and external views
differ greatly from vehicle to vehicle. For example, of a deep-diving vehicle equipped with mapping
 PLATFORMS: AUTONOMOUS UNDERWATER VEHICLES 165

Doppler velocity logger

and inertial navigation system

Sub-bottom Batteries
profiler
Fluid intake
Sonar electronics
pressure vessel Computer
enclosure
Water
sensor

Acoustic
transducers
Flotation

Antenna

Lifting eye

Sonar receive

Fairing Thruster and duct

Figure 2 A propeller-driven, modular AUV with labeled subsystems. The top view shows the interior in which an internal mechanical
frame supports pressure vessels, internal components, and the propulsion system. This AUV, called a Dorado, is a ‘flooded’ vehicle
because the fairings are not watertight, and thus the interior spaces fill with water. Consequently internal components must all be
capable of withstanding ambient pressures. Joining rings are visible between the yellow fairing segments on the lower figure. These
allow the vehicle to be separated along its axis, allowing replacement or addition of hull sections. This allows reconfiguration of the
vehicle with new payloads, and if desired, with new batteries. The propulsion system on this vehicle is a ducted thruster capable of
being tilted both vertically and horizontally, to steer the vehicle in the vertical and horizontal planes. Courtesy of Farley Shane, MBARI.

sonars. The anatomy of an AUV typically includes • navigation sensors to determine the vehicle
the following subsystems: position;
• software and computers capable of managing • communication devices to allow communication
of human operators with the AUV;
vehicle subsystems to accomplish specific tasks
and even complete missions in the absence of • locating devices to allow operators to track the
vehicle and locate it for recovery or in the case of
human control;
emergencies;
• energy storage to provide power;
• devices for monitoring vehicle health (e.g., leaks
• propulsion system;
or battery failure);
• a system for controlling vehicle orientation and
• emergency systems for ensuring vehicle recovery
velocity;
in the event of failure of primary systems.
• sensors for measuring vehicle attitude, heading,
and depth; The mechanical design of an AUV must address
• pressure vessels for housing key electrical issues such as drag, neutral buoyancy, the highly
components; dynamic nature of launch and recovery, and the need
166 PLATFORMS: AUTONOMOUS UNDERWATER VEHICLES

to protect many delicate electrical components from supporting a steady increase in software capabilities
seawater. The desire to operate large numbers of for AUVs.
sensors for long distances encourages the con- Most AUV missions involve sequential tasks such
struction of larger vehicles to hold the necessary as descending from the surface to a set depth, then
equipment and batteries. However, the need for ease transiting to a survey location at a set speed,
of handling and minimizing logistical costs en- and then conducting a survey which might involve
courages the design of smaller vehicles. Operational flying a lawn-mower pattern. The vehicle may be
demands will also create constraints on vehicle de- commanded to maintain constant altitude over the
sign; for example, launch and recovery factors will bottom if the vehicle is mapping the seafloor. A
impose the need for lift points and discourage ex- water-column mission might require the vehicle
ternal appendages that will be easily broken. The profile in the vertical plane, moving in a saw-tooth
need to service vehicle components imposes a re- pattern called a yo-yo. The mission will likely include
quirement that internal components be easily ac- a transit from the end of the survey to a recovery
cessible for servicing and testing, and when location, with a final ascent to the surface for re-
necessary, replacement. covery. During the mission, the vehicle will monitor
In addition, a host of supporting software and the performance of onboard subsystems, and in the
hardware are required to operate an AUV. De- event of detection of anomalies, like a low battery
pending on the nature of the operations, supporting level, or a failed mission sensor, may abort the
equipment will include: mission and return to the recovery point early. A
more catastrophic failure might lead to the vehicle
• software and computers for configuring vehicle
shutting down primary systems, and dropping a drop
mission plans, and for reviewing vehicle data;
weight so as to float to the surface, and calling for
• systems for communicating with the vehicle both
help via satellite or direct radio frequency (RF)
on deck and when deployed;
communications.
• systems for recharging and monitoring vehicle
More complex vehicle missions can involve cap-
batteries;
abilities such as adapting survey operations to obtain
• handling gear for transporting, deploying, and
better measurements, or managing tasks such as
recovering AUVs;
AUV docking. An example of the first might be as
• devices for detecting locating devices on the AUV;
simple as a yo-yo mission which cues its vertical in-
• acoustic tracking systems for monitoring the lo-
flections from water temperature in order to follow a
cation of the vehicle when in the vicinity of a
thermocline. Also in the category of adaptive, but
support vessel.
more demanding, surveys, is the capability of fol-
lowing a thermal plume to its source, for example,
when an AUV is used to search for hydrothermal
AUV Mission Software
vents. The docking of an AUV with an underwater
Functionally, AUV software must address a variety structure encompasses yet a different type of com-
of needs, including: allowing human operators to plexity, created by the large number of steps in the
specify objectives, managing vehicle subsystems to process, and the high likelihood that individual steps
achieve mission objectives, logging data for sub- will fail. For example, docking involves homing on a
sequent review, and ensuring safety of the vehicle in docking structure, orienting for final approach, en-
the event of failures or unexpected circumstances. gaging the dock, and making physical connections to
The software must be capable of managing vehicle establish power and communication links. Any one
sensors and control systems to maintain a set head- of these steps might fail due to external pertur-
ing, speed, and depth. The software might also need bations; for example, currents or turbulence in the
to support interacting with a human operator during marine environment might cause the vehicle to miss
a mission. In addition to software on the vehicle it- the dock. The vehicle must be able to detect failures
self, AUV operators rely on a suite of software ap- and execute a process to recover and try again.
plications to configure and validate missions, to Docking is representative of the increasingly com-
maintain vehicle subsystems such as batteries, to re- plex capabilities AUVs are expected to master with
view data generated by the vehicle, to prepare mis- high reliability.
sion summaries, and when possible, to track and
manage the vehicle while underway. The exponential
Navigation
growth of computational power available for both
onboard and off-board computers, as well as the The ability for an AUV to determine its location on
increasingly pervasive nature of the Internet, are the Earth is essential for most scientific applications.
 PLATFORMS: AUTONOMOUS UNDERWATER VEHICLES 167

However, navigation in the subsea environment is error with ancillary measurements of velocity and
complicated by the opacity of seawater to all but position. For example, combining an INS with a
very low frequency electromagnetic radiation, ren- DVL for constraining velocity can result in a system
dering ineffective the use of commonly used tech- which provides navigation accuracies better than
nologies such as the Global Positioning System (GPS) 0.05% of distance traveled.
and other radio-based navigation techniques. Con- The two acoustic navigation methodologies most
sequently, navigation underwater relies primarily on frequently used in AUV operations are ultrashort
various acoustic and dead-reckoning techniques and baseline navigation (USBL) and long baseline (LBL)
the occasional excursion to the surface where radio- navigation. A USBL system uses an array of hydro-
based methods can be used. There is no single phone separated by a distance comparable to the
method of underwater navigation that satisfies all wavelength of sound to measure the direction of
operational needs, rather a variety of methods are propagation of an acoustic signal. Most often, a
employed depending on the circumstances. USBL system is mounted on a ship, and used to track
Dead-reckoning methods integrate a vehicle’s ve- a vehicle relative to the ship. With knowledge of the
locity in time to obtain an updated location. In order ship’s location and orientation, the location of the
to dead-reckon, the vehicle must know both the AUV can also be determined. In contrast, LBL
direction and speed of its travel. The simplest navigation acoustically measures the range between
methods use a magnetic compass to determine dir- the vehicle and an array of widely separated devices
ection, and use speed through the water as a proxy of known location. A common LBL approach is to
for Earth-referenced speed. However, the large place transponders on the seafloor, and let the vehicle
number of error sources for magnetic compasses range off the transponders. The process of deter-
make measurement of heading to better than a de- mining location using ranges from known locations
gree accuracy technically challenging. Currents pose is called spherical navigation, as the vehicle should
even more of a problem, as they may be comparable be located at the intersection of spheres with the
to the vehicle speed in amplitude, yet are not sensed measured radius, centered on the respective trans-
by a water-relative measurement. Dead-reckoning is ponders. An alternative LBL navigation method is to
improved by measuring velocity relative to the sea- track a vehicle which pings at a preset time to an
floor, for example, using a Doppler velocity log array of hydrophones at known locations. If the time
(DVL) or a correlation velocity log. A DVL is com- of the ping is not known, the problem of solving for
monly used by AUVs to measure velocity by meas- the vehicle location is called hyperbolic navigation,
uring the Doppler shift of sound reflected off the as only the difference in time of arrival of the
seafloor. Correlation velocity logs are more complex ping at the various hydrophones can be determined,
in concept, involving measurement of the correlation and this knowledge constrains the vehicle to be on a
of two pulses of sounds transmitted by the vehicle, hyperbola between the respective receivers. If the
reflected off the seafloor, and received by a hydro- time of the ping is known, perhaps triggered at a
phone array. In practice, DVLs are used when a ve- preselected time by a carefully calibrated clock, then
hicle operates close to the seafloor, perhaps within the problem reduces to spherical navigation. In
200 m, while correlation velocity logs are used when practice, a wide variety of USBL and LBL systems
the vehicle is operating in mid-water columns or near have been implemented for underwater navigation.
the surface in deep water. They must all address the challenges of acoustic
Inertial navigation system (INS) technology is well propagation in the ocean, which include the ab-
developed, as it is widely used for platforms like sorption of sound by seawater, diffraction by speed
aircraft and missiles. However, INS units appropriate of sound variations in the underwater environment,
for underwater use are expensive enough that they scattering by reflecting surfaces, and acoustic noise
are used only when navigation requirements are generated by physical, geological, biological, and
stringent, for example, for producing high-accuracy anthropogenic processes.
maps. A modern INS includes an array of acceler- Other methods of navigation include using geo-
ometers for measuring acceleration on three axes and physical parameters, for example, water depth, to
a laser or fiber optic gyroscope for measuring chan- constrain the vehicle location in the context of
ges in orientation. Additionally, an INS will include a known maps. These geophysically based navigation
GPS for initializing the unit’s location and orien- methods, similar to terrain contour mapping (TER-
tation, and a computer for acquiring and processing COM) navigation used by cruise missiles, depend on
data from INS component sensors. The position re- having good maps ahead of time. There are software
ported by an INS will have an error which will grow approaches in development that simultaneously
in time, and thus it is important to constrain INS build maps and use those same maps for navigation.
168 PLATFORMS: AUTONOMOUS UNDERWATER VEHICLES

These methods are called SLAM for simultaneous towed body. If ultrashort baseline acoustic navi-
localization and mapping. gation techniques are used to determine the vehicle
position, then layback of the towed body behind the
ship introduces significant errors as compared with
Using AUVs for Ocean Science having the ship directly over the sonar platform. For
this reason, some commercial use of towed sonar
Mapping the Seafloor
platforms use two ships, one to tow the sonar
AUVs are becoming the platform of choice for high- platform, and one positioned directly over the
resolution seafloor maps. Obtaining high-resolution platform to determine its precise location. Finally,
maps requires operating mapping sonars near the surface motion of the ship will be efficiently coupled
seafloor. Alternatives to an AUV include crewed to the tow body by the tow cable. Thus, even near
submersibles and tethered platforms. Crewed sub- the seafloor, the tow body will be subject to sea state
mersibles are too valuable for routine mapping, and experienced by the ship. Consequently, attraction of
are reserved for other uses which require the pres- the use of AUVs includes more economical oper-
ence of humans. Towed vehicles are used for sonar ations and high data quality. Figure 3 shows a cost
mapping, but have disadvantages as compared with comparison of a commercial deep-water towed sur-
AUVs, especially in deeper water. The principal vey and an equivalent AUV survey.
problem is the high drag of the cable used for a tow Sonar systems used on AUVs for mapping include
sled, which in water depths of several thousand multibeam sonar, side scan sonar, and sub-bottom
meters will limit speeds to approximately half a profilers. Multibeam sonars, operating at frequen-
meter per second. Even at these slow speeds, a cies of hundreds of kilohertz in the case of
towed platform will stream behind the towing ship, AUV-mounted systems, allow measurement of
creating several problems. Controlling the position range to the seafloor in multiple sonar beams and
of the towed vehicle over the bottom is very dif- are used to build up three-dimensional maps such as
ficult, even when running on a constant heading. that in Figure 4. Side scan sonars used by AUVs
When surveying a defined area on the seafloor in a also typically operate at frequencies of hundreds of
series of passes, the turns between passes may take kilohertz, and are used to image seafloor features.
longer than the actual survey passes themselves, as it Side scan sonars are particularly useful for finding
is necessary to turn slowly to maintain control of the objects, for example, looking for a shipwreck resting

Report preparation

100% Tracking boat

90%
Cost (normalized to total towed survey cost)

Time spent turning to

return to survey site
80%
Survey
70%
Transit to work site
60%
Mobilization demob
50%

40%

30%

20%

10%

0%
Towed survey AUV

Figure 3 A comparison of the economics of deep survey taken from costs of a survey with a towed vehicle, and projected costs of
the same survey with an AUV. The principal cost saving derives from the ability of the AUV to turn much faster than a deep-towed
vehicle, reducing the total survey time. Also, the AUV can be acoustically tracked by its mother ship, while a towed vehicle requires a
second ship for tracking because the towed vehicle will trail far behind the tow ship. Finally, mobilization and demobilization costs for
the AUV can also be lower, although this depends on the size of the AUV employed.
 PLATFORMS: AUTONOMOUS UNDERWATER VEHICLES 169

122° 00' 50" W 122° 00' 40" W 122° 00' 30" W 122° 00' 20" W

km
36° 46' 50" N 36° 46' 50
0 0.1 0.2

36° 46' 45" N 36° 46' 45

36° 46' 40" N 36° 46' 40

36° 46' 35" N 36° 46' 35

36° 46' 30" N 36° 46' 30

122° 00' 50" W 122° 00' 40" W 122° 00' 30" W 122° 00' 20" W

−1012.5 −1000.0 −987.5 −975.0 −962.5 −950.0 −937.5 −925.0 −912.5 −900.0 −887.5
Topography (m)

Figure 4 A bathymetric survey produced by an AUV at a depth of about 1000 m. Note the very small size of the survey area and high
resolution of the bathymetry. Courtesy of Dave Caress, MBARI.

on the seafloor. Sub-bottom profilers use lower- ocean. The smaller, buoyancy-driven gliders are
frequency sound, ranging from 1 kHz to tens of unique in their combination of mobility and en-
kilohertz in the case of an AUV-mounted system, to durance, moving at about a quarter of a meter per
penetrate into the seafloor. Depending on the bot- second for periods of months. Larger vehicles carry
tom type (e.g., sandy, muddy, or rock), a sub- more comprehensive payloads at higher speeds, but
bottom system might penetrate tens of meters. for shorter periods. Such vehicles might operate at
Cumulatively sonar payloads will consume com- 1.5 m s1 for a day. A common flight profile is to fly
paratively large amounts of energy, perhaps hun- the vehicle on a constant heading, while moving
dreds of watts. Mapping also requires high-fidelity between two depth extremes in a saw-tooth pattern.
navigation, and thus sonar-equipped AUVs will Often the upper depth extreme will be close to the
often also use more sophisticated navigation ap- surface. This strategy allows the production of ver-
proaches, like inertial navigation. Consequently, tical sections of ocean properties, such as those in
mapping AUVs of today are larger, more sophisti- Figure 5. Variations of this strategy might have
cated AUVs. the vehicle moving in a lawn-mower or zigzag pat-
tern in the horizontal plane, to develop a full three-
dimensional map of ocean properties. Figure 6 shows
Observing the Water Column
a visualization of an internal wave interacting with a
AUVs provide a relatively new tool for observing the phytoplankton layer using such a three-dimensional
physical, chemical, and biological properties of the mapping strategy.
170 PLATFORMS: AUTONOMOUS UNDERWATER VEHICLES

Temperature
0 16

14
10
12
20 10

30 8

0 1000 2000 3000 4000 5000 6000 7000

Fluorometer
0

1500
10

1000
20
500
30
0
0 1000 2000 3000 4000 5000 6000 7000
Optical backscatter
0
1000

10 800

600
20
400
30 200
0 1000 2000 3000 4000 5000 6000 7000
Downtrack distance (m)

Figure 5 Vertical sections of water properties obtained by an Odyssey AUV operating in Massachusetts Bay. The y-axis of each
figure is depth, in meters, and the x-axis is horizontal distance in meters. The top section shows temperature in degrees Celsius, the
middle shows chlorophyll fluorescence in arbitrary units, and the bottom shows optical backscatter, also in arbitrary units. The path of
the vehicle is shown as a white line, and the interpolated values of the measured property are plotted in color. The vehicle alternated
between obtaining high-resolution observations of the thin layer of organisms at the thermocline with full water column profiles.

All AUVs are limited by the availability of sensors. operation has been carried out with AUVs in both
Temperature, salinity, currents, dissolved oxygen, ni- the Arctic and Antarctic. Sea ice poses special oper-
trate, optical backscatter properties, and chlorophyll ational challenges for seagoing ocean scientists. For
fluorescence are examples of the growing in situ sens- example, ships with ice-breaking capability can op-
ing capabilities available for AUVs. However, many erate in the ice pack, but will typically not be able to
important properties, for example, pH, dissolved car- hold station, or even assure that tethers and cables
bon dioxide, and dissolved iron, cannot be measured deployed over the side will not be severed. AUVs are
reliably from a small moving platform. Furthermore, attractive in that they provide horizontal mobility
detection of marine organisms is usually accomplished under ice, and the ability to conduct operations near
by proxy; for example, chlorophyll fluorescence pro- the seafloor without the complications intrinsic in
vides an indicator for phytoplankton abundance. tether management. Challenges of operating AUVs
In situ methods which directly detect, classify, and under ice revolve around the need to assure return of
quantify marine organism abundance are not avail- the AUV to the ship for recovery, the process of re-
able, yet are increasingly important for understanding covering the AUV through the ice onto the ship, the
the structure and dynamics of ocean ecosystems. potential for having an AUV fail and become trapped
under ice, and the difficulty of carrying out tasks that
Operations in Ice-Covered Oceans
would normally be accomplished having an AUV
AUVs offer unique operational capabilities for sci- surface (e.g., obtaining a GPS update). Most safety
ence in ice-covered oceans. Successful under-ice strategies for AUVs in ice-free oceans default to bring
 PLATFORMS: AUTONOMOUS UNDERWATER VEHICLES 171

Monterey Bay

Front

5m

10 m Phytoplankton
layer

Internal
15 m wave

20 m
2 km

1 km
0 km
2 km 0 km
4 km 6 km

Figure 6 Interaction of layer of phytoplankton with an internal wave in Monterey Bay. Both physical and biological properties were
measured by an Odyssey AUV, which moved in a horizontal zigzag pattern across the survey volume, while profiling constantly in the
vertical plane. The phytoplankton layer, shown in green, was detected by a chlorophyll fluorescence sensor on the AUV. The cyan
surface shows deflection of a level of constant density of seawater by a passing internal wave. Courtesy of John Ryan, MBARI.

the vehicle directly to the surface, for example, by time-constrained surveys are most often encountered
dropping a weight. In the Arctic or Antarctic, this when surveying a dynamic process. For example, if
strategy could result in the vehicle becoming trapped the temporal decorrelation of ocean fields associated
under very thick ice, making the vehicle harder to with upwelling off Monterey Bay is about 48 h, at-
find and potentially impossible to recover. The usual tempts to map the ocean fields need to be accom-
surface location devices such as RF beacons, strobes, plished within that time frame. Scales of spatial
and combinations of RF communication and satellite variability will also determine acceptable separation
navigation will not work. Clearly AUV operations of observations: for example, decorrelation lengths
within the ice pack entail higher risk and a more in Monterey Bay are on the order of 20 km, so ob-
sophisticated vehicle. servations need to be spaced significantly closer to
minimize errors in reconstructing the ocean field.
How does this relate to the number of vehicles re-
Observation Systems, Observatories, and AUVs
quired to accomplish such a survey? Consider a grid
An understanding of power consumption of AUVs survey of a 100 km  100 km area with a resolution
provides insight to the attractiveness of employing of 10 km. A single vehicle would have to
multiple vehicles for certain ocean observation travel c. 1000 km at a speed of nearly 6 m s1,
problems. In some circumstances a survey must be traveling in a lawn-mower pattern. Using the ex-
accomplished within a set period. In oceanography, ample vehicle values from the ‘Basics of AUV
172 PLATFORMS: AUTONOMOUS UNDERWATER VEHICLES

performance’ section, the AUV would consume Autonomous Ocean Sampling Network (AOSN) II
about 3500 W if it were capable of operating at such field program (Figure 7). Gliding vehicles, with an
a high speed. In contrast, six of the same vehicles endurance of weeks to months, provided a con-
operating at their optimum speed would consume a tinuous presence with a minimal sensor suite. A few
total of 270 W. In other words, the six vehicles would propeller-driven vehicles provided observations of
consume 12 times less energy for the complete 48-h chemical and biological ocean parameters, allowing
survey. tracking of ecosystem response to the upwelling
Autonomous mobile platforms are making ob- process. Observations were fed to two oceano-
servation of the interior of the ocean more afford- graphic models, which provided synoptic realization
able and more flexible, enabling the practical of ocean fields and predicted future conditions.
realization of coupled observation–prediction sys- Among the many lessons are an improved know-
tems. For example, in late summer 2003, a diverse ledge of the scales of variability of upwelling pro-
fleet of AUVs was deployed to observe and predict cesses, an understanding of how to scale observation
the evolution of episodic wind-driven upwelling in systems to these processes, and insights to strategies
the environs of Monterey Bay. Over 21 different for adaptive sampling of comparatively rapidly
autonomous robotic systems, three ships, an air- changing processes with comparatively slow ve-
craft, a coastal ocean dynamics application radar hicles. These lessons are particularly relevant today,
(CODAR), drifters, floats, and numerous fixed given the present emphasis on developing ocean-
(moored) observation assets were deployed in the observing systems.

Figure 7 Example of a distributed observing system using AUVs. This diagram depicts an AOSN deployment in Monterey Bay.
 PLATFORMS: AUTONOMOUS UNDERWATER VEHICLES 173

See also IEEE (2001) Special Issue: Autonomous Ocean Sampling

Networks. IEEE Journal of Oceanic Engineering 26(4):
Gliders. Remotely Operated Vehicles (ROVs). 437--446.
Jenkins SA, Humphreys DE, Sherman J, et al. (2003)
Further Reading Underwater glider system study. Scripps Institution of
Oceanography Technical Report No. 53. Arlington,
Allmendinger EE (1990) Submersible Vehicle Systems VA: Office of Naval Research.
Design. New York: SNAME. Rudnick DL and Perry MJ (eds.) (2003) ALPS:
Bradley AM (1992) Low power navigation and control for Autonomous and Lagrangian Platforms and Sensors,
long range autonomous underwater vehicles. Workshop Report, 64pp. http://www.geo-prose.com/
Proceedings of the Second International Offshore and ALPS (accessed Mar. 2008).
Polar Conference, pp. 473–478.
Fossen T (1995) Guidance and Control of Ocean Vehicles. Relevant Website
New York: Wiley.
Griffiths G (ed.) (2003) Technology and Applications of http://www.mbari.org
Autonomous Underwater Vehicles. London: Taylor and – Monterey Bay 2003 Experiment, Autonomous Ocean
Francis. Sampling Network, MBARI.
 PLATFORMS: BENTHIC FLUX LANDERS
R. A. Jahnke, Skidaway Institute of Oceanography, to examine important questions such as: What are
Savannah, GA, USA
the respiration rates of deep-sea-floor populations?
& 2009 Elsevier Ltd. All rights reserved. What is the seafloor dissolution rate of calcium car-
bonate? Estimating benthic fluxes (i.e., the net ex-
change rate of solutes across the sediment surface) on
the seafloor, at in situ temperature and pressures, is
Introduction one strategy for avoiding sampling artifacts and im-
proving the accuracy of deep-ocean measurements.
The need to better understand chemical and bio- There are two common techniques for estimating
logical processes and solute transport mechanisms benthic fluxes. Benthic flux chamber incubations can
across the deep-sea floor has fueled major engineering be performed in which a known volume of bottom
advances in seafloor instrumentation in the last few water is trapped above a known area of seafloor. Any
decades. In principle, the kinds of experiments that solute transported out of the sediments covered by
can be conducted on the seafloor are limited only by the chamber will be trapped within the chamber
researchers’ imaginations. In reality, the majority of waters. Hence, the concentration of this constituent
the instruments that have been developed seek to in the chamber waters will increase with incubation
accomplish two basic types of operations: deploy, time. Conversely, the chamber water concentration
sample, and retrieve benthic flux chamber systems to of any chemical constituent that is being transported
directly measure seafloor exchange rates and deploy into the sediments will decrease with incubation
in situ sensor arrays capable of measuring the vertical time. The benthic flux is directly proportional to the
distribution of solutes in near-surface pore waters rate of concentration increase or decrease, the vol-
from which exchange rates can be estimated based on ume of bottom water enclosed within the chamber,
pore water transport models. Recent advances have and area of the seafloor covered by the chamber. The
expanded the types of experiments and measurements main advantage of this method is that there is min-
that can be performed by these instruments. Gener- imal disturbance to the sediment system, maximizing
ically, the instrument frames used to deploy these the probability of accurate estimates. In addition, the
types of instruments have been called ‘bottom land- fluxes obtained will reflect the net exchange due to
ers’. This term was coined in the early 1970s in rec- all transport processes occurring within the spatial
ognition of the similarity between the approximate dimensions of the chamber incubations. Thus, this
shape and function of these devices and the more technique will include nondiffusive exchange such as
famous ‘lunar lander’ that carried the first men to the that driven by the active irrigation of organism bur-
moon. In the following, the design strategies and rows. A weakness of this approach is that other than
basic instrumentation for conducting benthic flux inferences about total integrated reaction rates
chamber incubations and sensor measurements at the within the sediment column supporting the observed
deep-sea floor are discussed. benthic flux, little information is gained about the
reaction processes and distributions themselves.
Additionally, if there is significant bottom current-
Benthic Flux Measurement Strategies
driven advective flow through the surface sediments,
In the 1960s and 1970s, it was increasingly recog- differences between hydrodynamic conditions within
nized that the temperature and pressure changes that the chamber and the natural setting may result in
occur in bringing a sediment sample from the deep- significant inaccuracies in flux estimates.
sea floor to the sea surface may alter the sample. The other common strategy for estimating seafloor
Examples include changes in metabolic rates of fluxes is to measure the concentration gradient of
benthic populations, changes in pore water concen- chemical constituents very near (preferably across)
tration gradients from which diffusive fluxes are the sediment–water interface. Knowing the transport
calculated, and changes in chemical concentrations processes, the flux can be calculated based on the
in pore waters due to pressure- or temperature- transport rates and measured concentration gradi-
driven reactions. Additional artifacts continue to be ents. In principle, this method can be applied to all
identified to this day. Thus, analyzing deep-sea types of transport processes. In practice, however, the
samples brought to the deck of a ship has been in- exact nature of nondiffusive transport processes is
creasingly recognized as being not accurate enough unknown and this calculation strategy is used almost

174
 PLATFORMS: BENTHIC FLUX LANDERS 175

exclusively to estimate exchange due to molecular mounted in the central part of the frame. It is in these
diffusion only. Thus, a limitation of this approach is components where the greatest differences between
that accurate benthic fluxes cannot be estimated in a individual designs occur. Chamber instruments have
location where pore water exchange due to processes been designed to accommodate one to four chambers
other than molecular diffusion is significant. On the simultaneously. While increasing desirable repli-
other hand, a strength of using concentration profiles cation, multiple chambers on a single instrument
is that the distribution of reactions can be assessed tend to increase variability in the data by decreasing
and inferences of reaction mechanisms can be made chamber area and increase the complexity of the
from variations in the fluxes estimated at different control, data storage, and sampling systems.
depths below the sediment surface. Early chamber systems like the FVGR relied on
oxygen electrodes to quantify the changes in oxygen
within the chambers. Thus, the early instruments
Benthic Chamber Landers were only capable of estimating benthic oxygen de-
mand. While electrodes continue to be widely used,
Design and Operations
many modern instruments also employ an elec-
The development of benthic chamber landers fol- tronically controlled sampling system so that benthic
lowed the experiences gained from conducting exchange of many types of solutes, such as nutrients,
chamber or ‘bell jar’ incubations in shallow-water trace elements, and organic and inorganic carbon,
areas where they could be tended by scuba divers. can be assessed.
The major challenge in developing this instru- The utility of these instruments is demonstrated by
mentation was simply to automate the required op- their proliferation. Today, there are more than 20
erations to function reliably in the harsh conditions different research groups that have developed in situ
of the deep-sea floor. The earliest instrument that benthic flux chamber instruments. These groups have
was routinely deployed in the deep sea and has designed and implemented numerous modifications
provided a large data set is the free vehicle grab and alterations to Smith’s initial design, yet the basic
respirometer (FVGR) developed in the late 1970s by characteristics and capabilities of the chamber lander
Dr. K.L. Smith, Jr., at Scripps Institution of Oceano- remain the same. Examples of the numerous im-
graphy in San Diego, California, USA (Figure 1). portant modifications that have been made to the
The basic instrument consists of a structural frame early designs include several types of stirring mecha-
upon which the critical components are mounted. To nisms and time-series sampling systems so that the
minimize weight, the frame is most often constructed exchange of solutes other than oxygen can be ad-
of tubular aluminum. Instrument flotation is mounted dressed. In addition to the basic benthic flux chamber
on the upper portion of the instrument frame. The operations, complementary sampling devices have
most common type of flotation is glass spheres (as also been added. For example, one chamber instru-
shown in Figure 1), although syntactic foam flotation ment design has incorporated an in situ whole core
has also been used. The latter is much more expensive squeezer. This device recovers a sediment core and
than glass spheres but is not susceptible to implosion, then sequentially squeezes the surface pore waters
which is critical if manned submersibles are ever re- from the sediments to provide high-resolution pore
quired to work in close proximity. Both types provide water samples. While not appropriate for all solutes
relatively constant buoyancy. At the top of the flota- due to surface exchange during squeezing, these
tion section are mounted devices such as a flag, strobe samples provide important information concerning
light, radio transmitter, or satellite transmitter to help the pore water gradients of selected metabolites, such
locate the instrument when it is at the sea surface. as oxygen and nitrate, which greatly enhances the
Expendable weights are mounted at the lower interpretation of the flux chamber results.
portion of the tubular frame, usually adjacent to the Examples of results from a benthic flux chamber
‘feet’. As these instruments are free vehicles, these deployment at c. 3000 m on the continental rise of
weights provide the negative buoyancy necessary to the eastern US seaboard is displayed in Figure 2. On
drive the instrument to the seafloor. A latching each plot, the concentration is on the vertical axis
mechanism permits the weights to be released at the and the horizontal axis represents incubation time.
end of the experiment. Upon release, the flotation These results demonstrate the range of possible re-
provides sufficient positive buoyancy to pull the in- sponses. Solutes that are taken up by the sediments
strument back to the sea surface where it may be tend to decrease with incubation time. An example
recovered by a surface research vessel. of this is oxygen which is consumed in the sediments
The specific instrument packages, controlling through benthic respiration. Most other compo-
electronics and other operational components, are nents are produced in the sediments through the
176 PLATFORMS: BENTHIC FLUX LANDERS

Transducer
Radio beacon and strobe light

Glass sphere
Flotation

Aluminum tubular frame

Acoustic release
Remote release
Battery cylinder
Hydrostatic piston
Electronic pressure case

Camera
Respirometer tray
Strobe
Cable release mechanism
Grab tray
Descent weight

Frame foot

Disturbance
reflector
Grab assembly

Figure 1 Schematic of the free vehicle grab respirometer developed by Dr. Kenneth L. Smith, Jr. (Scripps Institution of
Oceanography).

decomposition or dissolution of biogenic debris. This that produce or consume them. For example, nitrate
production supports a flux out of the sediments and is produced by the oxidation of ammonium (nitrifi-
the concentrations of these constituents increase with cation) and is consumed by denitrifiying bacteria
incubation time. Specific examples of these include below the oxic zone. Whether there is a flux out of or
phosphate, released from degrading organic tissue, into the sediments depends on the ratio of these rates.
total inorganic carbon (TIC) released through res- Direct estimates of benthic fluxes can be made
piration and the dissolution of CaCO3, titration al- from these results from the relationship
kalinity (TA) produced by the dissolution of CaCO3,
and silicate which is produced by the dissolution of Benthic flux ¼ ðSVÞ=A
opal. Some species, such as nitrate, may increase or
decrease with incubation time depending on the where S is the slope of the concentration versus time
relative rates of the competing sedimentary processes results, V is the volume of the bottom water trapped
 PLATFORMS: BENTHIC FLUX LANDERS 177

300 19

275 18.5

Oxygen (μM)

Nitrate (μM)
250 18

225 17.5

200 17
0 20 40 60 80 0 20 40 60 80

2350 2450

2325 2425

TA (μEq l−1)
TIC (μM)

2300 2400

2275 2375

2250 2350
0 20 40 60 80 0 20 40 60 80

2 40

1.75
Phosphate (μM)

30
Silicate (μM)

1.5

20
1.25

1 10
0 20 40 60 80 0 20 40 60 80
Time (h) Time (h)

Figure 2 Example benthic flux chamber results from 2927 m on the continental rise of the US eastern seaboard using the chamber
system of Dr. Richard A. Jahnke (Skidaway Institute of Oceanography).

in the chamber, and A is the sediment surface en- alter the near-surface gradients, altering the flux. In
closed by the chamber. the extreme case, where concentration changes are
Note that if the chamber has vertical sides, large, such as complete oxygen depletion within the
V/A ¼ height of the water column trapped within the chamber, changes in chamber water chemistry may
chamber. alter near-surface chemical reactions and/or ex-
change processes, greatly altering the chemical flux.
To minimize this source of uncertainty, chamber
Major Design Controversies and Differences
deployments must be designed to minimize the con-
Despite the numerous advantages of benthic flux centration changes and maintain the chamber sedi-
incubations for estimating seafloor exchange, there ments as near to their natural state as possible. Thus,
are a variety of limitations and concerns that need to it is important that the most precise analytical pro-
be addressed in evaluating results from chamber in- cedures be employed, so that accurate fluxes can be
cubations and in assessing the relative merits of the quantified without large concentration changes.
different instrument designs. Maintaining natural benthic fluxes requires that
As shown in the example results, concentration the sediment surface not be disturbed during
changes are required to evaluate the solute exchange chamber deployment. This need has resulted in sev-
rate. However, the concentration changes will also eral different types of deployment strategies and
178 PLATFORMS: BENTHIC FLUX LANDERS

instrument designs. Most instruments have been de- disturbed layer. Layers of 1- and 3-mm thickness will
signed to settle to the seafloor and then, after some recover on timescales of c. 10 and 80 min, respect-
preset time period, slowly insert the chambers into ively. If deployments are short relative to the gradient
the bottom, assuming that this waiting period will response times, accurate fluxes will require the
allow the sediments that are resuspended by the maintenance of near-natural hydrodynamic con-
impact of the instrument with the bottom to be ditions within the chamber. On the other hand, if the
swept away, leaving a natural surface on which to deployments are long relative to the pore water
conduct the incubation. Of course, if the pressure gradient response times, the fluxes will be insensitive
disturbance ahead of the descending instrument is to chamber hydrodynamics and a simple chamber
sufficiently large, the entire sediment surface within water-stirring mechanism, such as rotating rods, a
the perimeter of the instrument frame will be im- disk, or circulation through an external pump, is
pacted. Recognizing this potential artifact, some in- adequate. For most solutes in the deep sea, deploy-
strument designs positioned the chambers lower than ments greater than 10–20 h are sufficient to minimize
the main frame, so that the chamber would en- artifacts due to changes in the diffusive sublayer
capsulate the sediments upon which the flux incu- thickness and thus only require a relatively simple
bation will be performed before the ‘bow wave’ from mixing strategy.
the frame reaches the surface. Another approach is to The situation is more complex if the sediments have
suspend a descent weight below the instrument. This many open burrows or exhibit elevated permeability,
weight provides the negative buoyancy needed for as would be characteristic of a well-sorted sandy
the instrument to descend to the seafloor. Once the sediment. In these environments, bottom flows may
weight reaches the seafloor, the positive buoyancy of drive advective pore water exchange. Alternation of
the instrument itself causes it to remain suspended natural flow conditions by the presence of the cham-
above the bottom. A winch system is then activated ber will likely influence the accuracy of the flux esti-
and the instrument package is very slowly pulled to mates and must be considered in interpreting results.
the seafloor. A recent innovative approach for min-
imizing bottom disturbance has been implemented
on the ROVER lander, a benthic flux chamber device Sensor Landers
capable of ‘crawling’ around the seafloor. Thus, once
Design and Operations
the instrument has impacted the bottom, it simply
crawls laterally to an undisturbed location prior to The basic instrument design and field deployment
conducting benthic flux chamber incubations and operations of the sensor landers are the same as that
pore water profiling operations. already discussed for the benthic flux chamber. The
Perhaps the most controversial aspect of benthic instrument consists of a frame, identical to that of a
flux incubations and instrument design is focused on benthic flux chamber lander, with flotation mounted
mechanisms and requirements for simulating natural at the top and expendable descent weights attached
flow conditions within the chamber. Exchange across near the feet. The major difference is that the benthic
a homogeneous muddy sediment surface requires chamber is replaced with an instrument package
transport across the hydrodynamic boundary layer capable of inserting microelectrodes or optode with
including the molecular diffusive sublayer. Hydro- high vertical resolution into the surface sediments.
dynamic conditions within the chamber control the An example of the basic instrument package is pre-
thickness of this layer. Changes in the hydrodynamic sented in Figure 3.
regime will alter the thickness of the layer and, at Each of these types of packages consists of three
least temporarily, will alter the exchange rate. basic components: the sensing electrodes themselves;
However, since the benthic flux is supported by a mechanism for moving the electrodes vertically
metabolic and chemical reactions in the sediments across the sediment–water interface; and the data
that are not influenced by the diffusive sublayer storage and controlling electronics. In the design
thickness, the seafloor exchange rate will eventually pictured in Figure 3, the sensing electrodes are at-
revert to its natural rate. Thus, the need to accurately tached directly to the main pressure case. Located
reproduce the natural hydrodynamic conditions with within the pressure case are the power and elec-
the benthic flux chamber depends critically upon the tronics necessary to operate the motor and electrodes
response time of the surface pore water gradients and and to provide for data storage and retrieval. The
the length of the chamber incubations. If controlled motor drive system is positioned outside the case to
by molecular diffusion, the time required for pore move the case vertically, so that profiles of the
water concentration gradients to recover after dis- measured components are obtained. The vertical
turbance scales as the square of the thickness of the resolution of the profiles is controlled by the size of
 PLATFORMS: BENTHIC FLUX LANDERS 179

Threaded rod
DC
motor Supporting AL-frame
housing and tracks

Radial ball bearings

Oil-filled
bladder

Pressure cylinder

Microelectrodes

Figure 3 Schematic of the microelectrode deployment apparatus developed by Dr. Clare Reimers (Rutgers University).

the electrode tip and the precision of the motor drive deployment is provided in Figure 4. Because oxygen
assembly. is consumed within the sediments, primarily due to
Early designs primarily employed oxygen and the respiration by benthic organisms but possibly
electrical resistivity electrodes. The latter is required also due to chemical consumption, oxygen concen-
to assess the porosity and tortuosity that influence the trations decrease with increasing depth in the sedi-
rate of diffusion in sediments. Porosity is a measure of ments. This downward concentration gradient
the proportion of the sediment comprised of void implies a benthic flux into the sediment. Contrast-
space between sediment particles and is generally 70– ingly, in the example provided in Figure 4, pH first
90% in muds and 30–60% in sands. Tortuosity is the increases and then decreases with sediment depth.
actual distance a solute would have to travel around This more complicated profile shape is due to com-
sediment particles relative to the straight-line distance peting reactions downcore. In this example, pH first
between two points. In recent years, numerous other increases with depth due to the dissolution of cal-
sensors have been developed and implemented for cium carbonate and then decreases with depth due to
deep-sea lander use. These include pH, pCO2, total continued production of carbon dioxide.
CO2, calcium, ammonium, and nitrate. It is antici- Formation factor values increase with sediment
pated that the development of other sensors will depth due to the compaction of sediment particles.
continue and the types of measurements possible in This decrease in resistivity implies a decrease in
the future will continue to expand. sediment porosity and increase in sediment tortuosity,
An example of oxygen, resistivity (formation both of which tend to decrease the effective diffusion
factor), and pH results obtained from a sensor lander coefficient with depth in the sediments.
180 PLATFORMS: BENTHIC FLUX LANDERS

Station G, 4685 m
(a) Formation factor (b) Oxygen (μmol kg−1) (c)
∇
pH

100
120
140
160
180
200
220
240
260
80
0.5 1 1.5 2 2.5 3 −0.06 −0.02 0 0.02 0.06

−1

3
Depth (cm)

10

Figure 4 Examples of deep-sea-floor microlectrode oxygen, pH, and resistivity results obtained by Dr. Burke Hales (Oregon State
University). Different symbols display simultaneous results from separate, individual electrodes installed a few centimeters apart on
the instrument.

These results can be used to estimate the diffusive It is important to note that the flux equation
flux across the sediment–water interface and to shown above is not limited to the sediment surface
evaluate the reaction rate within the measured depth but rather can be used to evaluate the diffusive flux
interval. The benthic flux can be estimated from Fick’s at any depth horizon within the profile. Thus, it is
first law corrected for porosity and the effects of often useful to define a sediment layer of a small
sediment particles on diffusion rates as shown below: thickness and calculate the diffusive fluxes at the top
and bottom of the layer. Assuming that horizontal
Benthic flux ¼ fDs dC=dz diffusive exchange can be neglected and that the
profile is in steady state, the difference in these fluxes
where f ¼ surface porosity, Ds ¼ effective sediment is a measure of the net production or consumption
diffusion coefficient, and dC/dz ¼ vertical concen- rate of the measured solute within the layer. Thus, by
tration gradient at the sediment–water interface. interpreting the vertical variations in the flux, one
Porosity near the sediment surface is generally es- can evaluate the distributions of reactions in the
timated from a regression of the measured resistivity sediments and potentially make inferences about the
and directly measured porosities at wider-spaced processes and mechanisms controlling solute dia-
depth intervals. The latter are generally determined genesis and benthic flux.
from the weight loss upon drying bulk sediments.
The effective diffusion coefficient can be estimated
Advantages, Limitations, and Design Concerns
by dividing the molecular diffusion coefficient by
the porosity and tortuosity (i.e., Ds ¼ D/(fy), where Unlike benthic flux chambers that require a signifi-
D ¼ molecular diffusion in free solution, and cant incubation time, sensor landers can obtain a
y ¼ tortuosity). This expression requires independent profile relatively quickly, usually within 1–2 h. The
measurements of porosity and tortuosity at the same exact required time is determined by the number of
vertical scale as the concentration profile. Such sampling depths and the response time of the sensors
measurements are often not available. A common employed. Thus, sensor landers can be used to obtain
alternate but less-accurate strategy in fine-grained in situ profiles relatively rapidly and estimate dif-
sediments is to approximate the effective diffusion fusive fluxes at the deep-sea floor.
coefficient as the molecular diffusion coefficient times There are also several limitations to this approach.
the square of the porosity. This relationship has been Because the measurements are generally made within
empirically derived from numerous individual studies. several hours of the instrument reaching the seafloor,
 PLATFORMS: BENTHIC FLUX LANDERS 181

the accuracy of the flux estimate can be severely horizontal variations that may be caused by bur-
compromised by physical disturbance of the sedi- rowing organisms or physical inhomogeneities. Be-
ment surface. Potential disturbance can be caused by cause the geochemical questions being asked often
the instrument itself or by the bow wave that pre- require knowledge of the mean benthic flux for a
cedes the instrument as it settles. Video recordings of known area or region, numerous replicate profiles
these instruments settling onto the bottom reveal are often required to estimate the average profile and
significant resuspension of surface sediments. While benthic flux.
the resuspended sediments are generally allowed to
settle back to the sea floor or be advected away prior Special Landers and Lander-Based
to making measurements, the effect of this disturb-
ance on the profiles is still a concern. The only in-
Research Strategies
strument to completely avoid this potential problem In addition to the basic landers discussed above, a
is the ROVER lander (discussed in the next section). variety of instruments have been developed in the
Maintaining natural hydrodynamic conditions last decade for special purposes and it is likely that
within the benthic boundary layer is also critical to new types will continue to be developed in the future.
the accuracy of microelectrode flux estimates. Unlike For example, simple benthic flux chamber landers
benthic flux chamber incubations that extend over have been developed to measure advective pore
time intervals sufficient to return to initial conditions water flows around hydrothermal vent and mid-
if diffusive boundary layer thicknesses are altered, ocean ridge systems. In these types of chambers, os-
electrode measurements are rapid and would motic pumps are used to continuously add a tracer
record transient conditions if made directly after and remove a sample. Pore water advection rates as
altering bottom hydrodynamic conditions. Because low as 0.1 mm yr  1 can be measured with this sys-
the instrument frame and electrodes themselves may tem. For seafloor microbial studies, a lander has been
alter bottom flow, such changes are a concern. developed capable of injecting radiolabeled tracers
Since the electrode sensing tips are very small continuously throughout the upper 70 cm of the
(generally 5–20 mm in diameter) as required to sediment column. The sediments surrounding the
achieve fine vertical resolution, they also respond to line of injection are cored and recovered at the end of

Location flag

Time-lapse camera

Flotation sphere

Strobe

Microprofiler
Bumper
Benthic chamber

Roller

Treads

Figure 5 The ROVER lander developed by Dr. Kenneth L. Smith, Jr. (Scripps Institution of Oceanography).
182 PLATFORMS: BENTHIC FLUX LANDERS

a preset incubation period and returned to the ship of remote sampling procedures and incubation ex-
for analysis. Other lander instruments are being de- periments on the seafloor. These capabilities have
veloped to address specific research questions and to greatly improved our understanding of the benthic
operate in particularly demanding environments. For processes, especially at abyssal depths where pres-
example, the benthic exchange of certain trace metals is sure- and temperature-driven artifacts have hindered
sensitive to the redox conditions of the sediment–water earlier studies.
interface. To accurately measure benthic fluxes of these
metals, the ‘oxystat’ lander has been developed in
which the chamber waters are circulated through gas-
permeable tubing in contact with bottom waters.
Exchange of oxygen across the tubing wall keeps the See also
oxygen concentrations in the chambers at near-natural
levels, minimizing potential changes to the metal Grabs for Shelf Benthic Sampling. Sensors for
Micrometeorological and Flux Measurements.
fluxes. Another lander device is the ‘integrated sedi-
ment disturbing lander’, in which a set of small plow
blades is rotated through the surface sediments within
the chamber to evaluate the impact of sediment dis-
turbance, such as would occur following a bottom
Further Reading
trawl or large storm, on benthic solute exchange. Berelson WM, Hammond DE, Smith KL, Jr., et al. (1987)
Perhaps the most innovative lander recently de- In situ benthic flux measurement devices: Bottom lander
veloped is the ROVER (Figure 5). ROVER is capable technology. MTS Journal 21: 26--32.
of performing repeated benthic flux chamber incu- Berg P, Roy H, Janssen F, et al. (2003) Oxygen uptake by
bations and microelectrode profiling measurements. aquatic sediments with a novel non-invasive eddy-
Most importantly, after a measurement cycle is com- correlation technique. Marine Ecology Progress Series
plete, the instrument uses a tractor-tread propulsion 261: 75--83.
Greeff O, Glud RN, Gundersen J, Holby O, and Jorgensen
system to move c. 5 m, so that the next measurement
BB (1998) A benthic lander for tracer studies in the sea
is performed on a natural, undisturbed surface. Since bed: In situ measurements of sulfate reduction.
this instrument is crawling laterally on the sediment Continental Shelf Research 18: 1581--1594.
surface, it eliminates the potential disturbances dis- Jahnke RA and Christiansen MB (1989) A free-vehicle
cussed for free-fall instruments. ROVER was con- benthic chamber instrument for sea floor studies. Deep-
structed to examine temporal variations in seafloor Sea Research 36: 625--637.
fluxes and is capable of performing duplicate benthic Reimers CE (1987) An in situ microprofiling instrument
flux chamber and microelectrode profiling measure- for measuring interfacial pore water gradients: Methods
ments at 30 individual sites over a 6-month period on and oxygen profiles from the North Pacific Ocean.
a single deployment. There is currently a significant Deep-Sea Research 34: 2019--2035.
effort underway to develop and install seafloor cabled Sayles FL and Dickinson WH (1991) The ROLAI2D
lander: A benthic lander for the study of exchange
observatories. It is likely that the ROVER-type of
across the sediment–water interface. Deep-Sea Research
instrumented systems will in the future be attached to 38: 505--529.
seafloor cable nodes and observatories and provide Smith KL, Jr., Glatts RC, Baldwin RJ, et al. (1997) An
long-term, near-real-time observations. autonomous bottom-transecting vehicle for making
Finally, an eddy correlation instrument is being long time-series measurements of sediment community
developed in which the benthic oxygen flux is as- oxygen consumption to abyssal depths. Limnology and
sessed by simultaneously directly measuring the Oceanography 42: 1601--1612.
concentration and velocity normal to the seafloor of Tengberg A, De Bovee F, Hall P, et al. (1995) Benthic
a small volume of water above the sediment surface chamber and profiling landers in oceanography – a
by using an acoustic Doppler velocimeter and an review of design, technical solutions and functioning.
oxygen microelectrode in concert. By integrating, for Progress in Oceanography 35: 253--294.
Tengberg A, Stahl H, Gust G, et al. (2004) Intercalibra-
a sufficient length of time, the instantaneous fluxes
tion of benthic flux chambers. Part I: Accuracy of
toward and away from the sediments, the net benthic flux measurements and influence of chamber hydro-
flux can be estimated. dynamics. Progress in Oceanography 60: 1--28.

Conclusions Relevant Website

In conclusion, bottom lander technology developed http://www.cobo.org.uk
in the last several decades now permits a wide variety – Coastal Ocean Benthic Observatory.
 REMOTELY OPERATED VEHICLES (ROVS)
K. Shepherd, Institute of Ocean Sciences, British chemistry. Depths for this work range from a few
Colombia, Sidney, Canada metres to 10 000 m.
Copyright & 2001 Elsevier Ltd.

Basic Design Characteristics

ROV systems are built in many different configur-
ations and sizes. However, there are many common
Introduction design characteristics that consist of some or all of
the components described in the sections below.
Remotely operated vehicles (ROVs) are vehicles that
are operated underwater and remotely controlled
from the surface. All types of this vehicle are con- Vehicle
nected to the surface platform by a cable that pro- Vehicles range in size from 20 cm in length and a
vides power and control communication to the mass of a few kilograms, to several metres in length
vehicle. There are three basic types of vehicle: and masses of thousands of kilograms. The vehicle
free-swimming tethered vehicles; bottom-crawling itself can be broken down into several subsystems.
tethered vehicles; and towed vehicles. The free-
swimming vehicle is the most common. It has
thrusters that allow maneuvering in three axes, and Frame The vehicle frame is typically an open
provides visual feedback through onboard video frame constructed of aluminum. Components are
cameras. It is often used for mid-water or bottom bolted to the frame. The frame provides structural
observation or intervention. Bottom-crawling ve- support and protection, and provides a method of
hicles move with wheels or tracks and can only connecting the buoyancy, propulsion, and other
maneuver on the bottom. Visual feedback is pro- vehicle systems.
vided by onboard video cameras. Bottom crawlers
are usually used for cable or pipeline work, such as
Buoyancy Buoyancy control is critical to the
inspection and burial. Towed vehicles are carried
proper performance of the vehicle. ROVs typically
forward by the surface ship’s motion, and are man-
have fixed buoyancy provided by syntactic foam, or
euvered up and down by the surface-mounted winch.
some other type of noncompressible foam. This
Towed vehicles usually carry sonar, cameras, and
flotation counteracts the weight of the vehicle frame
sometimes sample equipment.
and mechanical components. Smaller variations in
Remotely operated vehicles were first introduced
buoyancy are provided by vertical thrusters. This
to the offshore community in 1953. Over the next 22
type of vehicle is usually ballasted so that it will
years, several more vehicles were built to fulfill
float to the surface if the tether is accidentally
military and other government research require-
severed. This also improves operations, as the
ments. In 1975, the first commercial vehicle was built
vertical thrusters are usually forcing the vehicle
for the offshore oil industry. Since 1975, over 90% of
down, with the thruster wash moving upward away
the ROVs produced have been developed for com-
from the bottom. If the thrust is directed toward the
mercial offshore work that includes oil and gas
seabed, the silt is easily stirred up, destroying
drilling support, as well as pipeline and tele-
visibility.
communications cable inspection, burial, and repair.
As the depths for oil exploration and production
have increased, the commercial ROV industry has Propulsion The propulsion system consists of
been pressed to keep pace. Current exploration thrusters that control the vehicle motion in three
depths are now reaching 3000 m. The remaining axes. A minimum of two fore/aft thrusters control
vehicles are used to support military and scientific forward and reverse motion and speed and, by the
research and intervention. Military applications in- direction of thrust, the vehicle heading. Vertical
clude submarine rescue, mapping, reconnaissance, thrusters control the vertical motion of the vehicle.
recovery, and mine countermeasures. Scientific ap- Lateral thrusters may be used to allow the vehicle to
plications are far-ranging and cover many different maneuver sideways while maintaining a constant
fields including biology, physics, geology, and heading.

183
184 REMOTELY OPERATED VEHICLES (ROVS)

Vision Video cameras contained in pressure-proof Tether Management System (TMS)

housings with acrylic or glass faceplates are the
Some ROVs operate with a neutrally buoyant tether
primary source of vision. Multiple cameras are used
cable connecting them directly to the ship or work
on larger vehicles to give a wider field of view, or a
platform. The buoyancy of this cable can be modified
different perspective. High-resolution, state-of-the-
by adding floats and weights. A common alternative
art television broadcast-quality cameras are now
approach for larger vehicles is to use a tether man-
being integrated to provide high-quality images that
agement system (see Figure 1). The TMS can be de-
some clients require. In some cases, stereo vision is
signed in several different configurations.
implemented to help improve spatial awareness and
operator efficiency. • One approach is to use a ‘cage’, which houses the
vehicle for launch and recovery, and has a winch
that pays out or retracts tether as needed. The
Control Control of the vehicle is most often vehicle is clamped into the cage and is launched
implemented with computer control. A computer from the support vessel. The main winch, moun-
on the surface communicates with a computer ted on the support vessel, lowers the complete
mounted on the vehicle. Control input, from human package to the working depth and then suspends
or computer, is fed into the surface control it several meters above the worksite. The vehicle is
computer. The vehicle computer then issues the released, tether is paid out by the TMS, and the
control commands and provides feedback to the ROV flies out to perform its work. Upon com-
operator. This system is referred to as a telemetry pletion of the work, the vehicle returns to the cage
system. A second type of control, most often found and is clamped in place, and the complete package
on small, less sophisticated vehicles, is hardwire is recovered.
control. In this case the vehicle thrusters, lights, etc., • The ‘top hat’ configuration has a smaller TMS
are wired directly to surface controls. This with an integral winch that sits on top of the ve-
eliminates the requirement for control computers hicle. Once at operating depth the ROV unlatches
but restricts the amount of control that can be from the TMS and descends to the worksite. Upon
implemented, and can also limit the tether length. completion of the work, the ROV latches to the
bottom of the TMS and the complete package is
recovered.
Manipulators ROVs are usually fitted with some
type of manipulator. Smaller vehicles, if fitted with a
manipulator, will often carry a small arm with one Tether Cable
or two functions. Large vehicles will be fitted with The term tether is usually used to refer to the cable
two powerful manipulators. These will range from directly connected to the vehicle. The vehicle tether is
simple five-function, rate (the function direction is the greatest advantage that an ROV has over other
either on or off) control arms to complex seven- or types of systems, such as autonomous underwater
eight-function arms with force feedback and vehicles (AUVs) and manned submersibles. It delivers
spatially correspondent control. Manipulator power continuously to the vehicle as well as de-
technology has evolved steadily during the history livering control data. It also allows a tremendous
of the ROV. Reliability and efficiency have volume of data to be transmitted in real time from
improved as a result. the vehicle to the surface. This includes many chan-
nels of high-resolution video, acoustic sonar data,
vehicle feedback information, and other data. The
Other sensors ROVs are usually fitted with tether is also the greatest liability of the ROV: it is
additional sensors. Scanning sonars are common adversely affected by currents, it has high drag, and it
and give an acoustic image of the area surrounding can easily be damaged during operations. It is most
the vehicle. The range of the sonar will vary often neutrally buoyant by design, or is made neutral
depending upon the system used, but generally it by adding floats.
will reach past 100 m – well beyond the visual range
of the cameras. Altimeters are similar to echo
Umbilical Cable
sounders and give vehicle height above the bottom.
Depth sensors are implemented on nearly every The term umbilical usually refers to the cable, com-
vehicle. They range from precision sensors to hand- monly steel armored, that connects the support vessel
held units strapped to the vehicle frame in front of to the TMS. This cable will have a fiberoptic bundle,
the camera. or coaxial cable for command, control, and data
 REMOTELY OPERATED VEHICLES (ROVS) 185

Launch and recovery system

(LARS)

CSSF

Umbilical

Tether management system

(TMS)

Tether

Remotely operated vehicle

(ROV)

Figure 1 Use of the tether management system for larger vehicles.

transmission. This core will be surrounded by power Surface Control Station(s)

conductors, used to provide the vehicle with power.
Surface control stations usually contain at a min-
Finally, it will have a protective jacket and steel or
imum a video monitor, videocassette recorder, and
synthetic strength member. This cable will be paid in
joystick for vehicle control. As systems become lar-
and out from a deck mounted winch, to control the
ger and more complex, the amount of surface
depth of the TMS or vehicle.
equipment grows to include electrical distribution
systems, surface control computers, and consoles for
copilots and navigators.
Launch and Recovery System (LARS) and Winch
Control System
Most ROV systems come complete with an inte-
Control systems cover as wide a range of design as
grated LARS. Small vehicles can be deployed and
there are vehicles. The control systems can be broken
recovered by hand, while medium to large vehicles
down into two basic types.
employ either a crane or an A-frame. Large systems
typically have a purpose-built LARS that is inte- • Hardwired control. In this configuration each in-
grated with the umbilical winch. With a self-con- dividual ROV component is connected directly to
tained system the vehicle can be installed upon many the surface, through the tether, with its own set of
different platforms that are not equipped with launch dedicated wires. This approach is simple, robust,
and recovery gear. and inexpensive. It does limit tether length and
186 REMOTELY OPERATED VEHICLES (ROVS)

increases the wire count in the tether and the components such as transformers or hydraulic com-
amount of control that can be implemented. ponents are mounted either in plastic or in thin-wall
• Computer telemetry system. Computer control aluminum housings. The housings are then oil filled
allows a tremendous increase in the control and connected to a soft bladder. As the external
available for the vehicle. Wiring for the vehicle pressure increases, it presses onto the soft bladder.
can be reduced to power and one pair of control The oil in the bladder compresses somewhat, thus
wires, or fiberoptic cable. Video and sonar data equalizing the internal and external pressures. The
are still typically brought back discretely on their advantages of this type of housing are reduced
own fiber or signal wires. weight and cost, both significant design constraints.
Vehicles must be built with corrosion-resistant
materials. Aluminum is commonly used owing to its
Portable Design light weight, but it will eventually corrode. Titanium,
Almost all ROV systems are designed to be portable. stainless steel, and plastics are much more corrosion-
This allows them to be installed on ships or plat- resistant, but may have problems in specific
forms of opportunity in various ports around the applications.
world. When an operation is complete, they can be The system components that remain at the surface
demobilized and returned to a shore-based work area also must perform reliably in an extreme environ-
for maintenance and storage. The term ‘portability’ ment. The high-vibration and corrosive, wet atmos-
is stretched when referring to the large systems that phere of the exposed deck has led to the design of
weigh tens of tonnes, but with proper port facilities many components rated for marine duty. While ex-
these systems can be removed and installed on a pensive, these components will operate reliably
variety of vessels. under such conditions.
The human operators of ROVs must also with-
stand these harsh conditions. ROV personnel must
work long hours in a continually moving environ-
Challenges and Solutions
ment, often in wet and cold conditions. The systems
Remotely operated vehicles work in an extreme en- use high voltages, harsh oils, lubricants, and other
vironment. While working at depth they are subject dangerous substances. The pressure to perform well
to high external pressure, particularly as depths in- is high because often ROV work is carried out upon
crease. Sea water is also corrosive and electrically expensive installations that cannot afford downtime
conductive. Ships also present a high-motion and for repairs and maintenance. The complete ROV
high-vibration environment. ROV manufacturers spread, including the support vessel, is expensive to
and operators have dealt with these challenges in hire and there is no tolerance for unreliable people or
several ways, as described below. vehicles.
Some components must be protected from the
pressure and water by being mounted in a pressure-
proof housing. Pressure-proof housings are typically
Scientific Research Vehicles
made of a corrosion-resistant material such as
stainless steel or anodized aluminum. As greater Remotely operated vehicles have been supporting
pressures are encountered, the strength of these two scientific operations since the mid-1980s. Some
materials is no longer adequate and housings are ROVs were originally funded to complement man-
made from more exotic materials, such as titanium, ned submersible work, but a few were developed as
composites, or ceramics. replacements for existing submersibles, or as stand-
Electrical components such as cameras, lights and alone vehicles for smaller institutions. The strengths
sonars are mounted outside the main pressure and weaknesses of ROVs do not allow them to be
housings. They must be connected to the main tele- direct replacements for manned submersibles.
metry pressure housing by an electrical cable. The Manned submersibles (See Manned Submersibles,
cable penetrations, where the wires enter the pres- Deep Water, Manned Submersibles, Shallow Water).
sure proof housings, must be carefully designed. refer to manned vehicle article) were the dominant
Improper design can result in cables being extruded technology for ocean floor scientific research for
into the housing or, worse, failure of the seal and decades. ROVs have entered the field, and have
flooding of the housing. gained acceptance because of their distinct advan-
Pressure-compensated housings are often used for tages in many areas. They have unlimited power and
components that can withstand the pressure but re- can therefore remain on the bottom for extended
quire protection from the water. In this case, periods, efficiently performing large surveys,
 REMOTELY OPERATED VEHICLES (ROVS) 187

extended time series experiments, and multi- between the many elements that can be incorporated
disciplinary operations. A tremendous volume of into an ROV.
data is transmitted to the surface, with many chan-
nels of real time video, sonar, CTD (conductivity–
temperature–depth) data, and other information. In
fact, properly managing the data can be a challenge.
Conclusion
Many scientists can participate in the operations, The efficiency of ROVs will continue to improve in
which is an advantage. Operations often cover many two major ways. (i) The efficiency of the work will
disciplines, often with unexpected results. Key peo- improve with better integration of ROV capabilities
ple can always be on hand to discuss and decide into offshore component design. (ii) The efficiency of
upon modifications to the operational plan as the the ROV itself will also improve with advances in
operation unfolds. Some of the advantages that hydraulic components and design, control system
manned submersibles have will be difficult to replace components and design, and higher-voltage cables
with the ROV. It is difficult to replace the human eye and motors. Electric vehicles that do not have large
with remote telepresence. The surface ship motion hydraulic systems are beginning to enter the market.
and control will always influence the ROV Altogether, this will result in smaller, lighter cables,
operations. which will reduce systems size and cost and will have
The current (as at 2000) high-profile science ROVs less effect upon the vehicle as it is operating in cur-
are briefly described in Table 1. Each of these ve- rents, or traveling at speed. The multidisciplinary
hicles is unique. Some have been developed with a vehicle will remain as the dominant vehicle type, but
specific focus, and are therefore better at some tasks specialized vehicles will also become more wide-
than others. Every vehicle design is a compromise spread as more and more tasks are assigned to ROVs

Table 1 Summary of some scientific ROVs and their characteristics

Propulsion TMS Operatora Power Manipulators Depth

Jason/Medea http://www.marine.whoi.edu/ships/rovs/jason_med.htm
Electric Depressor weight; WHOI 9 kW Single electric 6000 m; significant
50 m fixed tether length manipulator vehicle upgrades
planned for 2002

ROPOS http://www.ropos.com
Hydraulic Cage; 250 m tether CSSF 22 kW Two hydraulic 5000 m
manipulators

Tiburon http://www.mbari.org/dmo/vessels/tiburon.html)
Electric None MBARI 15 kW Two hydraulic 4000 m
manipulators

Ventanna http://www.mbari.org/dmo/ventanna/ventanna.html
Hydraulic None MBARI 30 kW Two hydraulic 2000 m
manipulators

Victor http://www.ifremer.fr/victor/victor_uk.html
Electric Depressor weight IFREMER 20 kW Two 6000 m

Dolphin 3K http://www.jamstec.go.jp/jamstec-e/rov/3k.html
Hydraulic None JAMSTEC ? Two 3300 m

HYPER-DOLPHIN http://www.jamstec.go.jp/jamstec-e/rov/hyper.html
Hydraulic None JAMSTEC 56 kW Two 3000 m

KAIKO http://www.jamstec.go.jp/jamstec-e/rov/kaiko.html
Hydraulic None JAMSTEC ? Two 11 000 m

a
WHOI, Woods Hole Oceanographic Institution; CSSF, Canadian Scientific Submersible Facility; MBARI, Monterey Bay Aquarium
Research Institute; IFREMER, l’Institut Francais de Recherche pour l’Exploitation de la Mer; JAMSTEC, Japan Marine Science and
Technology Center
188 REMOTELY OPERATED VEHICLES (ROVS)

and the scope of work increases. Designers of off- See also

shore equipment are more commonly incorporating
ROV intervention technology into the original Manned Submersibles, Deep Water. Manned
Submersibles, Shallow Water. Towed Vehicles.
equipment. This has great benefits in improving
ROV efficiency. For many years, ROVs have been
challenged with attempting to work with com-
ponents designed for human hands or for dry land Further Reading
manipulation. Once thought and design are applied
to ROV intervention techniques, all parties benefit Marine Technology Society (1984) ROV’84 Technology
Update – An International Perspective. Proceedings of
from the increased efficiency. ROVs have evolved,
ROV’84 Conference and Exposition, San Diego. San
and are still evolving, to fill a requirement for re- Diego: Marine Technology Society.
liable, efficient vehicles in an environment that is Vadus JR and Busby RF (1979) Remotely Operated
inaccessible to humans. As these vehicles develop, Vehicles: An Overview. Washington, DC: US
and the engineering progresses on the vehicles as well Department of Commerce, National Oceanographic
as on their worksites, they will continue to fulfill a and Atmospheric Administration, Office of Ocean
unique and expanding role in the underwater world. Engineering.
 SEISMOLOGY SENSORS
L. M. Dorman, University of California, San Diego, imaging experiments which have been heretofore
La Jolla, CA, USA associated with the petroleum exploration industry.
Copyright & 2001 Elsevier Ltd. This contrasts sharply with the severely under-
determined experiments which have characterized
earthquake seismology. One change over the past
decade has been the disappearance of analog re-
cording systems.
A side effect of this rapid change is that a review
Introduction such as this provides a snapshot of the technology,
A glance at the globe shows that the Earth’s surface is rather than a long-lasting reference. The technical
largely water-covered. The logical consequence of details reported below are for instruments at two
this is that seismic studies based on land seismic stages of development: existing instruments (UTIG,
stations alone will be severely biased because of two SIO/ONR, SIO/IGPP-SP, GEOMAR, LDEO-BB) and
factors. The existence of large expanses of ocean instruments still in design and construction (WHOI-
distant from land means that many small earth- SP, WHOI-BB SIO/IGPP-BB) (see Tables 1 and 2;
quakes underneath the ocean will remain un- Figures 1–6). The latter construction project has the
observed. The difference in seismic velocity structure acronym ‘OBSIP’ for OBS instrument pool, and
between continent and ocean intruduces a bias in sports a polished, professionally designed web site at
locations, with oceanic earthquakes which are lo- http://www.obsip.org.
cated using only stations on one side of the event OBS designs are roughly divided into two cat-
being pulled tens of kilometers landward. Addition- egories which for brevity will be called ‘short-period’
ally, the depths of shallow subduction zone events, (SP) and ‘broadband’ (BB). The distinction blurs at
which are covered by water, will be very poorly de- times because some instruments of both classes use a
termined. Thus seafloor seismic stations are neces- common recording system, a possibility which
sary both for completeness of coverage as well as for emerges when a high data-rate digitizer has the
precise location of events which are tectonically im-
portant. This paper summarizes the status of seafloor
seismic instrumentation.
The alternative methods for providing coverage
are temporary (pop-up) instruments and perman-
ently connected systems. The high costs of seafloor
cabling has thus far precluded dedicated cables of
significant length for seismic purposes, although ef-
forts have been made to use existing, disused wires.
Accordingly, the main emphasis of this report will be
temporary instruments.
Large ongoing programs to investigate oceanic
spreading centers (RIDGE) and subductions (MAR-
GINS) have provided impetus for the upgrading of
seismic capabilities in oceanic areas.
The past few years has seen a blossoming of
ocean bottom seismograph (OBS) instrumentation,
both in number and in their capabilities. Active
experimental programs are in place in the USA,
Europe, and Japan. Increases in the reliability of
electronics and in the capacity of storage devices has
allowed the development of instruments which are
Figure 1 The UTIG OBS, a particularly ‘clean’ mechanical
much more reliable and useful. Major construction design, which has been in use for many years, with evolving
programs in Japan and the USA are producing electronics. The anchor is 1.2 m on each side. (Photograph by
hundreds of instruments, a number which allows Gail Christeson, UTIG.)

189
190 SEISMOLOGY SENSORS

Table 1 Characteristics of short period ocean bottom seismometers

 SEISMOLOGY SENSORS 191

Table 2 Characteristics of broadband ocean bottom seismometerrs

capability of operation in a low-power, high en- commonly only the vertical component, and/or
durance mode. hydrophones, may have somewhat limited recording
capacity and endurance, and are typically used in
active-source seismic experiments and for micro-
Short Period (SP) Instruments earthquake studies with durations of a week to a few
The SP instruments (Table 1) are light in weight and months. Two types (UTIG and JAMSTEC) are sin-
easy to deploy, typically use 4.5 Hz geophones, gle-sphere instruments.
192 SEISMOLOGY SENSORS

Top view (floats removed)

L22 Acoustic
mount release Hydrophone mount

0.9 m

Data
logger Dual burnwire release

0.9 m
Top of
lifting ring

Float
(4 × 13′D
glass balls)
Acoustic
0.9 m release

Data
logger

Bar grate
anchor
(3′ × 3′)
0.9 m 0.9 m
Side view End view

Figure 2 The IGPP-SP instrument. (Figure from Babcock, Harding, Kent, and Orcutt.)

D2 OBH/S

VHF radio
antenna Hydrophone and
recall transducer
Flashers
inside

Electronics
104 cm sphere Polyethylene
housing
Battery
sphere

Anchor release

Geophone package

Figure 3 The WHOI-SP instrument. The change of orientation between seafloor and surface modes allows the acoustic transducer
an unobstructed view of the surface while on the seafloor and permits acoustic ranging while the instrument is on the surface. (Figure
by Beecher Wooding and John Collins.)
 SEISMOLOGY SENSORS 193

Figure 4 The GEOMAR OBH/S. The shipping/storage container is equipped with an overhead rail so that it serves as an instrument
dispenser. The OBS version is shown here. (Photograph by Michael Tryon, UCSD.)

Broadband (BB) Instruments instruments are designed in two parts, the main
section contains the recording package, and release
The BB instruments (Table 2) provide many features and recovery aids, while the sensor package is
of land seismic observatories, relatively high dy- physically separated from the main section. This
namic range, excellent clock stability –o1 ms d1 configuration allows isolation from mechanical noise
drift. This class of instruments can be equipped with and and permits tuning of the mechanical resonance
hydrophones useful down to a millihertz. The BB of the sensor–seafloor system.
194 SEISMOLOGY SENSORS

Figure 5 The SIO-ONR OBSs (Jacobson et al. 1991, Sauter et al, 1990) being launched in Antarctica. The anchor serves as a
collector for the CAT fluid fluxmeter (Tryon et al. 2001). The plumbing for the flowmeter is in the light-colored box at the right-hand end
of the instrument. (Photograph by Michael Tryon, UCSD.)

Burnwire release for

sensor sphere

Glass flotation
spheres
Sensor
sphere

Acoustic
transponder
release

Recording
package

Figure 6 The LDGO-BB OBS. This is based on the Webb design in use during the past few years. The earlier version established a
reputation for high reliability and was the lowest noise OBS and lightest of its time period. The main drawback of the earlier version
was its limited (16-bit) dynamic range. (Figure from S. Webb, LDEO.)
 SEISMOLOGY SENSORS 195

0.8 Pressure (0.01 to 0.08 Hz)

0.6

0.4
Amplitude (Pa)

0.2
2 km
0 3 km
5 km
_ 0.2

_ 0.4

_ 0.6

_ 0.8
0 20 40 60 80 100
Time (s)

Figure 7 Synthetic seismograms of pressure at three ocean depths.

Displacement (from 0.01 to 0.08 Hz)

0.25
0.20
0.15
Amplitude (μm)

0.10
0.05
2 km
0 3 km
_ 0.05 5 km
_ 0.10
_ 0.15
_ 0.20
0 20 40 60 80 100
Time (s)

Figure 8 Synthetic seismograms of vertical motion at the same depths as Figure 7; note the reduction in the effects of the water
column reverberations. (From Lewis and Dorwan, 1998).

Sensor Considerations lower frequencies. However, lower frequency sensors

are affected, since a soft foundation permits tilt ei-
Emplacement of a sensor on the seafloor is almost ther in response to sediment deformation by the
always suboptimal in comparison with land stations. weight of the sensor or in response to water currents.
Instruments dropped from the sea surface can land The existing Webb instruments combat this problem
tens of meters from the desired location. The seafloor by periodic releveling of the sensor gimbals. The
material is almost always softer than the surficial PMD sensors have an advantage here in that the
sediments (these materials have shear velocities as mass element is a fluid and the horizontal com-
low as a few tens of meters per second. The sensor is ponents self-level to within 51.
thus almost always poorly coupled to the seafloor Why not use hydrophones then? These make lev-
and sensor resonances can occur within the fre- eling unnecessary and are more robust mechanically.
quency range of interest for short period sensors. In terms of sensitivity, they are comparable to seis-
Fortunately, these resonances have little effect on mometers. The disadvantage of hydrophones lies in
196 SEISMOLOGY SENSORS

the physics of reverberation in the water layer. Dorman LM (1997) Propagation in marine sediments.
A pulse incident from below is reflected from the sea In: Crocker MJ (ed.) Encyclopedia of Acoustics,
surface completely, and when it encounters the sea- pp. 409--416. New York: John Wiley.
floor it is reflected to a significant degree. Since the Dorman LM, Schreiner AE, and Bibee LD (1991) The
seafloor has a higher acoustic impedance than water, effects of sediment structure on sea floor noise.
In: Hovem J, et al. (eds.) Proceedings of Conference on
the reflected pressure pulse has the same sign as the
Shear Waves in Marine Sediments, pp. 239--245.
incident pulse and the signal is large. However, the Dordrecht: Kluwer Academic Publishers.
seafloor motion associated with pressure pulses tra- Duennebier FK and Sutton GH (1995) Fidelity of ocean
veling in opposite directions is opposite in sign, so bottom seismic observations. Marine Geophysical
cancellation occurs. Unfortunately, the frequency Researches 17: 535--555.
range in which these reverberations are troublesome Jacobson RS, Dorman LM, Purdy GM, Schultz A,
is in the low noise region. Figures 7 and 8 show and Solomon S (1991) Ocean Bottom Seismometer
synthetic seismograms of pressure and vertical Facilities Available, EOS, Transactions AGU, 72,
motion illustrating this effect. pp. 506, 515.
Sauter AW, Hallinan J, Currier R et al. (1990) Proceedings
of the MTS Conference on Marine Instrumentation, pp.
Further Reading 99–104.
Tryon M, Brown K, Dorman L, and Sauter A (2001) A new
Barash TW, Doll CG, Collins JA, Sutton GH, and Solomon benthic aqueous flux meter for very low to moderate
SC (1994) Quantitative evaluation of a passively leveled discharge rates. Deep Sea Research (in press).
ocean bottom seismometer. Marine Geophysical Webb SC (1998) Broadband seismology and noise under
Researches 16: 347--363. the ocean. Reviews in Geophysics 36: 105--142.
 VEHICLES FOR DEEP SEA EXPLORATION
S. E. Humphris, Woods Hole Oceanographic explore the ocean depths. For many years, re-
Institution, Woods Hole, MA, USA searchers have towed instruments near the seafloor
& 2009 Elsevier Ltd. All rights reserved. to collect various kinds of data (e.g., acoustic, mag-
netic, and photographic) remotely. With the devel-
opment of sophisticated acoustic and imaging
systems designed to resolve a wide range of ocean
floor features, towed vehicle systems have become
Introduction increasingly complex. Some now use fiber-optic, ra-
Exploring the deep sea has captured the imagination ther than coaxial, cable as tethers and hence are able
of humankind ever since Leonardo da Vinci made to transmit imagery as well as data in real time.
drawings of a submarine more than 500 years ago, Examples of deep-towed vehicle systems are included
and Jules Verne published 20 000 Leagues under the in Table 1 and Figure 2, and they tend to fall into
Sea in 1875. Since the early twentieth century, people two categories. Geophysical systems, such as SAR
have been venturing into the ocean in bathyspheres (IFREMER, France), TOBI (National Oceanography
and bathyscaphs. However, it was not until 1960 that Centre, Southampton, UK), and Deep Tow 4KS
the dream to go to the bottom of the deepest part of (JAMSTEC, Japan), collect sonar imagery, bathym-
the ocean was realized, when Jacques Piccard and a etry, sub-bottom profiles, and magnetics data, as they
US Navy lieutenant, Don Walsh, descended to the are towed tens to hundreds of meters off the bottom.
bottom of the Mariana Trench (10 915 m or 6.8 mil) Imaging systems, such as TowCam (WHOI, USA),
in Trieste (Figure 1). This vehicle consisted of a float Scampi (IFREMER, France), and Deep Tow 6KC
chamber filled with gasoline for buoyancy, and a (JAMSTEC, Japan), are towed a few meters off the
separate pressure sphere for the personnel, allowing bottom and provide both video and digital imagery
for a free dive rather than a tethered one. Containers of the seafloor.
filled with iron shot served as ballast to make the However, since the 1960s, scientists have been
submersible sink. After a 5-h trip to the bottom, and transported to the deep ocean and seafloor in sub-
barely 20 min of observations there, the iron shot was mersibles, or human-occupied vehicles (HOVs), to
released and Trieste floated back to the surface. make direct observations, collect samples, and de-
Since that courageous feat almost 50 years ago, ploy instruments. More recently, two other types of
dramatic advances in deep submergence vehicles and deep submergence vehicles – remotely operated ve-
technologies have enabled scientists to routinely hicles (ROVs) and autonomous underwater vehicles
(AUVs) – have been developed that promise to
greatly expand our capabilities to map, measure, and
sample in remote and inhospitable parts of the
ocean, and to provide the continual presence neces-
sary to study processes that change over time.

Human-Occupied Vehicles
The deep-sea exploration vehicles most familiar to
the general public are submersibles, or HOVs. This
technology allows a human presence in much of
the world’s oceans, with the deepest diving vehicles
capable of reaching 99% of the seafloor.
There exist about 10 submersibles available world-
wide for scientific research and exploration that can
dive to depths greater than 1000 m (Table 2 and
Figure 3). All require a dedicated support ship. These
Figure 1 The bathyscaph Trieste hoisted out of the water
battery-operated vehicles allow two to four individuals
in a tropical port, around 1959. Photo was released by the US
Navy Electronics Laboratory, San Diego, California (US Naval
(pilot(s) and scientist(s)) to descend into the ocean
Historical Center Photograph). Photo #NH 96801: US Navy to make observations and gather data and samples.
Bathyscaphe Trieste (1958–63). The duration of a dive is limited by battery life, human

197
198 VEHICLES FOR DEEP SEA EXPLORATION

Table 1 Examples of deep-towed vehicle systems for deep-sea research and exploration (systems that can operate at depths
Z1000 m)

Vehicle Operating Maximum operating Purpose

organization depth (m)

TowCam WHOI, USA 6500 Photo imagery; CTD; volcanic glass samples;
water samples
Deep-Tow Survey System COMRA, China 6000 Sidescan, bathymetry; sub-bottom profiling
DSL-120A HMRG, USA 6000 Sidescan; bathymetry
IMI-30 HMRG, USA 6000 Sidescan; bathymetry; sub-bottom profiling
Scampi IFREMER, 6000 Photo and video imagery
France
Système Acoustique IFREMER, 6000 Sidescan; sub-bottom profiling; magnetics;
Remorqué (SAR) France bathymetry
SHRIMP NOC, UK 6000 Photo and video imagery
TOBI NOC, UK 6000 Sidescan; bathymetry; magnetics
BRIDGET NOC, UK 6000 Geochemistry
Deep Tow 6KC JAMSTEC, Japan 6000 Photo and video imagery
Deep Tow 4KC JAMSTEC, Japan 4000 Photo and video imagery
Deep Tow 4KS JAMSTEC, Japan 4000 Sidescan; sub-bottom profiling

WHOI, Woods Hole Oceanographic Institution, USA; COMRA, China Ocean Mineral Resources R&D Association; HMRG, Hawai’i
Mapping Research Group, USA; IFREMER, French Research Institute for Exploration of the Sea; NOC, National Oceanographic
Centre, Southampton, UK; JAMSTEC, Japan Marine Science & Technology Center.

(a) (b)

(c) (d)

Figure 2 Examples of deep-towed vehicle systems. (a) SHRIMP, (b) Deep Tow, (c) Tow Cam, and (d) DSL-120A. (a) Courtesy
of David Edge, National Oceanography Centre, UK. (b) & JAMSTEC, Japan, with permission. (c) Photo by Dan Fornari, WHOI, USA.
(d) Courtesy of WHOI, USA.
 VEHICLES FOR DEEP SEA EXPLORATION 199

Table 2 HOVs for deep-sea research and exploration (vehicles that can operate at depths Z1000 m)

Vehicle Operating organization Maximum operating depth (m)

HOV (under construction) COMRA, China 7000

Shinkai 6500 JAMSTEC, Japan 6500
Replacement HOV (in planning stages) NDSF, WHOI, USA 6500
MIR I and II P.P. Shirshov Institute of Oceanology, Russia 6000
Nautile IFREMER, France 6000
Alvin NDSF, WHOI, USA 4500
Pisces IV HURL, USA 2170
Pisces V HURL, USA 2090
Johnson-Sea-Link I and II HBOI, USA 1000

Abbreviations as in Table 1; NDSF, National Deep Submergence Facility; HURL, Hawaii Undersea Research Laboratory; HBOI,
Harbor Branch Oceanographic Institution, USA.

endurance, and safety protocols, and typically does and their exotic communities of organisms, and
not exceed 8–10 h, including transit time to and from continue to be used extensively for investigation of
the working depth (about 4 h for a seafloor depth of these extreme deep-sea environments.
4000 m). (The Russian MIR submersibles are an ex- HOVs will continue to provide important cap-
ception; they operate on a 100-kWh battery that can abilities for deep-sea research at least for the fore-
accommodate dive times in excess of 12 h.) Housed seeable future. Although rapid progress is being
in a personnel sphere (Figure 4), the divers are main- made in videography and photography to develop
tained at atmospheric pressure despite the ever- capabilities that match those of the human eye, there
increasing external pressure with depth (1 atm every is still no substitute for the direct, three-dimensional
10 m). Cameras on pan and tilt mounts with zoom and view that allows divers to make contextual obser-
focus controls are located on the exterior of the ve- vations and integrate them with the cognitive ability
hicles, as well as quartz iodide and/or metal halide of the human brain. In recognition of this continuing
lights to illuminate the area. Submersibles are also need, there are two submersibles that are under
equipped with robotic arms that can be used to ma- construction or in the planning stages. The China
nipulate equipment or pick up samples, and a basket, Ocean Mineral Resources R&D Association
usually mounted on the front of the vehicle, to trans- (COMRA) is constructing their first submersible that
port instruments, equipment, or samples. These ve- will have a maximum operating depth of 7000 m. It
hicles can handle heavy payloads, maintaining neutral is expected to be operational in 2007. In the United
buoyancy as their weight changes through a variable States, over 40 years after the submersible Alvin was
ballast control system. All these capabilities, together delivered in 1964, the National Deep Submergence
with their slow speeds (1–2 knots), make submersibles Facility at Woods Hole Oceanographic Institution is
best suited to detailed observations, imaging, and in the planning stages for a new and improved re-
sampling in localized areas, rather than operating in a placement HOV with an increased operating depth
survey mode. of 6500 m.
Many significant discoveries during the past four
decades of marine research have resulted from ob-
servations and samples taken from submersibles.
Through direct observations from submersibles,
Remotely Operated Vehicles
biologists have discovered many previously unknown Over the past 20 years, marine scientists have begun
animals, and have documented that gelatinous ani- to routinely use ROVs to collect deep-sea data and
mals (cnidarians, ctenophores, etc.) form a dominant samples. ROVs were originally developed for use in
ecological component of mid-water communities. the ocean by the military for remote observations, but
These soft-bodied, fragile animals would have been were adapted in the mid-1970s by the offshore energy
destroyed by the trawl nets used in earlier days to industry to support deep-water operations. There
sample these depths. Submersibles have enabled are many ROVs commercially operated today, ran-
geologists to explore the global mid-ocean ridge sys- ging from small, portable vehicles used for shallow-
tem, and have provided them with a detailed view of water inspections to heavy, work-class, deepwater
the nature of volcanic and tectonic activity during the ROVs used by the offshore oil and gas industry
formation of oceanic crust. Submersibles played an in support of subsea cable laying, retrieval, and
important role in the discovery of hydrothermal vents repair.
200 VEHICLES FOR DEEP SEA EXPLORATION

(a)

(b)

(c)

Figure 3 Examples of HOVs used to conduct scientific research. (a) Shinkai 6500, (b) Sea Link, and (c) Nautile. (a) & JAMSTEC,
Japan, with permission. (b) Courtesy of Harbor Branch Oceanographic Institution, USA. (c) & IFREMER, France, with permission;
O. Dugornay.
 VEHICLES FOR DEEP SEA EXPLORATION 201

Personnel Main ballast

hatch vent Communication High pressure Thrusters
transducer air spheres (2) (1 of 6)
Lifting
T

Video light Sonar

Video camera

Light bar

Variable
ballast
spheres (4)

35-mm
cameras

Video Strobes
cameras
Motor controller
for relay pressure
Starboard vessels
manipulator
Batteries

Variable
ballast sphere

Descent
weights

Pilot view port

Port
manipulator
Sample basket
Ski

Figure 4 Cutaway illustration of the submersible Alvin showing the major components of an HOV. Illustration by E. Paul Oberlander,
WHOI, USA.

There are about a dozen ROVs that are available and exploration and, through the shipboard control
to the international scientific community (Table 3 and room (Figure 6), allow a number of scientists and
Figure 5). While some of these have dedicated sup- engineers to discuss the incoming data and make
port ships, many can operate in the ‘flyaway’ mode; collective decisions about the operations. Another
that is, they can be shipped to, and operated on, a distinct advantage is their ability to remain under-
number of different ships. Unlike the HOVs, ROVs water for extended periods of time because power is
are unoccupied, and are tethered to a support ship provided continuously from the ship. This endurance
usually by a fiber-optic cable that has sufficient means that scientists can make observations over
bandwidth to accommodate a wide variety of periods of many days, instead of a few hours a day,
oceanographic sensors and imaging tools. The cable and gives them the flexibility to react to unexpected
provides power and communications from the ship to events. The disadvantage of an ROV is that its tether
the ROV, allowing control of the vehicle by a pilot on constrains operations because the range of the ve-
board the ship. The pilot can also use the manipu- hicle with respect to the ship cannot exceed a
lator arm(s) to collect samples and perform experi- few hundred meters. Movement of the ship must
ments. The cable transmits images and data from the therefore be carefully coordinated with the move-
ROV to the control room on board the ship where ments of the vehicle – this requires a ship equipped
monitors display the images of the seafloor or water with a dynamic positioning system. In addition, the
column in real time. These capabilities, together with tether is heavy and produces drag on the vehicle,
their excellent power and lift, allow ROVs to perform making it less maneuverable and vulnerable to en-
many of the same operations as HOVs. tanglement in rugged terrain. However, with careful
Obvious advantages of using ROVs are that they tether management, ROVs are well suited to map-
remove the human risk factor from deep-sea research ping and surveying small areas, as well as to making
202 VEHICLES FOR DEEP SEA EXPLORATION

Table 3 ROVs for deep-sea research and exploration (vehicles that can operate at depths Z1000 m)

Vehicle Operating organization Maximum operating depth (m)

Nereus (hybrid) (under construction) NDSF, WHOI, USA 11 000

Kaiko 7000 JAMSTEC, Japan 7000
Isis NOC, UK 6500
Jason II NDSF, WHOI, USA 6500
ATV SIO, USA 6000
CV (Wireline Reentry System) SIO, USA 6000
Victor 6000 IFREMER, France 6000
ROV (on order) NOAA Office of Ocean Exploration, USA 6000
ROPOS CSSF, Canada 5000
Tiburon MBARI, USA 4000
Quest Research Centre Ocean Margins, Germany 4000
Hercules Institute for Exploration, USA 4000
Sea Dragon 3500 COMRA, China 3500
Hyper Dolphin JAMSTEC, Japan 3000
Aglantha Institute of Marine Research, Norway 2000
Ventana MBARI, USA 1500
Cherokee Research Centre Ocean Margins, Germany 1000

Abbreviations as in Tables 1 and 2; SIO, Scripps Institution of Oceanography; CSSF, Canadian Scientific Submersible Facility;
MBARI, Monterey Bay Aquarium Research Institute; SIO, Scripps Institution of Oceanography.

more detailed observations, imaging, and sampling mission while paying out as much as 20 km (about
of specific features. 11 mi) of micro-cable. Once the mission is complete,
While many of the ROVs available to the scientific the HROV will detach from the micro-cable and
community have a wide range of capabilities, a few guide itself to the sea surface for recovery, while
are purpose-built. For example, the Wireline Reentry the micro-cable is recovered for reuse. In 2008–09,
System known as CV, and operated by Scripps In- almost 50 years after the dive of the Trieste, Nereus
stitution of Oceanography, is a direct hang-down will dive to the bottom of the Mariana Trench.
vehicle designed specifically for precision placement
of heavy payloads on the seafloor or in drill holes
(Figure 7). Unlike conventional, near-neutrally
buoyant ROVs, the Wireline Reentry System can
Autonomous Underwater Vehicles
handle payloads of a few thousand kilograms, de- Although the concept of AUVs has been around for
pending on the water depth. It has been used, for more than a century, it is only in the last decade or
example, to install seismometer packages in, and two that AUVs have been applied to deep-sea re-
recover instruments packages from, seafloor drill search and exploration. AUV technology is in a
holes in water depths up to 5500 m, as well as to phase of rapid growth and expanding diversity.
deploy precision acoustic ranging units on the axis of There are now more than 50 companies or insti-
the mid-ocean ridge. tutions around the world operating AUVs for a var-
Another ROV being built at Woods Hole iety of purposes. For example, the offshore gas and
Oceanographic Institution for a specific purpose is oil industry uses them for geologic hazards surveys
Nereus (Figure 8). More correctly referred to as a and pipeline inspections, the military uses them for
hybrid remotely operated vehicle, or HROV, because locating mines in harbors among other applications,
it will be able to switch back and forth to operate as and AUVs have been used to search for cracks in the
either an AUV or an ROV on the same cruise, Nereus aqueducts that supply water to New York City.
will be capable of exploring the deepest parts of the There are currently about a dozen AUVs being
world’s oceans, as well as bringing ROV capabilities used specifically for deep-sea exploration (Table 4
to ice-covered oceans, such as the Arctic. The HROV and Figure 9), although the numbers continue to
will use a lightweight fiber-optic micro-cable, only increase. These unoccupied, untethered vehicles are
1/32 of an inch in diameter, allowing it to operate at preprogrammed and deployed to drift, drive, or glide
great depth without the high-drag and expensive through the ocean without real-time intervention
cables typically used with ROV systems. Once from human operators. All power is supplied by
the HROV reaches the bottom, it will conduct its energy systems carried within the AUV. Data are
 VEHICLES FOR DEEP SEA EXPLORATION 203

(a) (b)

(c) (d)

Figure 5 Examples of ROVs used for deep-sea research. (a) ROV Kaiko, (b) ROV Jason II, (c) ROV Tiburon, and (d) ROV Victor
6000. (a) & JAMSTEC, Japan, with permission. (b) Photo by Tom Bolmer, WHOI, USA. (c) Photo by Todd Walsh & 2006, MBARI,
USA, with permission. (d) & IFREMER, France, with permission; M. Bonnefoy.

recorded and are then either transmitted via satellite Hence, AUVs are frequently used to identify regions
when the AUV comes to the surface, or are down- of interest for further exploration by HOVs and
loaded when the vehicle is recovered. They are gen- ROVs.
erally more portable than HOVs and ROVs and can Unlike HOVs and ROVs that are designed with the
be deployed off a wide variety of ships. By virtue of flexibility to carry different sensors and equipment
their relatively small size, limited capacity for scien- for different purposes, AUV system design and attri-
tific payloads, and autonomous nature, AUVs do not butes are driven by the specific research application.
have the range of capabilities of HOVs and ROVs. Some, such as the autonomous drifters and gliders
They are, however, much better suited than HOVs (essentially drifters with wings and a buoyancy
and ROVs to surveying large areas of the ocean that change mechanism that allow the vehicle to change
would take years to cover by any other means. They heading, pitch, and roll, and to move horizontally
can run missions of many hours or days on their while ascending and descending in the water col-
battery power and, with their streamlined shape, can umn), are designed for research in the water column
travel many kilometers collecting data of various to better understand the circulation of the ocean and
types depending on which sensors they are carrying. its influence on climate. While satellites provide
204 VEHICLES FOR DEEP SEA EXPLORATION

Figure 6 Portable control van for the ROV Jason II constructed

from two shipping containers assembled on board the R/V Knorr.
& Dive and Discover, WHOI, USA.

global coverage of conditions at the sea surface,

AUVs are likely to be the only way to continuously
access data from the ocean depths. Equipped with
oceanographic sensors that measure temperature,
salinity, current speed, and phytoplankton abun-
dance, drifters and gliders profile the water column
by sinking to a preprogrammed depth, and then rising
to the surface where they transmit their data via
satellite back to the scientist on shore. By deploying
hundreds to thousands of these vehicles, scientists Figure 7 The Wireline Reentry System, known as the CV,
will achieve a long-term presence in the ocean, and operated by Scripps Institution of Oceanography. This
will be able to make comprehensive studies of vast specialized ROV can precisely place heavy payloads on the
oceanic regions. seafloor and in drill holes. The control vehicle, which weighs
about 500 kg in water, is deployed at the end of a 17.3-mm
Other, more sophisticated AUVs are also used (0.68’’) electromechanical (coax) or electro-optico-mechanical
to investigate water column characteristics, and (three copper conductors, three optical fibers) oceanographic
ephemeral or localized phenomena, such as algal cable. The vehicle consists of a steel frame equipped with two
blooms. The first of the Dorado Class of AUVs, horizontal thrusters mounted orthogonal to each other to control
operated by Monterey Bay Aquarium Research lateral position. The vertical position is controlled by winch
operation. Instrumentation includes a compass, pressure gauge,
Institute, was deployed in late 2001 to measure the lights, video camera, sonar systems, and electronic interfaces to
inflow of water into the Arctic basin through the electrical releases and to a logging probe. Courtesy of Scripps
Fram Strait. Autosub, operated by the National Institution of Oceanography – Marine Physical Laboratory, USA.
Oceanography Centre, Southampton, UK, was de-
ployed to measure flow over the sills in the Strait of
Sicily. The REMUS (Remote Environmental Moni- electronic still cameras, and, most recently, a towed
toring UnitS) class of AUVs is extremely versatile and acoustic array.
they have been used on many types of missions. The Still other AUVs are designed specifically for near-
standard configuration includes an up- and down- bottom work. They have proved particularly useful
looking acoustic Doppler current profiler (ADCP), for near-bottom surveying and mapping, which can
sidescan sonar, a conductivity–temperature (CT) be accomplished autonomously while the support
profiler, and a light scattering sensor. However, many ship simultaneously conducts other, more traditional,
other instruments have been integrated into it for operations. One of the earliest vehicles to provide
specific missions, including fluorometers, biolumin- this capability was the Autonomous Benthic Explorer
escence sensors, radiometers, acoustic modems, for- (ABE) developed at Woods Hole Oceanographic In-
ward-looking sonar, altimeters, and acoustic Doppler stitution. ABE was designed to be extremely stable in
velocimeters. REMUS can also carry a video plank- pitch and roll and to be reasonably efficient in for-
ton recorder, a plankton pump, video cameras, ward travel. All the buoyancy is built into the two
 VEHICLES FOR DEEP SEA EXPLORATION 205

upper pods, while the majority of the weight placed between the hulls where they are protected.
(the batteries and the main pressure housing) is in the ABE is most efficient traveling forward, but it can
central lower section. The three-hull structure also also move backward, up or down, left or right, and
allows the seven vertical and lateral thrusters to be can hover and turn in place. Equipment that it usually
carries includes temperature and salinity sensors, an
optical backscatter sensor, a magnetometer to meas-
ure near-bottom magnetic fields, and an acoustic al-
timeter to make bathymetric measurements and for
its automated bottom following. ABE can dive to
depths of 5500 m for 16–34 h, and it uses acoustic
transponder navigation to follow preprogrammed
track lines automatically. Its capability to maintain a
precise course over rugged seafloor terrain gives it the
ability to make high-precision seafloor bathymetric
maps with features a few tens of centimeters tall and
less than a meter long being identifiable.
Other AUVs have been specifically developed for
high-resolution optical and acoustic imaging of the
seafloor. For example, SeaBED, also developed at
Woods Hole Oceanographic Institution, was designed
specifically to further the growing interests in seafloor
optical imaging – specifically, high-resolution color
imaging and the processes of photo-mosaicking and
three-dimensional image reconstruction. In addition
to requiring high-quality sensors, this imposes add-
itional constraints on the ability of the AUV to carry
out structured surveys, while closely following
the seafloor. The distribution of the four thrusters,
coupled with the passive stability inherent in a two-
hulled vehicle with a large metacentric height, allows
SeaBED to survey close to the seafloor, even in very
Figure 8 Schematic illustration of the HROV, Nereus, currently
under construction at Woods Hole Oceanographic Institution, rugged terrain.
USA, in its autonomous mode (upper) and its ROV mode (lower). In the future, AUVs will play an important role in
Illustration by E. Paul Oberlander, WHOI, USA. the development of long-term seafloor observatories.

Table 4 Examples of AUVs for deep-sea research and exploration (vehicles that can operate at depths Z1000 m)

Vehicle Operating organization Maximum operating depth (m)

Dorado Class MBARI, USA 6000

CR-01, CR-02 COMRA, China 6000
Sentry WHOI, USA 6000
REMUS Class WHOI, USA 6000
Autosub 6000 NOC, UK 6000
Autonomous Benthic Explorer NDSF, WHOI, USA 5500
Explorer 5000 Research Centre Ocean Margins, Germany 5000
Jaguar/Puma WHOI, USA 5000
Urashima (hybrid) JAMSTEC, Japan 3500
Aster x IFREMER, France 3000
Bluefin AUV Alfred Wegener Institute, Germany 3000
Bluefin 21 AUV SIO, USA 3000
Odyssey Class MIT, USA 3000
SeaBED WHOI, USA 2000
Autosub 3 National Oceanography Centre, UK 1600
Spray Gliders WHOI, USA 1500
Seaglider Univ. of Washington, USA 1000

Abbreviations as in Tables 1–3.

206 VEHICLES FOR DEEP SEA EXPLORATION

(a) (b)

(c) (d)

(e) (f)

Figure 9 Examples of AUVs used in oceanographic research. (a) The Spray Glider, (b) Urashima, (c) Autosub, (d) SeaBED,
(e) Dorado Class, and (f) ABE. (a) Photo by Jane Dunworth-Baker, WHOI, USA. (b) & JAMSTEC, Japan, with permission.
(c) Courtesy of Gwyn Griffiths, National Oceanography Centre, Southampton, UK. (d) Photo by Tom Kleindinst, WHOI, USA. (e) Photo
by Todd Walsh & 2004, MBARI, USA, with permission. (f) Photo by Dan Fornari, WHOI, USA.

Apart from providing the high-resolution maps docking stations for AUVs, and there are a number
needed to optimally place geological, chemical, and of research groups currently working on developing
biological sensors as part of an observatory, AUVs this technology. When an event – most likely a seis-
will also operate in a rapid response mode. It is en- mic event – is detected, scientists on shore will be
visaged that deep-sea observatories will include able to program the AUV, via satellite and a cable to
 VEHICLES FOR DEEP SEA EXPLORATION 207

a surface buoy, to leave its dock and conduct surveys The Future
in the vicinity of the event. The AUV will then return
to its dock and return the data to shore for assess- The technological breakthroughs in deep-sea vehicle
ment by scientists as to whether further investigation design over the last 40 years have resulted in un-
with ships is warranted. precedented access to the deep ocean. While each
type of vehicle has its own advantages and dis-
advantages, the complementary capacities of all
Navigating Deep-Sea Vehicles types of deep-submergence vehicles provide synergies
Unlike glider and drifter AUVs that can come to the that are revolutionizing how scientists conduct re-
sea surface and determine their positions using a search in the deep ocean. They are learning how to
Global Positioning System (GPS), deep-sea vehicles exploit those synergies by using a nested survey
working at the bottom of the ocean have no such strategy that employs a combination of tools in se-
reference system because the GPS system’s radio quence for investigations at increasingly finer scales:
frequency signals are blocked by seawater. The ship-based swath-mapping systems and towed ve-
technique that has been the standard for three- hicle systems for reconnaissance over large areas to
dimensional acoustic navigation of deep-sea vehicles identify features of interest, followed by more de-
is long-baseline (LBL) navigation – a technique de- tailed, high-resolution mapping, imagery, and
veloped more than 30 years ago. LBL operates on the chemical sensing with AUVs, and finally, seafloor
principle that the distance between an underwater observations and experimentation using HOVs and
vehicle and a fixed acoustic transponder can be re- ROVs. A demonstration of the power of such an
lated precisely to the time of flight of an acoustic approach occurred on a cruise to the Galápagos Rift
signal propagating between the vehicle and trans- in 2002. The investigative strategy was directed
ponder. Two or more acoustic transponders are toward ensuring that all potential sites of hydro-
dropped over the side of the surface ship and an- thermal venting in the rift valley were identified
chored at locations selected to optimize the acoustic and investigated visually with the HOV Alvin. The
range and geometry of planned seafloor operations. AUV ABE was deployed at night to conduct high-
Each transponder is a complete subsurface mooring resolution mapping of the seafloor and collect
comprised of an anchor, a tether, and a buoyant conductivity–temperature–depth (CTD) data in the
battery-powered acoustic transponder. The positions lower water column to detect sites of venting. Upon
of the transponders on the seafloor are determined by its recovery in the morning, micro-bathymetry maps
using the GPS on board the ship and ranging to them and temperature anomaly maps were quickly gener-
acoustically while the ship circles the point where ated, compiled with previous data, and then given to
each transponder was dropped. The positions of the the scientists diving in Alvin that day for their use in
transponders on the seafloor can be determined this directing the dive. Today, the vehicles are being de-
way with an accuracy of about 10 m. ployed in various combinations to attack a range of
Transponders have accurate clocks to measure multidisciplinary problems.
time very precisely, and they are synchronized with Deep-sea vehicles will also play indispensable roles
the clocks on the vehicle and on the ship. Each in establishing and servicing long-term seafloor ob-
transponder is set to listen for acoustic signals servatories that will be critical for time-series in-
(or pings) transmitted either from the deep-sea vestigations to understand the dynamic processes
vehicle or the ship at a specific frequency. When each going on beneath the ocean. AUVs will undertake a
transponder hears these acoustic signals, it is pro- variety of mapping and sampling missions while
grammed to transmit an acoustic signal back to the using fixed observatory installations to recharge
vehicle and the ship. Each transponder pings at a batteries, offload data, and receive new instructions.
different frequency, so the ship and the vehicle can They will be used to extend the spatial observational
discern which transponder sent it. The time of flight capability of seafloor observatories through sur-
of the acoustic signals gives a measure of distance to veying activities, and will document horizontal
each transponder, and using simple triangulation, the variability in seafloor and water column properties –
unique point in three-dimensional space where all necessary for establishing the context of point
distances measured from all the transponders and the measurements made by fixed instrumentation. HOVs
ship intersect can be calculated. More recently, con- and ROVs will be required to install, service,
ventional LBL navigation has been combined with and repair equipment and instrumentation on the
Doppler navigation data, which measures apparent seafloor and in drill holes, as well as collect samples
bottom velocity of the vehicle, for better short-term as part of time-series measurements. The additional
accuracy. capabilities that these vehicles will need for service
208 VEHICLES FOR DEEP SEA EXPLORATION

and repair activities will likely build on ROV tools Reves-Sohn R (2004) Unique vehicles for a unique environ-
that are currently being used in the commercial ment. Oceanus 42: 25--27.
undersea cable industry. Rona P (2001) Deep-diving manned research submersibles.
Deep-sea vehicles will clearly have a role to play in Marine Technology Society Journal 33: 13--25.
deep-sea research for the foreseeable future, and they Rudnick DL, Davis RE, Eriksen CC, Fratantoni DM, and
Perry MJ (2004) Underwater gliders for ocean research.
will be at the vanguard of a new era of ocean
Marine Technology Society Journal 38: 48--59.
exploration. Shank T, Fornari D, Yoerger D, et al. (2003) Deep
submergence synergy: Alvin and ABE explore the
Galápagos Rift at 86 1 W. EOS, Transactions of the
See also American Geophysical Union 84(425): 432--433.
Yoerger D, Bradley AM, Walden BB, Singh H, and
Gliders. Platforms: Autonomous Underwater Bachmayer R (1998) Surveying a subsea lava flow using
Vehicles. the Autonomous Benthic Explorer (ABE). International
Journal of Systems Science 29: 1031--1044.
Further Reading
Bachmayer R, Humphris S, Fornari D, et al. (1998)
Relevant Websites
Oceanographic exploration of hydrothermal vent sites http://auvlab.mit.edu
on the Mid-Atlantic Ridge at 371 N 321 W using – AUV Lab Vehicles, AUV Lab at MIT Sea Grant.
remotely operated vehicles. Marine Technology Society http://www.ropos.com
Journal 32: 37--47. – Canadian Scientific Submersible Facility.
Davis RE, Eriksen CE, and Jones CP (2002) Autonomous http://www.comra.org
buoyancy-driven underwater gliders. In: Griffiths G – China Ocean Mineral Resources R&D Association.
(ed.) The Technology and Applications of Autonomous http://divediscover.whoi.edu
Underwater Vehicles, pp. 37--58. London: Taylor and – Dive and Discover: Expeditions to the Seafloor.
Francis. http://www.soest.hawaii.edu
De Moustier C, Spiess FN, Jabson D, et al. (2000) Deep-sea – Hawai’i Undersea Research Laboratory (HURL),
borehole re-entry with fiber optic wireline technology. School of Ocean and Earth Science and Technology.
Proceedings of the 2000 International Sympo- http://www.ifremer.fr
sium on Underwater Technology, Tokyo, 23–26 May – IFREMER Fleet.
2000, pp. 379–384. http://www.mbari.org
Fornari D (2004) Realizing the dreams of da Vinci and – Marine Operations: Vessels and Vehicles, Monterey
Verne. Oceanus 42: 20--24. Bay Aquarium Research Institute.
Fornari DJ, Humphris SE, and Perfit MR (1997) Deep http://www.mpl.ucsd.edu
submergence science takes a new approach. EOS, – Marine Physical Laboratory, Scripps Institution of
Transactions of the American Geophysical Union 78: Oceanography.
402--408. http://www.noc.soton.ac.uk
Fryer P, Fornari DJ, Perfit M, et al. (2002) Being there: The – National Oceanography Centre, Southampton.
continuing need for human presence in the deep ocean http://www.jamstec.go.jp
for scientific research and discovery. EOS, Transactions – Research Vessels, Facilities, and Equipment,
of the American Geophysical Union 83(526): 532--533. JAMSTEC.
Funnell C (2004) Jane’s Underwater Technology 2004– http://www.apl.washington.edu
2005, 800pp, 23rd edn. Alexandria, VA: Jane’s Infor- – Seaglider, Applied Physics Laboratory, University of
mation Group. Washington.
National Research Council (2004) Exploration of the Seas: http://www.whoi.edu
Voyage into the Unknown. Washington, DC: National – Ships and Technology: National Deep Submergence
Academies Press. Facility, Woods Hole Oceanographic Institution.
National Research Council (2004) Future Needs of Deep http://www.rcom.marum.de
Submergence Science. Washington, DC: National – Technology page, MARUM.
Academies Press.
SENSORS: METEOROLOGY
 SENSORS FOR MEAN METEOROLOGY
K. B. Katsaros, Atlantic Oceanographic and worked out over a period of two centuries. With
Meteorological Laboratory, NOAA, Miami, FL, USA modern manufacturing techniques, the aneroid has
Copyright & 2001 Elsevier Ltd. become standardized and is the commonly used de-
vice, calibrated with transfer standards back to the
classical method. The fact that it takes a column of
about 760 mm of the heavy liquid metal mercury
(13.6 times as dense as water) illustrates the sub-
Introduction stantial weight of the atmosphere. Corrections for
the thermal expansion or contraction of the mercury
Basic mean meteorological variables include the fol- column must be made, so a thermometer is always
lowing: pressure, wind speed and direction, tem- attached to the device. Note that the word ‘weight’ is
perature, and humidity. These are measured at all used, which implies that the value of the earth’s
surface stations over land and from ships and buoys at gravitational force enters the formula for converting
sea. Radiation (broadband solar and infrared) is also the mercury column’s height to a pressure (force/unit
often measured, and sea state, swell, wind sea, cloud area). Since gravity varies with latitude and altitude,
cover and type, and precipitation and its intensity and mercury barometers must be corrected for the local
type are evaluated by an observer over the ocean. Sea value of the acceleration due to gravity.
surface temperature and wave height (possibly also Atmospheric pressure decreases with altitude. The
frequency and direction of wave trains) may be balancing column of mercury decreases or the ex-
measured from a buoy at sea; they are part of the set of pansion of the aneroid chamber increases as the
parameters required for evaluating net surface energy column of air above the barometer has less weight at
flux and momentum transfer. Instruments for meas- higher elevations. Conversely, pressure sensors can
uring the quantities described here have been limited therefore be used to measure or infer altitude, but
to the most common and basic. Precipitation is an must be corrected for the variation in the atmos-
important meteorological variable that is measured pheric surface pressure, which varies by as much as
routinely over land with rain gauges, but its direct 10% of the mean (even more in case of the central
measurement at sea is difficult because of ship motion pressure in a hurricane). An aneroid barometer is the
and wind deflection by ships’ superstructure and transducer in aircraft altimeters.
consequently it has been measured routinely over the
ocean only from ferry boats. However, it can be esti-
mated at sea by satellite techniques, as can surface Wind Speed and Direction
wind and sea surface temperature. Satellite methods
Wind speed is obtained by two basic means, both
are included in this article, since they are increasing in
depending on the force of the wind to make an object
importance and provide the only means for obtaining
rotate. This object comprises either a three- or four-
complete global coverage.
cup anemometer, half-spheres mounted to horizontal
axes attached to a vertical shaft (Figure 1A). The
cups catch the wind and make the shaft rotate. In
Pressure
today’s instruments rotations are counted by the
Several types of aneroid barometers are in use. They frequency of the interception of a light source to
depend on the compression or expansion of an produce a digital signal.
evacuated metal chamber for the relative change in Propeller anemometers have three or four blades
atmospheric pressure. Such devices must be com- that are turned by horizontal wind (Figure 1B). The
pensated for the change in expansion coefficient of propeller anemometer must be mounted on a wind
the metal material of the chamber with temperature, vane that keeps the propeller facing into the wind.
and the device has to be calibrated for absolute val- For propeller anemometers, the rotating horizontal
ues against a classical mercury in glass barometer, shaft is inserted into a coil. The motion of the shaft
whose vertical mercury column balances the weight generates an electrical current or a voltage difference
of the atmospheric column acting on a reservoir of that can be measured directly. The signal is large
mercury. The principle of the mercury barometer was enough that no amplifiers are needed.
developed by Evangilista Toricelli in the 17th cen- Both cup and propeller anemometers, as well as
tury, and numerous sophisticated details were vanes, have a threshold velocity below which they do

211
212 SENSORS FOR MEAN METEOROLOGY

(A)
(B)

(C) (D)

Figure 1 (A) Cup anemometer and vane; (B) propeller vane assembly; (C) three-dimensional sonic anemometer; (D) radiation
shield for temperature and humidity sensors. (Photographs of these examples of common instruments were provided courtesy of R.M.
Young Company.)

not turn and measure the wind. For the propeller in the tunnel. Calibration sensors can be fine cup
anemometer, the response of the vane is also crucial, anemometers or pitot tubes.
for the propeller does not measure wind speed off- The wind direction is obtained from the position
axis very well. These devices are calibrated in wind of a wind vane (a vertical square, triangle, or other-
tunnels, where a standard sensor evaluates the speed wise shaped wind-catcher attached to a horizontal
 SENSORS FOR MEAN METEOROLOGY 213

shaft, Figure 1A and B). The position of a sliding the temperature of the air, it must be well ventilated,
contact along an electrical resistance coil moved by which is sometimes assured by a protective housing
the motion of the shaft gives the wind direction and a fan pulling air past the sensor. Shielding from
relative to the zero position of the coil. The position direct sunlight has been done traditionally over land
is typically a fraction of the full circle (minus a small and island stations by the use of a ‘Stephenson
gap) and must be calibrated with a compass for ab- screen,’ a wooden-roofed box with slats used for the
solute direction with respect to the Earth’s north. sides, providing ample room for air to enter. Modern
Other devices such as sonic anemometers can de- devices have individual housings based on the same
termine both speed and direction by measuring the principles (Figure 1D).
modification of the travel time of short sound pulses The classic measurements of temperature were
between an emitter and a receiver caused by the done with mercury in glass or alcohol in glass ther-
three-dimensional wind. They often have three sound mometers. For sea temperature, such a thermometer
paths to allow evaluation of the three components of was placed in a canvas bucket of water hauled up on
the wind (Figure 1C). These devices have recently deck. Today, electronic systems have replaced most
become rugged enough to be used to measure mean of the glass thermometers. Table 1 lists some of these
winds routinely, and have a high enough frequency sensors (for details see the Further Reading section).
response to also determine the turbulent fluctuations. The sea surface temperature (SST) is an important
The obvious advantage is that the instrument has no aspect of air–sea interaction. It enters into bulk for-
moving parts. Water on the sound transmitter or mulas for estimating sensible heat flux and evapor-
receiver causes temporary difficulties, so a sonic an- ation. The temperature differences between the air at
emometer is not an all-weather instrument. The one height and the SST is also important for deter-
sound paths can be at arbitrary angles to each other mining the atmospheric stratification, which can
and to the natural vertical. Processing of the data modify the turbulent fluxes substantially compared
transforms the measurements into an Earth-based with neutral stratification.
coordinate system. The assumptions of zero mean The common measure of SST is the temperature
vertical velocity and zero mean cross-wind velocity within the top 1 or 2 m of the interface, obtained
allow the relative orientation between the instrument with any of the contact temperature sensors de-
axes and the Earth-based coordinate system to be scribed in Table 1. On ships, the sensor is typically
found. Difficulties arise if the instrument is experi- placed in the ship’s water intake, and on buoys it
encing a steady vertical velocity at its location due to may even be placed just inside the hull on the bot-
flow distortion around the measuring platform, for tom, shaded side of the buoy. Because the heat losses
instance. to the air occur at the air–sea interface, while solar
Cup and propeller anemometers are relatively in- heating penetrates of the order of tens of meters
sensitive to rain. However, snow and frost are (depth depending on sun angle), a cool skin, 1–2 mm
problematic to all wind sensors, particularly the ones in depth and 0.1–0.51C cooler than the lower layers,
described above with moving parts. Salt contamin- is often present just below the interface. Radiation
ation over the ocean also causes deterioration of the thermometers are sometimes used from ships or
bearings in cup and propeller anemometers. Proper piers to measure the skin temperature directly (see
exposure of wind sensors on ships is problematic Radiative Transfer in the Ocean).
because of severe flow distortion by increasingly
large ships. One solution has been to have duplicate
sensors on port and starboard sides of the ship and Humidity
selecting the valid one on the basis of the recording
The Classical Sling Psychrometer
of the ship’s heading and the relative wind direction.
An ingenious method for evaluating the air’s ability
to take up water (its deficit in humidity with respect
to the saturation value, see 68 on Evaporation and
Temperature Humidity) is the psychrometric method. Two ther-
The measurements of both air and water tempera- mometers (of any kind) are mounted side by side,
ture will be considered here, since both are important and one is provided with a cotton covering (a wick)
in air–sea interaction. Two important considerations that is wetted with distilled water. The sling psy-
for measuring temperature are the exposure of the chrometer (Figure 2) is vigorously ventilated by
sensor and shielding from solar radiation. The axiom swinging it in the air. The air passing over the sensors
that a ‘thermometer measures its own temperature’ is changes their temperatures to be in equilibrium with
a good reminder. For the thermometer to represent the air; the dry bulb measures the actual air
214 SENSORS FOR MEAN METEOROLOGY

Table 1 Electronic devices for measuring temperature in air or water

Name Principle Typical use

Thermocouple Thermoelectric junctions between two wires (e.g. Copper-Constantan) Good for measuring differences
set up a voltage in the circuit, if the junctions are at different of temperature
temperatures. The reference junction temperature must be measured
as well

Resistance R ¼ RRef(1 þ aT) Platinum resistance thermometers

thermometer are used for calibration and as
reference thermometers
Where R is the electrical resistance, RRef is resistance at a reference
temperature, and a is the temperature coefficient of resistance

Thermistor R ¼ a exp (b/T) Commonly used in routine sensor

systems
Where R is resistance, T is absolute temperature, and a and b are
constants

Radiation Infrared radiance in the atmospheric window, 8–12 mm, is a measure Usually used for measuring
thermometer of the equivalent black body temperature water’s skin temperature

temperature, the wet bulb adjusts to a temperature thermometers housed in ventilated cylindrical
that is intermediate between the dew point and air shields. Ventilation can be simply due to the nat-
temperature. As water from the wick is evaporated, ural wind (in which case errors at low wind speeds
it takes heat out of the air passing over the wick until may develop), or be provided by a motor and a fan
an equilibrium is reached between the heat supplied (typically an air speed of 3 m s1 is required). A
to the wet bulb by the air and the heat lost due to water reservoir must be provided to ensure con-
evaporation of water from the wick. This is the wet tinuous wetting of the wet bulb. The reservoir
bulb temperature. The Smithsonian Tables provide should be mounted below the psychrometer so that
the dew point temperature (and equivalent satur- water is drawn onto the wet bulb with a long wick.
ation humidity) corresponding to the measured ‘wet (This arrangement assures that the water has had
bulb temperature depression,’ i.e. the temperature time to equilibrate to the wet bulb temperature of
difference between the dry bulb and the wet bulb the air.)
thermometers at the existing air temperature. If these large wet bulbs collect salt on them over
time, the relative humidity may be in error. This
is not a concern for short-term measurements. A
Resistance Thermometer Psychrometer
salt solution of 3.6% on the wet bulb would result
A resistance thermometer psychrometer consists in an overestimate of the relative humidity of
of stainless steel-encased platinum resistance approximately 2%.

Wet bulb

Psychrometer revolves
Dry bulb about this axis

Handle

Figure 2 Sling pyschrometer. (Reproduced with permission from Parker, 1977.)

 SENSORS FOR MEAN METEOROLOGY 215

Capacitance Sensors of Humidity length of which depends both on the generation of

spray in the area, the height of the measurement, and
The synoptic weather stations often use hygrometers
the size of the transducer (i.e. the fraction of the
based on the principle of capacitance change as the
surface area that may be contaminated). One of the
small transducer absorbs and desorbs water vapor.
protective devices that was successfully used in the
To avoid contamination of the detector, special filters
Humidity Exchange over the Sea (HEXOS) experi-
cover the sensor. Dirty filters (salt or other con-
ment is the so-called ‘spray flinger.’
taminants) may completely mask the atmospheric
effects. Even the oil from the touch of a human hand
is detrimental. Two well-known sensors go under the Spray-Removal Device
names of Rotronic and Humicap. Calibration with
mercury in glass psychrometers is useful. The University of Washington ‘spray flinger’
(Figure 3) was designed to minimize flow modifi-
cation on scales important to the eddy correlation
Exposure to Salt
calculations of evaporation and sensible heat flux
As for wind and temperature devices, the humidity employing data from temperature and humidity
sensors are sensitive to flow distortion around ships sensors inside the housing. The design aims to ensure
and buoys. Humidity sensors have an additional that the droplets removed from the airstream do not
problem in that salt crystals left behind by evapor- remain on the walls of the housing or filter where
ating spray droplets, being hygroscopic, can modify they could evaporate and affect the measurements.
the measurements by increasing the local humidity The device has been tested to ensure that there are no
around them. One sophisticated, elegant, and ex- thermal effects due to heating of the enclosure, but
pensive device that has been used at sea without this would be dependent on the meteorological
success is the dew point hygrometer. It depends on conditions encountered, principally insolation. The
the cyclical cooling and heating of a mirror. The housing should be directed upwind.
cooling continues until dew forms, which is detected Although there is a slow draw of air through the
by changes in reflection of a light source off the unit by the upwind and exit fans (1–2 m s1), it is
mirror, and the temperature at that point is by def- mainly a passive device with respect to the airflow.
inition the dew point temperature. The problem with Inside the tube, wet and dry thermocouples or other
this device is that during the heating cycle sea salt is temperature and humidity sensors sample the air for
baked onto the mirror and cannot be removed by mean and fluctuating temperature and humidity.
cleaning. Wind tunnel and field tests showed the airflow inside
Several attempts to build devices that remove the the unit to be steady and about one-half the ambient
spray have been tried. Regular Stephenson screen- wind speed for wind directions o401 off the axis.
type shields provide protection for some time, the Even in low wind speeds there is adequate ventilation

Water reservoir
Supporting wire mesh
with nylon filter

Dry bulb
thermocouple
Air flow

Wet bulb
thermocouple

60 cm

Figure 3 Sketch of aspirated protective housing, the ‘spray flinger’, used for the protection of a thermocouple psychrometer by the
University of Washington group. The system is manually directed upwind. The spray flinger is a 60 cm long tube, 10 cm in diameter,
with a rotating filter screen and fan on the upwind end, and an exit fan and the motor at the downwind end. The filter is a single layer of
nylon stocking, which is highly nonabsorbent, supported by a wire mesh. Particles and droplets are intercepted by the rotating filter and
flung aside, out of the airstream entering the tube. The rotation rate of the filter is about 625 rpm. Inspection of the filter revealed that
this rate of rotation prevented build-up of water or salt. The nylon filter needs to be replaced at least at weekly intervals. (Reproduced
with permission from Katsaros et al., 1994.)
216 SENSORS FOR MEAN METEOROLOGY

for the wet bulb sensor. Comparison between data radiometers and the wind vector from scatte-
from shielded and unshielded thermocouples (re- rometers, active microwave instruments. Both pas-
spectively, inside and outside the spray flinger) show sive and active sensors depend on the changing
that the measurements inside are not noticeably af- roughness of the sea as a function of wind speed for
fected by the housing. their ability to ‘sense’ the wind. The first scatte-
A quantitative test of the effectiveness of the spray rometer was launched on the Seasat satellite in 1978,
flinger in removing aerosols from the sample air- operating for 3 months only. The longest record is
stream was performed during HEXMAX (the from the European Remote Sensing (ERS) satellites 1
HEXOS Main Experiment) using an optical particle and 2 beginning in 1991 and continuing to function
counter to measure the aerosol content with diam- well in 2000. Development of interpretation of the
eters between 0.5 and 32 mm in the environmental air radar returns in terms of both speed and direction
and at the rear of the spray flinger. Other devices depends on the antennae viewing the same ocean
have been constructed, but have had various dif- area several times at different incidence angles rela-
ficulties, and the ‘spray flinger’ is not the final an- tive to the wind direction. A recently launched sat-
swer. Intake tubes that protect sensors have also been ellite (QuikSCAT in 1999) carries a new design with
designed for use on aircraft. a wider swath, the SeaWinds instrument. Scatte-
rometers are providing surface wind measurements
with accuracy of 7 1.6 m s1 approximately in speed
and 7201 in direction at 50 km resolution for ERS
Satellite Measurements
and 25 km for SeaWinds. They view all of the global
With the global ocean covering 70% of the earth’s ocean once in 3 days for ERS and in approximately 2
surface, large oceanic areas cannot be sampled by in days for QuikSCAT. Microwave radiometers such as
situ sensors. Most of the meteorological measure- the Special Sensor Microwave/Imager (SSM/I), op-
ments are taken by Voluntary Observing Ships (VOS) erational since 1987 on satellites in the US Defense
of the merchant marine and are, therefore, confined Meteorological Satellite Program, have wider swaths
to shipping lanes. Research vessels and military ships covering the globe daily, but they are not able to
may be found in other areas and have contributed sense the ocean surface in heavy cloud or rainfall
substantially to our knowledge of conditions in areas areas and do not give direction. They can be as-
not visited by VOS. The VOS report their obser- similated into numerical models where the models
vations on a 3 hour or 6 hour schedule. Some mean provide an initial guess of the wind fields, which are
meteorological quantities such as SST and wind modified to be consistent with the details of the
speed and direction are observable by satellites dir- radiometer-derived wind speeds.
ectly, while others can be inferred from less directly Surface pressure and atmospheric surface air
related measurements. Surface insolation and pre- temperature are not yet amenable to satellite obser-
cipitation depend on more complex algorithms for vations, but surface humidity can be inferred from
evaluation. Satellite-derived surface meteorological total column water content. From the satellite-
information over the ocean are mostly derived from observed cloudiness, solar radiation at the surface
polar-orbiting, sun-synchronous satellites. The fam- can be inferred by use of radiative transfer models.
ous TIROS and NOAA series of satellites carrying This is best done from geostationary satellites whose
the Advanced Very High Resolution Radiometer sensors sweep across the Earth’s surface every 3
(AVHRR) and its predecessors has provided sea hours or more often, but only view a circle of useful
surface temperature and cloud information for more data extending 7501 in latitude, approximately.
than three decades (SST only in cloud-free con- Precipitation can also be inferred from satellites
ditions). This long-term record of consistent meas- combining microwave data (from SSM/I) with visible
urements by visible and infrared sensors has and infrared signals. For tropical regions, the Tropical
provided great detail with a resolution of a few Rainfall Measuring Mission (TRMM) on a low-orbit
kilometers of many phenomena such as oceanic eddy satellite provides precipitation estimates on a monthly
formation, equatorial Rossby and Kelvin waves, and basis. This satellite carries a rain radar with 500 km
the El Niño phenomenon. Because of the wide swath swath in addition to a microwave radiometer.
of these short wavelength devices, of the order of Developments of multispectral sensors and con-
2000 km, the whole earth is viewed daily by either tinued work on algorithms promises to improve the
the ascending or descending pass of the satellite accuracy of the satellite information on air–sea
overhead, once in daytime and once at night. interaction variables. Most satellite programs depend
Another mean meteorological variable observable on the simple in situ mean meteorological measure-
from space is surface wind speed, with microwave ments described above for calibration and validation.
 SENSORS FOR MEAN METEOROLOGY 217

A good example is the important SST record pro- de Leeuw G (1990) Profiling of aerosol concentrations,
vided by the US National Weather Service and used particle size distributions, and relative humidity in the
by all weather services. The analysis procedure em- atmospheric surface layer over the North Sea. Tellus
ploys surface data on SST from buoys, particularly 42B: 342--354.
small, inexpensive, free-drifting buoys that are Dobson F, Hasse L, and Davies R (eds.) (1980) Instruments
and Methods in Air–Sea Interaction, pp. 293--317. New
spread over the global oceans to ‘tie-down’ the cor-
York: Plenum.
rection for atmospheric interference for the satellite Geernaert GL and Plant WJ (eds.) (1990) Surface Waves
estimates of SST. The satellite-observed infrared ra- and Fluxes, 2, pp. 339--368. Dordrecht: Kluwer
diances are modified by the transmission path from Academic Publishers.
the sea to the satellite, where the unknown is the Graf J, Sasaki C, et al. (1998) NASA Scatterometer
aerosol that can severely affect the interpretation. Experiment. Asta Astronautica 43: 397--407.
The aerosol signal is not directly observable yet by Gruber A, Su X, Kanamitsu M, and Schemm J (2000) The
satellite, so the surface-measured SST data serve an comparison of two merged rain gauge-satellite
important calibration function. precipitation datasets. Bulletin of the American
Meteorological Society 81: 2631--2644.
Katsaros KB (1980) Radiative sensing of sea surface
Future Developments temperatures. In: Dobson F, Hasse L, and Davies R
(eds.) Instruments and methods in Air–Sea Interaction,
New measurement programs are being developed by pp. 293--317. New York: Plenum Publishing Corp.
international groups to support synoptic definition of Katsaros KB, DeCosmo J, Lind RJ, et al. (1994)
the ocean’s state similarly to meteorological meas- Measurements of humidity and temperature in the
urements and to provide forecasts. The program goes marine environment. Journal of Atmospheric and
under the name of the Global Ocean Observing Oceanic Technology 11: 964--981.
System (GOOS). It includes new autonomous buoys Kummerow C, Barnes W, Kozu T, Shiue J, and Simpson J
cycling in the vertical to provide details below the (1998) The tropical rainfall measuring mission
interface, a large surface drifter component, and the (TRMM) sensor package. Journal of Atmospheric and
Oceanic Technology 15: 809--817.
VOS program, as well as certain satellite sensors.
Liu WT (1990) Remote sensing of surface turbulence flux.
The GOOS is being developed to support a modeling
In: Geenaert GL and Plant WJ (eds.) Surface Waves and
effort, the Global Ocean Data Assimilation Experi- Fluxesyyy, 2, pp. 293--309. Dordrecht: Kluwer
ment (GODAE), which is an experiment in fore- Academic Publishers.
casting the oceanic circulation using numerical Parker SP (ed.) (1977) Encyclopedia of Ocean and
models with assimilation of the GOOS data. Atmospheric Science. New York: McGraw-Hill.
Pinker RT and Laszlo I (1992) Modeling surface solar
irradiance for satellite applications on a global scale.
See also Journal of Applied Meteorology 31: 194--211.
Reynolds RR and Smith TM (1994) Improved global sea
Sensors for Micrometeorological and Flux surface temperature analyses using optimum
Measurements. interpolation. Journal of Climate 7: 929--948.
List RJ (1958) Smithsonian Meteorological Tables 6th edn.
City of Washington: Smithsonian Institution Press.
Further Reading van der Meulen JP (1988) On the need of appropriate filter
techniques to be considered using electrical humidity
Atlas RS, Hoffman RN, Bloom SC, Jusem JC, and sensors. In: Proceedings of the WMO Technical
Ardizzone J (1996) A multiyear global surface wind Conference on Instruments and Methods of
velocity data set using SSM/I wind observations. Observation (TECO-1988), pp. 55–60. Leipzig,
Bulletin of the American Meteorological Society 77: Germany: WMO.
869--882. Wentz FJ and Smith DK (1999) A model function for the
Bentamy A, Queffeulou P, Quilfen Y, and Katsaros KB ocean-normalized radar cross-section at 14 GHz
(1999) Ocean surface wind fields estimated from derived from NSCAT observations. Journal of
satellite active and passive microwave instruments. Geophysical Research 104: 11 499--11 514.
IEEE Transactions Geosci Remote Sens 37:
2469--2486.
 SENSORS FOR MICROMETEOROLOGICAL & FLUX
MEASUREMENTS
J. B. Edson, Woods Hole Oceanographic Institution, the sensors used to meet these requirements, which
Woods Hole, MA, USA includes examples of some of the obstacles that
Copyright & 2001 Elsevier Ltd. marine researchers have had to overcome. These
obstacles include impediments caused by environ-
mental conditions and engineering challenges that
are unique to the marine environment. The dis-
cussion is limited to the measurement of velocity,
temperature, and humidity. The article concludes
Introduction with a description of the state-of-the-art sensors
currently used to measure the desired fluxes.
The exchange of momentum, heat, and mass be-
tween the atmosphere and ocean is the fundamental
physical process that defines air–sea interactions. Flux Measurements
This exchange drives ocean and atmospheric circu-
lations, and generates surface waves and currents. The exchange of momentum and energy a few meters
Marine micrometeorologists are primarily concerned above the ocean surface is dominated by turbulent
with the vertical exchange of these quantities, par- processes. The turbulence is caused by the drag (i.e.,
ticularly the vertical transfer of momentum, heat, friction) of the ocean on the overlying air, which
moisture, and trace gases associated with the mo- slows down the wind as it nears the surface and
mentum, sensible heat, latent heat, and gas fluxes, generates wind shear. Over time, this causes faster-
respectively. The term flux is defined as the amount moving air aloft to be mixed down and slower-
of heat (i.e., thermal energy) or momentum trans- moving air to be mixed up; the net result is a
ferred per unit area per unit time. downward flux of momentum. This type of turbu-
Air–sea interaction studies often investigate the lence is felt as intermittent gusts of wind that buffet
dependence of the interfacial fluxes on the mean an observer looking out over the ocean surface on a
meteorological (e.g., wind speed, degree of stratifi- windy day.
cation or convection) and surface conditions (e.g., Micrometeorologists typically think of these gusts
surface currents, wave roughness, wave breaking, as turbulent eddies in the airstream that are being
and sea surface temperature). Therefore, one of the advected past the observer by the mean wind. Using
goals of these investigations is to parametrize the this concept, the turbulent fluctuations associated
fluxes in terms of these variables so that they can be with these eddies can be defined as any departure
incorporated in numerical models. Additionally, from the mean wind speed over some averaging
these parametrizations allow the fluxes to be in- period (eqn [1]).
directly estimated from observations that are easier
to collect and/or offer wider spatial coverage. Ex- uðtÞ ¼ UðtÞ  Ū ½1
amples include the use of mean meteorological
measurements from buoys or surface roughness In eqn [1], u(t) is the fluctuating (turbulent) com-
measurements from satellite-based scatterometers to ponent, U(t) is the observed wind, and the overbar
estimate the fluxes. denotes the mean value over some averaging period.
Direct measurements of the momentum, heat, and The fact that an observer can be buffeted by the wind
moisture fluxes across the air–sea interface are cru- indicates that these eddies have some momentum.
cial to improving our understanding of the coupled Since the eddies can be thought to have a finite size, it
atmosphere–ocean system. However, the operating is convenient to consider their momentum per unit
requirements of the sensors, combined with the often volume, given by raU(t), where ra is the density of
harsh conditions experienced over the ocean, make air. In order for there to be an exchange of mo-
this a challenging task. This article begins with a mentum between the atmosphere and ocean, this
description of desired measurements and the oper- horizontal momentum must be transferred down-
ating requirements of the sensors. These require- ward by some vertical velocity. The mean vertical
ments involve adequate response time, reliability, velocity associated with the turbulent flux is nor-
and survivability. This is followed by a description of mally assumed to be zero. Therefore, the turbulent

218
 SENSORS FOR MICROMETEOROLOGICAL & FLUX MEASUREMENTS 219

transfer of this momentum is almost exclusively via momentum is transferred via wave drag and viscous
the turbulent vertical velocity, w(t), which we asso- shear stress caused by molecular viscosity. The ocean
ciate with overturning air. is surprisingly smooth compared to most land sur-
The correlation or covariance between the fluctu- faces. This is because the dominant roughness
ating vertical and horizontal wind components is the elements that cause the drag on the atmosphere are
most direct estimate of the momentum flux. This the wind waves shorter than 1 m in length. Although
approach is known as the eddy correlation or direct the longer waves and swell give the appearance of a
covariance method. Computation of the covariance very rough surface, the airflow tends to follow these
involves multiplying the instantaneous vertical vel- waves and principally act to modulate the mo-
ocity fluctuations with one of the horizontal com- mentum flux supported by the small-scale roughness.
ponents. The average of this product is then Therefore, the wave drag is mainly a result of these
computed over the averaging period. small-scale roughness elements.
Because of its dependence on the wind shear, the Turbulence can also be generated by heating and
flux of momentum at the surface is also known as the moistening the air in contact with the surface. This
shear stress defined by eqn [2], where î and ĵ are unit increases the buoyancy of the near-surface air, and
vectors, and v is the fluctuating horizontal com- causes it to rise, mix upward, and be replaced by less-
ponent that is orthogonal to u. buoyant air from above. The motion generated by
this convective process is driven by the surface
t0 ¼ îuw  ĵvw ½2 buoyancy flux (eqn [3]).

Typically, the coordinate system is rotated into the B0 ¼ ra cp wyv ½3

mean wind such that u, v, and w denote the longi-
tudinal, lateral, and vertical velocity fluctuations, In eqn [3] cp is the specific heat of air at constant
respectively. Representative time series of longi- pressure and yv is the fluctuating component of the
tudinal and vertical velocity measurements taken in virtual potential temperature defined in eqn [4],
where Y  and y are the mean and fluctuating com-
the marine boundary layer are shown in Figure 1.
The velocities in this figure exhibit the general trend ponents of the potential temperature, respectively,
that downward-moving air (i.e., wo0) is transport- and q is the specific humidity (i.e., the mass of water
ing eddies with higher momentum per unit mass (i.e., vapor per unit mass of moist air).
rau>0) and vice versa. The overall correlation is
therefore negative, which is indicative of a down- yv ¼ y þ 0:61Ȳq ½4
ward flux of momentum. These quantities also define the sensible heat (eqn
Close to the surface, the wave-induced momentum [5]) and latent heat (eqn [6]) fluxes.
flux also becomes important. At the interface, the
turbulent flux actually becomes negligible and the
H0 ¼ ra cp wy ½5

2.5
2.0 E0 ¼ Le wra q ½6
Velocity fluctuations (m s )
_1

1.5 where Le is the latent heat of evaporation. The par-

1.0 cels of air that are heated and moistened via the
0.5 buoyancy flux can grow into eddies that span the
0
entire atmospheric boundary layer. Therefore, even
_ 0.5
in light wind conditions with little mean wind shear,
these turbulent eddies can effectively mix the marine
_ 1.0
boundary layer.
_ 1.5 Conversely, when the air is warmer than the
_ 2.0 ocean, the flow of heat from the air to water (i.e., a
_ 2.5 downward buoyancy flux) results in a stably strati-
0 5 10 15 20 25 30 fied boundary layer. The downward buoyancy flux is
Time (s) normally driven by a negative sensible heat flux.
However, there have been observations of a down-
Figure 1 Time-series of the longitudinal (thick line) and vertical
velocity (thin line) fluctuations measured from a stable platform.
ward latent heat (i.e., moisture) flux associated with
The mean wind speed during the sampling period was the formation of fog and possibly condensation at
10.8 m s1. the ocean surface. Vertical velocity fluctuations have
220 SENSORS FOR MICROMETEOROLOGICAL & FLUX MEASUREMENTS

to work to overcome the stratification since upward- 2.5

moving eddies are trying to bring up denser air and 2.0
vice versa. Therefore, stratified boundary layers tend

Velocity fluctuations (m s )
1.5

_1
to dampen the turbulent fluctuations and reduce the 1.0
flux compared to their unstable counterpart under
0.5
similar mean wind conditions. Over the ocean, the
most highly stratified stable boundary layers are 0
usually a result of warm air advection over cooler _ 0.5
water. Slightly stable boundary conditions can also _ 1.0
be driven by the diurnal cycle if there is sufficient _ 1.5
radiative cooling of the sea surface at night. _ 2.0
_ 2.5
0 5 10 15 20 25 30
Sensors
Time (s)
Measurement of the momentum, sensible heat, and
latent heat fluxes requires a suite of sensors capable Figure 2 Time-series of the longitudinal (thick line) and vertical
velocity (thin line) fluctuations measured from a 3 m discus
of measuring the velocity, temperature, and moisture
buoy. The mean wind speed during the sampling period was
fluctuations. Successful measurement of these fluxes 10.9 m s1. The measured fluctuations are a combination of
requires instrumentation that is rugged enough to turbulence and wave-induced motion of the buoy.
withstand the harsh marine environment and fast
enough to measure the entire range of eddies that
platforms. Additionally, if the platform is moving,
transport these quantities. Near the ocean surface,
the size of the smallest eddies that can transport these the motion of the platform will be sensed by the in-
strument as an additional velocity and will con-
quantities is roughly half the distance to the surface;
taminate the desired signal (Figure 2). Therefore, the
i.e., the closer the sensors are deployed to the surface,
platform motion must be removed to accurately
the faster the required response. In addition, micro-
measure the flux. This requires measurements of the
meteorologists generally rely on the wind to advect
linear and angular velocity of the platform. The al-
the eddies past their sensors. Therefore, the velocity
ternative is to deploy the sensors on fixed platforms
of the wind relative to a fixed or moving sensor also
or to reduce the required motion correction by
determines the required response; i.e., the faster the
relative wind, the faster the required response. For mounting the sensors on spar buoys, SWATH vessels,
or other platforms that are engineered to reduced the
example, planes require faster response sensors than
wave-induced motion.
ships but require less averaging time to compute the
fluxes because they sample the eddies more quickly.
The combination of these two requirements results
in an upper bound for the required frequency re- Shear Stress
sponse (eqn [7]).
The measurement of momentum flux or shear stress
fz has a long history. The earliest efforts attempted to
E2 ½7 adapt many of the techniques commonly used in the
Ur
laboratory to the marine boundary layer. A good
Here f is the required frequency response, z is the example of this is the use of hot-wire anemometers
height above the surface, and Ur is the relative vel- that are well-suited to wind tunnel studies of turbu-
ocity. As a result, sensors used on ships, buoys, and lent flow. Hot-wire anemometry relies on very fine
fixed platforms require a frequency response of ap- platinum wires that provide excellent frequency re-
proximately 10–20 Hz, otherwise some empirical sponse and satisfy eqn [7] even close to the surface.
correction must be applied. Sensors mounted on The technique relies on the assumption that the
aircraft require roughly an order of magnitude faster cooling of heated wires is proportional to the flow
response depending on the sampling speed of the past the wire. Hot-wire anemometers are most
aircraft. commonly used in constant-temperature mode. In
The factors that degrade sensor performance in the this mode of operation, the current heating the wire
marine atmosphere include contamination, cor- is varied to maintain a constant temperature using a
rosion, and destruction of sensors due to sea spray servo loop. The amount of current or power required
and salt water; and fatigue and failure caused by to maintain the temperature is a measure of the
long-term operation that is accelerated on moving cooling of the wires by the wind.
 SENSORS FOR MICROMETEOROLOGICAL & FLUX MEASUREMENTS 221

Unfortunately, there are a number of problems successful design uses springs to attach a sphere and
associated with the use of these sensors in the marine its supporting structure to a rigid mount. The springs
environment. The delicate nature of the wires (they allow the sphere to be deflected in both the hori-
are typically 10 mm in diameter) makes them very zontal and vertical directions. The deflection due to
susceptible to breakage. Hot-film anemometers pro- wind drag on the sphere is sensed by proximity
vide a more rugged instrument with somewhat sensors that measure the displacement relative to the
slower, but still excellent, frequency response. Rather rigid mount. Carefully calibrated thrust anemom-
than strands of wires, a hot-film anemometer uses a eters have been used to measure turbulence from
thin film of nickel or platinum spread over a small fixed platforms for extended periods. They are fairly
cylindrical quartz or glass core. Even when these rugged and low-power, and have adequate response
sensors are closely monitored for breakage, aging for estimation of the flux. The main disadvantages of
and corrosion of the wires and films due to sea spray these devices are the need to accurately calibrate the
and other contaminants cause the calibration to direction response of each sensor and sensor drift due
change over time. Dynamic calibration in the field to aging of the springs.
has been used but this requires additional sensors. A very robust sensor for flux measurements, par-
Therefore, substantially more rugged anemometers ticularly for use on fixed platforms, relies on a
with absolute or more stable calibrations have gen- modification of the standard propellor vane an-
erally replaced these sensors in field studies. emometer used to measure the mean wind. The
Another laboratory instrument that meets these modification involves the use of two propellers on
requirements is the pitot tube; this uses two con- supporting arms set at 901 to each other. The entire
centric tubes to measure the difference between the assembly is attached to a vane that keeps the pro-
static pressure of the inner tube, which acts as a peller pointed into the wind. The device is known as
stagnation point, and the static pressure of the air a K-Gill anemometer from the appearance of the
flowing past the sensor. The free stream air also has a twin propeller-vane configuration (Figure 3). The
dynamic pressure component. Therefore, the differ- twin propellers are capable of measuring the in-
ence between the two pressure measurements can be stantaneous vertical and streamwise velocity, and the
used to compute the dynamic pressure of the air flow vane reading allows the streamwise velocity to be
moving past the sensor using Bernoulli’s equation broken down into its u and v components. This de-
(eqn [8]). vice is also very robust and low power. However, it
has a complicated inertial response on moving
1
Dp ¼ ra aU2 ½8
2

Here a is a calibration coefficient that corrects for

departures from Bernoulli’s equations due to sensor
geometry. A calibrated pitot tube can then be used to
measure the velocity.
The traditional design is most commonly used to
measure the streamwise velocity. However, three-axis
pressure sphere (or cone) anemometers have been
used to measure fluxes in the field. These devices use
a number of pressure ports that are referenced
against the stagnation pressure to measure all three
components of the velocity. This type of anemometer
has to be roughly aligned with the relative wind and
its ports must remain clear of debris (e.g., sea spray
and other particulates) to operate properly. Con-
sequently, it has been most commonly used on re-
search aircraft where the relative wind is large and
particulate concentrations are generally lower out-
side of clouds and fog.
Figure 3 The instrument at the far right is K-Gill anemometer
The thrust anemometer has also been used to dir-
shown during a deployment on a research vessel. The instrument
ectly measure the momentum flux in the marine at- on the left is a sonic anemometer that is shown in more detail in
mosphere. This device measures the frictional drag of Figure 4. Photograph provided by Olc Persson (CIRES/NOAA/
the air on a sphere or other objects. The most ETL).
222 SENSORS FOR MICROMETEOROLOGICAL & FLUX MEASUREMENTS

platforms and is therefore most appropriate for use However, the times of flight differ if there is a com-
on fixed platforms. Additionally, the separation be- ponent of the wind velocity along the path between
tween the propellers (typically 0.6 ) acts as a spatial the transducers. The velocity is directly computed
filter (i.e., it cannot detect eddies smaller than the from the two time of flight measurements, t1 and t2,
separation), so it cannot be used too close to the using eqn [9], where L is the distance between the
surface and still satisfy eqn [7]. This is generally not a transducers.
problem at the measurement heights used in most

field deployments. L 1 1
U¼  ½9
Over the past decade, sonic anemometers have 2 t1 t2
become the instrument of choice for most investi-
gations of air–sea interaction. These anemometers Three pairs of transducers are typically used to
use acoustic signals that are emitted in either a con- measure all three components of the velocity vector.
tinuous or pulsed mode. At present, the pulse type These devices have no moving parts and are there-
sonic anemometers are most commonly used in fore far less susceptible to mechanical failure. They
marine research. Most commercially available de- can experience difficulties when rain or ice covers the
vices use paired transducers that emit and detect transducer faces or when there is a sufficient volume
acoustic pulses (Figure 4). One transducer emits the of precipitation in the sampling volume. However,
pulse and the other detects it to measure the time of the current generation of sonic anemometers have
flight between them. The functions are then reversed proven themselves to be remarkably reliable in long-
to measure the time of flight in the other direction. term deployments over the ocean; so much so that
The basic concept is that in the absence of any wind two-axis versions of sonic anemometers are also
the time of flight in either direction is the same. beginning to replace cup and propellor/vane an-
emometers for mean wind measurements over the
ocean.

Motion Correction
The measurement of the fluctuating velocity com-
ponents necessary to compute the fluxes is compli-
cated by the platform motion on any aircraft, sea-
going research vessel, or surface mooring. This mo-
tion contamination must be removed before the
fluxes can be estimated. The contamination of the
signal arises from three sources: instantaneous tilt of
the anemometer due to the pitch, roll, and yaw (i.e.,
heading) variations; angular velocities at the an-
emometer due to rotation of the platform about its
local coordinate system axes; and translational vel-
ocities of the platform with respect to a fixed frame
of reference. Therefore, motion sensors capable of
measuring these quantities are required to correct the
measured velocities. Once measured, these variables
are used to compute the true wind vector from eqn
[10].

U ¼ TðUm þ Xm  RÞ þ Up ½10

Here U is the desired wind velocity vector in the

desired reference coordinate system (e.g., relative to
water or relative to earth); Um and Xm are the
measured wind and platform angular velocity vectors
Figure 4 A commercially available pulse-type sonic
respectively, in the platform frame of reference; T is
anemometer. The three sets of paired transducers are cabable
of measuring the three components of the velocity vector. This
the coordinate transformation matrix from the
type of device produced the time-series in Figure 1 and Figure 2. platform coordinate system to the reference co-
The sonic anemometer measures 0.75 m for top to bottom. ordinates; R is the position vector of the wind sensor
 SENSORS FOR MICROMETEOROLOGICAL & FLUX MEASUREMENTS 223

with respect to the motion sensors; and Up is the Thermistors and resistance wires are devices
translational velocity vector of the platform meas- whose resistance changes with temperature. Ther-
ured at the location of the motion sensors. mistors are semiconductors that generally exhibit a
A variety of approaches have been used to correct large negative change of resistance with temperature
wind sensors for platform motion. True inertial (i.e., they have a large negative temperature co-
navigation systems are standard for research aircraft. efficient of resistivity). They come in a variety of
These systems are expensive, so simpler techniques different forms including beads, rods, or disks.
have been sought for ships and buoys, where the Microbead thermistors are most commonly used in
mean vertical velocity of the platform is un- turbulence studies; in these the semiconductor is
ambiguously zero. These techniques generally use the situated in a very fine bead of glass. Resistance wires
motion measurements from either strapped-down or are typically made of platinum, which has a very
gyro-stabilized systems. stable and well-known temperature–resistance rela-
The strapped-down systems typically rely on a tionship. The trade-off is that they are less sensitive
system of three orthogonal angular rate sensors and to temperature change than thermistors. The probe
accelerometers, which are combined with a compass supports for these wires are often similar in design to
to get absolute direction. The high-frequency com- hot-wire anemometers, and they are often referred to
ponent of the pitch, roll, and yaw angles required for as cold-wires.
the transformation matrix are computed by inte- All of these sensors can be deployed on very fine
grating and highpass filtering the angular rates. mounts (Figure 5), which greatly reduces the adverse
The low-frequency component is obtained from the effects of solar heating but also exposes them to
lowpass accelerometer signals or, more recently, the harsh environments and frequent breaking. Add-
angles computed from differential GPS. The trans- itionally, the exposure invariably causes them to
formed accelerometers are integrated and highpass become covered with salt from sea spray. The coating
filtered before they are added to lowpass filtered GPS of salt causes spurious temperature fluctuations due
or current meter velocities for computation of Up to condensation and evaporation of water vapor on
relative to earth or the sea surface, respectively. The these hygroscopic particles. These considerations
gyro-stabilized system directly computes the orien- generally require more substantial mounts and some
tation angles of the platform. The angular rates are
then computed from the time-derivative of the
orientation angles.

Heat Fluxes
The measurement of temperature fluctuations over
the ocean surface has a similar history to that of the
velocity measurements. Laboratory sensors such as
thermocouples, thermistors, and resistance wires are
used to measure temperature fluctuations in the
marine environment.
Thermocouples rely on the Seebeck effect that
arises when two dissimilar materials are joined to
form two junctions: a measuring junction and a
reference junction. If the temperature of the two
junctions is different, then a voltage potential dif-
ference exists that is proportional to the temperature
difference. Therefore, if the temperature of the ref-
erence junction is known, then the absolute tem-
perature at the junction can be determined. Certain
combinations of materials exhibit a larger effect
(e.g., copper and constantan) and are thus commonly
used in thermocouple design. However, in all cases
Figure 5 A thermocouple showing the very fine mounts used
the voltage generated by the thermoelectric effect is for turbulence applications. The actual thermocouple is situated
small and amplifiers are often used along with the on fine wires between the probe supports and is too small to be
probes. seen in this photograph.
224 SENSORS FOR MICROMETEOROLOGICAL & FLUX MEASUREMENTS

sort of shielding from the radiation and spray. While a narrowband source for the radiation and a detector
this is acceptable for mean temperature measure- to measure the reduced transmission of that radi-
ments, the reduction in frequency response caused by ation over a known distance.
the shields often precludes their use for turbulence Early hygrometers of this type generated and de-
measurements. tected ultraviolet radiation. The Lyman a hygrometer
To combat these problems, marine micro- uses a source tube that generates radiation at the
meteorologists have increasingly turned to sonic Lyman a line of atomic hydrogen which is strongly
thermometry. The time of flight measurements from absorbed by water vapor. This device has excellent
sonic anemometers can be used to measure the speed response characteristics when operating properly.
of sound c along the acoustic path (eqn [11]). Unfortunately, it has proven to be difficult to operate
in the field due to sensor drift and contamination of


L 1 1 the special optical windows used with the source and
c¼ þ ½11 detector tubes. A similar hygrometer that uses
2 t1 t2
krypton as its source has also been used in the field.
The speed of sound is a function of temperature and Although the light emitted by the krypton source is
humidity and can be used to compute the sonic not as sensitive to water vapor, the device still has
temperature Ts defined by eqn [12], where UN is more than adequate response characteristics and
velocity component normal to the transducer path. generally requires less maintenance than the Lyman
a. However, it still requires frequent calibration and
c2 þ UN
2 cleaning of the optics. Therefore, neither device is
Ts ¼ Tð1 þ 0:51qÞ ¼ ½12 particularly well suited for long-term operation
403
without frequent attention.
The normal wind term corrects for lengthening of the Commercially available infrared hygrometers are
acoustic path by this component of the wind. This being used more and more in marine micro-
form of velocity crosstalk has a negligible effect on meteorological investigations (Figure 6). Beer’s law
the actual velocity measurements, but has a meas-
urable effect on the sonic temperature.
Sonic thermometers share many of the positive
attributes of sonic anemometers. Additionally, they
suffer the least from sea salt contamination com-
pared to other fast-response temperature sensors.
The disadvantage of these devices is that velocity
crosstalk must be corrected for and they do not
provide the true temperature signal as shown by eqn
[12]. Fortunately, this can be advantageous in many
investigations because the sonic temperature closely
approximates the virtual temperature in moist air
Tv ¼ T(1 þ 0.61q). For example, in many investi-
gations over the ocean an estimate of the buoyancy
flux is sufficient to account for stability effects. In
these investigations the difference between the sonic
and virtual temperature is often neglected (or a small
correction is applied), and the sonic anemometer/
thermometer is all that is required. However, due to
the importance of the latent heat flux in the total heat
budget over the ocean, accurate measurement of the
moisture flux is often a crucial component of air–sea
interaction investigations.
The accurate measurement of moisture fluctu-
ations required to compute the latent heat flux is
arguably the main instrumental challenge facing
marine micrometeorologists. Sensors with adequate
frequency response generally rely on the ability of Figure 6 Two examples of commercially available infrared
water vapor in air to strongly absorb certain wave- hygrometers. The larger hygrometer is roughly the height of the
lengths of radiation. Therefore, these devices require sonic anemometer shown in Figure 4.
 SENSORS FOR MICROMETEOROLOGICAL & FLUX MEASUREMENTS 225

(eqn [13]) provides the theoretical basis for the

transmission of radiation over a known distance.

T ¼ eðgþdÞD ½13

Here T is the transmittance of the medium, D is the

fixed distance, and g and d are the extinction co-
efficients for scattering and absorption, respectively.
This law applies to all of the radiation source de-
scribed above; however, the use of filters with infra-
red devices allows eqn [13] to be used more directly.
For example, the scattering coefficient has a weak
wavelength dependence in the spectral region where
infrared absorption is strongly wavelength
dependent. Filters can be designed to separate the
infrared radiation into wavelengths that exhibit
strong and weak absorption. The ratio of transmit-
tance of these two wavelengths is therefore a func-
tion of the absorption (eqn [14]), where the
subscripts s and w identify the variables associated
with the strongly and weakly absorbed wavelengths.

Ts
Eeds D ½14
Tw

Calibration of this signal then provides a reliable

measure of water vapor due to the stability of current
generation of infrared sources.
Infrared hygrometers are still optical devices and
can become contaminated by sea spray and other Figure 7 A sensor package used to measure the momentum,
airborne contaminants. To some extent the use of the sensible heat, and latent heat fluxes from a moving platform. The
transmission ratio negates this problem if the con- cylinder beneath the sonic anemometer/thermometer holds 3-
tamination affects the two wavelengths equally. axis angular rate sensors and linear accelerometers, as well as a
magnetic compass. Two infrared hygrometers are deployed
Obviously, this is not the case when the optics be-
beneath the sonic anemometer. The radiation shield protects
come wet from rain, fog, or spray. Fortunately, the sensors that measure the mean temperature and humidity.
devices recover well once they have dried off and are Photograph provided by Wade McGillis (WHOI).
easily cleaned by the rain itself or by manual flushing
with water. Condensation on the optics can also be
reduced by heating their surfaces. These devices re- for moisture fluctuations. Hot-wire and hot-film
quire longer path lengths (0.2–0.6 m) than Lyman a anemometers along with the finer temperature and
or krypton hygrometers to obtain measurable ab- humidity devices are also required to measure directly
sorption (Figure 6). This is not a problem as long as the viscous dissipation of the turbulent eddies that
they are deployed at heights bD. occurs at very small spatial scales.
Instruments for measuring turbulence are gener-
ally not considered low-power when compared to
Conclusions
the mean sensors normally deployed on surface
The state of the art in sensor technology for use in the moorings, so past deployments of these sensors were
marine surface layer includes the sonic anemometer/ mainly limited to fixed platforms or research vessels
thermometer and the latest generation of infrared with ample power. Recently, however, sensor pack-
hygrometers (Figure 7). However, the frequency ages mounted on spar and discus buoys have suc-
response of these devices, mainly due to spatial aver- cessfully measured motion-corrected momentum and
aging, precludes their use from aircraft. Instead, air- buoyancy fluxes on month- to year-long deployments
craft typically rely on gust probes for measurement of with careful power management. The use of these
the required velocity fluctuations, thermistors for sensor packages is expected to continue owing to the
temperature fluctuations, and Lyman a hygrometers desirability of these measurements and technological
226 SENSORS FOR MICROMETEOROLOGICAL & FLUX MEASUREMENTS

advances leading to improved power sources and Dobson FW, Hasse L, and Davis RE (1980) Air–Sea
reduced power consumption by the sensors. Interaction; Instruments and Methods. New York:
Plenum Press.
Edson JB, Hinton AA, Prada KE, Hare JE, and Fairall CW
See also (1998) Direct covariance flux estimates from mobile
platforms at sea. Journal of Atmospheric and Oceanic
Moorings. Satellite Remote Sensing Microwave Technology 15: 547--562.
Scatterometers. Sensors for Mean Meteorology. Fritschen LJ and Gay LW (1979) Environmental
Ships. Turbulence Sensors. Instrumentation. New York: Springer-Verlag.
Kaimal JC and Gaynor JE (1991) Another look at sonic
thermometry. Boundary-Layer Meteorology 56:
Further Reading 401--410.
Larsen SE, Højstrup J, and Fairall CW (1986) Mixed and
Ataktürk SS and Katsaros KB (1989) The K-Gill, a twin dynamic response of hot wires and measurements of
propeller-vane anemometer for measurements of turbulence statistics. Journal of Atmospheric and
atmospheric turbulence. Journal of Atmospheric and Oceanic Technology 3: 236--247.
Oceanic Technology 6: 509--515. Schmitt KF, Friehe CA, and Gibson CH (1978) Humidity
Buck AL (1976) The variable path Lyman-alpha sensitivity of atmospheric temperature sensors by salt
hygrometer and its operating characteristics. Bulletin of contamination. Journal of Physical Oceanography 8:
the American Meteorological Society 57: 1113--1118. 141--161.
Crawford TL and Dobosy RJ (1992) A sensitive fast- Schotanus P, Nieuwstadt FTM, and de Bruin HAR (1983)
response probe to measure turbulence and heat flux Temperature measurement with a sonic anemometer
from any airplane. Boundary-Layer Meteorology 59: and its application to heat and moisture fluxes.
257--278. Boundary-Layer Meteorology 26: 81--93.
SENSORS: TURBULENCE
 TURBULENCE SENSORS
N. S. Oakey, Bedford Institute of Oceanography, into large-scale eddies determined by flow boundary
Dartmouth, Nova Scotia, Canada conditions that characterize turbulent kinetic energy
Copyright & 2001 Elsevier Ltd. at its maximum scales. Further vortex stretching
creates smaller and smaller eddies resulting in a tur-
bulent cascade of energy (velocity fluctuations) to
smaller scales until viscous forces begin to dominate
where the energy is eventually dissipated as heat.
Introduction This article focuses on sensors to measure this dis-
sipation process directly by measuring the effect of
This article describes sensors and techniques used to viscosity on the turbulent cascade.
measure turbulent kinetic energy dissipation in the The irregular and aperiodic velocity fluctuations in
ocean. Dissipation may be thought of simply as the space and time characteristic of turbulence, accom-
rate at which turbulent mechanical energy is con- panied by energy transfer between scales and asso-
verted into heat by viscous friction at small scales. ciated fluid mixing, may be described mathematically
This is a complicated indirect measurement requiring through nonlinear terms in the Navier–Stokes equa-
mathematical models to allow us to envisage and tion. Nevertheless, it is difficult to solve numerically
understand turbulent fields. It will require using this in oceanographic applications. At the dissipation
theory to understand how sensors might be de- scales, typically a few meters and smaller, we nor-
veloped using basic principles of physics to measure mally assume that the turbulent field is homogenous
properties of a turbulent field to centimeter scales. and that it has definable statistical averages in all
Instruments must be used to carry these sensors into parts of the field. We further assume that direction is
the ocean so that the researcher can measure its unimportant (isotropy) and statistical distributions
turbulent characteristics in space and time. It is also depend only on separation distances between points.
this sensor–instrument combination that converts the With the turbulence controlled only by internal
sensor output into a quantity, normally a voltage parameters, we assume the nature of the nonlinear
varying in time, that is used by the experimenter to cascade of energy from large to small scales generates
calculate turbulent intensity. Thus, both the charac- a universal velocity spectrum. An example of this
teristics of sensors and the way in which the sensor– spectrum is shown schematically in Figure 1A. At
instrument combination samples the environment low wavenumbers, k, no energy is taken out by vis-
must be understood and will be discussed below. cous dissipation, so the energy flux, e, across each
wave number, or down the cascade, is constant.
Through dimensional arguments, the three-dimen-
Understanding Turbulence in the sional turbulent energy spectrum, EðkÞ in this region
Ocean (called the inertial subrange) as a function of wave-
There is no universally accepted definition of tur- number k is given by
bulence. Suppose that one stirs a bowl of clear water
and injects some colored dye into it. One sees that EðkÞ ¼ ae2=3 k5=3 ½1
filaments of dye become stretched, twisted and con-
torted into smaller and smaller eddies and eventually where a is a constant determined experimentally to
the bowl becomes a uniform color. This experiment be approximately 1.5. In practice the three-dimen-
leads to one definition of turbulence. It includes the sional spectrum given in eqn [1] cannot easily be
concept that eddies in the water are distributed measured and one must use the one-dimensional
randomly everywhere in space and time, that energy analogy where k is replaced by a component ki .
is transferred from larger to smaller eddies, and that At higher wavenumbers or smaller scales the vel-
over time the mean separation of the dyed particles ocity gradient spectrum (obtained by multiplying the
increases. In contrast, the ocean is typically stratified spectrum in eqn [1] by the square of the wave
through a density that is determined by the tem- number k2 ) shows more clearly where dissipation
perature and salt in the water as well as the pressure. occurs. Figure 1B shows the spectra of velocity shear
In this environment, a vertical shear in the velocity in for velocity fluctuations for one component of k for
the water column can be large enough to overcome values of e most typically found in the ocean. In this
the stability. Energy from the mean flow is converted case, the spectra of fluctuations transverse to the

229
230 TURBULENCE SENSORS

The factor 2p gives a length scale from the radian

wave number. This is an important scale for the de-
sign of instruments and sensors because it defines the
smallest diameter eddies that must be measured.
The dissipation e is given by integrating the spec-
trum shown in Figure 1B.

ZN ZN
e ¼ 15v k21 E1 ðk1 Þdk1 ¼ 7:5v k21 E2 ðk1 Þdk1 ½3
0 0

E1 ðk1 Þ is the one-dimensional wavenumber spectrum

of longitudinal velocity, and E2 ðk1 Þ is the one-
dimensional spectrum of transverse velocity and one
assumes isotropy to estimate the factors 15 and 7.5,
respectively. In practice, the upper integration limit
may be replaced with the viscous cutoff scale. For the
transverse turbulent velocity u, the shear variance in
the z direction, ðdu=dzÞ2 is equivalent to the integral
of equation [3] and e is given by


15 du 2
e¼ v ½4
2 dz

These assumptions are important to the way in

which sensors are designed. A common way to ob-
serve turbulent fields is by making measurements of
velocity and other mixing quantities along a trajec-
tory through a turbulent field assuming that it is
frozen in space and time. Measurement along a line,
recorded as a time-series (Figure 2), is interpreted as
Figure 1 (A) The universal, velocity spectra for dissipation spatial variability by assuming stationarity and using
rates that typically occur in the ocean. Power density in velocity is the known sensor velocity to convert into distance.
plotted as a function of wavenumber. The shape of the spectrum Standard Fourier transform techniques allow one to
remains the same but, as the energy in the turbulent field generate spectra similar to those in Figure 1 from
increases, the spectrum moves to higher wavenumbers and to
which dissipation, e, may be estimated.
higher intensities. (B) The equivalent universal, velocity shear
spectra. (A) and (B) both show the inertial subrange and If there is a temperature gradient in the water
dissipation region, but in (B), the dissipation portion is more column when turbulence is generated, the velocity
strongly emphasized. field strains the temperature field, creating strongly
interleaved temperature filaments over the vertical
measurement direction are shown but the picture for range of the overturn. The temperature micro-
along-axis fluctuations would look almost identical. structure intensity depends not only on the mean
At the highest wavenumbers (smallest scales), vis- gradient but also on the energy in the turbulent field,
cous dissipation reduces the energy per unit wave- in particular dissipation, e. Temperature fluctuations
number to zero. At small scale, it is assumed that recorded as a time-series (Figure 2) can be repre-
turbulent motion is determined only by kinematic sented by spectra similar to those shown in Figure 1.
viscosity, v( ¼ 1.3  106 m2 s1 at 101C), and the As with the velocity fluctuations, there is a subrange
rate, e, at which energy passed down from larger where diffusive and viscous effects are unimportant
eddies, must be dissipated. By dimensional argu- where temperature fluctuations are transferred
ments the length scale at which viscous forces equal towards higher wave numbers. Temperature spectra
inertial forces, and viscosity dissipates the turbulent persist to length scales smaller than the viscous cutoff
energy as heat, is given by viscous cutoff scale scale. In this range, not only kinematic viscosity, v,
 1=4 and dissipation, e, are important but also, thermal
Lv ¼ 2p v3 =e ½2 diffusion, kT (E1.4  107 ms s1). The cutoff
 TURBULENCE SENSORS 231

0 layers on continental shelves. (To put these numbers

into a simple perspective, energy dissipated in the
ocean may range from the almost insignificant rate of
10
100 W km3 to the very large rate of 100 MW km3.)
Present sensors and instruments are capable of meas-
uring over this range of dissipation. Regions of higher
uz Tz dissipation such as river outflows and tidal channels
20 are not normally measurable with sensors and instru-
ments described here.

30 Measuring Dissipation in the Ocean

Depth (m)

The most common technique of estimating dissi-

pation in the ocean involves measuring small-scale
40 velocity and temperature fluctuations. This may be
accomplished by dropping a profiler vertically, tow-
ing one horizontally or setting it at a fixed position
50 and measuring the fluctuations in velocity and tem-
perature as the water moves past the sensors. This
allows a time-series of turbulent velocity fluctuations
to be recorded. A typical platform used to measure
60
dissipation in the ocean is a vertical profiler that falls
typically at a speed of 0.5–1.0 ms1. There have been
many such instruments built and each one typically
70 carries a number of sensors to measure some com-
8 10 12 14 16 18 20 ponents of the turbulent velocity as well as tem-
Temperature (°C)
perature microstructure. A sample time-series for a
Figure 2 A representative vertical profile of temperature is vertical profiler is shown in Figure 2. Assuming that
shown from the surface to bottom obtained with a vertical falling the turbulent field is isotropic, homogeneous and
instrument. In panels at the right are shown expanded portions of stationary one can use the mean flow velocity to
the velocity shear (uz ) and the gradients in temperature (Tz ). The determine the wavenumber scale and calculate the
panels represent small sections of the vertical record that are
treated as time-series to calculate spectra similar to those in
one-dimensional turbulence spectrum, E1 ðk1 Þ or
Figure 1. The upper panel of Uz at mid-depth is a region of low E2 ðk1 Þ, as defined above and from this determine the
dissipation and the one below represents higher dissipation. dissipation, e, using eqns [3] and [4].
As the turbulent dissipation gets larger, the wave-
number at cut-off gets larger. Alternatively, Lv and
wavelength for temperature fluctuations is given by
LT become smaller as shown in Figure 3(A). As one
 1=4 tries to measure higher dissipation one must have a
LT ¼ 2p vk2T =e ½5 sensor with better spatial resolution and higher fre-
quency response. We convert from wavenumber, k
Under restricted circumstances, the temperature (cycles m1), to frequency using the relationship f ¼
gradients or temperature microstructure can be kV (Hz) where V (m s1) is the flow speed past the
measured in the ocean to this scale. Under these sensor. The cutoff frequencies corresponding to Lv
circumstances one can determine LT and hence es- and LT are given by fci ¼ V=Li. In practice, one does
timate dissipation e. not have to measure the microstructure variance to
The sensors used most commonly in oceanography the cutoff frequency because of the universal char-
to measure dissipation make use of the above ideas. acteristic of the dissipation curves. A usual com-
Velocity fluctuations may be used to determine dissi- promise is to consider that if 90% of the dissipation
pation e directly using eqns [3] or [4]. Measuring curve is measured then a satisfactory measure of
temperature fluctuations allows dissipation to be cal- dissipation can be achieved. This is summarized in
culated indirectly from eqn [5]. The units used to ex- Figure 3A which shows the sampling frequency that
press dissipation in the ocean are W kg1 (watts of must be achieved to resolve a particular dissipation.
mechanical energy converted into heat per kilogram of (It must be remembered that to resolve the energy at
sea water). Typical values range from 1010 W kg1 in any frequency one must sample at least twice that
the deep ocean to 104 W kg1 in active boundary frequency.)
232 TURBULENCE SENSORS

10 differentiated to emphasize the high frequency part

LT and LV (cm)

of the turbulence spectrum. This gives the velocity

5
shear, and analysis of this signal allows direct gen-
LV
eration of spectra similar to the theoretical ones
LT
shown in Figure 1B. For this reason, airfoil probes
0 are often called shear probes. These sensors measure
(A) _ 10 _8 _6 _4
10 10 10 10 the component of turbulence perpendicular to the
drop direction of the instrument. As such, it is eqn
Sampling freq. (Hz)

2000
[4]. Which is most relevant to calculating dissipation,
LT e. Figure 3B, shows that to measure dissipation to
1000 105 W kg1 in a flow speed of 1 m s1 (along the
LV
axis of the sensor) the output must be sampled to at
least as rapidly as 200 Hz.
0 _
10 _8 _6 _4 Of the many instruments that use this sensor to
10 10 10 10
(B) _
Dissipation (W kg 1) measure dissipation, the most common are vertical
profilers. Those used near the surface are often called
Figure 3 (A) The decrease in the cutoff scale with increasing tethered free-fall profilers because they have a light,
dissipation for viscous dissipation, Lv (eqn [2]) and thermal loose line attached to the instrument for quick re-
dissipation, LT (eqn [5]). (B) The sampling frequency that is
covery and redeployment. The line is usually a data
required for a particular e for viscous dissipation, Lv and for
thermal cutoff, LT. The upper and lower boundaries of the shaded link to the ship where data are recorded on com-
bands correspond to measurement at flow speeds of 1.0 and puters for analysis. Because of the intermittent nature
0.5 m s1, respectively. of turbulence, it is important to have many profiles
(or independent samples) in measuring dissipation to
be able to obtain a statistically robust average value.
Turbulence Dissipation Sensors For deeper measurements of dissipation, free-fall
profilers are used that have no tether line. They are
Airfoil Probes
deployed to a predetermined depth in the ocean
One of the most commonly used turbulence sensors where their buoyancy is changed to allow them to
to measure turbulent velocity fluctuations is called an return to the surface. These instruments record in-
airfoil probe. This sensor is an axially symmetrical ternally and can be inherently quieter than tethered
airfoil made of flexible rubber surrounding a sensi- free-fall instruments but are slower to recover and
tive piezoelectric crystal. The sensitive tip of the redeploy. In practice, both types of profilers can
probe approximates a parabola of revolution, several measure dissipation as low as 1010 W kg1. In
millimeters in diameter and about 1 cm long. The shallow regions of high dissipation such as the bot-
crystal generates a voltage proportional to the mag- tom boundary layer of tidally generated flow over
nitude of a force applied perpendicular to its axis. banks, in bottom river channels or in active regions
The crystal is rigid in one transverse direction so such as the Mediterranean outflow tethered free-fall
responds to a cross force only in one direction. Thus, instruments have been most successful. Where the
two sensors are required to measure the two trans- dissipation exceeds 105 W kg1, these profilers and
verse components of turbulent velocity fluctuations. the shear probe sensor give limited results.
The sensor is placed on the leading end of an in- The airfoil probe has also been used successfully to
strument that is moving relative to the water at a obtain dissipation measurements horizontally. It has
mean speed V. In a mean flow along the axis of the been used as a sensor on a towed fish pulled hori-
shear probe, no lift will be generated and no force zontally at speeds of order 1 m s1. The results look
applied to the crystal. If there is an off-axis turbulent similar to those in Figure 2 where the depth axis is
velocity, a lift will be generated which will apply a replaced by a horizontal axis and similar techniques
force to the piezoelectric crystal through the flexible to those described above are used to extract dissi-
rubber tip. Thus, the sensor will provide a voltage pation. Because of towline vibration, a towed in-
that is linearly proportional to the turbulent velocity. strument is generally noisier than a free-fall profiler.
The effective resolution of the sensor is of order If the vibration noise of the platform is transferred to
1 cm, the smallest scale of turbulence that can be the airfoil sensor, it will generate velocity signals
effectively measured by this sensor. From Figure 3, it relative to the sensor indistinguishable from turbu-
can be seen that for values above 105 W kg1 this lence in the water with the sensor not vibrating.
type of sensor will begin to underestimate dissi- Generally, a measurable dissipation lower limit for
pation. Normally the signal from the sensor is these instruments of 109 W kg1 would be
 TURBULENCE SENSORS 233

considered good. These shear probes have also been Pitot Tubes
mounted on submarines for horizontal measure-
Another recently developed sensor used to measure
ments. Nevertheless, this platform has had only
dissipation makes use of a Pitot tube. If a Pitot tube is
limited use because of vehicle noise and expensive
placed in water flowing at a speed W along its axis,
operating costs. More recently, unmanned sub-
the pressure generated by the flow is proportional
marines called autonomous underwater vehicles have
to W 2 . This technique has been applied to turbulence
been used as suitable platforms for turbulent kinetic
measurements by carefully designing an axisym-
energy measurements. They are expected to have
metric port a few millimeters in diameter on the tip
similar noise characteristics to towed instruments.
of a sensor of order 1 cm in diameter. By connecting
Another interesting way of obtaining horizontal
the port to a very sensitive differential pressure sen-
measurements is to place shear probes on a moored
sor, fluctuations in pressure along the axis of the
instrument. The turbulence in the water is measured
probe can be measured. Using suitable electronic
as it flows past the sensor at a speed V m s1. In this
circuits, a signal is produced that is linearly pro-
case, the water velocity must typically be faster than
portional to along-axis fluctuations in turbulent
0.1 m s1 for the measurements to be within the
velocity. In this sense, it is similar to the heated-film
sensor capabilities and mooring vibrations generate
sensor and different from the shear probe which
similar problems to towed instruments.
measures fluctuations perpendicular to the mean
flow. This sensor has been used in conjunction with a
Thin Film Sensors pair of shear probes to simultaneously measure all
three components of turbulent velocity fluctuations.
One of the original sensors used to measure turbu-
lence and dissipation is called a hot film sensor. In
Temperature Microstructure Sensors
these sensors, a platinum or nickel film is deposited
on the surface near the conical tip of a glass rod of As outlined above, if there is turbulent mixing oc-
order 1 mm diameter and covered with a thin film of curring in a region where there is a temperature
quartz to insulate it from the water. The film is gradient, the turbulent velocity will cause the tem-
heated to several degrees centigrade above the am- perature to be mixed. If, for example, warmer fluid
bient temperature and special electronics are used to overlays colder fluid, turbulence will move parcels of
maintain a constant thin film temperature. Water warm fluid down and cold fluid up. A temperature
flowing across the probe cools the platinum. Fluc- sensor that traverses a patch of fluid such as this will
tuations in the current, required to keep the sensor at measure fluctuations in temperature as shown in
a constant temperature, are a measure of the turbu- Figure 2. The spectrum of these fluctuations can be
lent velocity fluctuations along the axis of the sensor. used to determine the dissipation using eqn [5]. Be-
This sensor measures the E1 ðk1 Þ component of the cause the molecular diffusivity of heat for water is
turbulent field as opposed to the E2 ðk1 Þ component much smaller than the molecular viscosity, the scale
measured by the shear probe. Therefore, the first part at which temperature fluctuations cease is about a
of the eqn [3] is relevant to estimating dissipation. factor of three smaller than the scale at which vel-
The primary advantage of this sensor over the shear ocity fluctuations cease. This is shown clearly in
probe is that it has much smaller spatial resolution Figure 3A that compares LT and Lv . These facts
and a much higher frequency response. As one can place a severe restriction on the speed and size of a
see from Figure 3, this allows one to measure to temperature sensor compared to a shear probe, or
higher dissipation rates. The disadvantages of this alternatively limits the speed that an instrument may
probe are that the electronics to run it are much more fall. For the same fall speed, a temperature sensor
complicated than for shear probes and the sensors must be sampled at a much higher rate than a shear
are more difficult to fabricate and quite expensive. probe. The simplest temperature sensor with the
They also require a lot of power to heat since they precision and noise level to measure temperature
are very low in resistance (of order 5–10 O). Because microstructure in the ocean is the thermistor.
the quartz insulation must be extremely thin to The smallest thermistors that are used in sea water
provide good heat transfer, thin films are also very are a fraction of millimeter in diameter and have a
fragile and easily damaged by impact with particles frequency response of order 10 ms. At a flow speed of
in the water. These probes do not provide an output 1 m s1, one is able to delimit the spectrum of tem-
voltage that is linear with turbulent velocity fluctu- perature for dissipations up to about 107 W kg1.
ations. They also tend to be noisy and subject to Some success has been obtained by using very slow
fouling. They are seldom used today in ocean moving profilers that fall or rise at about 0.1 m s1.
measurements. An alternative to the thermistor is a thin film
234 TURBULENCE SENSORS

thermometer. It is similar to the hot film velocity fluctuations in the water change the transit time and
sensor described above and is constructed identically. allow water velocity fluctuations to be inferred. This
Used as a thermometer, the change in the resistance configuration of sensors is generally mounted as a
of this sensor is a measure of change of temperature. fixed array on a platform on the bottom and has been
Thin film sensors are faster than thermistors, typi- used to measure turbulent mixing in many places on
cally with a time-constant of 2 ms which means that continental shelves. This technique has the advan-
for any sensor velocity the temperature fluctuations tage over profiling dissipation sensors of measuring
may be measured to a higher wave number. These three components of velocity fluctuations over long
sensors are nevertheless at least an order of magni- periods of time at a single place. Dissipation is esti-
tude noisier than thermistors, which means that they mated from the k5=3 wavenumber range.
are suitable for measuring microstructure only in
regions where there are strong mean gradients. Using
thermometry to measure dissipation is subject to Conclusions
large errors because, as indicated in eqn [5], dissi-
pation is proportional to (LT Þ4 , and this requires The measurement of mixing rates in the ocean is
accuracy in determining LT that is seldom achieved. important to our understanding of the distributions
Some success has been made using sensors that of temperature, salinity, and nutrients in the ocean.
measure conductivity as a proxy for temperature. We need to understand this to include them correctly
These sensors make use of the fact that the con- in climate and biological ocean models. The way in
ductivity of sea water is determined by both salt which energy is converted from sources at large scale
and temperature and in most cases, the temperature and dissipated at small scales has required the de-
causes most of the fluctuations. The techniques velopment of a variety of ocean sensors. Some of
used are similar to those described above for tem- these are described briefly above. It is hoped that
perature. Some of the sensors are smaller and faster enough of the key words and ideas have been put
than thermistors and less noisy than thin film forward for the reader to understand some of the
thermometers. They are still limited to the same principles involved in turbulence measurement and
constraints as thermometers in that they must fully at least some of the sensors and techniques used.
resolve the spectrum in order to estimate LT and
utilize eqn [5]. Nomenclature
Acoustic Current Meters a An experimentally determined spectral
constant
Acoustic techniques have also been used to measure E(k) Energy spectral density
water velocity in the ocean and indirectly to infer E1(k1) One-dimensional energy wavenumber
dissipation rates. One such technique utilizes an spectrum – fluctuations along the axis of
acoustic Doppler current meter optimized to meas- measurement
ure vertical velocity fluctuations in the water column. E2(k1) One-dimensional energy wavenumber
In these instruments, a sound pulse is transmitted spectrum – fluctuations perpendicular to
into the water and the sound scattered back to a the axis of measurement
sound receiver. The back-scattered pulse contains e Dissipation of turbulent kinetic energy
information about the water velocity because of the f measurement or sampling frequency
Doppler shift in the sound frequency. This technique KT molecular diffusivity of heat
is unable to measure to dissipation scales but instead, Lv viscous cutoff scale
measures vertical velocities in the k5=3 wavenumber LT temperature cutoff scale
range defined by eqn [1]. By suitably defining a tur- u horizontal velocity fluctuation
bulent timescale, dissipation is estimated from the v kinematic viscosity
intensity in the fluctuations in the vertical velocity. V flow velocity along axis of sensor
This technique is very useful in studying turbulence W drop velocity
in regions of intense mixing such as tidally driven z distance coordinate (normally vertical)
flows.
In another technique, an array of small acoustic
transmitters and receivers is configured such that the See also
transit time of a pulse of sound can be measured over
a short distance of around 10–20 cm. Velocity Profiling Current Meters.
 TURBULENCE SENSORS 235

Further Reading Journal of Atmospheric and Oceanic Technology (1999)

16(11), Special Issue on Microstructure Sensors.
Bradshaw P (1971) An Introduction to Turbulence and Its Neumann G and Pierson WJ (1966) Principles of Physical
Measurement. Oxford, New York, Toronto, Sydney, Oceanography. Englewood Cliffs, NJ: Prentice Hall.
Paris, Braunschweig: Pergamon Press. Patterson GK and Zakin JL (1973) Turbulence in liquids.
Dobson F, Hasse L, and Davis R (1980) Air–Sea Proceedings of the Third Symposium, 414pp., Depart-
Interaction Instruments and Methods. New York and ment of Chemical Engineering, University of Missouri-
London: Plenum Press. Rolla.
Frost W and Moulden TH (1977) Handbook of Summerhayes CP and Thorpe SA (1996) Oceanography,
Turbulence, vol. 1: Fundamentals and Applications. pp. 280--299. New York: John Wiley.
New York: Plenum Press
Hinze JO (1959) Turbulence. New York, Toronto, London:
McGraw-Hill.
SENSORS: OPTICAL
 ABSORBANCE SPECTROSCOPY FOR CHEMICAL
SENSORS
R. Narayanaswamy, The University of Manchester, the optical fiber to the phototransducer. The photo-
Manchester, UK transducer generates an electrical signal that is re-
F. Sevilla, III, University of Santo Tomas, Manila, lated to the concentration of the analyte.
The Philippines This article focuses on chemical sensors that are
& 2009 Elsevier Ltd. All rights reserved. based on absorbance spectroscopy. The various
principles involved are reviewed, and some appli-
cations are presented.

Introduction Absorbance and Reflectance

Chemical sensors have introduced an alternative Spectroscopy
technology for chemical measurements. These ana-
Ultraviolet (UV)/visible absorbance spectroscopy has
lytical devices generate an electrical signal in re-
been employed extensively in analytical chemistry.
sponse to the presence of a specific substance, the The basis for analysis here is the ability of the analyte
signal being related to the concentration of the ana-
or its derivative to absorb radiation that impinges
lyte. The application of these sensors has simplified
into the measuring system. The radiation can have a
chemical analysis, since the need for obtaining a la-
wavelength occurring in the UV (200–400 nm), vis-
boratory sample is eliminated and a real-time and
ible (400–780 nm), near-infrared (780–3000 nm),
on-site measurement can be carried out. Further-
or infrared (3–50 mm) region. The absorption of
more, the configuration of these devices enables the
radiation causes a reduction in the intensity of the
detection with great sensitivity of low concentrations
radiation after it has passed through the system
of chemical species. (Figure 2). This phenomenon is mathematically de-
A number of analytical principles have been ex-
scribed by the Beer–Lambert law, as expressed by the
ploited in the development of chemical sensors.
following equation:
Among them are the optical methods of chemical
analysis, which rely on the interaction of electro-
magnetic radiation with matter for the quantitation I0
A ¼ log ¼ ecl ½1
of a large number of substances. These methods It
provide a rapid and nondestructive tool for the
measurement of chemical species. Optical methods where A is the absorbance, c is the concentration of
have indeed played an important role, and continue the absorbing species, I0 and It are the intensity of
to do so, in various field of chemical analysis. incident and transmitted light, respectively, e is the
One group of chemical sensors that are based on
the optical methods of chemical analysis is called
optodes (or optrodes). Other types of optical sensing
techniques that utilize waveguides, for example, Optoelectronic
surface plasmon resonance, also exist but are not instrumentation
reviewed here. The basic concept of optodes is de- Light Photo-
source transducer
picted diagrammatically in Figure 1. An optode
consists of an optical fiber, an optoelectronic instru-
mentation that incorporates a light source and a
Optical
phototransducer, and a solid-phase molecular rec- fiber
ognition element. Light from a suitable source is system
launched into the optical fiber and directed to the
sensing zone which contains the molecular recog-
nition element. The molecular recognition element Molecular
recognition
reacts with the analyte, resulting in a modification of element
its optical property. This change is probed by the
supplied radiation which is subsequently guided via Figure 1 Basic design of optical fiber chemical sensor.

239
240 ABSORBANCE SPECTROSCOPY FOR CHEMICAL SENSORS

Cladding
(a) (b)

Incident Transmitted
radiation radiation
(Io) (It)

Core
Analyte
solution
Figure 3 Structure of an optical fiber. (a) Longitudinal
cross-section view, and (b) latitudinal cross-section view.
Figure 2 Absorption of radiation.

terms of the molar absorptivity (e) and the concen-

molar absorptivity of the species, and l is the optical
tration (c) of the absorbing species. Thus, eqn [2] can
path length of the absorbing species.
be rewritten as
Absorbance spectroscopy is applicable only when
the measurand system containing the analyte is
transparent. However, if the medium is optically ð1  RÞ2 ec
FðRÞ ¼ ¼ ¼ kc ½3
dense or even opaque, in which case the absorbance 2R S
measurements would produce a high background, a
where k ¼ e/S, and S is assumed to be independent of
technique that is complementary to absorptiometry
concentration. Equation [3] is analogous to the Beer–
(viz., reflectometry) can be employed for analytical
Lambert relationship (eqn [1]) and holds true within
measurements. In this case, the radiation infringes on
a range of concentrations for solid solutions in which
the boundary interface of two media having different
the absorber is adsorbed onto the surface of a scat-
dielectric constants and reflection occurs.
tering particle. The reflectance values (R) are gener-
Two distinct types of reflection are possible,
ally evaluated relative to the reflectance of standard
namely (1) specular (or mirror-type) reflection and
reference materials such as barium sulfate.
(2) diffuse reflection. Specular reflection occurs at the
interface of a medium with no transmission through
it, and reflection is at the same angle as the incident Optical Fibers
light; whereas, in diffuse reflection, the light pene-
trates the medium and subsequently reappears at the Almost all optical sensors employ optical fibers to
surface after partial absorption and multiple scat- transmit light to and from the molecular recognition
tering within the medium. Of these two processes, element. The fiber couples the optoelectronic in-
diffuse reflection has found to be useful in chemical strumentation to the molecular recognition element,
measurements. Specular reflection is minimized or resulting in an integrated analytical system. This in-
eliminated through appropriate sample preparation tegration has contributed to the simplification of the
and optical engineering. chemical measurement process.
The distribution of diffusely reflected light is ra- Optical fibers consist of a cylinder (known as the
ther homogeneous and largely independent of size ‘core’) of transparent dielectric with a certain re-
and shape of the particles. The optical characteristics fractive index (n1), surrounded by a thin film of an-
of diffuse reflectance are dependent on the com- other dielectric (called the ‘cladding’) of a lower
position of the system. Among several theoretical refractive index (n2). Most optical fibers are then
models that have been proposed for diffuse reflec- covered with a protective jacket that has no influence
tance, the most widely used is the Kubelka–Munk on the wave-guiding properties of the optical fiber.
theory. Here, the scattering layer is assumed to be The basic structure of an optical fiber is shown in
infinitely thick, which may be effectively the case Figure 3. The common materials used in optical
with molecular recognition element utilized in op- fibers include plastic (poly(methyl methacrylate)),
tical sensors, and the reflectance (R) is related to the glass, and quartz.
absorption coefficient (K) and the scattering co- Incident light is transmitted through the fiber
efficient (S), as follows: when it impinges the core–cladding interface at an
angle greater than the critical angle, so that there is
ð1  RÞ2 K total internal reflection at the core–cladding inter-
FðRÞ ¼ ¼ ½2 face. A series of total internal reflection takes place
2R S
until the light reaches the other end of the fiber
where F(R) is known as the Kubelka–Munk function. (Figure 4). The optical fiber light transmission char-
The absorption coefficient K can be expressed in acteristics are described by its numerical aperture
 ABSORBANCE SPECTROSCOPY FOR CHEMICAL SENSORS 241

beam entering the fiber core. The distance dp is

typically of the order of a fraction of a wavelength of
light.
The incorporation of optical fibers in chemical
Figure 4 Transmission of a light beam inside an optical fiber sensors imparts a number of advantages to optical
through total internal reflection. sensors over the conventional devices in many areas
of application. Optical sensors are electrically pas-
Evanescent sive and immune to electromagnetic disturbances.
wave They are geometrically flexible, corrosion-resistant,
and capable of being miniaturized. They are com-
patible with telemetry and capable of operation in
remote and hostile environments. They can be of
low-cost, of rugged construction, and intrinsically
safe. The optical fibers used in these sensors are
capable of transmission of optical signals over great
distances with low attenuation of optical power.
Thus optical sensors are capable of measurements of
Figure 5 Evanescent wave due to a light beam being
samples in their dynamic environment, no matter
transmitted inside an optical fiber.
how distant, difficult to reach, or harsh that en-
vironment is. Intrinsic safety aspects are imparted to
(NA), which is directly proportional to the sine of the these sensing devices by the low optical power util-
half angle (a) of the acceptance cone of light entering ized in them. Furthermore, the chemical sensing
it which, in turn, is related to n1 and n2, as in eqn [4]: process itself is nonelectrical. With these sensors only
very small sample volumes are needed for analysis,
NA ¼ sin a ¼ ðn21  n22 Þ1=2 ½4 which has the advantages of nonperturbation of
samples in real-time monitoring applications.
where n0 is the refractive index of the surrounding However, optical sensing devices possess certain
medium, for example, air. limitations, such as interference from ambient light,
Basically there are three kinds of optical fibers in limited dynamic range, long response times, limited
use for sensing purposes. These are the multimode specificity, and nonreversibility. Many of these limi-
step index, the multimode graded index, and the tations can be eliminated or reduced by the use of
single-mode step index fibers. These fibers differ in appropriate instrumentation and sensing phases, and
the number of light beams that travel through the thus the sensor devices can be used advantageously in
length of the waveguide. The light rays accepted into specific applications.
the fiber interact, and only those which undergo
constructive propagation traverse through the fiber.
A small portion of light that is transmitted through
Optoelectronic Instrumentation
an optical fiber or a waveguide by total internal re-
flection extends outside core and is referred to as the Optical fiber sensors involve similar instrumentation
‘evanescent wave’. In an optical fiber, the evanescent as those involved in spectrophotometers. These sen-
wave penetrates the cladding material (Figure 5). The sors require both optical and electrical components,
intensity (I) of the evanescent wave decreases ex- including a light source; a wavelength selector; a
ponentially with increasing distance (d) from the photodetector; and a readout device. A block dia-
surface according to eqn [5]: gram of the basic instrumentation associated with
optical sensors is presented in Figure 6. The radiation
I ¼ I0 ed=dp ½5 supplied by the light source is first made mono-
chromatic by a wavelength selector before it enters
where I0 is the electric field intensity at the core– into optical fiber sensor system. Then, the radiation
cladding interface and dp, the depth of penetration, is emanating from the sensor is directed to the photo-
the distance from the interface of the point at which detector which subsequently generates an electrical
the electric field intensity has reduced to 1/e of its signal. This magnitude of the final signal is displayed
value at the interface. This characteristic depth of in the readout device.
penetration is dependent on wavelength of the Several types of sources, including incandescent
propagating light, refractive indices of the core and lamps (tungsten and quartz–halogen), lasers, light-
the cladding, and the acceptance angle of the light emitting diodes (LEDs), and laser diodes, have been
242 ABSORBANCE SPECTROSCOPY FOR CHEMICAL SENSORS

Light Signal Mono- Photo- Lock-in

source modulator chromator detector amplifier

Optical fiber

Readout
Measurand device
system

Figure 6 Schematic diagram of a typical instrumentation system associated with optical fiber chemical sensors.

used. Each of these types of source has its own ad- For absorbance- and reflectance-based chemical
vantages and disadvantages. The detection of light is sensors, a number of reagents employed in color-
carried out using a photocounting device, which imetry can serve as a molecular recognition element.
converts optical signals into electrical signals that Chromogenic reagents, such as pH colorimetric in-
can be amplified electronically. The photodetectors dicators and chelating reagents, have been used as
used in optical sensors include photomultiplier tubes, chemical transduction elements in optical sensors.
p–i–n photodiodes, avalanche photodiodes, and Novel materials, such as conducting polymers, mo-
photodiode arrays. lecularly imprinted polymers, and nanoparticles,
A wavelength selector, such as a filter or a mono- have also been exploited for molecular recognition in
chromator, isolates the desired wavelength for the optical fiber chemical sensors.
measurement. Optical couplers and lenses are used The molecular recognition element is often em-
to focus the beam of light to the optical fiber and to ployed in the solid state. The reagent(s) is usually
direct light to the detector. It should be emphasized immobilized onto an inert and stable solid material
that efficient coupling of light ensures the attainment by physical methods (adsorption, entrapment, and
of high sensitivity of sensor signals. It is also com- electrostatic attraction) or by chemical means (co-
mon practice to exclude extraneous light reaching valent bond formation). Among the solid supports
the detector by suitable modulation of the light that have been used to immobilize the molecular
source and by synchronizing the detector to this recognition element are glass, silica gel, or organic
modulation frequency so as to detect only those polymers. The physical methods are simple and
source signals. Instrumental drift may be eliminated economical to carry out, but they do not necessarily
or reduced by the use of a suitable referencing sys- produce stable reagent matrices. On the other hand,
tem. Drifts in sensor response due to aging of the the chemical means of immobilization produce a
sensing phase and variable light source may be en- strongly bound reagent, but this is often achieved
countered. But these contributions may not be sig- after several reaction steps in the synthesis or modi-
nificant due to short periods of measurement periods fication of the reagent and/or the support material in
used with the sensing devices. order to realize stable chemical bond(s) between
them.
The sensor response characteristics in chemical
Molecular Recognition Element transducer-based sensors depend on the manner in
The molecular recognition element constitutes the which the analyte–reagent interaction takes place. In
primary sensing unit of a chemical sensor. It inter- a simple system, where direct indicators are em-
faces the chemical sensor with the measurand sys- ployed, the analyte concentrations can be correlated
tem, generating an optical signal in response to the to the optical changes that occur in the reagent phase
presence of the analyte. It transforms the analyte into or in the product or both. The correlated sensor
another substance with distinct optical properties, so signal may also be dependent on the equilibrium
that it can be considered as a chemical transduction constant of the analyte–reagent reaction. For ex-
unit. The transformation can be detected and quan- ample, pH can be measured by monitoring the
tified by the instrumentation system via the optical changes in optical property of acid-base indicators.
fiber. However, if the analyte itself possesses a de- In many sensors, reversible reactions are preferred in
tectable optical characteristic, then that property can the chemical transduction process because they can
be directly measured using optical fibers and quan- be used in continuous monitoring applications. In
tified to the concentrations of the analyte. this case, the response time of these sensors (i.e., the
 ABSORBANCE SPECTROSCOPY FOR CHEMICAL SENSORS 243

time to reach the reaction equilibrium) is dependent transduction is employed (Figure 7(a)), the light
on mass transfer processes. Irreversible reactions interacts directly with the measurand system. This
may also be employed in sensors, which can result in type of sensors has been described as ‘spectroscopic
‘one-shot’ devices. Although such sensors would be sensors’, since they are based on the spectral
of limited merit, measurements with high sensitivity properties of the analyte. In sensors involving a
can be attained here. In certain cases, the reagent molecular recognition element, the reagent phase
phases can be regenerated by the use of another is often translucent and is set at one end of the
chemical reaction and the sensor reused. Indirect feed fiber and of the collector fiber (Figure7(b)). The
chemical reactions involving two or more reagents analyte reacts with the immobilized reagent, and
and/or reactions can also be adapted as optical the product absorbs the light that passes through the
chemical transducers. Many enzyme-based reactions reagent phase.
fall into this category. In reflectance-based sensors, chemical transduc-
tion is often employed and the molecular recognition
element is placed at one end of an optical fiber sys-
tem. A single optical fiber or optical fiber bundles can
Sensor Design be used in this type of optical sensors. In single-fiber
A variety of configurations have been utilized in sen- sensors (Figure 8(a)), the source light and detected
sors based on absorbance and other spectroscopic light travel through the same optical fiber, and are
measurements. The sensor designs can be classified discriminated either temporally or by wavelength
into two categories: extrinsic and intrinsic sensors. In with the aid of a beam splitter. Optical fiber bundles
extrinsic sensors, the optical fiber acts only as a light are often configured as a bifurcated system
guide between the light source and chemical sensor and (Figure 8(b)), wherein the incident radiation travels
between the sensor and the detector. In intrinsic sen- through one branch and the reflected light is directed
sors, the optical fiber becomes a part of the transducer. to the photodetector through the other branch.
In absorbance-based sensors, light is usually fed at Evanescent wave interactions have been exploited
one end of an optical fiber, guided to a sensing cell, in optical fiber chemical sensors. In these devices, the
collected through a second optical fiber, and detected cladding material of the optical fiber is removed and
at the other end of this fiber. If no chemical replaced with the analyte system itself (Figure 9(a))

(a) Flow cell

From light To photo-

source detector
Optical Optical
fiber fiber

Analyte

(b) Reagent
film

From light To photo-

source detector
Optical Optical
fiber fiber

Analyte
inlet

Figure 7 Typical configuration of absorbance-based chemical sensors: (a) employing no chemical transduction, and (b) with
chemical transduction.
244 ABSORBANCE SPECTROSCOPY FOR CHEMICAL SENSORS

(a) (b) From light

From light To photo-
source
source detector
Beam
To photo- splitter
detector

Optical Optical
fiber fiber

Molecular
Molecular
recognition
Membrane Membrane recognition
element
element

Figure 8 Typical configuration of reflectance-based chemical sensors: (a) employing a single fiber, and (b) employing a bifurcated
optical fiber bundle.

(a)
Cladding Optical
fiber

From light To photo-

source detector

Flow cell

Analyte
inlet

(b)

From light To photo-

source detector

Cladding Reagent Optical

phase fiber

Figure 9 Typical configuration of evanescent-wave chemical sensors: (a) not employing chemical transduction, and (b) with
chemical transduction.

or with a thin layer of the molecular recognition variations in the background caused by source, de-
element (Figure 9(b)). The presence of the analyte tector, and also intrinsic absorption of light by op-
affects the evanescent wave interaction and generates tical fibers. This can be done by subtracting the blank
changes in the optical signal. signal from the sensor signal electronically, or by the
In optical fiber absorptiometry or reflectometry, use of two identical optical fibers, one of which is
reference signals are employed for correcting used as a reference.
 ABSORBANCE SPECTROSCOPY FOR CHEMICAL SENSORS 245

Sensor Applications concentration. Organophosphate and carbamate

pesticides, and also many toxic metal ions, have been
A number of optical fiber pH sensors have been de- quantified using this principle.
veloped based on the absorption characteristics of a Absorbance measurement has also been used in
colorimetric acid-base indicator. The reagent is im- gas-phase sensing. A nitrogen dioxide sensor has
mobilized on a finely powdered solid support or on a been constructed using optical fibers and measure-
membrane that is deposited on the tip of the optical ment of absorbance of the gas at 0.5 mm using an
fiber and held in place by an enclosing membrane. argon-ion laser source. Real-time measurement in
One of the earliest reported optical fiber sensor for the lower ppm concentration range has been con-
pH was based on the measurement of the absorbance ducted at remote locations that are 20 km away. A
of phenol red covalently bound to polyacrylamide similar device has been used for the measurement of
mixed with polystyrene microspheres to scatter the methane at low concentrations by recording ab-
light within the reagent phase. The sensor was con- sorbance at 1.33 mm.
figured as a probe employing a bifurcated optical Many sensors for gases and vapors such as am-
fiber, with one arm for the probing radiation and the monia, carbon dioxide, humidity, hydrogen cyanide,
other arm for the scattered radiation from the re- hydrogen sulfide, etc., have developed based on the
agent phase. The sensor measures pH in the range use of analyte-specific immobilized reagents and
7.0–7.4 to 0.01 pH units. This example illustrates the measurements of absorbance or reflectance. For in-
feasibility of the approach though the sensors may stance, the optical sensing of ammonia can be carried
not be directly applicable to pH measurements in out through the change in color of polyaniline de-
seawater. Different indicators may be employed for posited on a polystyrene substrate. An optical sensor
such applications. for gaseous hydrogen sulfide was based on the de-
Absorbance-based optical sensing has been applied velopment of a grayish color on a cellulose mem-
for the measurement of metal ions. These ion sensors brane impregnated with lead acetate.
involve the use of immobilized metal-ion-selective
reagents interfaced to optical fibers. These sensors
rely on the fact that the metal ion (M) reacts with the
immobilized reagent (R) to form a metal complex
Conclusions
(MR), accompanied by either an enhancement or As described above, many types of optical sensor
change of color of the immobilized reagent. This designs have been studied for a variety of analytes
change can be correlated to metal ion concen- using different types of transduction reactions in the
trations. Thus, an optical sensor for micromolar development of absorbance- and reflectance-based
amounts of cobalt was developed employing a re- sensors. A current trend in this development is to
agent phase consisting of pyrogallol red held on a construct multianalyte sensing systems based on the
cellulose acetate film. Likewise, a fiber optic sensor use of single or a few reagent phases, together with
for copper ions occurring in ppm levels was devised the employment of appropriate signal-processing
based on a-benzoinoxime, a highly selective techniques such as pattern recognition and artificial
colorimetric reagent for copper ion, immobilized neural networks. Most of the optical sensors de-
on hydrophobic reagent Amberlite XAD-2 micro- scribed above can be designed for use in oceano-
spheres. graphic measurements in the analysis of heavy metal
Optical sensors based on absorbance measurement ions, dissolved gases, and other species.
have been developed for organic compounds, such as Applications of absorbance-based sensors to ocean
pesticides, occurring in environmental water systems. sciences have yet to demonstrate their potential.
In the case of pesticide sensors, more than one type Most of the applications published in the literature
of transduction system is involved – an enzyme re- describe ‘proof-of-concept’ studies with the sensors
agent and a pH indicator co-immobilized on suitable and with very little or no practical demonstration in
polymeric material. In the absence of pesticides, the such areas. Some of the recent studies in this area
reaction of the immobilized enzyme, such as acetyl- have been focused on the monitoring of dissolved
choline esterase, with its specific substrate will be CO2 in seawater using absorbance and fluorescence
accompanied by a change in pH that can then be measurements with transducers that incorporate
measured by the pH sensor using absorbance or re- pH-sensitive indicators. The use of infrared ab-
flectance. However, in the presence of the pesticides, sorption measured through the evanescent waves in
the substrate/immobilized enzyme reaction is in- optical fibers for subsea monitoring of organic
hibited and the degree of inhibition, transduced in compounds (at ppm levels) has also been demon-
the pH sensor, can be correlated to the pesticide strated. Parameters such as sensitivity, specificity,
246 ABSORBANCE SPECTROSCOPY FOR CHEMICAL SENSORS

lifetime, aging, etc., need to be investigated and ad- Narayanaswamy R (1991) Current developments in
dressed before the sensors can be used for seawater optical biochemical sensors. Biosensors and Bio-
measurements. These studies would present great electronics 6: 467--475.
challenges in order to realize practical sensors. Narayanaswamy R (1993) Chemical transducers based on
However, there is substantial interest in new sensors fibre optics for environmental monitoring. The Science
of the Total Environment 135: 103--113.
for oceanographic applications including monitoring
Narayanaswamy R (1993) Optical chemical sensors:
of nutrients and pollutants, and optical sensors have Transduction and signal processing. Analyst 118:
great potential here. The devices are clearly attractive 317--322.
in concept and require expertise from several scien- Narayanaswamy R and Sevilla FS, III (1988) Optical fibre
tific disciplines including analytical chemistry, poly- sensors for chemical species. Journal of Physics E:
mer chemistry, environmental chemistry, fiber optics, Scientific Instruments 21: 10--17.
and opto-electronics. Narayanaswamy R and Wolfbeis OS (2004) Springer Series
on Chemical Sensors and Biosensors, Vol. 1: Optical
Sensors – Industrial Environmental and Diagnostic
Applications. Berlin: Springer.
See also
Oehme I and Wolfbeis OS (1997) Optical sensors for
Fluorometry for Chemical Sensing. Inherent determination of heavy metal ions. Microchimica Acta
Optical Properties and Irradiance. Wet Chemical 126: 177--192.
Analysers. Orellana G and Moreno-Bondi MC (2005) Springer Series
on Chemical Sensors and Biosensors, Vol. 3: Frontiers
in Chemical Sensors – Novel Principles and Techniques.
Berlin: Springer.
Further Reading Rogers KR and Poziomek EJ (1996) Fiber optic sensors
for environmental monitoring. Chemosphere 33:
Andres R, Kuswandi B, and Narayanaswamy R (2001) 1151--1174.
Optical fiber biosensors based on immobilized enzymes – Seitz WR (1988) Chemical sensors based on immobilised
a review. The Analyst 126: 1469--1491. indicators and fiber optics. CRC Critical Reviews in
Eggins BR (2002) Chemical Sensors and Biosensors. Analytical Chemistry 19: 135--171.
London: Wiley. Sevilla F, III and Narayanaswamy R (2003) Optical
Hales B, Burgess L, and Emerson S (1997) An absorbance- chemical sensors and biosensors. In: Alegret S (ed.)
based fiber-optic sensor for CO2(aq) measurement in Integrated Analytical Systems, ch. 9. Amsterdam:
pure waters of sea floor sediments. Marine Chemistry Elsevier.
59: 51--62. Tokar JM and Dickey TDN (2000) Chemical sensor
Harmer AL and Narayanaswamy R (1988) Spectroscopic technology. Current and future applications. Ocean
and fibre-optic transducers. In: Edmunds TE (ed.) Science and Technology 1: 303--329.
Chemical Sensors, ch. 13. Glasgow: Blackie. Varney MS (2000) Chemical Sensors in Oceanography.
Lieberzeit PA and Dickert FL (2007) Sensor technology London: Taylor and Francis.
and its applications in environmental analysis. Wise DL and Wingard LB, Jr. (1991) Biosensors with
Analytical and Bioanalytical Chemistry 387: 237--247. Fiberoptics. Clifton, NJ: Humana.
Mizaikoff B (1999) Mid-infrared evanescent wave sensor – Wolfbeis OS (1991) Fiber Optic Chemical Sensors
a novel approach for sub-sea monitoring. Measurement and Biosensors, vols. I and II. Boca Raton, FL: CRC
Science and Technology 10: 1185--1194. Press.
 BIO-OPTICAL MODELS
A. Morel, Université Pierre et Marie Curie, content, and more generally the bio-optical state of
Villefranche-sur-mer, France ocean water. Once identified, and if recognized as
Copyright & 2001 Elsevier Ltd. statistically significant, such empirical relationships
(between optical properties and [Chl] can be inverted,
and thereafter used as predictive tools or model.
It is worth remarking that regular trends generally
vanish in so-called Case 2 waters (Table 1). Indeed,
in these waters the optical properties are no longer
Introduction influenced just by phytoplankton and related par-
The expression ‘bio-optical state of ocean waters’ ticles, as they are in Case 1 waters. They are also,
was coined, in 1978, to acknowledge the fact that in and independently, determined by other substances
many oceanic environments, the optical properties of of terrestrial origin, notably by sediments and col-
water bodies are essentially subordinated to the ored dissolved (organic) matter, carried from land
biological activity, and ultimately to phytoplankton into coastal zones and not correlated to [Chl].
and their derivatives. More recently the adjective Therefore, bio-geo-optical models, that might be
bio-optical has been associated with nouns like developed and locally useful in such areas, are not of
model or algorithms. At least two meanings can be general applicability.
distinguished under the term ‘bio-optical model.’ The two kinds of models are not disconnected. To
A bio-optical model can designate a tool used to the extent that the IOPs at the level of particles are
analyze, and then to predict, the optical properties of additive, the first models, in principle, may be util-
biological materials, such as phytoplanktonic or ized to reconstruct the IOPs of a water body con-
heterotrophic unicellular organisms, the most abun- taining any assemblage of organisms and other
dant living organisms in the ocean. Such models are (living or detritus) biogenic particles. Then these
based on various fundamental theories of optics bulk IOPs can be combined through the radiative
which apply to a single particle, and make use of a transfer equation (RTE) with the appropriate
set of rigorous equations. The optical properties boundary conditions (the illumination conditions at
which can be ‘modeled’ belong to the category of the the surface and the reflectance properties of the
inherent optical properties (IOP, see Radiative bottom, in particular), with a view to computing the
Transfer in the Ocean). Defined at the level of a AOPs at various depths within the water column. In
single cell, the extension of IOPs to a collection of this way, the result of the second category of models,
cells (a population) or to an assemblage of popu- the descriptive models, can be understood or
lations is straightforward from conceptual and nu- interpreted.
merical viewpoints. The computation of IOPs are Because empirical bio-optical models generally
carried out by using some physical characteristics of refer to the trophic level, depicted by [Chl], they are
the organisms, or of the population (such as cell size, in essence restricted to upper oceanic layers, where
size distribution, chemical composition which gov- the photosynthetic activity takes place, where the
erns the complex index of refraction). vegetal biomass is confined, and [Chl] is measurable.
Bio-optical models can also refer to various ways In addition, a considerable effort in developing bio-
of describing and forecasting the ‘bio-optical state’ of optical models originates from the need to interpret
the ocean, namely the optical properties of a water the satellite ocean color data in terms of chlorophyll
body as a function of the biological activity within concentration, which is only detectable in the upper
this water. Both the IOPs and the apparent optical oceanic layer. Possible relationships between optical
properties (AOPs) of the water are aimed at in such properties and heterotrophic activity (bacterial
approaches. In contrast to the first kind of theoretical abundance), or particulate organic carbon and
models, these models are essentially empirical, de- minerogenic contents in the interior of the ocean, are
scriptive, and actually derived from field measure- not examined here.
ments. They initially rest on observations of some Finally, it must be added that the spectral domain
regular variations in the oceanic optical properties encompassed by bio-optical models is that of visible
along with its algal content in ‘Case 1 waters’ (see (or photosynthetic) radiation, namely the 400–
Table 1). The chlorophyll concentration, [Chl], is 700 nm domain, occasionally slightly extended
commonly used as an index to quantify the algal toward the near infrared and near ultraviolet regions.

247
248 BIO-OPTICAL MODELS

Table 1 Concepts and quantities used in bio-optical models

Case 1/Case 2 water. Case 1 waters are those waters in which phytoplankton and their accompanying and covarying retinue of
material are the principal agents responsible for the variations in optical properties of the water bodies. The accompanying material
includes living heterotrophic organisms, such as bacteria or virus, various debris of biological origin, and dissolved organic matter
excreted by organisms or liberated by decaying detritus. Such waters are typical of the open ocean, far from land influence. Con-
versely, Case 2 waters are influenced not only by unicellular algae and related particles or substances, but also by other optically
significant components, from terrestrial origin, such as inorganic and organic particles in suspension, yellow substances resulting from
land drainage, and sediments resuspended from bottom

Quantity Units Symbol

Absorption coefficient m1 a

Scattering coefficient m1 b
Volume scattering function m1sr1 b(y)
Back-scattering coefficient m1 bb
Back-scattering efficiency (the ratio bb =b) – b̃b
Attenuation coefficient ðc ¼ a þ bÞ m1 c
Chlorophyll-specific (absorption or scattering) coefficients of phytoplankton m2 (mg Chl)1 af , bf
Efficiency factors for absorption and scattering (subscripts a, b, respectively), Qa , Qb
defined as the ratios of energy absorbed within the particle, or scattered out
from the particle, to the energy impinging onto its geometrical cross-section
Relative size of a spherical particle, defined as a ¼ pDnw ðl0 Þ1 D diameter, nw – a
refractive index of water, and l0 , wavelength in vacuo
Relative (complex) refractive index of the particle, defined as the ratio of the – m ¼ n  in 0
index of the substance forming the particle to the refractive index of water
(n, real part, n0 imaginary part)
Van de Hulst parameter, defined as r ¼ 2aðn  1Þ – r
Depth of the euphotic layer where PAR is reduced to 1% of its surface value m Zeu
Photosynthetic available radiation (within the 400–700 nm range) photons s1m2 PAR
Attenuation coeffcient for downward irradiance (also K) m1 Kd

Optical Models for Individual Particle of volume, having a given size x, and within a dx
or Population of Particles interval (around x), k is a scaling factor, and j is an
exponent, with typical values around 4 for oceanic
In the open ocean Case 1 waters, phytoplankton particles. Such a distribution means that an increase
with their accompanying retinue of living and de- by a factor of 10 in size corresponds to a reduction in
trital particles, are the principal agents responsible number (in frequency of occurrence) by a factor of
for the determination of the optical properties. The 10 000.
size of these particles extends from less than 0.1 mm As a consequence of this abruptly decreasing
(virus, colloids, debris), to less than 1 mm for het- number of particles with size (combined with optical
erotrophic bacteria and picoplanktonic algal species, theories, see below), the particles which are the most
and from 1 to tens or even several hundreds of mi- optically significant are in the size range of 1–10 mm,
crometers for phytoplankton, protists, and large and thus include most of the small heterotrophic
heterotrophic organisms (and actually up to tens of organisms, phytoplanktonic cells, and various small
meters for whales). It is well known that the size debris. This is true for the scattering properties, but
distribution function of marine particles is rather not for the back-scattering coefficient predominantly
monotonic, with numbers continuously increasing due to smaller particles (Figure 1). It is also true for
toward smaller sizes (Figure 1). A simple function, absorption, even if in this process heterotrophs play a
often sufficient to approximately describe the size minor role (they are rather colorless); in contrast,
distribution of oceanic particulate matter, is a power algal cells containing a variety of pigments (chlor-
law (known as Junge distribution), ophylls, carotenoids, and occasionally phycobilins)
are strongly absorbing bodies.
dnðxÞ=dx ¼ NðxÞ ¼ kxj ½1 The interaction between radiation and an optical
object like a particle is conveniently described by
where x is the size (e.g., the diameter, if the particles dimensionless numbers, called efficiency factors for
can be considered as spherical), NðxÞ is the distri- absorption and for scattering, and denoted Qa and
bution function, i.e., the number of particles per unit Qb , respectively (Table 1). One advantage of these
 BIO-OPTICAL MODELS 249


DC? 0 5 10 15 20
14
10 V
1.0

HB 3 Qb
μm )
_1

pP 1
10
10
_3
N (x) (m

nP 2
0.5

6 Qb 2
10
bb b
μP 1

MP 3
2
0.0 Qa
10
_2 _1 0.2 0 3.0 1 2
10 10 10 10 10 0
Size x (μm) 0 10 20 30
D (μm)
Figure 1 Schematic plot (shaded stripe, left-hand side ordinate
log-scale) of the approximate numerical concentration of particles Figure 2 Efficiency factor for scattering, Qb , as a function of
according to their size in oceanic waters (eqn [1] with j ¼ 4). the parameter r (defined in Table 1), or of the diameter, D, when
Approximate abundance versus mean size of major groups of the relative index of the particle is set equal to 1.05, and the
microorganisms or particles are also indicated, with notations as wavelength l is 675 nm. Curve represents Qb for a nonabsorbing
follows: DC, debris and colloids; V, viruses; HB, heterotrophic particle. Curve 2 represents Qb for an absorbing particle, with
bacteria; pP, picophytoplankton; nP nanoplankton; mP n 0 ¼ 0:0075, n ¼ 1:05, and l ¼ 675nm. Curve 3 represents the
microplankton; MP, large macroplankton. (Adapted from efficiency factor for absorption Qa for the same n, n 0 , and l values
Stramski and Kiefer (1991) Light scattering by microorganisms as for curve 2. Note that when the size increases, Qb oscillates
in the open ocean. Progress in Oceanography 28: 343–383.) around, and tends asymptotically, toward 2, if the particle is not
Linear plot (right-hand ordinate scale, from 0 to 1) of the absorbing, or toward 1 when it is absorbing; Qa tends also toward
progressive value of the scattering coefficient, b, when the 1.
upper limit of the integral (eqn [3]) is increasing; the progressive
value is relative and normalized by its final value (unity). A similar which contains these particles, are simply expressed
curve is drawn for the backscattering coefficient, bb . For these
as
computations, the relative refractive index of the spherical
particles is 1.05, and the wavelength is 550 nm. (Adapted from a or b ¼ NpðD2 =4Þ: Qa or b ½2
Morel A and Ahn Y-H (1991) Optics of heterotrophic
nanoflagellates and ciliates: atentative assessment of their where N is the number of particles per unit volume,
scattering role in oceanic waters compared to those of bacterial pðD2 =4Þ represents the geometrical cross-section of a
and algal cells. Journal of Marine Research 49: 177–202.)
single (spherical) particle. The Q factors are simul-
taneously functions of D and m, through the par-
factors lies in the fact that theories are available by ameter r (see Table 1).
which their values can be predicted as a function of For perfectly transparent particles ðn0 ¼ 0Þ, a and
the relative size (a, defined in Table 1), and the Qa are obviously 0. In this case, Qb after oscillations
relative complex index of refraction of the particle tends asymptotically toward 2 for increasing size
(m, see Table 1). The Mie-Lorenz theory provide for (Figure 2); such a particle is able to remove from the
spherical particles accurate Q values, and the angu- radiative field by scattering twice the amount of ra-
lar values of the volume scattering function (see diation intercepted by its geometrical cross-section
Radiative Transfer in the Ocean). When the refract- (this is often called the ‘extinction paradox’). For
ive index of the particle is close to that of the sur- absorbing particles, Qa increases with increasing
rounding medium (as is the case for most of the size, and Qb , after some oscillations tends toward 1,
watery oceanic particles in suspension in water), the like Qa (Figure 2).
so-called van de Hulst approximation can apply and If the particles are not uniform in size, and the
provide Q-factors more rapidly than via Mie population is characterized by a size distribution
computations. function, NðDÞ the eqn [2] must be integrated over
If the particles are made of the same substance the appropriate size interval, according to
(same refractive index, m), and are assumed to be Z
spherical, with the same diameter, D, the absorption a; or b ¼ ðp=4Þ N ðDÞD2 Qa or b ðD; mÞdðDÞ ½3
and scattering coefficients, a and b, of the medium
250 BIO-OPTICAL MODELS

The same parameters (size, complex refractive index) if the pigments were homogeneously distributed.
are used in the frame of the rigorous Mie theory, to This ‘packaging’ effect, and its corollary, the
compute the volume-scattering function (VSF, see ‘flattening’ of the absorption spectrum, both ori-
Table 1) of any individual particle. The VSF for the ginate from the behavior of the Qa factors with
entire particle population is simply obtained by varying size and n0 . These effects are well under-
adding the individual VSF with the appropriate stood, described by simple equations and accur-
weight, as derived from the distribution function ately modeled.
NðDÞ. All the bio-optical models of the scattering
The above models, which address the optics of
properties of marine particles rest on such an ap-
individual cells, or ultimately deal with populations,
proach. Their limitations do not originate from the-
allow the properties observed in oceanic waters
ory, but from the present lack of accurate
containing assemblages of these particles to be in-
information about the sizes and composition of the
terpreted. In this sense, they are able to support the
suspended material (in Case 1, and even more in
second category of bio-optical models, which are
Case 2 waters). If the actual pattern of the particle
examined below.
VSF is globally understood and can be reconstructed,
the predictive skill of the models (actually the
available information used as inputs) are still in- Modeling the Optical Properties of
sufficient to allow the evolution of the back-scat- Ocean Waters in Relation to Their
tering coefficient to be safely parameterized as a
Biological State
function of the bio-optical state (see reflectance
modeling). What can be predicted, however, is the In oceanic Case 1 waters, far from significant terri-
extremely low back-scattering efficiency exhibited by genous influences, the origin of all materials present
most of the unicellular organisms with low refractive is necessarily to be found in the first link of the food
index (such as algal cells, for instance). Also that this chain, namely in photosynthesizing phytoplanktonic
efficiency increases with the decreasing particle size is organisms. Heterotrophic organisms, as well as in-
a theoretical evidence; as a consequence, the particles animate detritus or dissolved organic matter are re-
responsible for the formation of the scattering and lated to algal biomass, and to the initial creation of
back-scattering coefficients do not belong to the organic matter (and particles) through photo-
same size range (Figure 1). synthesis. Therefore, the water optical properties are
In summary, theoretical models are available logically studied as a function of this vegetal
which account very well for most of the observed biomass.
optical properties. They have been validated in par- Because chlorophyll a is the single ultimate
ticular through in vitro experiments and by using photosynthetic pigment, the most abundant in all
various cells grown in culture (heterotrophic or living plants, and because it is easily determined, its
photo-autotrophic organisms). Several phenomena, concentration in the water is a convenient, albeit
predicted through theories and subsequent bio-op- imperfect, index of the bio-optical state. It is worth
tical models, are worth mentioning. recalling that the chlorophyll concentration in
oceanic waters, used as descriptor of the bio-optical
• Scattering by small (0.4–2 mm) organisms depends state, and to which the optical properties are to be
on the wavelength according to a l2 law. This related, varies within about 3 orders of magnitude
spectral dependency is perfectly verified for het- (say 0.02 and 20 mg m3, between oligotrophic
erotrophic bacteria (almost nonabsorbing bodies); zones and eutrophic conditions in upwelling areas).
for small phytoplanktonic cells (such as Pro- In practice, [Chl] and the optical properties within
chlorococcus and Synechococcus), the presence of the upper layers must be measured simultaneously at
pigments results in localized features (minima) sea to examine if statistically significant correlations
superimposed onto the general l2 spectral pattern. can be found between some IOPs or AOPs and [Chl].
• For larger organisms, the scattering spectrum may When such correlations are expressed mathematic-
exhibit various shapes, including a ‘flat’ (l0) shape ally (in general through nonlinear laws, and with a
when the size exceeds approximately 10 mm. For certain confidence interval), the corresponding ex-
algal cells, the various absorbing pigments always pression can be used as a model or an algorithm. In
influence the scattering spectrum, by introducing contrast to what occurred for the previous category
minima and maxima in scattering throughout the of bio-optical models (which are based on exact
absorption bands of these pigments. physical laws), the models examined below are in
• Absorption by pigmented cells in suspension dif- essence ‘empirical’. Varying uncertainties are there-
fers from that of a ‘solution’ of the same material, fore attached to each model; depending on inclusion
 BIO-OPTICAL MODELS 251

of new data, the numerical formulation or the input 0.08

parameters of these models are still liable to further
evolution.
0.06

(m (mg chla) )
_1
Inherent Optical Properties and [Chl] ap

The absorption coefficient of oceanic water Beside 0.04

the fixed contribution of water itself ðaw Þ to this

2
aφ
coefficient, the varying biological contribution
(sometimes denoted abio ) can itself be partitioned 0.02
anp
into a component due to particulate material, ap ,
and another due to colored dissolved organic
matter, acdom (all these coefficients are spectral 0.00
400 450 500 550 600 650 700
quantities, even if the symbol l is omitted, when not
(A) λ (nm)
necessary)

a ¼ aw þ acdom þ ap ½4 ap, aφ

In turn, ap can be divided according to 0.2

)
440
ap ¼ af þ anap ½5 a p(
(m ) 40)
a φ (4
_1

where af and anap are the partial absorption

coefficients by phytoplanktonic cells, and ‘nonalgal’ 0.1
particles, respectively. This nonalgal compartment
0)
includes colored debris and all kinds of a p (56
heterotrophic organisms. Techniques are available aφ (560)
to discriminate between ap and anap and to 0.0
determine their spectra; such measurements are 0 5 10
performed with particles retained on a filter, before (B) Chl (mg m )
_3

and after methanol extraction. The algal absorption

af spectrum is indirectly obtained by difference. Figure 3 (A) Mean spectral absorption values, as they result
Typical shapes of these spectra are shown in from statistical analyses, when the chlorophyll concentration
Figure 3(A) when [Ch1] is 1 mg m3. within the water body is 1 mg m3. The curves represents ap ðlÞ,
plotted inside a shaded area which represents 7 1SD, the
Statistical analyses of absorption measurements nonalgal particle absorption spectrum, anap ðlÞ, and
performed systematically in Case 1 waters with in- phytoplankton absorption spectrum, aj ðlÞ (see eqn [4]). (B)
creasing [Chl] have demonstrated that the partial Nonlinear evolution of the mean absorption coefficients ap ðlÞ and
coefficients (af and anap ) increase with [Chl] in a aj ðlÞ, with increasing chlorophyll concentration (see also eqns
nonlinear manner (Table 2, examples in Figure 3B). [2.1] and [2.2] in Table 2); the two selected wavelengths
correspond to the maximum (440 nm), and the minimum
As a consequence, the chlorophyll-specific absorption (560 nm) of algal absorption. (Adapted from Bricaud A, Morel A,
coefficient of phytoplankton (af , Table 1) is not a Babin M et al. (1998) Variation of light absorption by suspended
constant and decreases when [Chl] increases. Such a particles with chlorophyll a concentration in oceanic (case 1)
trend is, at least partly, due to the above-mentioned waters: analysis and implications for bio-optical models. Journal
packaging effect. In addition, a regular change in the of Geophysical Research 103: 31033–31044.)
pigment composition is also at the origin of this de-
crease; recall that algal absorption originates from all values of the exponents (in eqns [2.1] and [2.2],
accessory pigments, while normalization is made Table 2). In comparison, nonalgal particles are less
with respect to the sole chlorophyll a. absorbing, and their absorption, regularly increasing
Most of the recent studies in oceanic waters have toward the short wavelengths, is well modeled with
shown that algal cells are the dominant term in an exponential function (Table 2, eqn [2.3]).
forming ap. On average, af would represent about The measurements of acdom (performed on filtered
70% of ap within the absorption blue maximum of water) are rather scarce in oceanic waters; they have
algae (around 440 nm), and even more in the red peak not yet provided a clear relationship (if any) between
(around 675 nm). The spectral af and ap patterns in this term and [Chl], but have shown that the spectral
Figure 3(A) are slightly changing with [Chl], as a shape, namely a monotonous (exponential) increase
consequence of the differences between the spectral of acdom toward the short wavelengths, is rather
252 BIO-OPTICAL MODELS

Table 2 Bio-optical models for Case 1 waters, and corresponding equations by which the main inherent optical properties (IOP) of a
water body can be related to its chlorophyll concentrationa

ap ðlÞ ¼ Ap ðlÞ½ChlEpðlÞ ½2:1

aj ðlÞ ¼ Aj ðlÞ½ChlEjðlÞ ½2:2

Note that the terms Ap and Aj are displayed in Figure 3(A), as ap ðlÞ and aj ðlÞ are shown when [Chl] is set equal to 1 mgm3; the
exponents EpðlÞ, and EjðlÞ are similar in magnitude but not equal, and they vary within the range 0.6–0.9, approximately.
 
anap ðlÞ ¼ anap ðl0 Þexp Snap ðl  l0 Þ ½2:3

acdom ðlÞ ¼ acdom ðl0 Þexp½Scdom ðl  l0  ½2:4

Note that in these two last expressions, the reference wavelength ðl0 Þ is arbitrary (often 440 nm is adopted), and the slopes (S) of the
exponential decrease are approximately Snap ¼ 0.012 nm1 and Scdom ¼ 0.015 nm1.

b ðChl;l0 Þ ¼ bw ðl0 Þ þ bp ðChl;l0 Þ and bp ðChl;l0 Þ ¼ Bp ðl0 Þ½Chlx ½2:5

with the exponent x in the range 0.6–0.7, approximately; with Bðl0 Þ, at l0 ¼ 550 nm, statistically found to vary between 0.15 and
0.45 m2 (mgChl)1 (see also, Figure 4).
bp ðChl;lÞ ¼ bp ðChl;l0 Þðl=l0 Þy ½2:6

with the exponent y between  0.5 and  2 (the value  1 is commonly adopted).
0
bp ðC; l0 Þ ¼ Bp0 ðl0 Þ½Cx ½2:7

with the exponent x0 close also to 1 (see also Figure 4).

a
Subscripts: w, water; p, particles; f, phytoplankton; nap, nonalgal particle; cdom, colored dissolved organic matter.

stable (eqn [2.4]). It is worth noting that this spectral Bp ðl0 Þ (which represents the value of bp when
dependency is close to that typical of anap, apart from Chl ¼ 1 mg m3) has been found to vary within a
a small difference in slope. All models presently factor 3 for Case 1 waters (see also Figure 4; and eqn
proposed in the literature have the same exponential [2.5]).
structure. The nonlinearity in eqn [6] (a power law with
 o1) is such that when the algal biomass increases,
The scattering coefficient In open ocean waters, the bp increases more slowly. This effect is generally at-
scattering coefficient, b ¼ bw þ bp , is the sum of the tributed to a change (i.e., a decrease) in the relative
constant molecular scattering (bw), and of a varying contribution of detritus and of heterotrophic organ-
contribution, bp, resulting from the presence of all isms to scattering when [Chl] increases (in eutrophic
kinds of particles, living organisms, and detritus. waters). This explanation is corroborated by the
Unlike ap, bp cannot be split into algal and nonalgal following observation: when bp is studied as a
contributions, as there is no experimental technique function of the particulate organic carbon concen-
allowing such a discrimination. tration, [C], instead of [Chl], it varies linearly with
Bio-optical models aim at relating this bulk co- [C] (see Table 2, eqn [2.7]).
efficient bp to [Chl]. In situ measurements of bp ðlÞ, at Spectral measurements of bp in the open ocean are
fixed wavelength, l0 (or indirect determination via rather scarce. If the size distribution function obeys a
the particle attenuation coefficient cp ðlÞ, from which Junge distribution (eqn [1]), the spectral dependency
ap ðlÞ is subtracted), have led, through least-squares of the scattering coefficient can be theoretically pre-
analyses, to a nonlinear dependence upon [Chl] of dicted as being a power law, with an exponent (y, in
the form Table 2, eqn [2.6]) which is related to j (eqn [1]),
simply by: y ¼ 3  j. With typical values for j around
bp ðl0 Þ ¼ Bp ðl0 Þ½Chlx ½6 4, y would be around  1, as roughly observed, and
 BIO-OPTICAL MODELS 253

_
[C] (mg m 3 ) The first term can be (again approximately) ex-
0 1 2 3 pressed as a function of the IOPs of optically pure
10 10 10 10
1 seawater (aw and bw), whereas Kbio ðlÞ, resulting
10
from the presence of all kinds of biological materials,
can be related to [Chl].
On the basis of many field measurements in
0
10 oceanic waters, it has been shown that the spectral
Kbio ðlÞ values do not vary at random but are inter-
bp (550) [m ]

related. The proposed optical classification is based

_1

_1 on the realization that such a rather regular change

10 affects simultaneously all the wavelengths and pro-
gressively modifies the entire spectrum. Bio-optical
bp = f (Chl) algorithms derived from statistical analyses of these
10
_2
field data allow the entire KðlÞ spectrum to be spe-
cified, as soon as Kðl0 Þ, at a reference wavelength l0
bp = f (C)
is known (Table 3, eqn [3.1]).
_3
A second way of analyzing the field data consists
10
_3 _2 _1
of relating the Kbio ðlÞ values to [Chl]. In Case 1
0 1
10 10 10 10 10 waters, the diffuse attenuation coefficients appear to
_3
[Chl] (mg m ) be highly correlated to [Chl], and the statistical re-
lationships (linear regression on log-transformed
Figure 4 Particle-scattering coefficient as a function of the quantities) are expressed as
chlorophyll concentration in Case 1 waters, and at l ¼ 550 nm
(eqn [2.5] in Table 2); the natural variability in this relationship for
Case 1 waters is represented by the shaded band. The dashed
line represents the empirical relationship statistically obtained Kbio ðlÞ ¼ wðlÞ½ChleðlÞ ½8
between bp and the organic carbon concentration (upper
abscissa scale).
with exponents eðlÞ are always o1, whatever the
wavelength. The corresponding KðlÞ bio-optical
generally adopted in bio-optical modeling. Such a model consists of a set of such nonlinear expressions
monotonic decrease of bp throughout the spectrum is based on eqns [7] and [8] (eqn [3.2], Table 3; Fig-
probably oversimplifying, particularly at high algal ure 5). More complex models have also been used
concentration; indeed, phytoplankton scattering (eqn [3.4]).
spectra are, as mentioned before, featured in reponse To the extent that K is largely determined by ab-
to the pigment absorption. sorption, the nonlinear character of the correlation
between K and [Chl] is not surprising and resembles
Apparent Optical Properties and [Chl] that observed for a (Figure 3B). By integrating over
Downwelling irradiance The downwelling and the whole visible domain (the photosynthetic avail-
upwelling irradiances (radiant flux per unit of area, able radiation (PAR) domain, see Radiative Transfer
see Radiative Transfer in the Ocean) are convenient in the Ocean), a relationship between Zeu, the depth
measurements to make at sea to characterize the of the euphotic zone (cf. Table 1), and [Chl] can be
penetration of daylight into the water column. The obtained. This nonlinear relationship can also be
depth variations of these quantities are quantified derived through a direct analysis of the column-in-
by diffuse attenuation coefficients (Radiative tegrated chlorophyll content and of Zeu, observed at
Transfer in the Ocean). When dealing with sea by using a photometer able to determine the
downward irradiance Ed, the corresponding vertical PAR profile. This bio-optical algorithm is
coefficient is Kd (simply written K). It depends on useful to predict, in Case 1 waters, the depth of the
the IOPs of the water and on the geometrical euphotic zone when the vertical chlorophyll profile
structure of the light field. Only by approximation, has been determined (eqn [3.3], Table 3).
it can be seen as the sum of a term due to the water
itself and a varying contribution of all materials Irradiance reflectance This apparent optical
(particulate and dissolved) originating from property, RðlÞ, is crucial in the interpretation of the
biological activity, so that remotely sensed ocean color. It is defined as the ratio
of upwelling irradiance, Eu, to downwelling
KðlÞDKw ðlÞ þ Kbio ðlÞ ½7 irradiance at the same depth (actually just beneath
254 BIO-OPTICAL MODELS

Table 3 Bio-optical models, and corresponding equations by which the main apparent optical properties (AOP) in Case 1 waters can
be related to the chlorophyll concentration [Chl]

K ðlÞ ¼ Kw ðlÞ þ M ðlÞ½K ðl0 Þ  Kw ðl0  ½3:1

where Kw ðlÞ is the attenuation coefficient for downwelling irradiance in pure water, MðlÞ are statistically derived coefficients and l0 a
reference wavelength.
K ðlÞ ¼ Kw ðlÞ þ wðlÞ½ChleðlÞ ½3:2

wðlÞ and eðlÞ are empirical wavelength-dependent factors and exponents, derived from statistical analysis.

Zeu ¼ z ½Chlz ½3:3

z is about 38 m, and the exponent z is close to 1/2; this expression is computed by using eqn [3.2], combined with a standard spectrum
of solar radiation at sea level.
n  o
K ðl½ChlÞ ¼ k ðlÞexp k 0 ðlÞlog10 ½Chl=½Chl0 2 þ 0:001½Chl2 ½3:4

[Chl0] is a reference concentration (0.5 mg m3), and k ðlÞ and k 0 ðlÞ are empirical spectral parameters dervied from statistical analysis.

log10 ð½ChlÞ ¼ a0 þ a1 r þ a2 r 2 þ a3 r 3 ½3:5

where r ¼ log10 ½Rðl1 Þ=Rðl2 Þ is the decimal logarithm of a ratio of reflectances at two wavelengths; the cubic polynomial is an
example (first order polynomials have also been proposed). The inverse relationships, which express various r ratios as a function of
log10 ([Chl]), also have the same polynominal structure.

the surface in remote sensing applications) expressed through a polynomial with respect to the
log-transformed quantities (Table 3; linear approxi-
RðlÞ ¼ Eu ðlÞ=Ed ðlÞ ½9 mations have also been employed).
Thanks to radiative transfer studies, it has been
Other expressions involving upward radiance (in- shown that (to a first approximation) RðlÞ can be
stead of Eu) are also in use in ocean color remote expressed as a function of two inherent properties, a
sensing; they are geometrically related to RðlÞ as and bb, through
defined, and physically modeled in a similar way.
Historically, purely empirical models were the first
to have been developed. As for Kd, quasi-simul- RðlÞ ¼ f :bb =ða þ bb Þ ½11
taneous field measurements of RðlÞ and of [Chl]
were carried out, and then statistically analyzed. The where f is not a constant factor (because R is not an
relationships obtained through such regression ana- inherent property), but depends in a complex manner
lyses were used as algorithms when processing ocean on the structure of the light field, and on the VSF of
color data. Most of these algorithms consider the the particles. Its value (between about 0.3 and 0.5 for
ratios of reflectance at two wavelengths, l1 and l2 . common cases in oceanic waters) can only be com-
Statistical analyses have demonstrated that these puted by solving the RTE with the appropriate
ratios vary regularly with [Chl]. Therefore, they can boundary conditions. As a consequence of eqn [11], a
be operated as predictor of [Chl], through a certain bio-optical model for reflectance reduces to a com-
function F: bination of models for a and bb, separately con-
sidered. Because a and bb are inherent properties, the
½Chl ¼ F½Rðl1 Þ=Rðl2 Þ ½10 additivity principle applies (eqns [4] and [5], for a).
Therefore a purely analytical model would be built if
When such ratios of reflectances (called band ratio in a and bb (actually each term forming a and bb) could
ocean color science) and [Chl] are plotted in log-log be completely parameterized in terms of [Chl].
space, the scatterplot is sigmoid. In the so-called Such a model is not yet available. Modeling the
empirical ‘chlorophyll-algorithms’ in use in ocean back-scattering coefficient is presently an unresolved
color remote sensing, the functional dependency is problem. This coefficient is generally expressed as
 BIO-OPTICAL MODELS 255

K (675) 10

K (600)
K = K w + K bio (m 1)

5 λ1 = 443 nm
_

λ2 = 555 nm

R (λ1) / R (λ2)
)
0.1 K (560
2 λ1 = 490 nm
0)
(44 λ2 = 555 nm
K
1

0.5
0.01
0.1 1 10
_3
[Chl] (mg m ) 0.01 0.1 1 10
_3
[Chl] (mg m )
Figure 5 Diffuse attenuation coefficient for downward
irradiance, at selected wavelengths, as a function of the Figure 7 Example of outputs of models for Case 1 waters
chlorophyll concentration in Case 1 waters. The initial values providing the evolution with [Chl] of ratios of reflectances at two
(the ordinates when [Chl] ¼ 0.02 mg m3) are almost the pure sea wavelengths (as indicated) (see eqn [3.2], Table 3).
water values (Kw in eqn [7], see also eqn [3.1], in Table 3, for the
term Kbio).
value) have been proposed to express its dependency
on [Chl] and on l.
When dealing with absorption, the role of acdom in
0.03
open ocean is largely unknown, or more precisely its
0.1 magnitude has not yet been successfully related to
0.1
[Chl]. For this reason, in particular, a can advan-
0.3 tageously be replaced by its proxy, Kd, which cu-
mulates the influence of all absorbing substances,
1 and is easily modeled (see above). Some manipula-
3
tions of the RTE, however, are necessary to derive a
0.01 from Kd. Such models are called ‘semianalytical’ (see
Figure 6). They are also used to predict the ratios of
R (λ)

10
reflectances as a function of [Chl], and reciprocally
_
[Chl] (mg m 3)
(as through eqn [10]); they result in algorithms
having the same structure as those derived from
the empirical approach (eqn [3.4], Table 3, and
Figure 7). Other bio-optical models, involving re-
0.001
flectances at more than two wavelengths, have also
been proposed.
Reasonable agreements have been reached be-
tween such semi-analytical approaches, empirical
approaches, and independent data sets (not used
300 400 500 600 700 when developing the models). The sigmoidal shape is
λ (nm) well understood if not analytically reproduced with
accuracy.
Figure 6 Example of a spectral model for reflectance as a
function of [Chl] and for Case 1 waters; the various bio-optical
models presently in use are not identical but very similar.
Conclusions
In terms of accuracy or predictive capabilities, the
being the product b:b̃b , of the scattering coefficient first category of bio-optical models, dealing with the
(which can be related to [Chl] through eqn [6]), and properties of individual organisms, are robust par-
the dimensionless back-scattering efficiency b̃b. The ticularly because they rest on firm theoretical bases.
latter quantity is not presently well documented, and The second category, dealing with bio-optical prop-
rather divergent hypotheses (including a constant erties of oceanic water bodies, in essence empirical,
256 BIO-OPTICAL MODELS

cannot be as accurate, especially because they also local biological activity with its specific assemblages,
reflect the ‘natural’ variability around mean laws. will be more efficient in the future. For Case 2
This variability is expected to the extent that waters, bio-optical models based on [Chl], are in-
various substances, dissolved and particulate, living sufficient and must be supplemented by accounting
or inanimate, which are optically significant, are not for the presence of mineral and organic sediments,
strictly covarying with [Chl]. Therefore, the links and dissolved colored substance, not covarying with
between bulk optical properties and [Chl] cannot be the algal biomass.
tight. Actually, most of these properties may vary
within a factor 2 or 3 (or more) around the mean
value provided by the (nonlinear) models. This re- See also
mark holds true for coefficients like ap, aj , bp, and K. Optical Particle Characterization. Radiative
The situation is somewhat better when ratios of Transfer in the Ocean.
optical coefficients are involved, and when covaria-
tions of these coefficients reduce the amplitude of
fluctuations around the mean (as for reflectance and Further Reading
band ratios, for instance). In summary, the predictive
skill of such models is inevitably limited by the Demers S (1991) Particle Analysis in Oceanography.
tightness of the regressions on which they are based. NATO-ASI Series, vol. G 27. Berlin Heidelberg:
Springer-Verlag.
It is necessary to bear in mind that further studies at
Gordon HR, Brown OB, Evans RH, et al. (1988) A semi-
sea will likely change the result of the regression
analytical radiance model for ocean color. Journal of
analyses, and thus the numerical values of the tabu- Geophysical Research 93: 10909--10924.
lated parameters entering into the models. For this Gordon HR and Morel A (1983) Remote sensing of ocean
reason, only equations are given in Tables 2 and 3, color for interpretation of satellite visible imagery: a
the provisional numerical information to be found in review. Lecture Notes on Coastal and Estuarine Studies.
the literature, is likely amenable to modification in the New York, Berlin, Heidelberg, Tokyo: Springer-Verlag.
future). Jerlov NG and Nielsen ES (1974) Optical Aspects of
Most of the bio-optical models, by which optical Oceanography. London, New York: Academic Press.
properties and [Chl], are related, rely on nonlinear Morel A (1988) Optical modeling of the upper ocean in
relationships, and often on power laws of [Chl], with relation to its biogenous matter content (Case 1 waters).
Journal of Geophysical Research 93: 10749--10768.
exponents below 1. As a result, the optical properties
Morel A and Smith RC (1982) Terminology and units in
vary within a narrower interval than does [Chl] in
optical oceanography. Marine Geodesy 5: 335--349.
Case 1 waters, and do not span more than 2 orders of Smith RC and Baker KS (1978) The bio-optical state of
magnitude (instead of 3), which is still considerable. ocean waters and remote sensing. Limnology and
It can be envisaged that for Case 1 waters ‘regional’ Oceanography 23: 247--259.
bio-optical models, encompassing a less wide Van de Hulst HC (1957) Light Scattering by Small
chlorophyll concentration range, and accounting for Particles. New York: Wiley.
 FLUOROMETRY FOR BIOLOGICAL SENSING
D. J. Suggett and C. M. Moore, University of Essex, What Is Fluorescence?
Colchester, UK
Light absorption by chromophoric molecules (pig-
& 2009 Elsevier Ltd. All rights reserved.
ments) raises electrons within those molecules to an
excited state or higher energy level. As the molecules
de-excite, most of the energy is released as heat;
however, a proportion of the energy is also released
as light. Fluorescence is the emission of light at a
Introduction longer wavelength following light absorption at a
shorter wavelength. The fluorescence yield that arises
Oceanography has been transformed through the from an excitation light source is typically referred to
use of fluorescence to assay biology. Light-absorbing as F. Many factors including the optical geometry of
pigments cause many organisms to naturally fluo- the measurement system comprising the excitation
resce. Primarily, but not exclusively, these pigments light source, sample compartment, and detector will
are associated with photosynthesizing organisms, affect F. Typically, F is measured in relative units,
such as algae, aquatic vascular plants, and aerobic although, for a given optical geometry (C), F will also
anoxygenic photosynthetic (AAP) bacteria. Fluo- vary according to the absorptivity (A) of the sample
rescence is most commonly detected as the emission and the photon flux density (PFD) of the excitation
that follows ‘active’ excitation using an actinic light light source:
source. One of the major breakthroughs for ocean-
ography occurred in the 1960s with the detection of F ¼ fF  C  A  PFD
actively induced chlorophyll a fluorescence in situ.
Chlorophyll a is contained by all algae and cyano- The value of C depends on the proportion of fluo-
bacteria and thus provides a measure of abundance. resced light that is intercepted by the detector and the
However, at typical environmental temperatures the efficiency of conversion of photons into an electric
chlorophyll a fluorescence emission signature largely signal. Importantly, fF is the quantum yield of
originates from oxygen evolving photosystem II fluorescence, or the molar ratio of light emitted to
(PSII). Consequently, the chlorophyll a fluorescence light absorbed, a value that is unique to the inherent
signal contains information that can be used to physical properties of the fluorescent molecule.
characterize the photosynthetic activity of this
complex. Recent technological advances have pro-
Estimating Abundance
vided twenty-first-century oceanography with an
array of active chlorophyll a induction fluorometers Fluorescence is most commonly used to estimate
that are used routinely to assess photosynthetic the abundance of chlorophyll a. In the case of
physiology. In addition, enhanced capacities within chlorophyll a, relaxation from excited state 1 to
remote sensing platforms has enabled researchers to ground state results in dissipation of a small pro-
make major steps in using ‘passive’ fluorescence portion (3–5%) of the excitation energy at wave-
from chlorophyll a, the fluorescence that is stimu- lengths greater than c. 650 nm, that is, as red
lated as a result of natural excitation by the sun fluorescence. Most fluorometers deliver narrow-band
(solar-stimulated), to assess global photosynthetic blue light to correspond with wavelengths at which
activity. algae and plants exhibit maximum rates of light
In addition to the natural (auto-)fluorescent mol- absorption (Figure 1). Subsequent detection of
ecules, the process of introducing compounds into fluorescence is centered toward 680–700 nm to
cells for binding to, and thus labeling, specific mol- coincide with the peak emission by chlorophyll a.
ecules has further expanded the tools available to Almost all (499%) of fluorescence at these wave-
oceanographic research. Such compounds are either lengths arises from PSII for the majority of algae and
themselves fluorescent or become fluorescent fol- aquatic plants in nature.
lowing a specific biological reaction and have opened When algal and plant pigments are extracted into
up many new avenues for exploring physiology and solvents, the fluorescence that arises (F) is expected
taxonomy. Here we provide a brief synthesis of to exhibit a one-to-one relationship with the
fluorescence techniques used to examine the biology concentration of chlorophyll a (and their break-
of marine systems. down products). However, such a relationship is not

257
258 FLUOROMETRY FOR BIOLOGICAL SENSING

1
These organisms contain green fluorescence-like
Chl a
PSC proteins (GFPs), a family of chromophoric molecules
Abs ( )

PC that contribute to the colorfulness of corals and

fluoresce at wavelengths between 450 and 600 nm.
Although the function of GFPs is widely debated,
0 researchers have recently employed GFP fluorescence
1 in situ to determine the abundance of polyps re-
Fex ()′ em = 730

cruited onto coral reefs.

0 Phylogenetic Discrimination
1
Fex = 478′ em ()

In addition to chlorophyll a, algae and aquatic plants

contain many other chromophoric molecules termed
Synechococcus sp.
accessory pigments that act to supplement light
capture. Light absorption by chlorophyll a is re-
0 Chaetoceros muelleri stricted to narrow wavebands centered on c. 440 and
400 500 600 700
 (nm)
670 nm; however, light spectra within aquatic sys-
tems are relatively broad (c. 300–750 nm). Accessory
Figure 1 Absorption and fluorescence spectra for a diatom pigments absorb at wavelengths not targeted by
(Chaetoceros muelleri ) and a phycocyanin-containing cyano- chlorophyll a (Figure 1). Some accessory pigments
bacteria (Synechococcus spp.): top panel – optical absorption have a high Ft with chlorophyll a and actively sup-
spectra on intact cellular suspensions. Overlaid are absorption
plement the light that is harvested for photo-
spectra for extracts of the predominant pigments, chlorophyll a
(chl a) and photosynthetic carotenoids (PSC) for Chaetoceros synthesis, whereas others have a relatively low Ft
and chl a and phycocyanin (PC) for Synechococcus; middle and are termed photoprotective pigments. Import-
panel – the fluorescence emission at 730 nm following antly, a specific accessory pigment array is unique to
hyperspectral excitation. The two arrows are to demonstrate each algal family providing a first-order taxonomic
that excitation in the blue yields a higher F730 for Chaetoceros
discrimination of complex algal communities. An
since excitation of chl a and PSC is favored while excitation in
the orange yields a higher F730 for Synechcoccus since excitation accessory pigment that is excited with a light source
of chl a and PC is favored. In this way, fluorometers with tuned to the wavelength of peak absorption will in-
fluorescence yields recorded for excitation at various wave- duce a higher chlorophyll a fluorescence yield.
lengths can provide some taxonomic discrimination; bottom panel Therefore, the chlorophyll a emission signature from
– hyperspectral fluorescence emission following excitation by
sequential excitation of multiple wavelengths pro-
blue light, a wavelength commonly used for excitation sources of
many commercial fluorometers. Most fluorometers are designed vides information on the relative abundance of spe-
to measure fluorescence emission at c. 680 nm (chl a, e.g., cific accessory pigments and thus on certain algal
Chaetoceros); however, fluorescence from PC can contaminate families.
the chl a signal as is the case for Synechococcus. Maximum In a similar manner, fluorometers can be tuned to
values for all spectra are scaled to a value of 1 for unity.
target-specific accessory pigments to assess changes
in abundance of a specific algal group. Cyanophytes
expected for natural intact samples. Viable cells are a group of prokaryotic organisms that are also
modify the relationship between F and the chloro- known as blue-green algae but are in fact bacteria
phyll a concentration as a result of variability of the that possess photosynthetic machinery, including
amount of light absorbed, the proportion of ab- chlorophyll a and accessory pigments. Blooms of
sorbed light that is transferred to PSII, referred to as cyanophytes occur frequently in coastal waters and
the transfer efficiency (Ft), as well as of fF . As a lakes with many species producing substances that
result, the ratio of F to chlorophyll a concentration can prove highly toxic to other organisms, including
can alter according to taxonomic (genetic) as well as humans if ingested in large quantities. The most
physiological (acclimation and stress) variability. abundant pigments in these organisms, collectively
Consequently, chlorophyll a fluorescence in nature termed phycobilins, have a unique auto-fluorescence
does not provide an absolute measurement of the signature that can be easily detected in addition to
amount of chlorophyll a in a water sample but can be that of chlorophyll a (Figure 1).
used to infer changes of abundance. Prochlorophytes are oxygenic phototrophic pro-
Similar arguments hold for pigmented organisms karyotes and, along with the cyanophyte Synecho-
other than aquatic plants and algae that fluoresce, coccus, represent the most abundant phytoplankton
most notably, corals and anemones (anthazoa). cells throughout the world’s oceans. Cells of both
 FLUOROMETRY FOR BIOLOGICAL SENSING 259

phytoplankton groups are extremely small (0.2– use of fluorescent tags. Consequently, fluorescence-
2 mm) and require use of their fluorescence signatures based approaches can be considered to be ‘everyday’
for enumeration. Traditional microscope-based epi- tools for examining physiology and molecular-based
fluorescence techniques rely on the fluorescence ecology. However, these approaches require removal
naturally generated by cells under actinic light; of samples from nature. Perhaps the greatest advance
however, the fluorescence yielded by these phyto- for examining physiology in situ again takes advan-
plankton groups is often too dim to make this tage of the unique fluorescence characteristics of
approach viable for oceanography. Instead, flow chlorophyll a.
cytometry (FCM), a technique of biomedical origin, Chlorophyll a fluorescence contains a measure of
is now commonly used. FCM employs a combination fF but represents just one of several pathways that
of light scattering by cells and auto-fluorescence by can be used to dissipate absorbed excitation energy,
natural photosynthetic pigments (chlorophyll a and the others being photochemistry (fP) and heat (fH).
the orange fluorescing phycobilin, phycoerythrin) to Conservation of energy requires that preferential use
enable both identification and enumeration of the of one pathway decreases the use of the others such
different phytoplankton groups. that fF þ fP þ fH ¼ 1. Therefore, an excitation
Modification of FCM via introduction of fluo- protocol that changes the quantum (photon) yields of
rescence tags into water samples can be used to competing processes that affect fF can thus provide
discriminate other functional groups of aquatic information on photosynthetic physiology. Such an
organisms. For example, highly sensitive nucleic acid- approach is termed ‘variable’ fluorescence induction.
specific fluorescent stains that target DNA or rRNA Absolute quantum yields are not typically measured,
have also made possible the detection and enumer- nor do they need to be, provided that the fluo-
ation of heterotrophic bacteria and most recently of rescence signals that are measured accurately report
viruses. Similarly, a more advanced technique, re- changes in the relative quantum yield of fluorescence.
ferred to as fluorescence in situ hybridization (FISH), One variable fluorescence induction approach that
can be used to selectively target regions of DNA or has become popular in oceanography is fast repe-
rRNA that consist of evolutionarily conserved and tition rate (FRR) fluorometry and its derivatives.
variable nucleotide sequences, thus enabling dis- Here, the excitation that is delivered cumulatively
crimination of cells at any taxonomic level ranging closes the PSII reaction centers that dissipate excitons
from kingdom to species. For example, the FISH via photochemistry to raise F from an initial back-
technique has been modified to enable identification ground (F0) to a maximum (Fm) value (Figure 2).
of the taxonomic and life cycle status of single This process occurs within a single trapping event of
coccolithophore cells collected from the ocean. the reaction center pool, termed a ‘single turnover’
(ST). A biophysical model describing the process of
sequential reaction center closure is fit to this fluor-
escence transient to yield various parameters that
Physiological Applications
describe the photochemical process (Figure 2). An-
Many of the fluorescence-based approaches used to other protocol commonly favored to assay PSII
examine abundance and taxonomy have been further physiology is multiple turnover (MT) induction using
modified to provide insights into physiology. Appli- an excitation pulse of longer duration to turn over
cations of fluorescent tags, molecules, and dyes that the pool of reaction centers more than once
can be covalently bound to sensing biomolecules are (Figure 2). ST and MT protocols provide very dif-
now widespread within online fiber-optic biosensors, ferent physiological information.
DNA sequencing, DNA chips, protein detection, and Use of variable fluorescence techniques to pro-
immunoassays. Near-infrared (NIR) fluorescent dyes vide measures of photochemical efficiency, termed
are often most desirable since nonspecific back- Fv/Fm ¼ (Fm  F0)/Fm, for algae and aquatic plants
ground fluorescence is considerably reduced in the has become commonplace in oceanography (see
NIR spectral range and hence sensitivity can be sig- Figure 3). Some of the earliest in situ measurements
nificantly improved. For example, NIR fluorescence were made in the mid-1990s to demonstrate lower
dyes attached to esters can be used to examine the photosynthetic viability under limitation of nutri-
nature of intracellular signaling mechanisms. Recent ents, such as iron and nitrogen. Chlorophyll a
advances in understanding the biological production fluorescence yields the same action spectrum as
of reactive oxygen species, such as singlet oxygen that of O2 evolution confirming that it can pro-
(1O2), superoxide (O 2 ), and hydrogen peroxide vide a representative description of PSII photo-
(H2O2), which can be destructive to proteins and chemical efficiency. Several investigations have also
lipids, has been achieved almost exclusively through demonstrated good agreement between Fv/Fm and
260 FLUOROMETRY FOR BIOLOGICAL SENSING

MT Measurements of PSII physiology as photo-

Fm
chemical efficiency, effective absorption and electron
Fluorescence yield (F )

turnover Fv/Fm, t, and sPSII (Figures 2 and 3) are

(ex = 478, em = 685)

ST
Fm
strictly obtained (1) in darkness either in situ at night

II
and (2) in the laboratory upon discrete samples re-
σPS
moved from the environment. Consequently, all
parameters yield the characteristics inherent to the
potential to process excitation energy. However,
these parameters are all modified in the light as a
PSII result of changes to fP, fH, and fS, and hence the
F0
minimum and maximum fluorescence. Such mea-
Flashlet number
surements are important since they inform the
Figure 2 Fast repetition rate (FRR) style chl a-fluorescence ‘functional’ capacity of PSII to process electrons.
induction for the microalga Symbiodinium spp. Each induction Here, these parameters are denoted as Fq0 =Fv0 , t0 , and
consists of a four-step sequence: (1) single turnover (ST) s0PSII , respectively. Measurements of Fq0 =Fv0 represent
excitation from a short (c. 100 ms) pulse; (2) ST relaxation from
a weak modulated light (c. 500 ms); (3) MT excitation from a
the product of two other parameters of physiologi-
longer (c. 600 ms) pulse; (4) MT relaxation from a weak modu- cal interest, Fq0 =Fm0
(the PSII operating efficiency) and
0 0
lated light (over 1 s). In this way the fluorescence is modulated in Fv =Fm (the maximum PSII yield under actinic light),
a controlled manner between the initial background and which describe the contribution of photochemistry
maximum yield, termed F0 and Fm, respectively. A biophysical and heat to changes of Fv/Fm, respectively. Increas-
model describing the oxidation–reduction of PSII is fit to the
induction profile to yield measures of various physiological
ingly, these fluorescence parameters are being em-
parameters, notably, the effective absorption by PSII (sPSII) and ployed to address complex aspects of photosynthetic
the turnover time of electrons by PSII (t). sPSII describes the physiology that are observed in nature.
target area available for light harvesting by the PSII reaction
centers t describing the rate at which acceptor molecules within
PSII re-oxidize following excitation and thus the rate at which Algal and Aquatic Plant Productivity
electrons can be funneled out of PSII. Both s and t characterize a
wide range of acclimation and adaptation properties inherent to Changes of abundance over time provide a simple
microalgae and AAPs. index of productivity. Numerous laboratory studies
have employed fluorescence to monitor phyto-
the quantum yield of photosynthesis based on ‘con- plankton abundance within cultures; however, such
ventional’ O2 measurements. However, as has been measurements rarely equate to the actual cellular
identified from phytoplankton cultures in the la- growth rate since fF can vary according to the cel-
boratory, natural variability of Fv/Fm is notoriously lular growth phase. An alternative measurement of
difficult to interpret since fF can vary between taxa productivity can be obtained from transformation of
largely as a result of differences in photosynthetic the physiological parameters afforded from variable
architecture (Figure 3). Measurements of Fv/Fm also chlorophyll a fluorescence induction of PSII into
become problematic where a proportion of fluo- photosynthesis rates.
rescence that is induced does not originate from Simple algorithms have been introduced that use
(active) chlorophyll a. One extreme example comes FRR-style induction techniques to generate the gross
from cyanobacteria species containing a phycobilin photosynthetic electron transfer by PSII (PET, mol
called phycocyanin that often bloom in the Baltic electrons (mol PSII reaction center)  1 s  1):
Sea. Fluorescence emitted by phycocyanin is centered
between c. 615 and 645 nm and has a waveband that PET ¼ s0PSII  ðFq0 =Fv0 Þ  PFD
overlaps with the fluorescence emitted by chloro-
phyll a (see Figure 1). Consequently, phycocyanin Measurements of PET in situ are relatively easy to
can lead to an elevated measure of the minimum make but surprisingly rarely reported, primarily
fluorescence yield (F0) that is detected and con- since PET measurements are expressed in units that
sequently to a lower Fv/Fm. Recently, fluorescence are difficult to interpret ecologically. Conversion of
lifetime imaging (FLIM) has been introduced to PET to a measure of productivity expressed in con-
avoid some of these issues associated with (what are ventional units, such as O2 evolution or CO2 fix-
essentially empirical) chlorophyll a induction kinetic ation, is not trivial for natural samples. Within PSII,
measurments of Fv/Fm. FLIM is a technique for four electron steps are required to evolve 1 mol of
producing an image based on differences in the ex- O2; hence, PET should be proportional to 4 times the
ponential decay rate of fluorescence and thus pro- molar O2 evolution by all functional PSII reaction
vides mechanistic information inherent to fF. centers (RCIIs). In the laboratory, the amount of O2
 FLUOROMETRY FOR BIOLOGICAL SENSING 261

(a) 65° N (c) 65° N

35.2
64° N 64° N

35.1 2

Salinity
12° W 9° W 12° W 9° W
35
(b) 65° N (d) 65° N
1

34.9

34.8 0
64° N 64° N

12° W 9° W 12° W 9° W

(e) 65° N (g) 65° N

0.6
700
64° N 64° N
650

0.5 600
F √Fm

12° W 9° W 12° W 9° W

(f) 65° N (h) 65° N 550

500
0.4
450
64° N 64° N

12° W 9° W 12° W 9° W

Figure 3 An example of small spatial and temporal resolution of phytoplankton physiology using active fluorescence: maps of
salinity (a, b), chlorophyll (c, d), Fv/Fm (e, f), and sPSII (g, h), across a section of the Iceland-Faroes front during two surveys separated
by around 2 weeks during June 2001 (Moore, unpublished). A bloom is observed along the frontal boundary and is associated with
marked changes in physiological parameters (e–h) measured by active fast repetition rate fluorometry. These different ‘physiological
acclimation states’ were associated with communities containing different phytoplankton species. Therefore, fluorescence-based
physiological parameters contain information on both taxonomy (adaptation) and physiology (acclimation and/or stress).

evolved per unit chlorophyll a per ST saturating the variability of PSUPSII in nature. Direct de-
flash, termed the photosynthetic unit size of PSII termination of PSUPSII for natural phytoplankton
(PSUPSII), can be determined relatively easily. This samples has been achieved but only when the bio-
measurement accounts both the yield of O2 evolved mass was considerably high, such as within shelf seas
per electron step and the ratio of RCIIs per unit or during blooms, to provide an adequate O2 signal.
chlorophyll a. Therefore, gross O2 production by Alternative (indirect) methods have been proposed to
PSII per unit chlorophyll a (PO 2
chl a
) can be obtained as estimate PSUPSII from knowledge of the spectral ab-
sorption and fluorescence excitation characteristics,
PO
chl a ¼ PET  PSUPSII
2
sPSII and the pigment array. However, such alter-
native methods have only received limited validation.
Of major concern for PO chl a determinations in
2
Most oceanographic investigations to date have
oceanography is thus an accurate understanding of simply assumed values for PSUPSII, an approach that
262 FLUOROMETRY FOR BIOLOGICAL SENSING

can introduce considerable error where the phyto- relatively under-sampled but largest of aquatic sys-
plankton community structure is known to change. tems, the open ocean (Figure 3). Fluorometers can be
Much of the development of fluorescence-based programmed to log continuously and integrated into
algorithms for gross O2 productivity has come via sophisticated bio-optical packages as part of undu-
laboratory-grown phytoplankton suspensions. Ben- lating instruments or moorings to provide a multi-
thic habitats are extremely productive and similar dimensional picture of productivity. Indeed, many
algorithms have been applied to algae inhabiting research programs now employ these larger-scale
substrates but with additional caveats. Most prob- approaches to better understand the effects of en-
lematic is the phenomenon whereby microalgae in- vironmental change upon primary productivity.
habiting the benthos (microphytobenthos) have a However, the sampling scale achieved in situ is still
frustrating tendency to migrate away from the ac- ultimately constrained, in the case of active fluo-
tinic light source of a fluorometer, thereby making rescence and FCM, by power requirements (Table 1).
prolonged measurements inaccurate. In addition, Remote sensing enables assessment of productivity
symbiotic microalgae of corals inhabit an environ- at scales greater than can be achieved in situ. Active
ment where the light environment can be radically fluorometry can be applied remotely provided that
altered by the host anthazoa’s pigments, polyp shape, the excitation required to induce fluorescence is
and CaCO3 skeletal morphology. Consequently, the sufficient. Two such techniques, light detection and
amount of light absorbed by symbiotic algae is ex- ranging (LIDAR) fluorosensing and laser-induced
tremely difficult to determine accurately. fluorescence transient (LIFT) fluorescence, were ini-
Despite the various drawbacks outlined above, the tially developed for remote sensing large-scale (eco-
major promise proposed for fluorescence-based logical) areas of terrestrial forest canopies but have
productivity is as strong as ever. Conventional O2 recently been applied to oceanography. Both LIDAR
evolution or CO2 fixation measurements must be and LIFT techniques are modified induction fluo-
performed upon lengthy and expensive incubations of rometers that excite PSII using targeted lasers. In the
discrete samples, a process that introduces consider- case of LIFT, the laser excitation signal is used to
able error notably via internal recycling of O2 or CO2. both manipulate the level of photosynthetic activity
However, fluorescence measurements can be made in and to measure the corresponding changes in the
situ and in real time. Most laboratory and field PO 2
chl a fluorescence yield. As with FRR fluorescence, the
determinations do not correspond one-to-one with LIFT technique generates fluorescence transients that
simultaneous O2 evolution or CO2 fixation measure- enable measurements of Fv/Fm, t, and sPSII. Ulti-
ments, a trend that is not surprising given difficulties
associated with the various approaches. Consequently, Table 1 Common scales of sampling (meters) with current
there remains some way to go before the considerable fluorescence techniques for marine biological applications
promise of real-time productivity estimation from
active chlorophyll a fluorescence is achieved. o10  6 to 10  3 10  3 to 10 3 410 3
New Optode methods for measuring the net
Analysis of discrete Small-scale Large-scale
community O2 evolution directly in situ also employ samples: in situ in situ
fluorescence. Optode technology is based on the campaigns: campaigns*
ability of selected substances to act as dynamic Fluorescence Induction Remote
fluorescence quenchers in the presence of oxygen. imaging fluorometers sensing:
Not only do such methods provide highly accurate microscopes Optodes LIDAR and LIFT
(epifluorescence Flow cytometers Satellite
and sensitive O2 measurements but also, upon and active chl retrieval
coupling to in situ fluorescence-based PO chl a de-
2
a fluorescence
terminations, may potentially provide important new induction)
insights into ecosystem function, metabolism, and Flow cytometers
trophic coupling. Intracellular molecular
tags

The smallest scales are dominated by techniques that perform

measurements at the single-cell (or intracellular) scale upon discrete
Scales of Productivity Measurements water samples removed from nature. Mesoscales are dominated by
A second advantage afforded from fluorescence over instruments that can be deployed in situ, for example, variable
chlorophyll a fluorometers and modified flow cytometers (Cytobuoy
conventional O2 evolution or CO2 fixation pro-
and FlowCytobot technologies), while the largest scales are
ductivity determinations is an increase of both spa- dominated by remote sensing techniques. The asterisk indicates
tial and temporal scale that can be investigated. Such that any of the techniques could be applied but that the scale
an advantage is particularly important for the afforded is limited by sampling effort and power requirements.
 FLUOROMETRY FOR BIOLOGICAL SENSING 263

mately, power constraints and safety considerations our most powerful monitoring tools as the Earth is
also constrain measurement scales for these laser- subjected to a time of extreme climate change.
based techniques.
Remote sensing of ocean productivity largely relies Nomenclature
on satellite retrieval of ocean color reflectance by
algal pigments. Productivity algorithms most sim- A absorptivity
plistically combine chlorophyll a concentration and Achl absorption specific to chlorophyll
PFD; however, such an approach has proven difficult C geometry of fluorometer optics
to retrieve quantitative information on algal C atmospheric effects
physiology. F fluorescence yield
Fluorescence can be observed passively via remote Fm maximum fluorescence yield
sensing since algae and aquatic plants naturally F0 initial fluorescence yield
fluoresce when they are excited by sunlight (solar Fq0 =Fm
0
PSII operating efficiency
stimulation). Chlorophyll a fluorescence is inherent Fq0 =Fm
0 PSII photochemical efficiency (under
to the water-leaving radiance with peak emission at actinic light)
683–685 nm. By analogy to the fluorescence yield Fv/Fm PSII photochemical efficiency (dark
induced actively, the amount with which chlorophyll adapted)
a fluorescence increases the water-leaving radiance Fv0 =Fm0
PSII maximum yield under actinic light
(F) depends on several factors, including the specific PO 2 gross O2 production by PSII per unit
chl a
absorption of chlorophyll (Achl), fluorescence quan- chlorophyll a
tum efficiency (fF), amount of incident sunlight sPSII PSII effective absorption cross section
(PFD), in addition to various atmospheric effects (dark adapted)
(C ): s0PSII PSII effective absorption cross section
(under actinic light)
F ¼ fF  C  Achl  PFD t turnover time of electrons (dark adapted)
t0 turnover time of electrons (under actinic
Chlorophyll a fluorescence (F) is quantified ac- light)
cording to the fluorescence line height (FLH) above fF quantum yield of fluorescence
background levels at 678 nm, several nanometers fH quantum yield of heat
below the wavelength of peak emission to avoid at- fP quantum yield of photochemistry
mospheric oxygen absorption at 687 nm. Estimation Ft energy transfer efficiency between pig-
of fF is achieved as F/(C  Achl  PFD), where ment molecules
(C  Achl  PFD) is the instantaneous absorbed radi-
ation by phytoplankton, termed ARP. A semi-
analytical model that also retrieves chlorophyll a See also
concentration from measurements of water-leaving
Aircraft Remote Sensing. Bio-Optical Models.
radiance in all bands from 412 to 551 nm is used to
Fluorometry for Chemical Sensing. Inherent
determine the ARP. Primary productivity (PP) can Optical Properties and Irradiance. Optical Particle
then be determined as Characterization. Photochemical Processes.
Satellite Oceanography, History and Introductory
PP ¼ fP  ARP Concepts.

Therefore, the relationship between fF and the Further Reading

quantum yield of PSII photochemistry (fP) must be
known. As shown above, this relationship will be Andersen RA (ed.) (2005) Algal Culturing Techniques.
modified by fH and fS, parameters that are not easy New York: Elsevier Academic Press.
to constrain for natural samples as a result of the Barbini R, Colao F, Fantoni R, Palucci A, and Ribezzo S
huge taxonomic and physiological (both acclimation (2001) Differential LIDAR fluorosensor system used for
phytoplankton bloom and seawater quality monitoring
and inhibition) variability. Consequently, consider-
in Antarctica. International Journal of Remote Sensing
able effort is currently invested in trying to under- 22: 369--384.
stand the relationship between fH and environment Falkowski PG and Raven JA (1997) Aquatic Photo-
for key algal taxa. Despite these potential draw- synthesis. New York: Blackwell.
backs, passive measurements of global chlorophyll a Geider RJ and Osborne BA (1992) Algal Photosynthesis:
fluorescence show great potential for future obser- The Measurement of Algal Gas Exchange. Boca Raton,
vations of global productivity and may offer one of FL: Chapman and Hall.
264 FLUOROMETRY FOR BIOLOGICAL SENSING

Groben R and Medlin L (2005) In situ hybridization of Moore CM, Suggett DJ, Hickman AE, et al. (2006)
phytoplankton using fluorescently labeled rRNA Phytoplankton photo-acclimation and photo-adaptation
probes. Methods in Enzymology 395: 299--310. in response to environmental gradients in a shelf sea.
Hoge FE, Lyon PE, Swift RN, et al. (2003) Validation of Limnology and Oceanography 51: 936--949.
Terra-MODIS phytoplankton chlorophyll fluorescence Papageorgiou GC and Govindjee (eds.) (2004) Chlorophylla
line height. I. Initial airborne LIDAR results. Applied Fluorescence: A Signature of Photosynthesis.
Optics 42: 2767--2771. Amsterdam: Springer.
Huot Y, Brown CA, and Cullen JJ (2005) New algorithms Suggett DJ, Maberly SC, and Geider RJ (2006) Gross
for MODIS sun-induced chlorophyll fluorescence and a photosynthesis and lake community metabolism during
comparison with present data products. Limnology and the spring phytoplankton bloom. Limnology and
Oceanography Methods 3: 108--130. Oceanography 51: 2064--2076.
Jeffrey SW, Mantoura RFC, and Wright SW (1997) Veldhuis MJW and Kraay GW (2000) Application of flow
Phytoplankton Pigments in Oceanography. Paris: cytometry in marine phytoplankton research: Current
UNESCO Publishing. applications and future perspectives. Scientia Marina
Kolber ZS, Klimov D, Ananyev G, Rascher U, Berry J, and 64: 121--134.
Osmond B (2005) Measuring photosynthetic para-
meters at a distance: Laser induced fluorescence Relevant Websites
transient (LIFT) method for remote measurements of
photosynthesis in terrestrial vegetation. Photosynthesis http://www.cytobuoy.com
Research 84: 121--129. – Environmental applications of flow cytometry,
Kolber ZS, Prášil O, and Falkowski PG (1998) CytoBuoy.
Measurements of variable chlorophyll fluorescence http://www.esa.int
using fast repetition rate techniques: Defining – European Space Agency Living Planet.
methodology and experimental protocols. Biochimica http://www.chelsea.co.uk
Biophysica Acta 1367: 88--106. – Fast Repetition Rate Methods Manual, Chelsea
Labas YA, Gurskaya NG, Yanushevich YG, et al. (2002) Technologies Group.
Diversity and evolution of the green fluorescent protein http://modis.gsfc.nasa.gov
family. Proceedings of the National Academy of – MODIS web, NASA.
Sciences of the United States of America 99:
4256--4261.
 FLUOROMETRY FOR CHEMICAL SENSING
S. Draxler and M. E. Lippitsch, excitation and emission, the so-called ‘Stokes shift’,
Karl-Franzens-Universität Graz, Graz, Austria occurs owing to rapid decay of excitation energy to
Copyright & 2001 Elsevier Ltd. the lowest vibrational level of the excited state, as
well as from subsequent decay to higher vibrational
levels of the ground state. As the relaxation to the
lowest vibrational level usually occurs within 1012 s,
the fluorescence emission spectrum does not depend
Introduction on the excitation wavelength. The fluorescence can be
described using the parameters spectral distribution
In the context of ocean sciences, ‘chemical sensing’ of the emission, quantum yield, decay time, and
denotes the acquisition of data on the concentration polarization.
of certain chemical species (the analytes) in sea If the molecule is excited to its first excited elec-
water. This acquisition process may be achieved by a tronic state, fluorescence is usually emitted directly
‘chemical sensor’ in the strict sense, i.e., a device that from this state, and the fluorescence emission spec-
continuously and reversibly exhibits a change in trum appears to be a ‘mirror image’ of the ab-
some of its properties as a function of the concen- sorption spectrum (Figure 1). Deviations from the
tration of a respective analyte. As an alternative, mirror image rule occur if the excitation goes to
sensing may be done directly, without the detour higher-lying states, if the molecule undergoes major
over the properties of a sensor, by measuring some geometrical rearrangements in the excite state, or if
effect displayed by the analyte itself. Intermediate excited-state chemical reactions occur.
between these two methods is the use of an indicator Fluorescence decay times usually range from some
that is added to a sample of sea water and changes its picoseconds to nanoseconds, or up to milliseconds in
properties depending on the analytes’ concentration. the case of phosphorescence.
An example of a sensor is the Clark electrode, which The fluorescence decay time t (sometimes also
delivers a current proportional to the concentration called ‘fluorescence lifetime’) describes the average
of oxygen. An example of direct sensing is the de- time a molecule spends in the excited state before it
termination of salinity by conductivity measure- returns to the ground state. In the simplest case, the
ments. The well-known use of litmus to monitor pH fluorescence intensity follows an exponential law:
is an example for the indicator method.
‘Fluorometry’ refers to the measuring of fluor- IðtÞ ¼ I0 et=t ½1
escence, which is the reemission of light from a where t is the time it takes for the fluorescence in-
compound upon exposition to light of a shorter tensity to decrease to a value of I0 e1 . The decay
wavelength (in this context, phosphorescence is in- time is connected to rate constants via
cluded under the term fluorescence, although strictly 1
speaking these are two different processes). Fluor- t¼ ½2
Gþk
escence occurs from a singlet excited state, while
phosphorescence is originating from a triplet. This
emission is characteristic for the emitting compound, 1.0
but may be subject to modification by the environ-
Absorption or relative

0.8
ment. Thus, fluorometry may be applied to both
fluorescence

sensing schemes outlined above: In a sensor device, 0.6

its characteristic fluorescence properties may be
modified by the concentration of the analyte. Direct 0.4
sensing preferentially will be applied if the analyte
itself is fluorescent. 0.2

0
Fluorescence 300 350 400 450 500 550
Wavelength (nm)
The fluorescence emission displays a number of
general characteristics. Except for atoms in the vapor Figure 1 Typical absorption and fluorescence of an organic
phase, the emission is shifted to lower wavelengths compound, showing ‘mirror image’ relationship between the
relative to the excitation. The energy loss between absorption spectrum (- - - -) and the fluorescence spectrum (- - - -).

265
266 FLUOROMETRY FOR CHEMICAL SENSING

and therefore depends on all processes affecting the certain amount of reliability if an ‘internal standard’
radiative or nonradiative rate constant. can be used for comparison. If water samples are
Fluorescence decay times usually range from some analyzed in the laboratory, a fluorophore (e.g. quin-
picoseconds to nanoseconds. If triplet states are in- ine sulfate) added in a well-defined concentration can
volved in the emission process, the measured decay serve as a standard. This is not possible for in situ
time can be of milliseconds to seconds, and this measurements. The only substance available as a
emission usually is called phosphorescence. standard in any case is water itself, but water is
Excitation with polarized light results in polarized nonfluorescent. Also, scattered light cannot be used,
emission, since fluorophores preferentially absorb since scattering strongly depends on the amount of
photons whose electric vectors are parallel to the small particles (dust, microorganisms, etc.) in the
transition dipole moment of the fluorophores. The sample.
fluorescence polarization P after excitation with There is, however, a weak wavelength-shifted
vertically polarized light is defined as component, called Raman scattering, that is useful as
a standard. Raman scattering does not, like fluor-
In  Ip escence, produce a specific wavelength but rather a
P¼ ½3 difference in wavelengths between excitation and
In þ Ip
emission. Thus, in shifting the excitation wavelength
where Ip and In are the fluorescence intensities of the the scattered wavelength is also shifted. The intensity
vertically (n) and horizontally (p) polarized emission. of the Raman band of water may be used as the
In a solid environment at low fluorophore concen- standard against which the fluorescence intensity of
trations, the polarization is determined only by the the analyte is calibrated (Figure 2). The quinine sul-
relative directions between absorption and emission fate standard is related to the Raman standard by
dipole moment in the fluorophore. In liquids, where defining the ratio of the integrated fluorescence of a
the fluorophore molecules may move freely, a de- 1 mg l1 quinine sulfate solution to the integrated
crease in polarization can be observed owing to ro- H2O Raman band as the ‘quinine sulfate unit’
tational diffusion or excitation transfer between (QSU). Results in QSUs should be directly compar-
fluorophores. able among different instruments and different
laboratories.
A much more refined method is three-dimensional
Techniques fluorescence (or excitation–emission matrix spec-
troscopy, EEM). The excitation as well as the emis-
Direct Sensing
sion wavelengths are scanned over a board range and
Direct measurement of the concentration of fluor- the fluorescence intensity is plotted as a function of
escent materials in sea water is usually done with both (Figure 3). Sophisticated mathematical evalu-
samples taken in the field and transported to a la- ation allows identification and relative quantification
boratory, either on board a research ship or on land. of multiple analytes.
The procedure of measuring involves three steps: Another technique for distinguishing between
excitation of the analyte, detection and identification multiple analytes is time-resolved fluorometry.
of fluorescent light, and determination of concen- Excitation is done with very short light pulses
tration of fluorescent analyte.
To excite a given analyte a source emitting light
within the analyte’s absorption band is necessary. 1.0
1.00
Fluorescence intensity

A monochromatic or at least narrowband source 0.8 0.35

(arbitrary units)

serves to discriminate the analyte against all com- 0.15

pounds with different absorption. In any case, the 0.6
detected fluorescence may be the superposition of the
emissions from various different analytes, and to add 0.4
*
further specificity, the fluorescence light has to be 0.2
spectrally analyzed. Most analytes have a charac-
teristic ‘fingerprint’ spectrum; thus, it is possible to 0
identify the analyte from its spectrum provided a 350 400 450 500 550
limited choice of analytes is known to be present. Wavelength (nm)
The main difficulty is in extracting from the intensity Figure 2 Fluorescence spectra of perylene at three different
of the detected fluorescence the concentration of the concentrations with the water Raman line (*) as the reference
respective analyte. This can only be done with a (maximum concentration 100 ng/l).
 FLUOROMETRY FOR CHEMICAL SENSING 267

Fluorescence intensity (arbitrary units)

29
0
34
0
Em 39
iss
ion 0 500
wa
vel 44 450
e 0
ngt 400
h(
nm 490 350 h (nm)
lengt
)
54 300 t io n wave
0 50
2 Excita

Figure 3 Three-dimensional or excitation-emission matrix spectrum. The small ridge running across the spectrum is the water
Raman line. (Courtesy of M.J.P. Leiner and K. Kniely.)

(durationo1 ns), and the fluorescence spectrum is Fluorescent chemical sensor devices consist essen-
recorded after a certain delay. Since the fluorescence tially of three main building blocks: an illumination
decay times of various compounds are different, the device, the sensor element proper, and a detecting
spectrum will change with increasing delay. With apparatus. Since the fluorescence to be excited and
short delays, all excited compounds contribute to the detected is that of the sensor element alone, much
spectrum. After a short while, the fluorescence of less effort needs to be made to discriminate between
some of the compounds has vanished while that of various sources of fluorescence. This opens the pos-
others persists (Figure 4). A careful analysis of a se- sibility of using small and cheap light sources such as
quence of successively delayed spectra provides the light-emitting diodes. Usually the sensor is used in
relative concentration of the various fluorescent situ rather than in a laboratory and the sensor
compounds contained in the sample. element is dipped directly into the sea water. How-
Fluorescence polarization measurement is a rather ever, this implies that the excitation light has to be
unusual technique in ocean sciences. The only ex- brought to the sensor element. This can be done by
perimental difference from simple fluorometry is that implementing a submersible sensor unit consisting of
excitation is done with polarized light and the the light source, the sensor element, and even the
emission is recorded separately for two perpendicu- detector. Alternatively, only the sensor element may
lar directions of polarization. be submersed and light can be transported via optical
fiber to and from the element. In some cases even the
sensor element may be part of an optical fiber.
Indicators and Sensors
The sensor element in most cases is a polymer
Indicators and sensors have in common that the membrane containing a fluorescent indicator dye.
fluorescent compound is not a constituent of the sea The fluorescence of the indicator can be altered by
water under investigation but is added deliberately the presence of the analyte in several different ways:
during the measuring process. From the standpoint The analyte may interact with the excited molecule
of fluorometry it makes no difference whether the in a way that brings about changes in the fluor-
fluorescent indicator is dissolved in the sea water escence intensity. On the other hand, it binds to the
sample or is incorporated in a sensor element that is indicator and thus alters the molecular properties in
brought into contact with the water. Therefore, the such a way that the absorption and/or emission are
following technical description applies to indicators shifted spectrally. And finally, binding of the analyte
and sensors as well. may not alter the spectral properties of an indicator
268 FLUOROMETRY FOR CHEMICAL SENSING

1.0 6
0 ns
Fluorescence intensity

0.8 100 ns
(arbitrary units)

5
0.6

_1
4

(I0 /I ) 1 or (0 / )
0.4

0.2 3

_
0
350 400 450 500 550 2
Wavelength (nm)
1
Figure 4 Time-resolved fluorometry of a mixture of two
compounds with decay times 4 ns and 300 ns. Spectra taken
immediately after excitation (0 ns) and with a considerable delay 0
(100 ns). 0 200 400 600 800
p O2 (torr)

molecule but may change its mass and size. This Figure 5 Stern–Volmer plot (see text) of fluorescence
yields a different moment of inertia and hence a quenching by oxygen.
different rotational diffusion and a change in fluor-
escence depolarization.
presents difficulties for in situ measurements, espe-
A reduction in fluorescence intensity is called
cially when the sensor equipment is intended to
quenching. Two types of quenching can be dis-
measure autonomously over an extended time. This
tinguished. Quenching is called dynamic if the
difficulty can be overcome by measuring the fluor-
interaction between the quencher (analyte) and the
escence decay time rather than the intensity. The
luminophore (indicator) occurs while the latter is in
decay time also obeys a Stern–Volmer law but it is
the excited state. In this case the rate constant of the
independent of drifts in fluorescence intensity that
nonradiative decay is increased, shortening the decay
may be brought about by leaching or bleaching of the
time and diminishing the quantum efficiency. Dy-
indicator dye or by aging effects in the light source or
namic quenching is not associated with a chemical
the detector. Thus, decay time-based sensors, despite
reaction between indicator and analyte. If, on the
requiring a more sophisticated optoelectronics, have
other hand, the interaction occurs while the indicator
much superior long-term stability and, in fact, are the
is in its electronic ground state, the quenching is
only promising candidates for unattended monitoring
called static. Static quenching may or may not in-
purposes.
clude a chemical reaction.
Static quenching implies some close association
In the simplest case, the reduction of the fluor-
between the indicator molecule in its ground state
escence intensity or the decay time is described by a
and the analyte. Usually an equilibrium is reached
linear function, the so called Stern–Volmer equation,
between associated indicator–analyte pairs and free
indicator, which is determined by the association
I0 constant characteristic for the partners. The higher
¼ 1  kSV ½A ½4
I the concentration of the analyte, the more the equi-
librium is shifted toward the association.
where I0 and I are the unquenched and quenched The associated and the free indicator may have
intensity, kSV is the Stern–Volmer constant (a quan- distinct fluorescence spectra or distinct decay times.
tity characteristic of a certain combination of indi- The spectra of the two forms intersect at a certain
cator and quencher), and [A] is the concentration of wavelength, the so-called isosbestic point. The total
the quenching analyte (Figure 5). fluorescence intensity at that wavelength is in-
The Stern–Volmer equation shows that we have to dependent of the relative concentrations and hence
measure two quantities, I and I0 , to extract the ana- independent of the analyte. Thus, this intensity can
lyte’s concentration. Because of the more reproducible be conveniently used as an internal standard. For two
conditions in a sensor membrane, instead of using an distinct decay times, the decay function becomes the
internal standard it is sufficient to recalibrate the sum of two exponentials
sensor from time to time by applying a standard so-
lution of the respective analyte. This is rather easy if  
the sensor is used in a laboratory environment, but IðtÞ ¼ I0 aet=ta þ bet=tb ½5
 FLUOROMETRY FOR CHEMICAL SENSING 269

where a and b represent the relative concentrations Table 1 Dissolved chemical species in sea water measurable
and ta and tb the respective decay times of the as- directly via fluorescence
sociated and the free indicator. The concentration of
Analyte lex/lem (nm) Decay
the analyte can be extracted from the decay curve by time (ns)
a fitting algorithm.
If the associated and free indicator have different Dissolved organic 220–390/300–500 o2
absorption spectra, the result in fluorometry is sim- matter (DOM/DOC)
ply a different fluorescence intensity because of a Dissolved proteinaceous 220–280/300–350 o2
different probability of excitation. If, on the other materials
Humic acids 230, 300–350/420–450 o2
hand, the species have different emission spectra, the Polycyclic aromatic 300–450/400–550 4–130
result will be the superposition of the two spectra. In hydrocarbons
many cases there exists an isosbestic point, that is, a Trace plastic and epoxy UV/UV o2
wavelength where the two emission spectra cross and compounds
where the emission intensity is independent of the Heavy metals X-ray —

relative concentrations of two species. This point lex, excitation wavelength; lem, emission (fluorescence) wave-
may be used for internal referencing. The concen- length.
tration of the analyte can be deduced from the ratio
of emission intensities at two wavelengths that are fluorescent DOC was determined at about 150 mmol
outside the overlapping region of the two spectra. in the Mid-Atlantic Bight. DOC from rivers seems to
be more strongly fluorescent than DOC from marine
systems. Fluorescence intensity is 2–2.5 times higher
Applications in deep waters, where photodecomposition and
oxidation of high-molecular-weight matter is low.
Direct Measurement of Fluorescent Species
There may, in certain extreme cases, even be a
in Sea Water
negative correlation between DOC and fluorescence
The most important marine fluorophore is chloro- intensity. This shows that the use of fluorescence data
phyll. However, this article is devoted only to sensing to determine DOC quantitatively is not straight-
of dissolved compounds (see Fluorometry for Bio- forward. Recalibration by comparison with chemical
logical Sensing) for fluorophores in living organ- DOC quantification is necessary if water conditions
isms). Table 1 summarizes the most commonly change significantly.
measured substances as well as their excitation and Among the components of DOM, the nutrients
emission wavelengths. (nitrate and phosphate) play a major role because of
Dissolved organic matter (DOM) is the catch-all their importance for metabolism. They are usually
denotation for all sorts of organic products from nonfluorescent, but their concentrations often cor-
metabolism and biological decomposition. The term relate well with overall DOC and thus also with
dissolved organic carbon (DOC) is more specific and fluorescence intensity under equivalent conditions.
quantitative as it is related to the carbon content in However, the ratio of carbon to nitrogen or phos-
the dissolved organic matter. Consequently, DOM phorus may vary in a wide range (observed C:N
can be quantified only in mass per unit water vol- ratios are in the range 16–38). Specific detection of
ume, while DOC can be specified in moles of carbon nutrients is usually impossible with direct fluor-
per unit volume. DOC nevertheless subsumes a escence methods.
multitude of compounds. They are important com- The situation is more favorable for proteinaceous
ponents of the global carbon cycle. Little is known material. The characteristic absorption and emission
about the detailed chemical composition of DOM, wavelengths (lex 220–270 nm, lem 300–350 nm, re-
but recent investigations have shown that acyl spectively) allow differentiation from other com-
oligosaccharides may form a significant fraction. pounds. The same is true for humic acids, which can
A large part of DOM is formed by components be distinguished by their unusually long-wavelength
that can be detected via fluorescence (fluorescent emission (420–450 nm).
organic matter, FDOM). The ratio FDOM/DOM The second large bulk of fluorescent compounds in
seems to be rather constant in surface waters; ac- sea water is made up of pollution products. The main
cordingly, the correlation between DOC and fluor- sources are sewage, oil spills, runoff, and atmos-
escence intensity is fairly robust. However, pheric deposition. A class of special interest is that of
compounds of low molecular weight essentially lack polycyclic aromatic hydrocarbons (PAHs), which are
fluorescence, at least when excited in the usual released into the environment in large quantities.
wavelength range for DOC (250–350 nm). This low- They are quite persistent, and some of them are
270 FLUOROMETRY FOR CHEMICAL SENSING

potent carcinogens. Their fluorescence spectra are Table 2 Dissolved chemical species in sea water measurable
not significantly different from those of other organic via fluorescence sensor devices
material, but they are outstanding by their long
Analyte Sensor type Range, sensitivity
fluorescence decay times. Thus they can be dis-
tinguished by time-resolved spectroscopy. The de- Oxygen Decay time 0–60 ppm, 1 ppb
tection limit for PAHs by time-resolved fluorometry pH Decay time 6–10, o0.1
has been shown of the order of nanograms per liter Carbon dioxide Intensity, 200–1000 ppm, 1 ppm
in the presence of DOC at milligrams per liter. decay time
Three-dimensional fluorometry has also been used Aluminum Intensity 10–1000 nmol l1, 10 nmol l1

to differentiate between various constituents of

fluorescent matter in sea water. For humic acids,
different excitation emission maxima were found for
Fluorescent sensors for other analytes have been
different types of water: coastal waters peaked at
made commercially available only in the field of
340/440 nm, shallow transitional waters at 310/
medical diagnostics. Fluorescent optical sensors have
420 nm, eutrophic waters near 300/390 nm, and
been developed by several laboratories for marine
deep-sea samples at 340/440 nm (exitation/emission
research, but these have not reached the market yet.
wavelengths). Although the chemical nature of the
A sensor system for CO2 was developed by C. Goyet
different humic species is not yet clear, these results
and colleagues. Using a combination of fluorescent
provide means of distinguishing between water mass
and absorptive indicator dyes, they succeeded in
sources in the ocean. Three-dimensional fluorometry
measuring pCO2 in sea water with a mean relative
was found to be advantageous for identifying pol-
error of about 2% compared to results from a gas
lutants. It was used, for example, to detect trace
chromatograph.
plastic and epoxy compounds in the presence of
Luminescent pH sensing has also been reported. In
other organic materials.
the range from pH 6 to 9, a resolution of better than
X-ray fluorescence is the method of choice for
0.1 pH is readily achieved.
measuring metal contaminants. Excitation as well as
A fluorescent indicator rather than a true sensor
emission wavelengths are in the X-ray region. The
has been used to determine dissolved aluminum in
spectra obtained are rather specific for certain
sea water at the nanomolar level. The indicator sal-
metals. Portable X-ray fluorescence spectrometers
icylaldehyde picolinoylhydrazone complexed with
are available for measurement in the field.
aluminum gives a fluorescence peaking at 486 nm
when excited at 384 nm. In the concentration range
Indicators and Sensors from 38 to 930 nmol l1, a detection limit of
9.8 nmol l1 and a precision of about 2% were
Table 2 summarizes the sensor type, the range cov-
achieved.
ered, and the sensitivity obtained for several
Numerous other sensors have been tested in the
analytes.
laboratory, but no proof of suitability for marine
Dissolved oxygen is the analyte most readily
applications has been given. There are remarkable
measured using a fluorescent optical chemical sensor.
new approaches to the detection of nutrients (nitrate,
Oxygen, because of its triplet ground state, is a no-
phosphate), but in these cases the specifications
torious fluorescence quencher. Various organic
needed for marine research have not yet been
fluorescence indicators have been proposed for
accomplished.
measuring oxygen, but in recent years metalloor-
ganic complexes have been found to be most suit-
able, especially ruthenium diimine complexes and Conclusion
platinum or palladium porphyrins. These indicators
are outstanding because of their long fluorescence Numerous fluorescent sensing methods have been
lifetimes and good quenchability by oxygen. The developed, but only a limited number have become
sensors cover the whole range of dissolved oxygen standard methods in marine research. The most re-
concentration encountered in aquatic environments. cent developments (decay-time sensors, three-di-
Compared to electrochemical sensors (Clark elec- mensional fluorescence, lifetime imaging) offer up
trode) they have the advantage that they do not interesting prospects. In particular, the high long-
consume oxygen and hence may be used even in term stability of decay time-based sensors make them
stagnant water. The technique of lifetime imaging prime candidates for instrumentation on unattended
has been applied to the study of spatial oxygen dis- measuring stations needed for global ocean ob-
tribution in sediments. serving systems.
 FLUOROMETRY FOR CHEMICAL SENSING 271

Symbols used Further Reading

[A] analyte concentration Aluwihare LI, Repeta DJ, and Chen RF (1997) A major
G radiative rate constant biopolymeric component to dissolved organic carbon in
I fluorescence intensity surface sea-water. Nature 387: 166--169.
I0 fluorescence intensity immediately after Glud RN, Gunderden JK, and Ramsing NB (2000)
excitation Electrochemical and optical oxygen microsensors for in
In fluorescence intensity polarized per- situ measurements. In: Buffle J and Horvai G (eds.) In-
situ Monitoring of Aquatic Systems. Chichester: Wiley.
pendicular to the plane of incidence
Goyet C, Walt DR, and Brewer PG (1992) Development of
Ip fluorescence intensity polarized parallel a fiber optic sensor for measurement of pCO2 in sea
to the plane of incidence water: design criteria and sea trials. Deep-Sea Research
k radiationless rate constant 39: 1015--1026.
kSV Stern–Volmer constant Karabashev GS (1996) Fluorometric methods in studies
lex excitation wavelength and development of the ocean (a review). Okeanologiya
lem emission wavelength 36: 165--172 (In Russian.).
Q fluorescence quantum yield Lakowicz JR (1999) Principles of Fluorescence Spectros-
P fluorescence polarization copy. Dordrecht: Kluwer Academic.
t fluorescence decay time Manuelvez MP, Moreno C, Gonzalez DJ, and Garciasvargas
t time M (1997) Direct fluorometric determination of dissolved
aluminium in seawater at nanomolar level. Analytica
Chimica Acta 355: 157--161.
Wolfbeis OS (ed.) (1991) Fiber Optic Chemical Sensors
See also and Biosensors. Boca Raton, FL: CRC Press.
Fluorometry for Biological Sensing.
 INHERENT OPTICAL PROPERTIES AND
IRRADIANCE
T. D. Dickey, University of California, Santa Barbara, ecology, red tides, pollution, biogeochemistry in-
CA, USA cluding carbon cycling, photochemistry, light-keyed
Copyright & 2001 Elsevier Ltd. migration of organisms, heating of the oceans, global
climate change, and remote sensing of ocean color. It
is worth noting that optical oceanographers are also
interested in the ultraviolet (UV) range and the in-
frared. For example, UV radiation can damage
phytoplankton and can also lead to modifications in
Introduction genetic material, and thus evolution, whereas the
penetration of the infrared is critical for near surface
Light is of great importance for the physics, chem-
radiant heating.
istry, and biology of the oceans. In this article, a brief
Solar radiation is clearly a primary driver for
introduction is provided to the two subdisciplines
ocean physics, chemistry, and biology as well as their
focusing on light in the ocean: ocean optics and bio-
complex interactions. Thus, it is not surprising that
optics. A few of the problems addressed by these
virtually all important oceanographic problems re-
subdisciplines are described. Several of the in situ
quire interdisciplinary approaches and necessarily
sensors and systems used for observing the sub-
atmospheric, physical, chemical, biological, optical,
surface light field and optical properties are intro-
and geological data sets. These data sets should be
duced along with general explanations of the
collected concurrently (concept of synopticity) and
operating principles for measuring optical variability
span sufficient time and space scales to observe the
in the ocean. Some of the more commonly used
relevant processes of interest. For local studies in-
ocean platforms and optical systems are also dis-
volving optics, variability at timescales of one day
cussed. Finally, some examples of oceanographic
and one year are especially important. For global
optical data sets are illustrated.
problems, variability extends well over ten orders of
Solar radiation, which includes visible radiation or
magnitude in space [O(millimeters) to O(104 kilo-
light, impinges on the surface of the ocean. On
meters)] and much longer in time for climate prob-
average, a small fraction or percentage is reflected
lems. Multiplatform observing approaches are
back into the atmosphere (roughly 6% on average)
essential. Capabilities for obtaining atmospheric and
while a high fraction penetrates into the ocean. This
physical oceanographic data are relatively well ad-
fraction (or percentage), defined as the albedo, varies
vanced in contrast to those for chemical, biological,
in time and space as a function of several factors
optical, acoustical, and geological data. This is not
including solar elevation, wave state, surface rough-
surprising, because of the greater complexity and
ness, foam, and whitecaps. Radiative transfer is a
nonconservative nature of the chemistry and biology
branch of oceanography termed ‘ocean optics,’ a
of the oceans. Yet, remarkable advances are being
term that denotes studies of light and its propagation
made in these areas as well. In fact, several bio-op-
through the ocean medium. Radiative transfer pro-
tical, chemical, geological, and acoustical variables
cesses depend on the optical properties of the com-
can now be measured on the same time and space
ponents lying between the radiant source (e.g., the
scales as physical variables; however many more
sun) and the radiation sink (the ocean and its con-
variables still need to be measured.
stituents). Another commonly used term is ‘bio-op-
tics,’ which invokes the notion of biological effects
on optical properties and light propagation and vice
Fundamentals of Ocean Optics
versa. Some of the data used for examples here focus
on bio-optical and physical interactions. A few operation definitions need to be introduced
Solar radiation spans a broad range of the elec- before discussing the various optical sensors. Light
tromagnetic energy spectrum. The visible portion, entering the ocean can be absorbed or scattered. The
roughly 400–700 nm, is of primary concern for details of these processes comprise much of the study
ocean optics for many practical problems. These in- of ocean optics. First, it is convenient to classify bulk
clude: underwater visibility, photosynthesis and pri- optical properties of the ocean as either inherent or
mary production of phytoplankton, upper ocean apparent. Inherent optical properties (IOPs) depend

272
 INHERENT OPTICAL PROPERTIES AND IRRADIANCE 273

only on the medium and are independent of the am- ct ðlÞ ¼ cw ðlÞ þ cph ðlÞ þ cd ðlÞ þ cg ðlÞ ½3
bient light field and its geometrical distribution. In-
herent optical properties include: spectral absorption The subscripts indicate contributions by pure sea
coefficient, aðlÞ, spectral scattering coefficient, bðlÞ, water (w), phytoplankton (ph), detritus (d), and
and spectral beam attenuation coefficient (or, some- gelbstoff (g) with units of m1 for all variables. De-
times denoted as ‘beam c’), cðl). To define these co- tritus is the term for particulate organic debris (e.g.,
efficients, consider a beam of monochromatic light fecal material, plant and animal fragments, etc.) and
impinging perpendicularly on a thin layer of water. gelbstoff (sometimes called yellow matter or gilvin) is
The fraction of the incident radiant flux, F0 (energy or the term for optically active dissolved organic ma-
quanta per unit time), which is absorbed by the me- terial. Note that eqns [1]–[3] are intended for use
dium, Fa/F0, divided by the thickness of the layer is where bottom and coastal sediment contributions are
the absorption coefficient, a, in units of m1. Simi- minimal. The spectral absorption coefficient of pure
larly, the analogous scattering coefficient, b (m1) is sea water is well characterized and is nearly constant
the fraction of the incident radiant flux scattered from in space and time with greater absorption in the red
its original path (primarily forward), Fb/F0, divided than blue portions of the visible spectrum. The
by the thickness of the layer. The beam attenuation magnitudes of the detrital and gelbstoff absorption
coefficient is simply c ¼ a þ b. spectra tend to decrease monotonically with in-
The spectral volume scattering function, bðc; l) creasing wavelength and can be modeled. Phyto-
represents the scattered intensity of light per unit plankton spectral absorption varies significantly in
incident irradiance per unit volume of water at some relation to community composition and environ-
angle c (with respect to the exiting, nonscattered mental changes; however characteristic peaks are
incident beam) into solid angle element DO. Units for typically found near wavelengths of 440 nm and
bðc; l) are m1 sr1. This is essentially the differ- 683 nm. Much bio-optics research focuses on the
ential scattering cross-section per unit volume in the temporal and spatial variability of aph ðl). Ship-based
parlance of nuclear physics. The spectral scattering ocean water samples have often been used for studies
coefficient is obtained by integrating the volume of the IOPs, however, it is quite preferable to obtain
scattering function over all directions (solid angles). in situ measurements to insure representative local
The forward scattering coefficient, bf , is obtained by values as well as to characterize temporal and spatial
integrating over the forward-looking hemisphere variability, preferably with simultaneous physical,
(c ¼ 0 to p/2) and the backward scattering co- chemical, and biological measurements.
efficient, bb, is calculated by integrating the back-
looking hemisphere (c ¼ p=2 to p). The spectral
Instrumentation
volume scattering phase function is the ratio of
bðc; l) to bðlÞ with units of sr1. Other important Beam transmissometers have been the most com-
IOPs include spectral single-scattering albedo, monly used instruments for measuring IOPs with a
o0 ðlÞ ¼ bðlÞ=cðl), and fluorescence, which can be variety of applications ranging from determinations
considered a special case of scattering. (Note: fluor- of suspended sediment volume to phytoplankton
escence is not strictly an IOP; however it is often used biomass and productivity to particulate organic car-
as a proxy for chlorophylla.) The proportion of light bon (POC). The principle of operation involves the
which is scattered versus absorbed is characterized measurement of the proportion of an emitted beam,
by o0 ðl); that is, if scattering prevails then o0 ðl) C, which is lost through both absorption and scat-
approaches a value of 1 and if absorption dominates, tering as it passes to a detector through some path-
o0 ðl) approaches 0. length DL. The beam attenuation coefficient, c, is
IOPs obey simple mathematical operations. For then given as ½lnð1  CÞ=DL; or light intensity, I,
many applications, it is often convenient to partition is given as IðDLÞ ¼ Ið0Þ expðcDL). Similar ex-
the total absorption coefficient, at ðl), the total scat- pressions apply for absorption coefficient, a, and
tering coefficient, bt ðl), and the total beam attenu- scattering coefficient, b. Beam transmissometers have
ation coefficient, ct ðlÞ, in terms of contributing often used a red light emitting diode (660 nm) for the
constituents such that: collimated light source and pathlengths of 25 or
100 cm (long pathlengths are preferable for clearer
at ðlÞ ¼ aw ðlÞ þ aph ðlÞ þ ad ðlÞ þ ag ðlÞ ½1 waters). The wavelength of 660 nm was selected in
order to minimize attenuation by gelbstoff, which
attenuates strongly at shorter wavelengths but min-
imally in the red. One of the measurement compli-
bt ðlÞ ¼ bw ðlÞ þ bph ðlÞ þ bd ðlÞ þ bg ðlÞ ½2 cations for beam c is that most scattering in ocean
274 INHERENT OPTICAL PROPERTIES AND IRRADIANCE

waters is in the forward direction. The acceptance instrument developed in the 1970s were commonly
angle of the detector is thus made as small as possible used for many problems. The challenge arises be-
(o 1–21) to minimize underestimation of beam c. cause dominant scattering typically occurs at small
Corrections are done for this effect as well as for angles (roughly 50% in the forward direction be-
instrument temperature and pressure effects. tween 0 and 2–61) making it difficult to sense the
In situ multispectral attenuation–absorption (ac- weak scattering light, which is near the intense illu-
meters) instruments have recently become com- minating beam. Ideally, measurements should be
mercially available. These concurrently measure done at several angles to better resolve the function.
spectral absorption and attenuation coefficients, However, the backscatter signal, bb, can be estimated
which are spectral signatures of both particulate and with a few measurement angles provided the form of
dissolved material. These devices use dual light the volume scattering function is relatively well
sources and interference filters on a single rotating known. Recently developed volume scattering and
wheel (nine wavelengths from 400 to 715 nm with backscatter instruments take advantage of this
10 nm bandwidth). Whereas earlier beam trans- function. An important consideration is that back-
missometers were open to sea water, the newer ac- scattering is related to the size distribution, shapes,
meters use two tubes through which sea water is and composition of the particles sampled (e.g.,
pumped. The inside of the c-tube is flat black to coastal versus open ocean particles). Thus, this type
minimize reflections whereas the a-tube is reflective of information can be obtained in principle. Radia-
(shiny) in order to maximize internal reflection to tive transfer models need both absorption and vol-
better estimate absorption. The deployment of these ume scattering function information to estimate the
meters requires minimal lateral and torsional stress- propagation of light. Also, an important remote
ing in mounting (sensitivity of beam alignment) and sensing parameter, remote sensing reflectance (dis-
good flow through the tubes. Calibrations involve cussed below) is related to bb and a. Key wavelengths
temperature, salinity, and pure water measurements. are selected for the spectral measurements on the
A correction is also needed to account for the fact basis of the application and water types (e.g., coastal
that not all scattered light is collected (causing a versus open ocean). Particle size distributions can
biased overestimate of absorption). also be measured using laser (Frauenhofer) dif-
Similar, though more capable instruments, have fraction instruments. These devices employ charged
recently been developed to measure a and c at 100 coupled device (CCD) array photodetectors (linear
wavelengths from 400 to 726 nm wavelengths with or circular ring geometries). Modified versions of
3.3 nm resolution. A single white light source is used some of these instruments can also measure particle
with fiberoptics to provide light to each of the two settling velocities. Particles in the range of 5–500 mm
tubes (for a and b) as well as for a reference path (for can be measured with resolution dependent on the
correcting changes in lamp output). Each light path number of individual detector rings. These devices
has its own spectrometer (using a 256 pixel photo- can also measure beam transmission by sensing the
diode array). Biofouling is an issue for all optical undeviated light.
instruments. Closed (pumped) systems are less sus- Irradiance is the radiant flux per unit surface area
ceptible to fouling as they are primarily in the dark. (units of W m2 or quanta (or photons) s1 m2 or
A variety of special antifouling methods have been mol quanta (or photons) s1 m2. Note that 1 mol of
attempted with mixed success. These have included photons is 6.02  1023 (Avogadro’s number) and that
copper screens, copper flow tubes, bromide solu- one mole of photons is commonly called an einstein.
tions, and other cleaning agents. Spectral scattering It is convenient to define downwelling irradiance, Ed,
coefficient can be computed from ac-meter data by as the irradiance of a downwelling light stream im-
simply performing the difference b ¼ c  a. Spectral pinging on the top face of a horizontal surface (e.g.,
a  c meter data and relevant spectral decomposition ideally flat light collector oriented perpendicular to
models can be used to provide in situ estimates of the local gravity vector) and upwelling irradiance,
aph ðl) and other components (e.g., gelbstoff) for a Eu, as the irradiance of an upwelling light stream
broad range of environmental conditions as a func- impinging on the bottom face of a horizontal surface.
tion of time at fixed depths using moorings or as a Irradiance reflectance, R, is Eu/Ed. Irradiance meas-
function of depth from ships or profilers deployed urements are fundamentally important for quantify-
from moorings. ing the amount of light available for photosynthesis
The volume scattering function is one of the more and for radiative transfer theory and computations.
important and challenging measurements of optical Another useful radiometric variable is radiance,
oceanography. Until recently, few instruments exis- L ¼ Lðy; FÞ, which is defined as the radiant flux at a
ted for this measurement and data collected with an specified point in a given direction per unit solid
 INHERENT OPTICAL PROPERTIES AND IRRADIANCE 275

angle per unit area perpendicular to the direction between IOPs and AOPs break down. It is important
of light propagation (units of W (or quanta s1) to mention that spectral radiometric measurements
m2 sr1). Zenith angle, y, is the angle between a of AOPs have been far more commonly made than
vertical line perpendicular to a flat plane and an in- measurements of IOPs.
cident light beam, and azimuthal angle, F, is the The relationships between IOPs and AOPs are
angle with respect to a reference line in the plane of central to developing quantitative models of spectral
the flat plate. By integrating Lðy; FÞcosy over the irradiance in the ocean. Radiative transfer theory is
solid angle ðdo ¼ siny dy dFÞ, of the upper hemi- used to provide a mathematical formalism to link
sphere (say of a flat plate from y ¼ 0 to p/2 and F IOPs and the conditions of the water environment
from 0 to 2p), one obtains Ed. Similarly, upwelling and light forcing to the AOPs of the water column.
irradiance, Eu, is calculated by integrating over the One motivation of this research is the desire to pre-
bottom hemisphere. Net downward irradiance is the dict AOPs given environmental forcing conditions
difference Ed  Eu, or the integral of Lðy; FÞcosy over and estimates or actual measurements of IOPs. An-
the entire sphere (full solid angle 4p). Scalar irradi- other related, inverse goal is to be able to estimate or
ance, E0, is defined as the integral of Lðy; FÞ over the determine the vertical (and ideally horizontal)
entire sphere. It should be noted that all definitions structure of IOPs and their temporal variability given
are for a specific wavelength of light. If one integrates normalized water-leaving radiance, Lwn ðlÞ, meas-
scalar irradiance over the visible wavelengths ured remotely from satellite or airplane color ima-
(roughly 400–700 nm; note that the visible range is gers. The extrapolation of subsurface values of LðlÞ
sometimes defined as 350–700 nm), then the bio- to the surface has been the subject of considerable
logically important quantity called photosynthetic- research, because a major requirement of ocean color
ally available radiation (PAR) is obtained. remote sensing is to match in situ determinations of
Apparent optical properties (AOPs) depend on those from satellite- or plane-based spectral radi-
both the IOPs and the angular distribution of solar ometers which necessarily must account for the ef-
radiation (i.e., the geometry of the subsurface am- fects of clouds and aerosols. As mentioned above, the
bient light field). To a reasonable approximation, the important remote sensing parameter, spectral radi-
attenuation of spectral downwelling incident solar ance reflectance, Lu ðlÞ=Ed ðl), has been found to be
irradiance, Ed ðl; zÞ, can be described as an ex- proportional bb/(a þ bb) or bb/a from Monte Carlo
ponential function of depth: calculations for waters characterized by b/a values
ranging from 1.0 to 5.0. The spectral radiance re-
Ed ðl; zÞ ¼ Ed ðl; 0 Þexp½Kd ðl; zÞz ½4 flectance evaluated just above the ocean surface is
defined as the remote sensing reflectance Rrs ðl). It
where z is the vertical coordinate (positive down- should also be noted that remote sensing estimates of
ward), Ed ðl; 0 Þ is the value of Ed just below the air– pigment (typically chlorophyll a or chlorophyll
sea interface, and Kd ðl; zÞ is the spectral diffuse at- a þ phaeopigments) concentrations use empirical re-
tenuation coefficient of downwelling irradiance. lations involving ratios such as Lwn(443 nm)/
Kd ðl) (here depth dependence notation is suppressed Lwn(550 nm) or Rrs(490 nm)/Rrs(555 nm) or these
for convenience) is one of the important AOPs. The ratios with other wavelength combinations.
contributions of the various constituents (e.g., water, One of the most commonly used instruments for
phytoplankton, detritus, and gelbstoff) to Kd ðl) are measuring AOPs is a broadband scalar irradiance E0
often represented in analogy to the absorption co- or PAR sensor, because of the need for determining
efficients given in eqn [1][1]. This leads to the term the availability of light for phytoplankton and its
‘quasi-inherent’ optical property as it is suggestive relative simplicity. A spherical light collector made of
that inherent optical properties, i.e., like at ðl), are diffusing plastic receives the light from approxi-
closely related to Kd ðlÞ, which is often described as a mately 4p sr. The light is transmitted via a fiberoptic
quasi-inherent optical property. However, a key connector or quartz light conducting rod to a pho-
point is that Kd ðlÞ is in fact dependent on the am- todetector which records an output voltage. The in-
bient light field. Analogous spectral attenuation co- strument is calibrated with a standard lamp.
efficients are defined for diffuse upwelling irradiance, Analogous sensors use flat plate cosine or hemi-
downwelling radiance, and upwelling radiance, spherical collectors. Spectroradiometers are irradi-
KL ðlÞ. It should be noted that for periods of low sun ance or scalar irradiance meters which use a variable
angle and/or when highly reflective organisms or monochromator placed between a light collector and
their products (e.g., coccolithophores and coccoliths) the photodetector. Light separation can be achieved
are present, then multiple scattering becomes in- using sets of interference filters (usually 10 nm in
creasingly more important and simple relations bandwidth) selected for particular purposes (e.g.,
276 INHERENT OPTICAL PROPERTIES AND IRRADIANCE

absorption peaks and hinge points for pigments). methods and data assimilation models for
Alternatively, higher spectral resolution (B3 nm predictions.
from 400 to 800 nm) can be achieved using grating
monochromators. Cosine (flat-plate), spherical, or
hemispherical collectors are used for several different
Optical Experiment Data Sets
measurements or calculations of the AOPs described Finally, a few examples of data, which have been
above (e.g., Ed, Lu, Kd, etc.). Irradiance cosine col- collected with some of the optical instruments
lectors are designed such that their responses to described above (Figure 1), are presented. An inter-
parallel radiant flux should be proportional to the disciplinary experiment devoted to the understand-
angle between the normal to the collector surface ing of relations between mixing processes and optical
and the direction of the radiant flux. This is neces- variability was conducted south of Cape Cod, Mas-
sary as the angle between the incoming radiant flux sachusetts on the continental shelf. Several spectral
and the collector is variable, as is the area of the optical instruments collected IOP and AOP data sets
projection. To account for the differences in re- using ship-based profiling and towed systems,
fractive indices of air and water, an immersion cor- moorings, and bottom tripods. The period of
rection factor (typically 1.3–1.4) is applied. It should the experiment covered almost one year (July 1996–
be noted that radiance sensors are designed to accept June 1997). Some instruments sampled at several
light in a small solid angle (typically a viewing angle minute to hourly intervals whereas others sampled
of a few degrees). Irradiance and radiance instru- with vertical resolution on the scale of centimeters
ments are calibrated using well-characterized, for a few weeks during two intensive field cam-
standard light sources. Because of the growing con- paigns. Several interesting observations were en-
cern about UV radiation, a number of instruments abled. These included: sediment resuspension forced
are designed to measure into the UV portion of the by two passing hurricanes (hurricane Edouard and
electromagnetic spectrum as well. hurricane Hortense), spring and fall phytoplankton
A grand challenge of oceanography is to greatly blooms, water mass intrusions, and internal solitary
increase the variety and quantity of ocean measure- waves. Time series data collected using the BIOP
ments. Optical and other ocean measurements are system (e.g. BIOPS and MORS in Figure 1) which
expensive. Thus, a major goal is to develop new was deployed from a mooring and a bottom tripod
sensors and systems, which can be efficiently de- located near each other in 70 m waters, are shown in
ployed from a host of available ocean platforms in- Figure 2. Figure 2(A) shows the 37-m time series of
cluding ships, moorings, drifters, floats, and total spectral absorption (water absorption has been
autonomous underwater vehicles (AUVs). A further subtracted) using a nine-wavelength ac-meter. The
need is to telemeter the data in near real-time. Ship- major feature is related to hurricane Edouard. The
based observations are useful for detailed profiling spectral absorption contributions due to phyto-
(high vertical resolution), but moorings are better plankton are shown for two days in the summer in
suited for high temporal resolution, long-term Figure 2(B); the peaks at 440 and 683 nm are caused
measurements. Drifters, floats, and AUVs can pro- by phytoplankton. Figure 2(C, D) shows ac-meter
vide horizontal coverage unattainable from the other time series of spectral (nine wavelengths) scattering
in situ platforms. Ultimately, all of these platforms, and attenuation coefficients. The record is dominated
along with satellite- and plane-based systems, are by sediment resuspension caused by mixing and
needed to fill in the time–space continuum. Several waves created by hurricane Edouard and the more
optical systems, which have been used for mooring- distant hurricane Hortense. The complete mooring
based time series studies, are illustrated in Figure 1. data set showed that sediments were lifted more than
These collective instrument systems include most of 30 m above the ocean floor.
the sensors described earlier. The sampling for these The second example highlights optical data col-
instruments is typically done every few minutes (in lected from a ship-based profiling system (similar to
some cases once per hour) for periods of several the system shown in Figure 1A). The purpose of the
months. At this point in time, biofouling, rather than measurements was to study the dispersion of con-
power or data storage, is the limiting factor. How- taminants (treated wastewaters) discharged about
ever, even this aspect is becoming less problematic 4 km offshore (70 m water depth) into Mamala
with several new antibiofouling methods. In add- Bay, off the coast of Honolulu, Hawaii in the fall of
ition, more optical sensors are being deployed from 1994. Optical and physical measurements were
other emerging platforms such as AUVs. The optimal made from shipboard using both profile and towing
utilization of optical data from the various platforms modes in the vicinity of the outfall plume. Optical
will require the use of advanced data merging instrumentation included a beam transmissometer
 INHERENT OPTICAL PROPERTIES AND IRRADIANCE 277

(660 nm), spectral ac-meters (nine wavelengths for a shallow layer versus deep layer phytoplankton, and
and c), PAR sensor, chlorophyll fluorometer, a par- old versus newly discharged sewage plume waters.
ticle size analyzer (laser diffraction method), and a Briefly, a profile (see Figure 3) taken near the end of
spectral absorption and fluorescence instrument. the outfall diffuser (water depth B 80 m) displayed the
Physical measurements of temperature, salinity and following features: (1) sewage plume waters centered
pressure were also done. Sampling was designed to near 60 m as characterized by low salinity and very
track both the horizontal and vertical structure of high values of spectral attenuation and absorption
effluent as manifest in the optical signals. The water (greater in the shorter wavelengths), and (2) a shallow
contributions to a, b, and c were removed for the phytoplankton layer near 20 m with modest relative
analyses. The collective optical and physical meas- maxima in chlorophyll fluorescence and spectral
urements enabled the partitioning of particle types into absorption and attenuation coefficients in the blue.
categories: particulate versus dissolved components, The profile of spectral single-scattering albedo,
phytoplankton opposed to detrital components, o0 ðlÞ ¼ bðlÞ=cðlÞ, shows a general trend of increased

(A) (B)
Par sensor PAR
sensor

Pump
Transmissometer
HiSTAR
Pump
9  Wetlabs Hydroscat-6
ac-9 Temperature
Temperature sensor
sensor Pressure sensor

BIOPS HiROS
(C) PAR
sensor
6  Ex, 16  Em
Wetlabs
safire
Pump

Temperature
sensor

(D) SAFIRE (E)

7  Ed's
Anti-fouling shuttered
PAR sensor
spectral radiometers
Data
Spectral 3  Ed's
logger
radiometers
UCSB data
Acoustic telemetry logger/
module Battery pack
Temperature
3  Lu's
sensor
7  Lu's
Fluorometer & VSF
sensors
MORS MOSS

Figure 1 Schematic showing a variety of optical systems for mooring applications. (A) System used for measuring inherent optical
properties (IOPs) including beam c (660 nm), spectral attenuation and absorption coefficients at nine wavelengths, along with PAR
and temperature. (B) System used for measuring inherent optical properties including spectral attenuation and absorption coefficients
at 100 wavelengths, spectral backscatter at six wavelengths, along with PAR, temperature, and pressure. (C) System measuring
spectral fluorescence with 6 excitation wavelengths and 16 emission wavelengths; also PAR and temperature sensors are included.
(D) System for measuring apparent optical properties (AOPs) including spectral downwelling irradiance and upwelling radiance at
seven wavelengths and PAR along with temperature. A telemetry module is also included. (E) System for measuring apparent optical
properties (AOPs) including spectral downwelling irradiance and upwelling radiance at three wavelengths along with instruments for
measuring chlorophyll fluorescence, volume scattering function. An antifouling shutter system is also utilized.
278 INHERENT OPTICAL PROPERTIES AND IRRADIANCE

37 m 0.10
o = YD 239
0.3 E 0.08 + = YD 204

aph () (m 1)
_
0.06
a t_w() (m 1)

0.2
_

0.04

0.1
0.02
715
0 650 0
412 510 650 715
190
(n m)
210 532 th (B) Wavelength (nm)
230
488 le ng
Year da ve
(A)
y (1996) 250 412 Wa

25

68 m
E
20
H
b t_w() (m 1)

15
_

10
 (nm)

5 412
440
488
(C) 0
30 510
532
68 m E
25 555
H 650
676
20
715
c t_w() (m 1)
_

15

10

0
235 240 245 250 255 260 265
(D) Year day (1996)

Figure 2 Illustrations showing data collected from the BIOPS system shown in Figure 1(A) on the continental shelf south of Cape
Cod Massachusetts in the summer and fall of 1996. (A) A time series of the total spectral absorption coefficient of light (after
subtracting the clear water component) at a depth of 37 m (total water depth is 70 m). (B) The spectral absorption on YD 204 and
239. (C) Time series of spectral scattering coefficient computed by differencing the total spectral attenuation and absorption
coefficients (again, the clear water coefficient has been subtracted) at 68 m. The large peaks are attributed to passages of hurricanes
Edouard (E) and Hortense (H). (D) as for (C) except for spectral beam attenuation coefficients. (Figures based on Chang and Dickey,
1999.)
 INHERENT OPTICAL PROPERTIES AND IRRADIANCE 279

0.15

a () (m 1)
_
0.10

0.05

0
10 20 700 750
30 40 50 550 600 650
60 500 m)
Depth 70 80 400 450 ngth (n
(m) Wavele
(A)

0.8
c () (m 1)

0.6
_

0.4
0.2
0
10 20 650 700 750
30 40 50
500 550 600
60 m)
Depth 70 80 400 450 ngth (n
(B) (m) Wavele

1.2
1.0
0

0.8
0.6
0.4
10 20 700 750
30 40 50 550 600 650
60 500 m)
Depth 70 80 400 450 ngth (n
(C) (m) Wavele

Figure 3 Data collected near an ocean sewage outfall in Mamala Bay off Honolulu, Hawaii in the fall of 1994. Vertical profiles of (A)
spectral absorption coefficient, (B) spectral attenuation coefficient, and (C) single scattering albedo. Data were obtained from an ac-
meter system similar to the one shown in Figure 1(A). Eight wavelengths are displayed and the major peak near 60 m is caused by
sewage plume waters with high scattering and absorption coefficents. (Figures based on Petrenko et al., 1997.)

importance of scattering for greater wavelengths with at wavelengths, which are coincident with those of the
the most pronounced effect for the sewage plume SeaWiFS ocean color satellite, and sample at hourly
waters centered at 60 m (Figure 3C). intervals. Two or more systems are deployed at dif-
The final example describes an important linkage ferent depths and another radiometer system is de-
between in situ and remote sensing observations of ployed from the surface buoy to collect incident
ocean color. The Bermuda Testbed Mooring (BTM) is spectral downwelling irradiance. Derived quantities
used to test new oceanographic instrumentation (in- include time series of several of the AOP quantities
cluding systems shown in Figure 1), for scientific introduced earlier (e.g., spectral diffuse attenuation
studies devoted to biogeochemical cycling and climate coefficients, spectral reflectances including remote
change, and for groundtruthing (verification using in sensing reflectance, and water-leaving radiance). Time
situ data) and algorithm development for ocean color series of spectral upwelled radiance from the BTM
satellites such as SeaWiFS. The latter aspect is the radiometer system (Figure 1D) located at 14 m are
focus here. Two BTM optical systems utilizing radi- shown for the period April through July, 1999 in the
ometers for measurements of spectral downwelling top set of panels of Figure 4. The variability is pri-
irradiance and spectral upwelling radiance are illus- marily caused by the daily cycle of solar insolation,
trated in Figure 1(D) and (E). These systems measure cloud cover, and optical properties associated with
280 INHERENT OPTICAL PROPERTIES AND IRRADIANCE

3
412 nm BTM 14 m Radiance (Lu)
2
1
0
3
443 nm
Radiance (W cm 2 nm 1 sr 1) 2
_

1
0
_

3
490 nm
_

2
1
0
3
510 nm
2
1
0
0.3
555 nm
0.2
0.1
0
March 31 April 22 May 14 June 5 June 27 July 21
Day (1999)
L w (W cm 2 nm 1 sr 1) L w (W cm 2 nm 1 sr 1)
_

1.6 April 21 1.6 1.6

April 26 May 10
_

1.2 1.2 1.2

0.8 0.8 0.8
_

0.4 0.4 0.4

0 0 0
400 450 500 550 600 400 450 500 550 600 400 450 500 550 600 SeaWiFs
BTM
_

1.6 July 17 1.6 July 19 1.6 July 20

_

1.2 1.2 1.2

0.8 0.8 0.8
_

0.4 0.4 0.4

0 0 0
400 450 500 550 600 400 450 500 550 600 400 450 500 550 600
Wavelength Wavelength Wavelength

Figure 4 BTM radiometer data collected with the system shown in Figure 1(D). The top set of panels shows a time series of spectral
upwelling radiance (five wavelengths) from the 14-m spectral radiometer system (April–July, 1999). The bottom set of panels shows
spectral water-leaving radiance derived from the BTM radiometers and from the SeaWiFS ocean color satellite for six days of the
sampling period.

phytoplankton and their products. The bottom panels Toward the Future
of Figure 4 show a subset of the water-leaving radi-
ance data as determined from the BTM radiometers The last two decades have been marked by the
(extrapolation to surface) and SeaWiFS satellite sen- emergence of a host of new optical instruments and
sors for six days during the particular sampling per- systems; their applications are growing at a rapid
iod. These two types of data are critical for rate. Looking toward the future, it is anticipated that
quantifying both the temporal and horizontal spatial optical instrumentation will be (1) more capable in
variability of water clarity, PAR, phytoplankton bio- spectral resolution, (2) smaller and require less
mass, and primary productivity. power, (3) suitable for deployment from a variety of
 INHERENT OPTICAL PROPERTIES AND IRRADIANCE 281

autonomous platforms, (4) designed for ease in tel- Dickey T, Frye D, Jannasch H, et al. (1998) Initial results
emetering of data for real-time applications, and (5) from the Bermuda Testbed Mooring Program. Deep-Sea
less costly as demand increases enabling higher vol- Research I 45: 771--794.
umes of data collection. Dickey T, Marra J, Weller R, et al. (1998) Time-series of
bio-optical and physical properties in the Arabian Sea:
October 1994–October 1995. Deep-Sea Research II 45:
2001--2025.
See also
Dickey TD, Chang GC, Agrawal YC, Williams AJ 3rd, and
Absorbance Spectroscopy for Chemical Sensors. Hill PS (1998) Sediment resuspension in the wakes of
IR Radiometers. Optical Particle Characterization. Hurricanes Edouard and Hortense. Geophysical
Satellite Remote Sensing Microwave Research Letters 25: 3533--3536.
Scatterometers. Dickey T, Zedler S, Frye D, et al. (2001) Physical and
biogeochemical variability from hours to years at the
Bermuda Testbed mooring site: June 1994–March
Further Reading 1998. Deep-Sea Research.
Foley D, Dickey T, McPhaden M, et al. (1998) Longwaves
Agrawal YC and Pottsmith HC (1994) Laser diffraction and primary productivity variations in the equatorial
particle sizing in STRESS. Continental Shelf Research Pacific at 01, 1401W February 1992–March 1993.
14: 1101--1121. Deep-Sea Research II 44: 1801--1826.
Bartz R, Zaneveld JRV, and Pak H (1978) A Gentien P, Lunven M, Lehaitre M, and Duvent JL (1995)
transmissometer for profiling and moored observations In situ depth profiling of particle sizes. Deep-Sea
in water. Ocean Optics V, SPIE 160: 102--108. Research 42: 1297--1312.
Booth CR (1976) The design and evaluation of a Griffiths G, Knap A, and Dickey T (1999) Autosub
measurement system for photsynthetically active quantum experiment near Bermuda. Sea Technology, December.
scalar irradiance. Limnology and Oceanography 19: Jerlov NG (1976) Marine Optics. Amsterdam: Elsevier.
326--335. Kirk JTO (1994) Light and Photosynthesis in Aquatic
Chang GC and Dickey TD (1999) Partitioning in situ total Ecosystems. Cambridge: Cambridge University Press.
spectral absorption by use of moored spectral absorption Mobley CD (1994) Light and Water: Radiative Transfer in
and attenuation meters. Applied Optics 38: 3876--3887. Natural Waters. San Diego: Academic Press.
Chang GC and Dickey TD (2001) Optical and physical Moore CC, Zaneveld JRV, and Kitchen JC (1992)
variability on time-scales from minutes to the seasonal Preliminary results from in situ spectral absorption
cycle on the New England continental shelf: July 1996– meter data. Ocean Optics XI, SPIE 1750: 330--337.
June 1997. Journal of Geophysical Research 106: O’Reilly JE, Maritorena S, Mitchell BG, et al. (1998)
9435--9453. Ocean color chlorophyll algorithms for SeaWiFS.
Chang GC, Dickey TD, and Williams AJ 3rd (2001) Journal of Geophysical Research 103: 24937--24953.
Sediment resuspension over a continental shelf during Petrenko AA, Jones BH, Dickey TD, Le Haitre M, and
Hurricanes Edouard and Hortense. Journal of Moore C (1997) Effects of a sewage plume on the
Geophysical Research 106: 9517--9531. biology, optical characteristics, and particle size
Dana DR, Maffione RA, and Coenen PE (1998) A new in distributions of coastal waters. Journal of Geophysical
situ instrument for measuring the backward scattering Research 102: 25 061--25 071.
and absorption coefficients simultaneously. Ocean Petrenko AA, Jones BH, and Dickey TD (1998) Shape and
Optics XIV 1: 1--8. near-field dilution of the Sand Island sewage plume:
Dickey T (1991) Concurrent high resolution physical and observations compared to model results. Journal of
bio-optical measurements in the upper ocean and their Hydraulic Engineering 124: 565--571.
applications. Reviews in Geophysics 29: 383--413. Petzold TJ (1972) Volume scattering functions for selected
Dickey T, Granata T, Marra J, et al. (1993) Seasonal waters, Scripps Institution of Oceanography Reference
variability of bio-optical and physical properties in the 72–78. La Jolla, California: Scripps Institution of
Sargasso Sea. Journal of Geophysical Research 98: Oceanography.
865--898. Smith RC, Booth CR, and Star JL (1984) Oceanographic
Dickey T, Marra J, Stramska M, et al. (1994) Bio-optical bio-optical profiling system. Applied Optics 23:
and physical variability in the sub-arctic North Atlantic 2791--2797.
Ocean during the spring of 1989. Journal of Spinrad RW, Carder KL, and Perry MJ (1994) Ocean
Geophysical Research 99: 22541--22556. Optics. Oxford: Oxford University Press.
 NEPHELOID LAYERS
I. N. McCave, University of Cambridge, which may be internal such as ridges and sea-
Cambridge, UK mounts). Most concentrated nepheloid layers occur
& 2009 Elsevier Ltd. All rights reserved. on the continental shelf, upper slope, or deep con-
tinental margin. They indicate the locus of active
resuspension and redeposition by strong bottom
currents and internal waves.

Introduction
Optics of Nephelometers: What They
A remarkable feature of the lower water column in
most deep parts of the World Ocean is a large increase
‘See’
in light scattering and attenuation conferred by the Detection of deep-ocean nepheloid layers has been
presence of increased amounts of particulate material. mainly through measurement of light scattering. The
This part of the water column is termed the bottom Lamont nephelometer has made the largest number of
nepheloid layer (BNL). Another class of nepheloid profiles in all oceans but is no longer in use. It used an
layers found especially at continental margins are incandescent bulb as the source and photographic film
intermediate nepheloid layers (INLs) (Figures 1 and 2). as the detector of the light scattered from angles be-
These occur frequently at high levels off the upper tween y ¼ 81 and 241 from the forward axis of the
continental slope and at the depth of the shelf edge. light beam. The film was continuously wound on as
From here, they spread out across the continental the instrument was lowered, resulting in an averaging
margin. These INLs are similar to the inversions ob- of the received signal over about 25-m depth. The
served in the BNL on some profiles (Figure 1). (The short-lived Geochemical Ocean Section Study (GEO-
surface nepheloid layer (SNL), not treated here, is SECS) nephelometer used a red (l ¼ 633 nm) laser
simply the upper ocean layer in which particles are source and a photoelectric cell to detect light scattered
produced by biological activity, and which may have from y ¼ 3–151 off the axis of the beam. The SeaTech
material from river plumes close to shore.) (now WetLabs, Inc.) LSS measures infrared light
The increase in light scattering is perceived relative (880 nm) backscattered (1801) from particles in the
to minimum values found at mid-water depths of sample volume using a solar-blind silicon detector.
2000–4000 m (shallower on continental margins). Because of multiple scattering, nephelometers do
The increased scattering is due to fine particles. This not yield precise optical parameters. Optical trans-
has been determined by particle-size measurements mission with a narrow beam can yield the attenu-
and filtration of seawater with determination of con- ation coefficient (c). The most commonly used
centration by weight and volume. Most data on the SeaTech and WetLabs transmissometers have a red
distribution and character of nepheloid layers have light source (l ¼ 660 or 670 nm) and usually a 0.20–
been acquired by optical techniques, principally by the 0.25-m path length.
Lamont photographic nephelometer and the SeaTech Most of the contribution to the total scattering b
transmissometer, and more recently the WetLabs comes from near-forward angles (low values of y).
transmissometer and light scattering sensor (LSS). Jerlov (1976) shows that, for surface waters, 47% of
The optical work has revealed that the BNL is up b occurs between y ¼ 01 and 31, 79% between 01 and
to 2000-m thick (can be more in trenches) and gen- 151, and 90% between 01 and 301. The GEOSECS
erally has a basal uniform region, the bottom mixed instrument records about 32% of b and the Lamont
nepheloid layer (BMNL), corresponding quite closely nephelometer about 16%. The total scattering is
to the bottom mixed layer defined by uniform po- given by Mie theory (assumed spherical particles) as
tential temperature (Figure 1). Above the BMNL a function of particle size d and relative refractive
there is a more or less exponential fall-off in intensity index (relative RI) n, and wavelength of light l. The
of light scattering up to the clear-water minimum relative indices of refraction of suspended material
marking the top of the BNL. are dominated by components with n ¼ 1.05 and
Both bottom and INLs are principally produced by 1.15, values probably characteristic of organic and
resuspension of bottom sediments. Their distribution mineral matter, respectively (e.g., RI of seawater
indicates the dispersal of resuspended sediment in the 1.34, quartz 1.55, ratio n ¼ 1.15).
ocean basins and is thus a signature of both material Particles from clear ocean waters and weak
and water transport away from boundaries (some of nepheloid layers (concentration Co40 mg m3) tend

282
 NEPHELOID LAYERS 283

c
1.0 0.8 0.6 0.4


1.80 1.85
41
24 300
4700

88
16
109

157

Meters above bottom

4800 200
179

215
Pressure (dbar)

234

252
4900 100

227

BMNL

5000
2 4 8 16 32 0
μm

Figure 1 Data from the SeaTech transmissometer in the Atlantic showing the BMNL and a nepheloid layer comprising multiple steps
in temperature and turbidity. Turbidity is given as c the attenuation coefficient, and y is potential temperature in 1C. Also shown on the
right are particle-size spectra determined by Coulter counter. Reproduced from McCave IN (1983) Particulate size spectra, behavior
and origin of nepheloid layers over the Nova Scotian Continental Rise. Journal of Geophysical Research 88: 7647–7666.

to have particle-size distributions by volume which Morel has calculated scattering according to Mie
are flat, equivalent to k ¼ 3 in a particle number dis- theory for suspensions with Junge distributions and
tribution of Junge type, N ¼ Kd  k where N is the several indices of refraction. Recalculation into cu-
cumulative number of particles larger than diameter d mulative curves of percentage scattering in Figure 3
and K and k are constants. However, this distribution illustrates the fact that most recorded scattering is
does not appear to be maintained at sizes finer than produced by fine particles. The cases shown are for
about 2 mm where k decreases toward 1.5. In con- values of k of 2.1, 3.2, and 4.0 and a two-component
centrated nepheloid layers, this distribution does not (peaked) distribution with k ¼ 2.1 up to a ¼ pdn/
occur at all and a peaked distribution with a peak l ¼ 32 and k ¼ 4.0 for larger sizes (a ¼ 32 is equiva-
between 3 and 10 mm is encountered. lent to d ¼ 5.6 mm for l ¼ 633 nm). In each case, three
284 NEPHELOID LAYERS

(a)
0
Knorr 51 sta. 698
Rockall Trough
54° 23.1′ N, 15° 18.7′ W
INL.

1000
Depth (m)

2000
BNL
900m
{ INV 2
INV 1

Nephels, arbitrary scale

3000


(b) 2.6 2.8 3.0 3.2
2200
Nephel inversion Higher-salinity core
2


2300

NEADW

2400
Depth (m)

Nephel inversion
1
2500
Knorr 51
Station 698
Rockall Trough
54° 28.1′ N AABW
2600 15° 18.7′ W
24 Aug. 1975
Source
_ High silica_

2700
40 50 60 70 80 90 15 20 25 30 35
Ne Silica (μg at/l)

Figure 2 (a) Full-depth profile taken in the Rockall Trough with the GEOSECS nephelometer. An INL and two inversions are
apparent. (b) Detail of the lower 500 m of the profile in (a) showing the relationship between the nephel (turbidity) inversions and
hydrography. Reproduced from McCave IN (1986) Local and global aspects of the bottom nepheloid layers in the world ocean.
Netherlands Journal of Sea Research 20: 167–181.

forward-scattering angles are given. It is clear that smaller sizes (ao32) at y ¼ 20 1. Thus the curves for
scattering close to the beam is more sensitive to large y ¼ 10 1 and 20 1 are generally representative, and the
particles than that at 201. (At yo0.51 we are essen- distribution for both k ¼ 3.2 and the composite case
tially dealing with a transmissometer.) In the case of show that 95% of the scattering is by particles
k ¼ 2.1 only about 22% of the scattering is from o5 mm for l ¼ 633 nm.
 NEPHELOID LAYERS 285

16 32 64 128 256 4 8 16 32
100
k = 3.2
k = 2.1

32 32
Cumulative percent scattering

32
20° k = 4.0
10° 2°
20° 20° 2°
2° 10°
2°
50
32 20° k = 2.1 for
< 32
10°
k = 4.0 for
10° < 32

0
1 2 4 8 16 32 64 128 256 0.5 1 2 4 8 1 2 4 8 16 32
= Πdnw / 

Figure 3 Cumulative percentage of scattering calculated from the data of Morel (1973) by McCave (1986). The right-hand case is for
a peaked distribution with the peak at a ¼ 32, equivalent to 5.6 mm for l ¼ 633 nm and n ¼ 1.15. Note that material larger than the peak
contributes virtually nothing to the scattering in this case. Reproduced from McCave IN (1986) Local and global aspects of the bottom
nepheloid layers in the world ocean. Netherlands Journal of Sea Research 20: 167–181.

So it is dominantly the fine fraction of particles in also found; steeper gradients in concentration as well
nepheloid layers that is seen and recorded by neph- as inversions are often found at the boundaries of
elometers. These particles have very low settling vel- distinct density (temperature and/or salinity) changes,
ocities, less than B5  106 m s1. Although larger but their frequency decreases upward.
particles are present, they are rare and their properties The thickness of the BNL is generally in the region of
and behavior cannot be invoked to explain features of 500–1500 m, and exceptionally up to 2000 m. This is
distribution shown by nephelometers. clearly greater than the thickness of the bottom mixed
The SeaTech 0.25 m path length transmissometer layer (Figures 2 and 4), a fact which rules out the
has been used for most of the modern work on the possibility of simple mixing by boundary turbulence
structure and behavior of nepheloid layers. The being a sufficient mechanism for BNL generation.
transmission T is related to beam attenuation co- In several cases, the nepheloid layer is seen to
efficient c over path length l as T ¼ecl. The major transcend water masses. That is to say, the nepheloid
control of c is due to particles. Attenuation is due to layer shows a general decline in turbidity upward
absorption a and scattering b, thus c ¼ a þ b. The through interleaved water masses of differing sources
value of c for pure seawater is about 0.36 m1 for this and temperature/salinity characteristics though there
instrument operating at l ¼ 660 nm, and any excess is may be a steeper turbidity gradient at the boundary
due to particulate effects. Because scattering very close between water masses.
to the beam is more sensitive to large particles, the The highest suspended sediment concentrations
transmissometer is more sensitive to larger particles occur in the BMNL in regions of strong bottom
and the nephelometer to smaller ones (Figure 4). currents where they are typically 100–500 mg m3
(this is the same as mg l1). In general, deep western
boundary currents and regions of recirculation carry
high particulate loads. However, high turbidity is
Nepheloid Layer Features also found beneath regions of high surface eddy
The principal features of nepheloid layers that must kinetic energy (variance of current speed) when
be accounted for are the facts that the concentration is located over strong thermohaline bottom currents.
generally highest close to the bed, decreasing upward, High surface eddy kinetic energy is connected
but also that this is not universally the case, because: with high bottom eddy kinetic energy; thus, inter-
inversions, upward increases of concentration, are mittent variability, when added to a strong steady
286 NEPHELOID LAYERS

PMC (μg l )
−1 up to a kilometer above the bed. It is not possible to
0 50 100 150 200 250
mix sediment across sharp density steps without
0
breaking them down. However, these features are
explicable if the layers are recently separated from the
Surface nepheloid layer
200
Attenuation excess bottom. Some layers marked by steps in potential
Coarser/organic-
dominated particles
temperature contain excess radon-222 (originating
400 from bottom sediments) with a 3.8 day half-life,
suggesting detachment of bottom layers within 2–3
600 weeks before sampling. It is anticipated that with time
Shallow intermediate these layers become thinner by mixing at their
nepheloid layer
800 boundaries and by lateral spreading to yield, eventu-
ally, a uniform stratification. In this, the upper part of
the BNL, sheared-out mixed layers that have, on
Depth (m)

1000

t
average, come further from the sloping sides of the
Clear water (~10 μg l )
−1 basins and from regions with less frequent resuspen-
1200
Good agreement sion, have lower concentrations. The basal layers are
between instruments
on average more recently resuspended and also gain
1400
material by fall-out from above; thus, there is an
overall decreasing particulate concentration, and in-
1600
creasing age upward.
Deep intermediate nepheloid layer

1800
Bottom nepheloid layer
Scattering excess
Finer/inorganic dominated
Decay of Concentration: Aging of
2000
particles Particulate Populations
26.00 26.25 26.50 26.75 27.00 27.25 27.50 27.75 28.00
3
The particles composing the nepheloid layer may be

t (kg m )
modified due to aggregation with similar-sized par-
ticles and scavenging by larger rapidly settling ones.
Figure 4 Profiles of particulate matter concentration calculated
from beam attenuation (solid line), and light scattering (dotted Aggregation may also be caused through biological
line) against depth, together with the density structure (st) of the activity, although little is known about such pro-
water column (dashed line). Reproduced from Hall IR, Schmidt S, cesses at great depths. Particles tend to settle and to
McCave IN, and Reyss JL (2000) Particulate matter distribution be deposited onto the bed, from the bottom mixed
and Th-234/U-238 disequilibrium along the Northern Iberian
layer. The larger particles should be deposited in a
Margin: Implications for particulate organic carbon export. Deep-
Sea Research I 47: 557–582. few weeks to months, 10–20 mm particles taking 50–
20 days to settle from a 60-m-thick layer. This will
not affect the layer perceived by nephelometers so
component, may be responsible for very high current quickly because the timescale of fine particle removal
speeds which produce intense sediment resuspension. initially involves Brownian aggregation with a ‘half-
life’ of several months to years.
The direct rate of deposition of very fine particles
Separated Mixed-Layer Model
(0.5–1 mm) from a layer which remained in contact
Nepheloid layer structure is consistent with a quasi- with the bed would be very slow. Concentration
vertical transport mechanism involving turbulent would halve in about 8 years. Thus the rate of de-
mixing in bottom layers of B10–50-m thickness (see crease in concentration of 0.5–1-mm particles is due
), followed by their detachment and lateral advection more to their being moved to another part of the size
along isopycnal (equal density) surfaces The detach- spectrum by aggregation (and then deposited) than
ment occurs in areas of steep topography as well as in to their being deposited directly. The fine material in
areas of lower gradient at benthic fronts where slop- dilute nepheloid layers has a mean residence time
ing isopycnals intersect the bottom. In many nephe- measured in years, demonstrated by the residence
loid layers, there are sharp upward increases in time of particle-reactive short half-life radionuclides
sediment content associated with steps in other such as 210Pb (t1/2 ¼ 22.3 years). In more concen-
properties such as temperature and salinity (Figure 1). trated nepheloid layers, a large proportion of this
Both the step structure and inversions in particulate material will be removed in under a year, and in the
matter concentration are incompatible with vertical BMNL residence times are tens to a hundred or so
turbulent mixing (which occurs mainly in the BMNL) days estimated via 234Th (t1/2 ¼ 24.1 days). The
 NEPHELOID LAYERS 287

dilute nepheloid layers in tranquil parts of the oceans in concentration. Present understanding of particle
could thus contain material that was resuspended aggregation and sinking rates suggests that this takes a
very far away. The contribution of this material to few years to achieve.
the net sedimentation rate of these tranquil regions
may not be negligible. The rate of deposition in the
central South Pacific of only 0.5–2 mm ky1 could Chemical Scavenging by Particles in
include up to 1 mm ky1 of fine material from the
nepheloid layer.
Nepheloid Layers
With aging, the individual detached layers com- Many chemical species are particle-reactive and
prising the nepheloid layer lose material by aggre- rapidly become adsorbed onto surfaces. This is why a
gation and settling and lose their identity by being number of elements are present in only trace quan-
thinned through shearing. An originally discontinuous tities in seawater as outlined by Robert Anderson in
vertical profile of concentration with inversions is 2004. The phenomenon of ‘boundary scavenging’,
converted to one of relatively smooth upward decline preferential removal of particle active species at

40° E 60° 80° 100° 120°

20° N Mid-ocean and
aseismic ridges
and plateaus

0.2
E
2

˚
0.

Excess turbidity
[log(Eb /Ec)] values:
0°
0.2

<0.2
0.2
0.2_ 0.4
0.4

0.4_ 0.6

20° 0.6 _ 1.0

>1.0
0.2

0.6 0.2
2
0.

4
0.
2

40
0.
0.6

0.4

0.4
0.
2

0.6
0.6
0.4

1.0 1.0

1.0
0.8

0.6
0.6

0.4 0.
60° S 6
0.6
0.8 0.4

4
0. 0.6
0.4 0.8

Figure 5 Distribution of excess turbidity for the Indian Ocean expressed as log(E/Ec) where E is the maximum light scattering near
the bed and Ec is the value at the clear water minimum. A value of 1 thus represents a factor of 10 increase from the clear-water value.
Reproduced from McCave IN (1986) Local and global aspects of the bottom nepheloid layers in the world ocean. Netherlands Journal
of Sea Research 20: 167–181; based on Lamont nephelometer data presented by Kolla V, Sullivan L, Streeter SS, and Langseth MG
(1976) Spreading of Antarctic bottom water and its effects on the floor of the Indian Ocean inferred from bottom water potential
temperature, turbidity and sea-floor photography. Marine Geology 21: 171–189; and Kolla V, Henderson L, Sullivan L, and Biscaye PE
(1978) Recent sedimentation in the southeast Indian Ocean with special reference to the effects of Antarctic bottom Water circulation.
Marine Geology 27: 1–17.
288 NEPHELOID LAYERS

60 30 0 30

60 60

45
45

0
200 00
30 5
30
100
50

15 15

0 0
>2000 μg cm−2
500_ 2000
15
100 _ 500 15

50 _ 100
<50 μg cm−2
30 30

50
45 100 45

500

60 60
60 30 0 30

Figure 6 Distribution of excess suspended sediment load in the nepheloid layer. The load is computed for concentration excess over
the value of the clear water minimum and integrated over the height of the nepheloid layer. Reproduced from Biscaye PE and Eittreim
SL (1977) Suspended particulate loads and transports in the nepheloid layer of the abyssal Atlantic Ocean. Marine Geology 23: 155–
172.

continental margins, has been proposed to account of the material involved in this scavenging becomes
for the deficit of several elements in the open ocean. aggregated and falls to the bed on the continental
Short half-life radionuclides have shown that the slope and rise in the model of Amin and Huthnance.
particles in the BNL are responsible for trace element Turnewitsch and Springer observed that the most
scavenging, and that the age of the suspensions in the likely reason for the difference between measured
BMNL is rather short, a few tens of days. The fact and calculated thorium fluxes over the Porcupine
that the BMNL is almost always present means that Abyssal Plain (NE Atlantic) is lateral advection of
234
this intense scavenging activity is continuous. Most Th-depleted water, a loss likely due to particle
 NEPHELOID LAYERS 289

scavenging in the BNL on the nearby continental seaward. Tidal motion on the rough topography of
margin. The 234Tht (‘t’ means total, i.e., dissolved mid-ocean ridges is known to yield enhanced mixing
and particle-associated) profile showed a 234Tht/238U as explained by Kurt Polzin and colleagues in 1997,
disequilibrium (the difference between 234Tht and and is also probably responsible for resuspension.
238
U activity) up to at least 500 mab (meters above Between these zones are depths of 1000–3000 m
bottom), and that secular 234Tht/238U equilibrium where stratification and currents are weak, there is
was reached between 500 and 1000 mab. That is to no primary production of organic matter, and the
say radiochemical evidence suggests lateral advection ocean is often undersaturated with respect to calcite,
on the 100-day timescale (4–5 234Th half-lives) aragonite, and opal. There is thus little particle
below 500 mab at a site quite close to the continental supply from the side, and a decrease in downward
margin (about 250 km away). Above that, the time- transport due to dissolution and bacterial con-
scale is longer. sumption of CaCO3, SiO2, and organic carbon,
leading to a nepheloid minimum where (in the At-
lantic) concentrations are 5–20 mg m3.
The Turbidity Minimum The concentrations given by nephelometers at the
The source of nepheloid layers at heights over a few clear-water minimum differ by a factor of about 3
hundred meters above the bed is believed to be the between areas under high surface productivity (high)
sides of the ocean basins and protrusions from the and mid-gyre regions (low). The flux of large bio-
bottom of the oceans such as seamounts, ridges, and genic particles from the surface in these areas ap-
rises. Major source regions are (1) deep zones of pears to provide a net supply of smaller particles to
stronger currents, which in most areas are at depths mid-depth (rather than scavenging them). This is also
greater than 3000 m; and (2) shallow areas where seen to be the case in temporal variability at a point,
material resuspended at the shelf edge and upper such that higher mid-water turbidity occurs in sum-
slope by surface and internal wave motions, spreads mer under higher productivity, and is lower in winter.

71° W 59° W
22° N 17° N
2.2°

4
 = 2.0°

1.8°
5
1.6°
Depth (km)

Caicos
outer ridge
6
log E/ED
0.7_ 0.8
Puerto Rico
trench axis

7 0.6_ 0.7 1.4°

South slope of Puerto Rico trench

0.5_ 0.6

8 0.4

Figure 7 Interflow of suspended sediment generated in the Western Boundary Undercurrent, having detached from the bed on
Caicos drift, flowing over the Puerto Rico trench. Reproduced from Tucholke BE and Eittreim SL (1974) The western boundary
undercurrent as a turbidity maximum over the Puerto Rico Trench. Journal of Geophysical Research 79: 4115–4118.
290 NEPHELOID LAYERS

Concentration and Spreading in the Not all intermediate layers detach on slopes. Per-
Atlantic and Indian Oceans haps the biggest INL in the ocean occurs where the
turbid plume of the North Atlantic Western Bound-
The broad picture of high-speed inflows on the ary Undercurrent leaves the bottom at the end of
western sides of ocean basins with a distributed re- Caicos Drift off the Bahamas at about 5200 m depth
turn flow proposed by Henry Stommel has been and proceeds at that level over the 8000-m-deep
confirmed by many hydrographic studies and current Puerto Rico trench (Figure 7).
measurements. These broad patterns of the inflow of Pronounced inversions are seen at depths between
cold bottom water are very similar to the excess BNL 1 and 4 km along the continental margin close to the
concentration (excess over the clear-water minimum) Congo (Zaire) River. A comparable situation is
for the Indian Ocean, and net BNL particulate load found over Nitinat Fan off the NW USA. This area
(mass per unit area) for the Atlantic based on lies off two submarine canyons which are the most
Lamont nephelometer data (Figures 5 and 6). It is likely sources of the intermediate layers. Thus,
also apparent that the concentration or load of
sediment generally declines going northward along
the paths of the major inflows of Antarctic bottom
water. % Transmission
However, there are several points of very high 62 64 66 62 64 66
concentration or load which are not obviously re- 0
lated to likely increases in mean bottom flow vel-
ocity, for example, the high in the center of the
Argentine Basin (Figure 6). Richardson and col-
leagues showed in 1993 that these turbidity highs are 1
sometimes caused by intermittently high velocities
under regions where high surface eddy kinetic energy
was propagated downward, resulting in high abyssal
eddy kinetic energy, but at other times and places 2
there was no correlation with current speeds. Thus
some uncertainty remains over the role of high
abyssal eddy kinetic energy versus locally accelerated
thermohaline flow or deep recirculation loops in 3
producing high concentration. Nevertheless, the
Depth (km)

major feature of concentrated BNLs is that they do

delineate the paths of high-velocity bottom currents
extremely well and they decline in concentration 4
away from high-velocity boundary regions and away
from the energetic Southern Ocean.

5
Boundary Mixing, INLs, and Inversions
Steps and inversions are best shown by modern high-
sampling-rate sensors (Figures 1, 2, and 4), though
striking examples may also be found in older data. 6
INLs are most common off the continental slope and
submarine canyons. In both cases, focusing of in-
ternal waves (see and ) or tides is believed to be
responsible, due to the amplification of bottom flow 7
velocities at points where the slope of the seabed S 9° 17.7′ S, 80° 36.1′ W 8° 20.4′ S, 81°01.8′ W
matches the wave characteristic slope CThis causes (Peru trench) (Peru trench)
both resuspension and boundary mixing (see and
)The suspensions so generated have both fine and Figure 8 Profiles of the nepheloid layer in the Peru trench
made with a 1 m path length transmissometer. Reproduced from
coarse particles at the time of generation but as they
Pak H, Menzies D, and Zaneveld JRV (1979) Optical and
spread seaward they rain out the larger ones over the Hydrographical Observations off the Coast of Peru during May–
slope, leaving only finer material with slow sinking June 1977, Data Report 77, Ref. 79-14, pp. 1–93. Corvallis, OR:
speeds. Oregon State University.
 NEPHELOID LAYERS 291

canyons may also act as point sources for supply of layer of the ocean. Earth and Planetary Science Letters
turbid layers which mix out into the ocean interior. 92: 157--164.
Baker ET and Lavelle J W (1984) The effect of particle size on
the light attenuation coefficient of natural suspensions.
Journal of Geophysical Research 89: 8197--8203.
Trenches and Channels
Biscaye PE and Eittreim SL (1977) Suspended particulate
A final example of the influence of the sides of a loads and transports in the nepheloid layer of the
basin on its nepheloid layers comes from the ex- abyssal Atlantic Ocean. Marine Geology 23: 155--172.
tremely thick nepheloid layers found in trenches and Dickson RR and McCave IN (1986) Nepheloid layers on
some deep-sea passages. The nepheloid layer thick- the continental slope west of Porcupine Bank. Deep-Sea
ness is that of the trench depth plus the thickness of Research 33: 791--818.
Gardner WD (1989) Periodic resuspension in Baltimore
the nepheloid layer over the adjacent seafloor for the
Canyon by focusing of internal waves. Journal of
Kuril–Japan trench. The result is a layer 2600-m
Geophysical Research 94: 185--194.
thick. Nepheloid layers 2700-m thick have been re- Hall IR, Schmidt S, McCave IN, and Reyss JL (2000)
corded in the Peru trench between 81 and 101 S Particulate matter distribution and Th-234/U-238
(Figure 8). Flow through Vema Channel between the disequilibrium along the Northern Iberian Margin:
Argentine and Brazil basins yielded a well-mixed Implications for particulate organic carbon export.
water column in temperature and turbidity with ex- Deep-Sea Research I 47: 557--582.
cess radon-222 up to 400 m above the bed. Models Jerlov NG (1976) Marine Optics. Amsterdam: Elsevier.
of boundary layer development show that this can- Johnson DA, McDowell SE, Sullivan LG, and Biscaye PE
not occur by vertical turbulent diffusion. Turbid (1976) Abyssal hydrography, nephelometry, currents
layers must have spread quickly from the sides to the and benthic boundary layer structure in Vema Channel.
center of the channel, giving added support to the Journal of Geophysical Research 81: 5771--5786.
detached mixed-layer model. Kolla V, Henderson L, Sullivan L, and Biscaye PE (1978)
The fact that there is more suspended material at Recent sedimentation in the southeast Indian Ocean
depths greater than about 4000 m partly reflects the with special reference to the effects of Antarctic bottom
water circulation. Marine Geology 27: 1--17.
fact that waters at those depths are in contact with a
Kolla V, Sullivan L, Streeter SS, and Langseth MG (1976)
much greater area of seabed in proportion to their
Spreading of Antarctic bottom water and its effects on
volume than the shallower parts of the oceans. For the floor of the Indian Ocean inferred from bottom
4–5 km depth the value is 0.83 km2 km3, whereas water potential temperature, turbidity and sea-floor
for 2–3 km it is 0.11 km2 km3. One might say that photography. Marine Geology 21: 171--189.
deeper waters feel more bed. The global distribution McCave IN (1983) Particulate size spectra, behavior and
of nepheloid layers is, for instrumental reasons, the origin of nepheloid layers over the Nova Scotian
global distribution of fine particles (o2 mm), because continental rise. Journal of Geophysical Research 88:
that is what the instruments that map them ‘see’, but 7647--7666.
they also contain larger particles which play an im- McCave IN (1986) Local and global aspects of the bottom
portant role in sediment dynamics, especially in the nepheloid layers in the world ocean. Netherlands
lowest 10 m of the BMNL. Journal of Sea Research 20: 167--181.
McPhee-Shaw E (2006) Boundary–interior exchange:
Reviewing the idea that internal-wave mixing enhances
Further Reading lateral dispersal near continental margins. Deep-Sea
Research II 53: 42--59.
Amin M and Huthnance JM (1999) The pattern of cross- Menard HW and Smith SM (1966) Hypsometry of ocean
slope depositional fluxes. Deep-Sea Research I 46: basin provinces. Journal of Geophysical Research 71:
1565--1591. 4305--4325.
Anderson RF (2003) Chemical tracers of particle transport. Morel A (1973) Indicatrices de Diffusion Calculées par la
In: Elderfield H (ed.) Treatise on Geochemistry, Vol. 6: Théorie de Mie pour les Systèmes Polydisperses, en Vue
The Oceans and Marine Geochemistry, pp. 247--273. de l’Application aux Particules Marines, Report 10, pp.
Oxford, UK: Pergamon. 1–75. Paris: Laboratoire d’Océanographie Physique,
Anderson RF, Bacon MP, and Brewer PG (1983) Removal University of Paris VI.
of 230Th and 231Pa at ocean margins. Earth and Pak H, Menzies D, and Zaneveld JRV (1979) Optical and
Planetary Science Letters 66: 73--90. Hydrographical Observations off the Coast of Peru
Armi L and D’Asaro E (1980) Flow structures of the during May–June 1977, Data Report 77, Ref. 79-14,
benthic ocean. Journal of Geophysical Research 85: pp. 1–93. Corvallis, OR: Oregon State University.
469--484. Polzin KL, Toole JM, Ledwell JR, and Schmitt RW (1997)
Bacon MP and Rutgers van der Loeff MM (1989) Removal Spatial variability of turbulent mixing in the abyssal
of thorium-234 by scavenging in the bottom nepheloid ocean. Science 276: 93--96.
292 NEPHELOID LAYERS

Richardson MJ, Weatherly GL, and Gardner WD (1993) Turnewitsch R and Springer BM (2001) Do bottom mixed
Benthic storms in the Argentine Basin. Deep-Sea layers influence 234Th dynamics in the abyssal near-
Research 40: 957--987. bottom water column? Deep-Sea Research I 48:
Thorndike EM (1975) A deep sea, photographic nephelo- 1279--1307.
meter. Ocean Engineering 3: 1--15. Zaneveld JRV, Roach DM, and Pak H (1974) The
Tucholke BE and Eittreim SL (1974) The western determination of the index of refraction of oceanic
boundary undercurrent as a turbidity maximum over particulates. Journal of Geophysical Research 79:
the Puerto Rico Trench. Journal of Geophysical 4091--4095.
Research 79: 4115--4118.
 OPTICAL PARTICLE CHARACTERIZATION
P. H. Burkill and C. P. Gallienne, microscopy, or bacteria, protozoa, or larger zoo-
Plymouth Marine Laboratory, West Hoe, plankton, which are analyzed by light microscopy.
Plymouth, UK As an adjunct to light microscopy, fluorescence-
Copyright & 2001 Elsevier Ltd. based techniques are used increasingly to character-
ize, and sometimes quantify, the chemical properties
of cells. Such approaches can be extremely powerful,
particularly when used in conjunction with fluores-
cently labeled molecular probes. Such probes can be
Particles and Their Properties tailored to target specific taxonomic groups. Al-
though microscopy remains the benchmark, for the
Particles are ubiquitous in ocean waters, where they simple reason that ‘what you see, you believe,’ it is
are intimately involved in defining the optical prop- time consuming and costly. A wide range of techni-
erties, productivity, and biogeochemistry of our seas. ques offer rapid analysis of particles. However, these
Marine particles exist in a wide range of sizes and tend to be ‘black box’ techniques and should always
concentrations, and exhibit an inverse relationship be used with appropriate controls. No single tech-
between size and concentration, and a positive rela- nique provides a panacea in particle analysis, and it
tionship between concentration and ambient nutrient is often useful to combine two or more com-
concentration (the ‘trophic status’) in surface waters plementry techniques.
of the ocean. Rapid optical techniques for analyzing plankton-
Particles range in size from the largest marine or- sized particles may be based on scattered, fluorescent
ganisms (blue whales, c. 70 m length) down to the or transmitted light. Scattering and fluorescence
size that arbitrarily divides particles from dissolved methods are applicable to smaller particles
materials. In biological oceanography, this is defined (o500 mm equivalent spherical diameter (ESD)),
operationally as 0.2 mm. But here we will focus on where as larger particles, such as zooplankton, are
particles that fall within the size range of the usually analyzed by transmission techniques. Two
plankton. Plankton organisms range from viruses techniques that have been developed rapidly in the
(c. 0.05 mm) up to larger zooplankton such as last decade, are analytical flow cytometry (AFC) and
euphausiids (c. 2 cm). However, even within this size optical plankton counting (OPC). AFC and OPC are
range, many particles are not living but instead particularly suitable for the analysis of smaller par-
contribute to the large pools of detritus that often ticles (viruses to protozoa) and larger particles
predominate over living particles in the ocean. (metazoa), respectively.

Particle Characterization
Analytical Flow Cytometry
A wide array of techniques is available for the char-
acterization of marine particles. Although most are Technique Analytical flow cytometry (AFC) is a
based on optical properties, nonoptical techniques, generic technique based on the multiparametric
such as the acoustic doppler current profiler (ADCP) analysis of single particles at high speed. Originally
and the multifrequency echosounder, can also be developed for medical hematology and oncology,
used to quantify and characterize particles such as AFC is used increasingly in biological oceano-
large zooplankton and fish in sea water. Optical graphy. Its strengths are derived from its quanti-
characterization techniques vary considerably in tative capability, versatility, sensitivity, speed,
their resolution. At one extreme, satellite-based re- statistical precision, and ability to identify and, in
mote sensing can be used to quantify and charac- many instruments, sort particle subsets from
terize the marine phytoplankton across whole ocean heterogeneous populations. Its drawbacks are its
basins. At the other extreme, microscope-based cost and, for commercial instruments, the small
techniques and analytical flow cytometry resolve volume of sample (c. 0.5 cm3) analyzed. Particle
single particles. For the biologist, microscopy is the characterization and quantification in flow
benchmark procedure for identification of plankton. cytometry relies on cellular fluorescence and light
This is true whether the particles of interest are vir- scatter, and the power of the technique derives from
uses, which are typically analyzed by electron the ability to make multiple measurements

293
294 OPTICAL PARTICLE CHARACTERIZATION

Beam Sorted cells

stop
Forward light Waste
scatter detector tank

Light Cell sorter

collection catcher tube
lens Flow cell
560nm DM
± 11.3 nm
Laser
640 nm LP beam
filter
Sample Sheath
Brewster fluid
window tank
Orange Red
fluorescence 585nm fluorescence
Depolarized detector filter ± 21nm 650 nm detector
light scatter LP filter
detector
Horizontal Side
polarizer scatter
detector
488 nm
filter

Figure 1 Operating principles of AFC in which samples containing the particles of interest are passed singly across a laser beam.
Each particle scatters light and this is collected by forward and side light scatter detectors. Birefringent particles will tend to depolarize
the vertically polarized laser light and this is measured at the appropriate detector. Fluorescence from each particle is collected and
spectrally filtered so the wavelength of interest is detected by photomultiplier tubes. Output from each sensor is digitized and the data
are transferred to a computer. Particle size and refractive index are determined by the light scatter and the chemical properties are
determined by fluorescence. Particles exhibiting appropriate properties can be collected by sorting, whereas other particles pass to
waste. (Figure produced by Glen Tarran, Plymouth Marine Laboratory.)

simultaneously on each cell at high speed. Typically, into an electrical signal. The signal is amplified,
up to 5000 cells can be analyzed per second and digitized, and stored transiently in computer mem-
sorting rates of 410 000 s1 with 498% purity, ory. The data are then displayed on a computer
can be achieved. screen and stored onto disk as ‘list mode’ data. The
The principles of AFC (Figure 1) are based on list mode data can be considered to be analogous to a
hydrodynamically focusing a suspension that is spreadsheet in which each row represents a particle,
streamed coaxially through a flow chamber so that each column represents a different AFC sensor with
individual particles pass singly through the focus of a values that represent quantitative optical signatures
high intensity light source. Suitable light sources in- of each particle. The advantage of list mode data, as
clude coherent wave lasers since these provide a very with data in any spreadsheet, is that it can be re-
stable light beam that can easily be focused to small played, reanalyzed, and redisplayed.
dimensions. Light flux from the highly focused light Commercial cytometers are usually equipped with
source generates enough fluorescence from indi- light-scatter detectors that are situated in the narrow
vidual cells for this to be measured in the few forward and an orthogonal angle, as well as two or
microseconds taken to traverse the light beam. As the three fluorescence PMTs. Although a single laser is
particles traverse the beam, they scatter light and normal, AFC instruments can be equipped with two
may also produce fluorescence. Cellular fluorescence or more lasers to increase the number of fluor-
arises from autofluorescent cells or cells stained with ochrome excitation wavelengths. Some cytometers
a fluorochrome, and this is collected by a high nu- may use arc-lamp excitation particularly if UV ir-
merical aperture lens located orthogonally to the ir- radiation is required. There is also a move to the use
radiation source and the sample stream. The light of diode lasers for applications that demand low
collected is spectrally filtered sequentially by dichroic power use. Flow chambers may vary in their
mirrors which reflect specific wavelengths into hydrodynamic, optical, mechanical, and electrical
photomultiplier tubes (PMT). The PMTs are optic- characteristics to achieve high sensitivity and good
ally screened by band-pass filters. The quantity of stream stability. AFC instruments often have quartz
light incident upon each PMT, responding within a cuvette sensing zones to improve sensitivity and to
given color band, is then proportionally converted allow the application of UV irradiation. Specialized
 OPTICAL PARTICLE CHARACTERIZATION 295

cytometers can also measure particle volumes based Cellular fluorescence may be derived from two
on the Coulter principle of electrical impedance al- basic categories:
teration as particles flow through a restricted orifice.
1. autofluorescence in which the fluorescent mol-
Other specialized instruments may generate images
ecule of interest occurs naturally in the cell;
of particles in the sensing zone. Data processing and
2. applied fluorescence in which the fluorescent dye
display procedures have been developed which han-
is applied, or otherwise generated, and fluor-
dle fully crossed-correlated multidimensional data
escence is accumulated within the cell.
and this is achieved by microcomputers. Consider-
able developments have taken place in the last few
years to apply sophisticated procedures such as Phytoplankton. Analysis of phytoplankton by
multiparametric statistics or neural net generation, AFC is based on the presence of chlorophyll, a
to identify and characterize particles from within highly autofluorescent compound that is found in
heterogeneous mixtures. all viable plants. Chlorophyll is the phytoplankton’s
Many AFC instruments are able to sort cells and principal light-harvesting pigment, and absorbs
this is invaluable for identification, manipulation or light strongly in the blue and red regions of the
as a gateway to other analysis procedures. High visible spectrum. Blue light cellular absorption
speed ‘sorting-in-air’ is based on developments in coincides with the emission of the argon ion laser at
ink-jet printing. Two populations may be sorted 488 nm that is commonly used in flow cytometers.
from the sample stream that undergoes oscillation, Chlorophyll fluorescence is emitted in the far red
driven by a piezo-electric crystal that is mechanically (lem ¼ 680 nm), thereby offering a useful Stokes’
coupled to the flow chamber. The crystal, driven at shift of some 200 nm. This window means that
30–40 kHz, produces uniform liquid droplets of phytoplankton can be readily characterized and
which a small percentage contain single cells. The quantified by flow cytometers equipped with an
‘sort logic’ circuitry compares processed signals from argon laser (or other blue light source) and suitable
the sensors with pre-set, operator-defined ranges. spectral filtration (such as a 650 nm longpass filter)
When the amplitude falls within the pre-set range, an of fluorescent light emitted by cells onto a sensitive
electronic time delay of a few microseconds is acti- photomultiplier tube.
vated. This triggers an electrical droplet-charging As well as chlorophyll, some phycobiliproteins are
pulse at the moment the cell arrives at the droplet also autofluorescent. One of these is phycoerythrin
formation break-off point. The droplet-charging which is found in cyanobacteria and cryptophytes.
pulse causes a group of droplets to be charged, and Although phycoerythrin absorbs light in the green–
subsequently, deflected by a static electric field into a blue end of the spectrum, the 488 nm emission of the
collection vessel. Cells failing the pre-set sort criteria argon is sufficiently close to excite this compound. In
do not trigger droplet-charging, and so pass studies of phytoplankton, fluorescence from phy-
undetected into the waste collector. Other sorting coerythrin is measured by a separate photomultiplier
procedures include one in which a collecting arm tube that is spectrally filtered to collect emissions at
moves into the sample stream to pick up particles 585 nm. Based on this differentiation and coupled
that meet the programmed sort criteria. Sorting is an with light scatter measurements (the magnitude of
essential adjunct to AFC and is crucial to verifying which is roughly proportional to cell size), AFC can
both satisfactory instrument and analytical protocol readily differentiate and quantify the phytoplankton
operation. groups shown in Table 1.
In recent years, the application of powerful
Applications Oceanographic applications of AFC multivariate statistical and neural net procedures
are now diverse and continue to expand rapidly. have been applied to further characterize algal taxa
Although the fundamental principle of AFC remain
constant, recent developments in optical sensitivity Table 1 Routine AFC analysis of phytoplankton based on
optical characteristics
and design of AFC instruments have aided new
applications. Fluorochrome chemistry and mole- Differentiation AFC criteria
cular biology are both richly endowed fields and
developments in fluorescent assays of biochemical Phytoplankton Chlorophyll autofluorescence
constituents, coupled with the ability to target Prochlorococcus Low chlorophyll and light scatter
individual taxa have proved invaluable for AFC Synechococcus Low phycoerythrin and light scatter
Cryptophytes Phycoerythrin and light scatter
applications in marine biology. Detection limits are
Coccolithophores High orthogonal light scatter and laser
adequate to measure cellular attributes of many depolarization
planktonic cells.
296 OPTICAL PARTICLE CHARACTERIZATION

from within the complex mixtures that are typical of DNA as well as enzymes such as ribulose-1,5-
sea water. Multivariate statistics that have been used bisphosphate carboxylase. AFC instruments are
include quadratic discriminant analysis and canon- capable of great sensitivity and are able to quantify
ical variate analysis. The latter is a useful graphical concentrations of cellular chlorophyll in phyto-
technique for analyzing and displaying data, whereas plankton in the range of about 1–2000 fg cell1.
quadratic discriminant analysis can discriminate over In practice, marine phytoplankton are typically
two-thirds of mixtures of 22 algal taxa, with classi- analyzed using a fresh sample of sea water without
fication rates 470%. Such approaches are more than pretreatment. The sample is analyzed at a constant
two orders of magnitude faster than conventional rate so sample volume can be determined from an-
flow cytometric analyses for discriminating and alysis time. Chlorophyll-containing phytoplankton
enumerating phytoplankton species. and those containing phycoerythrin are registered by
Artificial neural nets (ANN) have proved to be the red and orange fluorescence emitted from single
extremely powerful in increasing AFC capability for cells as they traverse the laser beam. Typical sample
differentiating algal taxa. This approach is based on analysis time is generally 4–5 min. Examples of data
training an ANN to recognize the optical character- generated by AFC protocols for the analyses of nat-
istics of individual taxa. This is achieved by pre- ural waters are shown in Figure 2.
senting the net with AFC data derived from unialgal
cultures. The unknown samples are then analyzed by Bacteria. Bacteria, traditionally quantified by
the AFC under the same conditions and the data epifluorescence microscopy, can now be
passed through the trained ANN. The net outputs differentiated from other particles and analyzed by
identification probabilities for each cell analyzed. AFC (Figure 2). Both approaches are based on the
Several types of ANN have been used and it is now intercalation of a fluorochrome with the cell’s
possible for nets to differentiate and recognize 470 nucleic acid. However, such intercalation is
taxa with high accuracy. Considerable developments universal and often does not differentiate between
are anticipated in this field in the coming years. autotrophic and heterotrophic bacteria. A range of
As well as providing procedures for differentiation fluorochromes have been used including 488 nm
of phytoplankton from other particles, there are AFC absorbing YOYO-1, YO-PRO-1, PicoGreen and
protocols for quantifying cellular attributes of SYBR Green as well as the more traditional
phytoplankton. These include the cellular concen- UV excited bis-benzimide Hoechst 33342 or
trations of chlorophyll, phycoerythrin, protein and 40 ,6-diamidino-2-phenylindole (DAPI). Of these,

Figure 2 AFC characterization and differentiation of bacteria and phytoplankton in lab culture and in natural communities carried out
at the Plymouth Marine Laboratory. Bacteria are measured using SYBR Green fluorescence and the different phytoplankton are
measured by chlorophyll autofluorescence. Analysis can be verified by microscopic analysis of flow-sorted particles. (Figure produced
by Glen Tarran, Plymouth Marine Laboratory.)
 OPTICAL PARTICLE CHARACTERIZATION 297

SYBR Green offers the practical advantage that

autotrophic and heterotrophic bacteria can be
differentiated readily.
It is now possible to Quantify the cellular protein
and DNA content of bacteria in natural waters. Such
AFC techniques use the intensity of SYPRO-protein
or DAPI–DNA fluorescence of individual marine
bacteria. Cultures of various marine bacteria have
been measured in the range 60–330 fg protein cell1,
but the amount of natural bacterioplankton from the
North Sea in August 1998 was shown to be only (A)
24 fg protein cell1. The total DNA of natural bac-
teria has been estimated to be about 3 fg cell1 by
AFC techniques.
Changes in bacterioplankton community com-
position have also been assessed by molecular bio-
logical AFC techniques. The combination of AFC
analysis and sorting combined with denaturing gra-
dient gel electrophoresis of polymerase chain re-
action (PCR)-amplified 18S rDNA fragments and
fluorescence in situ hybridization has been shown to
be a rapid method of analyzing the taxonomic
composition of bacterioplankton. Experimental ma- (B)
nipulation of natural water samples resulted in a
bacterial succession from members of a Cytophaga
flavobacterium cluster, through gamma-proteo-
bacteria and finally alpha-proteobacteria.

Protozoa. AFC-based techniques for the analysis

of protozoa have, so far, been based on molecular
probes. Ribosomal RNA species-specific probes to
various members of the common heterotrophic
flagellate genus, Paraphysomonas, have been
developed (Figure 3). However, they have been
restricted to laboratory applications since naturally (C)
occurring organisms exhibit cellular fluorescence
Figure 3 Photomicrographs showing a mixture of four
levels that are often too low to distinguish from Paraphysomonas species after hybridization with a mixture of
background. This observation may either be a PV1 and EUK probes tagged with fluorescein and rhodamine,
reflection of poor probing efficiency or it may be respectively. PV1 is specific for P. vestita whereas EUK labels all
due to the organisms low growth rates in situ. eukaryotes. The mixture was irradiated at (A) 488 nm to show P.
vestita labeled with PV1, (B) 568 nm to show organisms labeled
with EUK, and (C) both 488 and 568 nm to reveal both probes.
Viruses. Viruses are now thought to be one of the Scale bar is 10 mm. (Reproduced from Rice et al. (1997) with
most abundant types of particles in the ocean. AFC permission from the Society for General Microbiology.)
protocols are now available for enumerating natural
marine viruses based on staining with the nucleic
acid-specific dye SYBR Green-I. Interestingly AFC- Siphoviridae) have also been stained with a variety
based counts are often higher than those obtained of highly fluorescent nucleic acid-specific dyes.
by microscopy, suggesting that further development Highest fluorescence is achieved using SYBR Green
work is needed. However, this AFC protocol reveals I, allowing DNA viruses with genome sizes between
two, and sometimes three, virus populations in 48.5 and 300 kb (kilobases) to be detected. Small
natural samples, whereas microscopy would only genome-sized RNA viruses (7.4–14.5 kb) are at the
differentiate one pool of viruses. Cultures of several current limit of detection by AFC.
different marine virus families (Baculoviridae,
Herpesviridae, Myoviridae, Phycodnaviridae, Zooplankton and larval fish. Although commer-
Picornaviridae, Podoviridae, Retroviridae, and cially available AFC instruments are directly applicable
298 OPTICAL PARTICLE CHARACTERIZATION

for the analysis of microbial cells, it is also possible to a taxonomic identification watershed may soon be
adapt the generic AFC concept for the analysis of reached. So far, the focus of AFC protocols has
metazoan organisms. This involves a scaling up of flow generally conformed to traditional taxonomic cri-
chambers and the associated fluidics system. The teria. But there may be another route that remains to
Macro Flow Planktometer built within the EU be explored. This involves an approach to classifying
MAROPT project has been applied to organisms as particles based directly on flow cytometry variables
large as larval fish. An inherent property of this system of light scatter, fluorescence, time of flight etc. Such
is that it incorporates ‘imaging in flow’ as part of the an approach might prove worthwhile because of its
analysis. Images may be stored either photo- direct simplicity. How such an approach would
graphically or electronically and then be available for compare to traditional taxonomic identification re-
subsequent image analysis. mains to be addressed, and that remains an exciting
challenge for the future.
In-situ AFC. As we move towards operational
oceanography, there is an increasing need for
autonomous, in situ instrumentation. An important Optical Plankton Counting
step towards this has been made recently with the
development of CytoBuoy, an AFC instrument The Optical Plankton Counter (OPC) has been de-
housed in a moored buoy and capable of wireless signed to analyze those zooplankton called the
data transfer. CytoBuoy is one of the very few AFC mesozooplankton whose size range conventionally
instruments to have been designed and built purely spans 0.2–20 mm.
for aquatic use. Its characteristics include enhanced
optics and electronics designed to obtain maximal Technique The OPC uses a collimated beam of
information on particle characteristics. Whereas light through an enclosed volume, received by a
standard cytometers reduce these to single peak or photosensor. When a particle interrupts this beam
area ‘list-mode’ numbers, time resolved signals are of light the sensor produces an electronic response
preserved fully and transferred to the computer as proportional to the cross-sectional area of the
raw data. Pulse shape signals aid identification particle (Figure 4). This response is digitized and
considerably and allow, for the first time, a true this digital size is converted into ESD using a
measurement of particle length. The CytoBuoy semiempirical formula. This is the diameter of a
concept has also been taken a step further and has sphere having the same cross-sectional area as the
been redesigned as a functional module of the UK particle being measured, and can be simply
Autosub autonomous underwater vehicle. converted into volume. The OPC comes in two
versions: a towed instrument for in situ use and a
Future trends. The generic capability of AFC lends benchtop laboratory version. The towed version has
itself to a variety of applications. The thrust in a sampling tunnel of 22 cm  2 cm. The laboratory
recent years has been towards greater taxonomic version has a glass flow cell 2 cm square.
resolution and this has been aided by the
application of molecular techniques particularly
LED array Monitor
involving oligonucleotide probes. These have many 640 nm photodiode
advantages including the ability to be tailored to
target particular groups of organisms. Here the level
of taxonomic targeting may range from general (e.g.
differentiation of classes of phytoplankton) down to Lens + aperture
individual species. Molecular probes may also be
coupled with chemotaxonomic capability that can Target
analyze cellular function such as specific enzyme
production. Reference
LED control Linear
Flow cytometry has opened up our ability to char- detector
acterize marine particles with a greater degree of
taxonomic resolution and further development of FSK out
techniques such as ANN will increase this capability Pulse
considerably. This should allow characterization of
natural populations objectively in close to real-time. Figure 4 Schematic of the operating principle of the optical
The quest for greater analytical resolution will un- plankton counter. LED, light-emitting diode; FSK; frequency shift
doubtedly continue. However, it is also possible that keying.
 OPTICAL PARTICLE CHARACTERIZATION 299

Applications The OPC is capable of large-scale, overestimated. Coincidence in the towed OPC can
rapid and continuous sampling of zooplankton, be reduced in areas of high abundance by inserting
providing a reliable measure of size distributed into the sampling tunnel a transparent plate,
abundance and biovolume, between 0.25 and reducing the sampling volume to one-fifth. There is
20 mm ESD, at data rates up to 200 s1. The use of a finite probability of coincidence occurring at all
the in situ OPC on various towed platforms is now concentrations, increasing with abundance and flow
well established. The laboratory version is intended rate through the instrument. It has been determined
for characterizing preserved samples. It has also experimentally that at a count rate of 30 s1, more
been deployed at sea in pump-through mode than 90% of particles will be counted. Smaller
producing continuous real-time data on surface particles far outnumber larger ones, so coincidence
zooplankton abundance and size distribution, will result in a loss of these smaller particles, and
permitting near continuous sampling of epipelagic biovolume is underestimated to a lesser degree than
zooplankton across ocean basin scales. Figure 5 abundance. Towed at 4 m s1, the standard OPC
shows data from the OPC in this mode on the will pass 17.6 l s1 through the sampling tunnel.
Atlantic Meridional Transect. The data in Figure 5 Coincidence will therefore begin to have a
show one of 11 transects completed to date, each significant effect on estimated abundance above
comprising a near-continuous 13 000 km transect of concentrations of 1700 m3, or 8500 m3 with the
size distributed biovolume in the North and South flow insert.
Atlantic, illustrating the power of this kind of Object orientation can also present problems in this
instrument. Data at this level of detail and spatial type of counter. Elongated organisms present a very
resolution acquired autonomously and continuously different cross-sectional area depending upon whe-
could not have been gathered over such large spatial ther they are side-on or end-on. Biovolume may be
scales by any other means. considerably underestimated in the latter case. It has
been shown that on average the cross-sectional area
Operational considerations Bias and coincidence measured for a randomly oriented object will be
require calibration of the instrument against some greater than 70% of the true value. Biovolume may
other sampling device. Initial calibration uses also be overestimated by the spherical model assumed
spherical glass beads of known size. Several in the OPC calibration – most zooplankton have a
researchers have noted nonlinearity in this shape closer to that of an oblate spheroid. We use an
calibration at the extremes of the size range, and ellipsoidal model based on cross-sectional area of the
have suggested that the operational size range particle. From cross-sectional area the length and
should be reduced. Sensor response time and width of the ellipsoid can be calculated (assuming a
coincidence limit the densities at which the OPC can length to width ratio typical for copepods).
operate. Coincidence occurs when more than one Several cases have been reported of OPCs pro-
particle is in the beam at the same time. They ducing counts many times higher than those from
register as one larger particle and abundance will, concurrent net samples. Comparison between OPC
therefore, be underestimated, and biomass and a Longhurst Hardy Plankton Recorder (LHPR)

600

500
Biovolume (mm3 m )
_3

> 2000 µm
400 1000−2000 µm
500−1000 µm
300
250−500 µm
200

100

0
47 43 35 27 18 9 −1 − 8 −15 _ 24 _ 29 _ 35 _ 46
Latitude (˚N)

Figure 5 A 13 000 km transect of size-distributed epipelagic zooplankton biovolume in the North and South Atlantic from the OPC in
continuous surface sampling mode on the Atlantic Meridional Transect.
300 OPTICAL PARTICLE CHARACTERIZATION

showed that abundance and biovolume recorded by system was subsequently developed. Although this
the OPC were consistently four times higher than for system avoided some of the problems associated
the LHPR (200 mm mesh). Using a 53 mm mesh with the deployment and use of net systems, and
showed that there had been significant under- preservation of samples, a human investigator still
sampling of zooplankton o350 mm ESD by the carried out counting and identification of the
LHPR. Studies like these show that care is required samples. Automated image processing of digitized
when comparing any two sampling methods. When photographic images could alleviate this problem,
comparing the OPC to net systems, careful selection but duration and spatial resolution is limited by the
of mesh size or OPC lower size threshold is required. amount of film that can be carried.
Our own investigations indicate that the most suit-
able mesh size for a net used in comparison with the Video methods The most advanced video method
standard OPC is 125 mm (250 mm ESD, 3:1 ellips- for producing abundance and size distribution
oidal model, mesh size 75% of width of smallest indicators of good temporal and spatial resolution,
animal to be quantitatively sampled). together with near-real-time classification to
It must be emphasized that OPC cannot dis- taxonomic groups, is the video plankton recorder.
criminate living from nonliving particles. All par- This consists of a towed frame 3 m long with four
ticles within its operating range will be detected, cameras each having concentric fields of view
including detritus, marine aggregates, air bubbles, (5–100 mm) covering the size range of interest
etc. A recent study found average abundance in the (0.5–20 mm. ESD). The imaged volume is defined
Faroe–Shetland Channel and north–western North by the oblique intersection of the cone defining the
Sea to be 2.5 times higher for an OPC than for net camera field of view, and that produced by a
samples and up to 40 times higher at extremely low collimated, strobed 80-watt xenon light beam pulse
concentrations. This was put down to detrital par- 1 ms duration, producing dark field illumination.
ticles/marine aggregates, which may in some cases be Sample volume varies from 1 ml to 1 liter. Fiberoptic
the most abundant particle type in the water, and are telemetry is used to send video data to the surface,
often too fragile to be sampled by nets. Detrital where information is recorded to tape. Subsequent
particles may be considered part of the plankton, upgrading of this system is expected to permit
being significant in marine food webs, but their preprocessing to be done in real time, and near real
presence in large numbers will bias comparisons be- time production of size and species distributions.
tween OPC and net derived estimates of abundance. Concurrent calibration of the device is provided by
an integrated LHPR system. The device can
differentiate detrital material and zooplankton,
Particle Imaging Instruments unlike optical and acoustic systems. The video
plankton recorder has been used extensively around
Automated particle counters (electronic, acoustic,
George’s Bank in the north-west Atlantic.
and optical) can give reliable data on zooplankton
Video cameras used in scientific imaging typically
abundance and size distribution, but tell us little or
produce 10–100 million bytes s1, and image-pro-
nothing about the species present, except where
cessing algorithms are highly computer intensive. For
dominant species are already known and well sep-
real-time acquisition systems to be used at sea in
arated in size. Larger autotrophic particles may be
continuous sampling mode, assuming 1 m3 min1
counted as zooplankton, and in areas where high
as the required sampling volume, and at typical
concentrations of detritus or marine aggregates are
oceanic zooplankton abundance of 50–5000 m3, we
present, they are not discriminated from zoo-
need to be able to process 3000–300 000 animals per
plankton. To make these discriminations, infor-
hour. Although currently available technology can
mation on shape as well as size is needed, requiring
resolve this problem to some degree, the solution is
the use of imaging devices.
not yet likely to be economical in size or cost.

Photographic methods High-speed silhouette ROV devices Remotely operated vehicles (ROVs)
photography has been undertaken at sea without carrying video cameras have been used for in situ
the need for sample preservation. Samples in a studies of zooplankton abundance and behavior.
shallow tray on top of 8  10-inch (20  25 cm) film Gelatinous organisms are notoriously difficult to
were exposed by a xenon strobe. Samples could sample using traditional methods, and an ROV
then be counted and identified from the film, to carrying a video camera has been successfully
provide a permanent record. A towed version of the deployed for their study. ROVs are usually
system using concentrating nets ahead of a camera restricted to small-scale, observational studies, but
 OPTICAL PARTICLE CHARACTERIZATION 301

may be useful for the location of zooplankton Jonker R, Groben R, Tarran G, et al. (2000) Automated
patches. Small imaging volume may preclude the identification and characterisation of microbial
study of rarer taxa, and poor image quality may populations using flow cytometry: the AIMS project.
preclude the study of smaller taxa. Zooplankton Scientia Marina 64: 225--234.
also exhibit attraction/avoidance responses to the Kachel V and Wietzorrek J (2000) Flow cytometry and
integrated imaging. Scientia Marina 64: 247--254.
presence of ROV systems.
Olson RJ, Zettler ER, and DuRand MD (1993)
Phytoplankton analysis using flow cytometry. In: Kemp,
See also et al. (eds.) Handbook of Methods in Aquatic Microbial
Ecology, pp. 175--186. Boca Raton: Lewis.
Acoustic Scattering by Marine Organisms. Reckermann M and Colijn F (2000) Aquatic flow
Fluorometry for Biological Sensing. Inherent cytometry: achievements and prospects. Scientia Marina
Optical Properties and Irradiance. Remotely 64(2): 119--268.
Operated Vehicles (ROVs). Rice J, Sleigh MA, Burkill PH, et al. (1997) Flow cytometric
analysis of characteristics of hybridization of species-
specific fluorescent oligonucleotide probes to rRNA of
Further Reading marine nanoflagellates. Applied and Environmental
Microbiology 63: 938--944.
Boddy L, Morris CW, and Wilkins MF (2000) Identification Rice J, O’Connor CD, Sleigh MA, et al. (1997) Fluorescent
of 72 phytoplankton species by radial basis function oligonucleotide rDNA probes that specifically bind
neural network analysis of flow cytometric data. Marine to a common nanoflagellate. Paraphysomonas vestita.
Ecology Progress Series 195: 47--59. Microbiology 143: 1717--1727.
Brussaard CPD, Marie D, and Bratbak G (2000) Flow Schultze PC, Williamson CE, and Hargreaves BR (1995)
cytometric detection of viruses. Journal of Virological Evaluation of a remotely operated vehicle (ROV) as a
Methods 85: 175--182. tool for studying the distribution and abundance of
Davis CS, Gallager SM, Marra M, and Stewart WK (1996) zooplankton. Journal of Plankton Research 17:
Rapid visualisation of plankton abundance and 1233--1243.
taxonomic composition using the video plankton Wood-Walker RS, Gallienne CP, and Robins DB (2000)
recorder. Deep-Sea Research II 43: 1947--1970. A test model for optical plankton counter (OPC)
Dubelaar GBJ and Gerritzen PL (2000) CytoBuoy: a step coincidence and a comparison of OPC derived and
forward towards using flow cytometry in operational conventional measures of plankton abundance. Journal
oceanography. Scientia Marina 64: 255--265. of Plankton Research 22: 473--484.
Gallienne CP and Robins DB (1998) Trans-oceanic Zubkov MV, Fuchs BM, Sturmeyer H, Burkill PH, and
characterisation of zooplankton community size Amann R (1999) Determination of total protein content
structure using an Optical Plankton Counter. Fisheries of bacterial cells using SYPRO staining and flow
Oceanography 7: 147--158. cytometry. Applied and Environmental Microbiology
Herman AW (1992) Design and calibration of a new 65: 3251--3257.
optical plankton counter capable of sizing small
zooplankton. Deep Sea Research 39(3/4): 395--415.
 PHOTOCHEMICAL PROCESSES
N. V. Blough, University of Maryland, College Park, including: (1) reduction of potentially harmful UV-B
MD, USA and UV-A radiation within the water column;
Copyright & 2001 Elsevier Ltd. (2) photo-oxidative degradation of organic matter
through the photochemical production of reactive
oxygen species (ROS) such as superoxide (O 2 ),
hydrogen peroxide (H2O2), the hydroxyl radical (OH)
and peroxy radicals (RO2); (3) changes in metal ion
speciation through reactions with the ROS or through
Introduction direct photochemistry, resulting in the altered bio-
Life on Earth is critically dependent on the spectral logical availability of some metals; (4) photochemical
quality and quantity of radiation received from the production of a number of trace gases of importance
sun. The absorption of visible light (wavelengths in the atmosphere such as CO2, CO, and carbonyl
from 400 to 700 nm) by pigments within terrestrial sulfide (COS), and the destruction of others such as
and marine plants initiates a series of reactions that dimethyl sulfide (DMS); (5) the photochemical pro-
ultimately transforms the light energy to chemical duction of biologically available low molecular
energy, which is stored as reduced forms of carbon. weight (LMW) organic compounds and the release of
This complex photochemical process, known as available forms of nitrogen, thus potentially fueling
photosynthesis, not only provides all of the chemical the growth of microorganisms from a biologically
energy required for life on Earth’s surface, but also resistant source material (the CDOM). These pro-
acts to decrease the level of a major greenhouse gas, cesses provide the focus of this article.
CO2, in the atmosphere. By contrast, the absorption
of ultraviolet light in the UV-B (wavelengths from Optical Properties of the Abiotic
280 to 320 nm) and UV-A (wavelengths from 320 to
400 nm) by plants (as well as other organisms) can
Constituents of Sea Waters
produce seriously deleterious effects (e.g. photo- CDOM is a chemically complex material produced
inhibition), leading to a decrease in the efficiency of by the decay of plants and algae. This material,
photosynthesis and direct DNA damage (UV-B), as commonly referred to as gelbstoff, yellow substance,
well as impairing or destroying other important gilvin or humic substances, can be transported from
physiological processes. The level of UV-B radiation land to the oceans by rivers or be formed directly in
received at the Earth’s surface depends on the con- marine waters by as yet poorly understood processes.
centration of ozone (O3) in the stratosphere where it CDOM is the principal light-absorbing component
is formed photochemically. The destruction of O3 in of the dissolved organic matter (DOM) pool in sea
polar regions, leading to increased levels of surface waters, far exceeding the contributions of discrete
UV-B radiation in these locales, has been enhanced by dissolved organic or inorganic light-absorbing com-
the release of man-made chlorofluorocarbons (CFCs), pounds. CDOM absorption spectra are broad and
but may also be influenced in part by the natural unstructured, and typically increase with decreasing
production of halogenated compounds by biota. wavelength in an approximately exponential fashion
These biotic photoprocesses have long been recog- (Figure 1). Spectra have thus been parameterized
nized as critical components of marine ecosystems using the expression [1].
and air–sea gas exchange, and have been studied ex-
tensively. However, only within the last decade or so aðlÞ ¼ aðl0 Þ  eSðll0 Þ ½1
has the impact of abiotic photoreactions on the
chemistry and biology of marine waters and their aðlÞ and aðl0 Þ are the absorption coefficients at
possible coupling with atmospheric processes been wavelength l and reference wavelength l0 , respect-
fully appreciated. Light is absorbed in the oceans not ively, and S defines how rapidly the absorption in-
only by phytoplankton and water, but also by colored creases with decreasing wavelength. Absorption
dissolved organic matter (CDOM), particulate detrital coefficients are calculated from relation [2], where A
matter (PDM), and other numerous trace light- is the absorbance measured across pathlength, r.
absorbing species. Light absorption by these con-
stituents, primarily the CDOM, can have a number 2:303  AðlÞ
a ð lÞ ¼ ½2
of important chemical and biological consequences r

302
 PHOTOCHEMICAL PROCESSES 303

6.0
(A) (B) 0.40
0.50
0.30

a (m 1)

a (m 1)
_

_
4.0 0.20
0.25
a (m 1)
_

0.10

0.00 0.00
300 400 500 600 300 400 500 600
2.0 Wavelength (nm) Wavelength (nm)

0.0

(C) 0.20 (D) 0.04

0.15 0.03
1.5
a (m 1)

a (m 1)
_

_
0.10 0.02
a (m 1)

0.05 0.01
_

1.0
0.00 0.00
300 400 500 600 300 400 500 600
Wavelength (nm) Wavelength (nm)
0.5

0.0
300 400 500 600 300 400 500 600
Wavelength (nm) Wavelength (nm)

Figure 1 Absorption spectra of CDOM (—), PDM (– – –) and phytoplankton (– – –) from surface waters in the Delaware Bay and
Middle Atlantic Bight off the east coast of the USA in July 1998: (A) mid-Delaware Bay at 39o 9.070 N, 75o 14.290 W; (B) Mouth of the
Delaware Bay at 381 48.610 N, 751 5.070 W; (C) Mid-shelf at 381 45.550 N, 741 46.600 W; (D) Outer shelf at 381 5.890 N, 741 9.070 W.

Due to the exponential increase of aðlÞ with de- compromise the determination of phytoplankton
creasing l, CDOM absorbs light strongly in the UV- biomass through satellite ocean color measurements.
A and UV-B, and thus is usually the principal con- As described below, the absorption of sunlight by
stituent within marine waters that controls the CDOM also initiates the formation of a variety of
penetration depth of radiation potentially harmful to photochemical intermediates and products. The
organisms (Figure 1). Moreover, for estuarine waters photochemical reactions producing these species ul-
and for coastal waters strongly influenced by river timately lead to the degradation of the CDOM and
inputs, light absorption by CDOM can extend well the loss, or bleaching, of its absorption. This process
into the visible wavelength regime, often dominating can act as a feedback to alter the aquatic light field.
the absorption by phytoplankton in the blue portion Particulate detrital material (PDM), operationally
of the visible spectrum. In this situation, the amount defined as that light-absorbing material retained on
and quality of the photosynthetically active radiation a GFF filter and not extractable with methanol, is a
available to phytoplankton is reduced, thus decrea- composite of suspended plant degradation products
sing primary productivity and potentially affecting and sediment that also exhibits an exponentially rising
ecosystem structure. High levels of absorption by absorption with decreasing wavelength (Figure 1);
CDOM in these regions can also seriously eqn [2] has thus been used to parameterize this
304 PHOTOCHEMICAL PROCESSES

material as well. However, the values of S acquired for dominated by reactions with dioxygen (O2) in a
this material are usually smaller than those of the process known as photo-oxidation. In this process,
CDOM. In estuarine and near-shore waters, and in O2 can act to accept electrons from excited states,
shallow coastal waters subject to resuspension of radicals (highly reactive species containing an un-
bottom sediments, PDM can contribute substantially paired electron) or radical ions generated within the
to the total water column absorption. However, in CDOM by the absorption of light. This leads to the
most marine waters, the PDM is a rather minor production of a variety of partially reduced oxygen
constituent. Little is known about its photochemical species such as superoxide (O 2 , the one-electron re-
reactivity. duction product of O2), hydrogen peroxide (H2O2,
Other light-absorbing trace organic compounds the two-electron reduction product of O2), peroxy
such as flavins, as well as inorganic compounds such radicals (RO2, formed by addition of O2 to carbon-
as nitrate, nitrite, and metal complexes, do not centered radicals, R) and organic peroxides (RO2H),
contribute significantly to the total water column along with the concomitant oxidation of the CDOM
absorption. However, many of these compounds are (Figure 2). Many of these reduced oxygen species as
quite photoreactive and will undergo rapid trans- well as the hydroxyl radical (OH), which is gener-
formation under appropriate light fields. ated by other photochemical reactions, are also quite
reactive. These reactive oxygen species or ROS can
undergo additional secondary reactions with them-
Photochemical Production selves or with other organic and inorganic seawater
constituents. The net result of this complex series of
of Reactive Oxygen Species reactions is the light-induced oxidative degradation
CDOM is the principal abiotic photoreactive con- of organic matter by dioxygen (Figure 2). This pro-
stituent in marine waters. Available evidence sug- cess leads to the consumption of O2, the production
gests that the photochemistry of this material is of oxidized carbon gases (CO2, CO, COS), the

(fluorescence) h  ′ + heat + CDOM

+h 
_ _
NO2 , NO3

+h 
CDOM* _
CDOM· + e (aq)
+

+ O2
NO, NO2 ·OH ±:
_ R· CDOM + O2
+ Br Radical ions
_
Carbon-centered radicals 2O2 + 2H + H2O2 + O2
(eg. CH3, CH3CO) _ H+ Hydrogen peroxide
_ Superoxide
Br2 (?)
+ O2
_
R'H
+ HCO3
RO2
CDOM·
Peroxy radicals
Carbon-centred radicals
_ CDOMoxidized + H+
CO3· Secondary reactions Secondary reactions
+ O2
RO + RO2H
Alkoxy radicals Organic peroxides
Secondar y reactions CDOM-OO·
Secondary reactions Peroxy radicals
Secondary reactions
Oxidized products
(CO, CO2,COS, LMW organic compounds)

Figure 2 Schematic representation of the photochemical and secondary reactions known or thought to occur following light
absorption by CDOM. For a more detailed description of these reactions see the text, Blough and Zepp (1995), and Blough (1997). Not
shown in this diagram are primary and secondary reactions of metal species; for a description of these processes, see Helz et al.
(1994) and Blough and Zepp (1995).
 PHOTOCHEMICAL PROCESSES 305

formation of a variety of LMW organic compounds, smaller than those for fluorescence, ranging from
the release of biologically available forms of nitro- B0.1% at 300 nm to B0.01% at 400 nm. The F for
gen, and the loss of CDOM absorption. Through other intermediates and products range even lower,
direct photochemical reactions and reactions with from B0.01% to 0.000 0001% (see below). The F
the ROS, the speciation of metal ions is also affected. for most of the intermediates and products created
These photochemical intermediates and products from the CDOM are highest in the UV-B and UV-A,
are produced at relatively low efficiencies. About and fall off rapidly with increasing wavelength;
98–99% of the photons absorbed by CDOM are yields at visible wavelengths are usually negligible
released as heat, while another B1% are re-emitted (see for example, Figure 3).
as fluorescence. These percentages (or fractions) of The hydroxyl radical, a very powerful oxidant,
absorbed photons giving rise to particular photo- can be produced by the direct photolysis of nitrate
responses are known as quantum yields ðFÞ. The F and nitrite (eqns [I]–[III]).
for the production of H2O2 and O 2 (the two reduced
oxygen species produced with highest efficiency), are
NO3  þ hv-O þ NO2 ½I
approximately one to two orders of magnitude

2.0 NO2  þ hv-O þ NO ½II

Lake Valkea-kotiten

1.5

O þ H2 O-OH þ OH ½III

Φ ( × 103)

1.0
The F values for these reactions are relatively high,
0.5
about 7% for nitrite and about 1–2% for nitrate.
However, because of the relatively low concen-
CO2
trations of these compounds in most marine surface
0.0 waters, as well as their low molar absorptivities in
River water the ultraviolet, the fraction of light absorbed is gen-
3.0 Suwanee erally small and thus fluxes of OH from these sources
Houghton also tend to be small. Recent evidence suggests that
OH, or a species exhibiting very similar reactivity, is
Φ ( × 104)

Kinoshe
2.0 produced through a direct photoreaction of the
Okefenokee
CDOM; quinoid moieties within the CDOM may be
responsible for this production. Quantum yields are
1.0
low, B0.01%, and restricted primarily to the ultra-
CO violet. In estuarine and near-shore waters containing
0.0 higher levels of iron, the production of OH may also
occur through the direct photolysis of iron–hydroxy
Sea water
complexes or through the Fenton reaction (eqn [IV]).
4.0
Φ ( × 107)

Fe2þ þ H2 O2 -Fe3þ þ OH þ OH ½IV

2.0 Compounds that do not absorb light within the

surface solar spectrum are also subject to photo-
chemical modification through indirect or ‘sensitized’
COS
0.0
photoreactions. In this case, the ROS or intermedi-
300 320 340 360 380 400 420 440 ates produced by direct photoreactions of a light-
Wavelength (nm) absorbing constituent such as CDOM can react
secondarily with the nonabsorbing compounds.
Figure 3 Wavelength dependence of the quantum yields (F)
DMS and COS, two trace gases of some importance
for the photochemical production of CO2, CO, and COS. Data
have been replotted from those dependencies originally reported
to the atmosphere, are thought to be destroyed and
in Vähätalo et al. (2000), Valentine and Zepp (1993), and Weiss created, respectively, by sensitized photoreactions in
et al. (1995) for CO2, CO, and COS, respectively. marine surface waters.
306 PHOTOCHEMICAL PROCESSES

Photochemical Production and 15–20-fold. The yields for CO2 production must thus
Consumption of LMW Organic approach, if not exceed, those for O 2 and H2O2.

Compounds and Trace Gases This result is somewhat surprising, since it implies
that about one CO2 is produced for each electron
The photolysis of CDOM produces a suite of LMW transferred from the CDOM to O2, further implying
organic compounds and a number of trace gases. The a high average redox state for CDOM. Although
production of these species presumably occurs CDOM (i.e. humic substances) is known to contain
through radical and fragmentation reactions arising significant numbers of carboxyl moieties that could
from the net oxidative flow of electrons from CDOM serve as the source of the CO2, the yield for CO2
to O2 (see above), although the exact mechanism(s) production, relative to O 2 and H2O2, would be ex-
have yet to be established. Most LMW organic pected to fall rapidly as these groups were removed
compounds produced contain three or fewer carbon photochemically; available evidence suggests that
atoms and include such species as acetaldehyde, this does not occur. An alternative explanation is that
acetate, acetone, formaldehyde, formate, glyoxal, other species, perhaps the CDOM itself, is acting as
glyoxalate, methylglyoxal, propanal, and pyruvate. an electron acceptor. Regardless of mechanism,
The F values for the production of individual com- existing information indicates that CO2 is the dom-
pounds are low, B0.001–0.0001%, with wave- inant product of CDOM photolysis (Figure 3).
lengths in the UV-B the most effective; efficiencies A recent estimate suggests that the annual global
decrease rapidly with increasing wavelength. Avail- photoproduction of CO in the oceans could be as
able evidence indicates that the F for O 2 and H2O2 high as 0.82  1015 g C. Assuming that CO2 photo-
production are about one to two orders of magnitude production is 15–20 times higher than that for CO,
larger than those for the LMW organic compounds, values for CO2 formation could reach from 12 to
so it appears that the sum of the production rates for 16  1015 g C y1. To place these numbers in per-
the known LMW organic compounds is small with spective, the estimated annual input of terrestrial
respect to the flux of photochemical equivalents from dissolved organic carbon to the oceans (0.2  1015 g
CDOM to O2. C y1) is only 1.3–1.7% of the calculated annual
Most, if not all, of these products are rapidly taken CO2 photoproduction, which is itself about 2–3% of
up and respired by bacteria to CO2. Numerous in- the oceanic dissolved organic carbon pool. These
vestigators have presented evidence supporting en- calculated CO2 (and CO) photoproduction rates may
hanced microbial activity in waters exposed to be high due to a number of assumptions, including
sunlight, with bacterial activities increasing from (1) the complete absorption of UV radiation by the
1.5- to almost 6-fold depending presumably on the CDOM throughout the oceans, (2) constant quan-
length and type of light exposure, and the concen- tum yields (or action spectra) for production in-
tration and source of the CDOM. Recently, bio- dependent of locale or light history, and (3)
logically labile nitrogen-containing compounds such neglecting mass transfer limitations associated with
as ammonia and amino acids have also been reported physical mixing. Nevertheless, these estimates clearly
to be produced photochemically from CDOM. Be- highlight the potential impact of abiotic photo-
cause CDOM is normally considered to be bio- chemistry on the oceanic carbon cycle. Moreover, the
logically refractive, this recent work highlights the products of this photochemistry are generated in
important role that abiotic photochemistry plays in near-surface waters where exchange with the at-
the degradation of CDOM, not only through direct mosphere can take place readily.
photoreactions, but also through the formation of Like the LMW organic compounds, bacteria can
biologically available products that can be respired oxidize CO to CO2; this consumption takes place in
to CO2 or used as nutrients by biota. A recent esti- competition with the release of CO to the atmos-
mate suggests that the utilization of biologically la- phere. Due to its photochemical production, CO
bile photoproducts could account for as much as exists at supersaturated concentrations in the surface
21% of the bacterial production in some near-surface waters of most of the Earth’s oceans. Recent esti-
waters. mates indicate that global oceanic CO emissions
Carbon dioxide and carbon monoxide are major could range from 0.013  1015 g y1–1.2  1015 g y1
products of the direct photolysis of CDOM (Fig- (see above). The upper estimate is based on calcu-
ure 3). Quantum yields for CO production are about lated photochemical fluxes (see below) and the as-
an order of magnitude smaller than those for O 2 and sumption that all CO produced is emitted to the
H2O2 production, ranging from B0.01% at 300 nm atmosphere. The lower estimate was calculated using
to B0.001% at 400 nm. Available data indicate that air–sea gas exchange equations and extensive meas-
the F for CO2 range even higher, perhaps as much as urements of CO concentrations in the surface waters
 PHOTOCHEMICAL PROCESSES 307

and atmosphere of the Pacific Ocean. The source of emission rates of these compounds to the atmosphere
the significant discrepancy between these two esti- via this source are negligible with respect to global
mates has yet to be resolved. Depending on the an- volatile organic carbon emissions, although this
swer, CO emitted to the atmosphere from the oceans production may play some role in certain restricted
could play a significant role in controlling OH levels locales exhibiting stronger source strengths, or in the
in the marine troposphere. marine environment remote from the dominant ter-
Carbonyl sulfide (COS) is produced primarily in restrial sources.
coastal/shelf waters, apparently by the CDOM-pho- The photolysis of nitrate and nitrite in sea water
tosensitized oxidation of organosulfur compounds. produces nitrogen dioxide (NO2) and nitric oxide
UV-B light is the most effective in its formation, with (NO), respectively (eqns [I] and [II]). Previous work
F decreasing rapidly from B6  107 at 300 nm to indicated that the photolysis of nitrite could act as a
B1  108 by 400 nm (Figure 3). The principal sinks small net source of NO to the marine atmosphere
of seawater COS are release to the atmosphere and under some conditions. However, this conclusion
hydrolysis to CO2 and H2S. Accounting for perhaps seems to be at odds with estimates of the steady-state
as much as one-third of the total source strength, the concentrations of superoxide and the now known
photochemical production of COS in the oceans is rate constant for the reaction of superoxide with
probably the single largest source of COS to the at- nitric oxide (6.7  109 M1 s1) to form peroxyni-
mosphere, although more recent work has revised trite in aqueous phases (eqn [V]).
this estimate downward. Smaller amounts of carbon
disulfide (CS2) are also generated photochemically O 
2 þ NO- OONO ½V
in surface waters through CDOM sensitized re-
action(s); F values decrease from B1  107 at
The peroxynitrite subsequently rearranges in part to
313 nm to 5  109 at 366 nm. The CS2 emitted to
form nitrate (eqn [VI]).
the atmosphere can react with OH to form add-
itional COS in the troposphere. Although it was

previously thought that the oxidation of COS in the OONO-NO3  ½VI
stratosphere to form sulfate aerosol could be im-
portant in determining Earth’s radiation budget and Even assuming a steady-state concentration of O 2
perhaps in regulating stratospheric ozone concen- (1012 M) that is about two orders of magnitude
trations, more recent work suggests that other sour- lower than that expected for surface sea waters
ces contribute more significantly to the background (B1010 M), the lifetime of NO in surface sea waters
sulfate in the stratosphere. would be only B150 s, a timescale too short for
Dimethyl sulfide (DMS), through its oxidation to significant exchange with the atmosphere except for
sulfate in the troposphere, acts as a source of cloud a thin surface layer. Moreover, even in this situation,
condensation nuclei, thus potentially influencing the the atmospheric deposition of additional HO2 rad-
radiative balance of the atmosphere. DMS is formed icals to this surface layer (to form O 2 ) would be
in sea water through the microbial decomposition of expected to act as an additional sink of the NO (flux
dimethyl sulfonioproprionate (DMSP), a compound capping). It appears that most if not all water bodies
believed to act as an osmolyte in certain species of exhibiting significant steady-state levels of O2 , pro-
marine phytoplankton. The flux of DMS to the at- duced either photochemically or thermally, should
mosphere is controlled by its concentration in surface act as a net sink of atmospheric NO and probably of
sea waters, which is controlled in turn by the rate of NO2 as well. Further, although less is known about
its decomposition. Estimates indicate that 7–40% of the steady-state levels of peroxy radicals in sea
the total turnover of DMS in the surface waters of waters due largely to their unknown decomposition
the Pacific Ocean is due to the photosensitized de- routes, their high rate constants for reaction with
struction of this compound, illustrating the potential NO (1–3  109 M1 s1) indicate that they should
importance of this pathway in controlling the flux of also act as a sink of NO. In fact, methyl nitrate,
DMS to the atmosphere. a trace species found in sea waters, may in part
In addition to these compounds, the photo- be produced through the aqueous phase reactions
chemical production of small amounts of non- (eqns [VII] and [VIII]) with the methylperoxy
methane hydrocarbons (NMHC) such as ethene, radical (CH3OO) generated through a known
propene, ethane, and propane has also been re- photochemical reaction of CDOM (or through at-
ported. Production of these compounds appears to mospheric deposition) and the NO arising from
result from the photolysis of the CDOM, with F the photolysis of nitrite (or through atmospheric
values of the order of 107–109. The overall deposition).
308 PHOTOCHEMICAL PROCESSES

CH3 OO þ NO-CH3 OONO ½VII Photochemical Calculations

Global and regional estimates for the direct photo-
CH3 OONO-CH3 ONO2 ½VIII chemical production (or consumption) of a particular
photoproduct (or photoreactant) can be acquired with
The concentrations of NO and NO2 in the tropo- knowledge of the temporal and spatial variation of the
sphere are important because of the involvement of solar irradiance reaching the Earth’s surface combined
these gases in the formation of ozone. with a simple photochemical model (eqn [3]).
The atmospheric deposition of ozone to the sea
surface can cause the release of volatile iodine com- Fðl; zÞ ¼ ED ðl; zÞ  Fi ðlÞ  aDi ðlÞ ½3
pounds to the atmosphere. There is also evidence
Here Fðl; zÞ is the photochemical production (or
that methyl iodide can be produced (as well as des-
consumption) rate; ED ðl; zÞ is the downwelling ir-
troyed) by photochemical processes in surface sea
radiance at wavelength, l, and depth, z, within the
waters. The release of these volatile iodine species
water column; aDi is the diffuse absorption co-
from the sea surface or from atmospheric aqueous
efficient for photoreactive constituent i; Fi ðlÞ is the
phases (aerosols) by these processes may act as a
quantum yield of this ith constituent. ED ðl; zÞ is well
control on the level of ozone in the marine tropo-
approximated by eqn [4].
sphere via iodine-catalyzed ozone destruction.
ED ðl; zÞ ¼ ED0 ðlÞ  eKd ðlÞz ½4
Trace metal photochemistry
ED0 ðlÞ is the downwelling irradiance just below the
A lack of available iron is now thought to limit sea surface and Kd ðlÞ is the vertical diffuse attenu-
primary productivity in certain ocean waters con- ation coefficient of downwelling irradiance. Kd ðlÞ
taining high nutrient, but low chlorophyll concen- can be approximated by eqn [5].
trations (the HNLC regions). This idea has spurred P P
interest in the transport and photochemical reactions ai ðlÞ þ bbi ðlÞ
Kd ðlÞE ½5
of iron in both seawaters and atmospheric aerosols. mD
Very little soluble Fe(II) is expected to be available at P P
the pH and dioxygen concentration of surface sea- where ai ðlÞ and bbi ðlÞ are the total absorption
waters due to the high stability of the colloidal iron and backscattering coefficients, respectively, of all
(hydr)oxides. The photoreductive dissolution of absorbing and scattering constituents within the
colloidal iron oxides by CDOM is known to occur at water column, and mD is the average cosine of the
low pH; this process is also thought to occur in angular distribution of the downwelling light. This
seawaters at high pH, but the reduced iron appears factor accounts for the average pathlength of light in
to be oxidized more rapidly than its detachment the water column, and for direct solar light is ap-
from the oxide surface. However, some workers have proximately equal to cos y, where y is the solar
found that CDOM-driven cycles of reduction fol- zenith angle (e.g. mD B1 when the sun is directly
lowed by oxidation increases the chemical avail- overhead). The diffuse absorption coefficient, aDi , is
ability, which was strongly correlated with the given by eqn [6].
growth rate of phytoplankton. Significant levels of ai
Fe(II) are also known to be produced photo- aDi ¼ ½6
mD
chemically in atmospheric aqueous phases (at lower
pH) and could serve as a source of biologically This model assumes that the water column is
available iron upon deposition to the sea surface. homogeneous, that Kd ðlÞ is constant with depth, and
Manganese oxides are also subject to reductive that upwelling irradiance is negligible relative to
dissolution by light in surface seawaters. This process ED ðl; zÞ. Combining eqn [3,4 and 6] gives eqn [7].
produces Mn(II), which is kinetically stable to oxi-
dation in the absence of bacteria that are subject to ED0 ðlÞ:eKd ðlÞz  Fi ðlÞ  ai ðlÞ
photoinhibition. These two effects lead to the for- Fðl; zÞ ¼ ½7
mD
mation of a surface maximum in soluble Mn(II), in
contrast to most metals which are depleted in surface This equation allows calculation of the spectral de-
waters due to biological removal processes. Other pendence of the production (consumption) rate as a
examples of the impact of photochemical reactions function of depth in the water column, assuming
on trace metal chemistry are provided in Further knowledge of ED0 ðlÞ, Kd ðlÞ, ai ðlÞ and Fi ðlÞ, all of
Reading. which can be measured or estimated (Figure 4).
 PHOTOCHEMICAL PROCESSES 309

0.03 2.0 x 1014

Solar irradiance
(A)
CDOM

(photons cm s nm )
1.5 x 1014

_1
Solar irradiance
0.02

a CDOM (cm 1)

_2 _1
_ 1.0 x 1014

0.01
0.5 x 1014

0.00 0.0

(B)
2x107
nm 1)
_
_3 _1
s
F (molecules cm

7
1x10

107 (C)
6
10

105

104
log F

103

102

10

1
300 350 400 450
Wavelength (nm)

Figure 4 Spectral dependence of CO photoproduction rates with depth, plotted on a linear (B) and logarithmic (C) scale. Depths in
(B) are (from top to bottom): surface, 0.5, 1, 1.5, and 2 m. Depths in (C) are (from top to bottom): surface, 0.5, 1, 1.5, 2, 4, 6, 8, and
10 m. These spectral dependencies were calculated using eqn [7], the wavelength dependence of the quantum yield for CO shown in
Figure 3, and the CDOM absorption spectrum and surface solar irradiance shown in (A). The attenuation of irradiance down the water
column in this spectral region was assumed to be only due to CDOM absorption, a reasonable assumption for coastal waters (see
Figure 1). Note the rapid attenuation in production rates with depth in the UV-B, due to the greater light absorption by CDOM in this
spectral region.

Integrating over wavelength provides the total pro- which upon substitution of eqn [4] becomes,
duction (consumption) rate at each depth. 
Integration of eqn [7] from the surface to depth z Y ðl; zÞ ¼ ED0 ðlÞ: 1  eKd ðlÞz  Fi ðlÞ
provides the spectral dependence of the photo- ai ð lÞ
chemical flux ðYÞ over this interval P P ½9
ai ðlÞ þ bbi ðlÞ
 
ED0 ðlÞ: 1  eKd ðlÞz  Fi ðlÞ  ai ðlÞ=Kd ðlÞ In most, but not all seawaters, the total absorption
Y ðl; zÞ ¼ ½8
mD will be much greater than the total backscatter,
310 PHOTOCHEMICAL PROCESSES

3.0 × 1010
CO2

2.5 × 1010

2.0 × 1010

1.5 × 1010

1.0 × 1010

0.5 × 1010

0
CO

1.5 × 109
nm 1)
_
_2 _1
s
Y (molecules cm

1.0 × 109

0.5 × 109

0
COS
6
6 × 10

4 × 106

2 × 106

0
300 350 400 450
Wavelength (nm)

Figure 5 Spectral dependence of the photochemical flux with depth for CO2, CO, and COS. Fluxes with depth are from the surface
to 0.25, 0.5, 1.0, 2.0, and 4 m, respectively (bottom spectrum to top spectrum). Below 4 m, increases in the flux are nominal. These
spectral dependencies were calculated using [10], the wavelength dependence of the quantum yields for CO2, CO and COS shown in
Figure 3, and the surface solar irradiance shown in Figure 4A. CDOM is assumed to absorb all photons in this spectral region (see
Figures 1 and 4).
 PHOTOCHEMICAL PROCESSES 311

P P
ai ðlÞb bbi ðlÞ, and thus the backscatter can be organic compounds (Figure 2), as well as affecting
ignored; this approximation is not valid for most trace metal speciation. The magnitude and impact of
estuarine waters and some coastal waters, where a these processes on upper ocean biogeochemical
more sophisticated treatment would have to be ap- cycles and their coupling with atmospheric processes
plied. This approximation leads to the final ex- are just beginning to be fully quantified and under-
pression for the variation of the spectral dependence stood. There remains the need to examine possible
of the flux with depth (Figure 5), couplings between atmospheric gas phase reactions
and photochemical reactions in atmospheric aqueous
 ai ð lÞ phases.
Fðl; zÞ ¼ ED0 ðlÞ: 1  eKd ðlÞz  Fi ðlÞ  P ½10
ai ð l Þ
The spectral dependence of the total water column
flux (z-N) is then given by, Further Reading
ai ð lÞ Blough NV (1997) Photochemistry in the sea-surface
FðlÞ ¼ ED0 ðlÞ  Fi ðlÞ  P ½11 microlayer. In: Liss PS and Duce R (eds.) The Sea
ai ð lÞ
Surface and Global Change, pp. 383--424. Cambridge:
with the total flux obtained by integrating over Cambrige University Press.
wavelength, Blough NV and Green SA (1995) Spectroscopic
characterization and remote sensing of non-living
Z organic matter. In: Zepp RG and Sonntag C (eds.) The
ai ð lÞ
F ED0 ðlÞ:Fi ðlÞ  P dl ½12 role of Non-living Organic Matter in the Earth’s
ai ð lÞ Carbon Cycle, pp. 23--45. New York: John Wiley.
l
Blough NV and Zepp RG (1995) Reactive oxygen species
To obtain global estimates of photochemical fluxes, in natural waters. In: Foote CS, Valentine JS, Greenberg
many investigators assume that the absorption due to A, and Liebman JF (eds.) Reactive Oxygen Species in
CDOM, aCDOM, dominates the absorption of all Chemistry, pp. 280--333. New York: Chapman & Hall.
other seawater constituents in the ultraviolet, and de Mora S, Demers S, and Vernet M (eds.) (2000) The
P Effects of UV Radiation in the Marine Environment.
thus that aCDOM ðlÞ= ai ðlÞE1.While this approxi-
Cambridge: Cambridge University Press.
mation is reasonable for many coastal waters, it is Häder D-P, Kumar HD, Smith RC, and Worrest RC (1998)
not clear that this approximation is valid for all Effects of UV-B radiation on aquatic ecosystems.
oligotrophic waters. This approximation leads to the Journal of Photochemistry and Photobiology B 46:
final expression for flux, 53--68.
Z Helz GR, Zepp RG, and Crosby DG (eds.) (1994) Aquatic
and Surface Photochemistry. Ann Arbor, MI: Lewis
Y ED0 ðlÞ:Fi ðlÞdl ½13 Publishers.
l Huie RE (1995) Free radical chemistry of the atmospheric
aqueous phase. In: Barker JR (ed.) Progress and
which relies only on the surface downwelling irradi- Problems in Atmospheric Chemistry, pp. 374--419.
ance and the wavelength dependence of the quantum Singapore: World Scientific Publishing Co.
yield for the photoreaction of interest. Uncertainties in Kirk JTO (1994) Light and Photosynthesis in Aquatic
the use of this equation for estimating global photo- Ecosystems. Cambridge: Cambridge University Press.
chemical fluxes include (1) the (usual) assumption Moran MA and Zepp RG (1997) Role of photoreactions in
that FðlÞ acquired for a limited number of samples the formation of biologically labile compounds from
is representative of all ocean waters, independent of dissolved organic matter. Limnology and Oceano-
locale or light history, and (2) differences in the spa- graphy 42: 1307--1316.
Thompson AM and Zafiriou OC (1983) Air–sea fluxes of
tially and temporally averaged values of ED0 ðlÞ
transient atmospheric species. Journal of Geophysical
utilized by different investigators. Research 88: 6696--6708.
Vähätalo AV, Salkinoja-Salonen M, Taalas P, and Salonen
K (2000) Spectrum of the quantum yield for
Conclusions photochemical mineralization of dissolved organic
carbon in a humic lake. Limnology and Oceanography
The absorption of solar radiation by abiotic sea 45: 664--676.
water constituents initiates a cascade of reactions Valentine RL and Zepp RG (1993) Formation of carbon
leading to the photo-oxidative degradation of or- monoxide from the photodegradation of terrestrial
ganic matter and the concomitant production (or dissolved organic carbon in natural waters. Environ-
consumption) of a variety of trace gases and LMW mental Science Technology 27: 409--412.
312 PHOTOCHEMICAL PROCESSES

Weiss EW, Andrews SS, Johnson JE, and Zafiriou OC Zepp RG, Callaghan TV, and Erickson DJ (1998) Effects of
(1995) Photoproduction of carbonyl sulfide in south enhanced solar ultraviolet radiation on biogeochemical
Pacific Ocean waters as a function of irradiation cycles. Journal of Photochemistry and Photobiology B
wavelength. Geophysical Research Letters 22: 215--218. 46: 69--82.
Zafiriou OC, Blough NV, Micinski E, et al. (1990)
Molecular probe systems for reactive transients in
natural waters. Marine Chemistry 30: 45--70.
 RADIATIVE TRANSFER IN THE OCEAN
C. D. Mobley, Sequoia Scientific Inc., WA, USA scattering is high compared to that by other sub-
Copyright & 2001 Elsevier Ltd.
stances. Absorption by chlorophyll and related pig-
ments therefore plays the dominant role in
determining the total absorption in such waters, al-
though covarying detritus and dissolved organic
matter derived from the phytoplankton also con-
Introduction tribute to absorption and scattering in case 1 waters.
Case 1 water can range from very clear (oligo-
Understanding how light interacts with sea water is a trophic) to very productive (eutrophic) water, de-
fascinating problem in itself, as well as being fun- pending on the phytoplankton concentration.
damental to fields as diverse as biological primary ‘Case 2 waters’ are ‘everything else,’ namely,
production, mixed-layer thermodynamics, photo- waters where inorganic particles or dissolved organic
chemistry, lidar bathymetry, ocean-color remote matter from land drainage contribute significantly to
sensing, and visual searching for submerged objects. the IOPs, so that absorption by pigments is relatively
For these reasons, optics is one of the fastest growing less important in determining the total absorption.
oceanographic research areas. Roughly 98% of the world’s open ocean and coastal
Radiative transfer theory provides the theoretical waters fall into the case 1 category, but near-shore
framework for understanding light propagation in and estuarine case 2 waters are disproportionately
the ocean, just as hydrodynamics provides the important to human interests such as recreation,
framework for physical oceanography. The article fisheries, and military operations.
begins with an overview of the definitions and ter- Table 1 summarizes the terms, units, and symbols
minology of radiative transfer as used in ocean- for various quantities frequently used in optical
ography. Various ways of quantifying the optical oceanography.
properties of a water body and the light within the
water are described. The chapter closes with ex-
amples of the absorption and scattering properties of Radiometric Quantities
two hypothetical water bodies, which are charac- Consider an amount DQ of radiant energy incident
teristic of the open ocean and a turbid estuary, and a in a time interval Dt centered on time t, onto a sur-
comparison of their underwater light fields. face of area DA located at position (x,y,z), and ar-
riving through a set of directions contained in a solid
angle DO about the direction (y, j) normal to the
Terminology
area DA, as produced by photons in a wavelength
The optical properties of sea water are sometimes interval Dl centered on wavelength l. The geometry
grouped into inherent and apparent properties. of this situation is illustrated in Figure 1. Then an
operational definition of the spectral radiance is
• Inherent optical properties (IOPs) are those
properties that depend only upon the medium and
DQ
therefore are independent of the ambient light Lðx; y; z; t; y; j; lÞ 
field. The two fundamental IOPs are the ab- DtDADODl
sorption coefficient and the volume scattering ½Js1 m2 sr1 nm1  ½1
function. (These quantities are defined below.)
In the conceptual limit of infinitesimal parameter
• Apparent optical properties (AOPs) are those
intervals, the spectral radiance is defined as
properties that depend both on the medium (the
IOPs) and on the directional structure of the am-
bient light field, and that display enough regular @4Q
Lðx; y; z; t; y; j; lÞ  ½2
features and stability to be useful descriptors of a @t @A @O @l
water body. Commonly used AOPs are the ir-
Spectral radiance is the fundamental radiometric
radiance reflectance, the remote-sensing reflect-
quantity of interest in optical oceanography: it
ance, and various diffuse attenuation functions.
completely specifies the positional (x,y,z), temporal
‘Case 1 waters’ are those in which the contribution (t), directional (y, j), and spectral (l) structure of the
by phytoplankton to the total absorption and light field. In many oceanic environments, horizontal

313
314 RADIATIVE TRANSFER IN THE OCEAN

Table 1 Quantities commonly used in optical oceanography

Quantity SI units Symbol

Radiometric quantities
Quantity of radiant energy J nm1 Q
Power W nm1 F
Intensity W sr1 nm1 I
Radiance W m2 sr1nm1 L
Downwelling plane irradiance W m2 nm1 Ed
Upwelling plane irradiance W m2 nm1 Eu
Net irradiance W m2 nm1 E
Scalar irradiance W m2 nm1 Eo
Downwelling scalar irradiance W m2 nm1 Eou
Upwelling scalar irradiance W m2 nm1 Eou
Photosynthetic available radiation Photonss1 m2 PAR

Inherent optical properties

Absorption coefficient m1 a
Volume scattering function m1 sr1 b
Scattering phase function sr1 b̃
Scattering coefficient m1 b
Backscatter coefficient m1 bb
Beam attenuation coefficient m1 c
Single-scattering albedo – oo

Apparent optical properties

Irradiance reflectance (ratio) – R
Remote-sensing reflectance sr1 Rrs
Attenuation coefficients m1
of radiance L(z, y, j) m1 K(y, j)
of downwelling irradiance Ed(z) m1 Kd
of upwelling irradiance Eu(z) m1 Ku
of PAR m1 KPAR

variations (on a scale of tens to thousands of meters)

ΔQ
of the IOPs and the radiance are much less than
variations with depth, in which case it can be as-
sumed that these quantities vary only with depth z.
(An exception would be the light field due to a single
ΔΩ
light source imbedded in the ocean; such a radiance
distribution is inherently three-dimensional.) More-
ΔA over, since the timescales for changes in IOPs or in
the environment (seconds to seasons) are much
greater than the time required for the radiance to
x reach steady state (microseconds) after a change in
IOPs or boundary conditions, time-independent ra-
 diative transfer theory is adequate for most oceano-
graphic studies. (An exception is time-of-flight lidar

bathymetry.) When the assumptions of horizontal
homogeneity and time independence are valid, the
y spectral radiance can be written as L(z, y, j, l).
Although the spectral radiance completely speci-
z fies the light field, it is seldom measured in all dir-
ections, both because of instrumental difficulties and
Figure 1 Geometry used to define radiance. because such complete information often is not
 RADIATIVE TRANSFER IN THE OCEAN 315

needed. The most commonly measured radiometric Turning this instrument upside down gives the
quantities are various irradiances. Suppose the light spectral upwelling plane irradiance Eu(z, l). Ed and
detector is equally sensitive to photons of a given Eu are useful because they give the energy flux
wavelength l traveling in any direction (y, j) within (power per unit area) across the horizontal surface at
a hemisphere of directions. If the detector is located depth z owing to downwelling and upwelling pho-
at depth z and is oriented facing upward, so as to tons, respectively. The difference Ed  Eu is called the
collect photons traveling downward, then the de- net (or vector) irradiance.
tector output is a measure of the spectral down- Photosynthesis is a quantum phenomenon, i.e., it
welling scalar irradiance at depth z, Eod(z, l). Such is the number of available photons rather than the
an instrument is summing radiance over all the amount of radiant energy that is relevant to the
directions (elements of solid angle) in the downward chemical transformations. This is because a photon
hemisphere; thus Eod(z, l) is related to L(z, y, j, l) of, say, l ¼ 400 nm, if absorbed by a chlorophyll
by molecule, induces the same chemical change as does
Z a photon of l ¼ 600 nm, even though the 400 nm
photon has 50% more energy than the 600 nm
Eod ðz; lÞ ¼ Lðz; y; j; lÞdO ½Wm2 nm1  ½3
2pd photon. Only a part of the photon energy goes into
photosynthesis; the excess is converted to heat or is
Here 2pd denotes the hemisphere of downward re-radiated. Moreover, chlorophyll is equally able to
directions (i.e., the set of directions (y, j) such that absorb and utilize a photon regardless of the pho-
0ryrp/2 and 0rjo2p, if y is measured from the ton’s direction of travel. Therefore, in studies of
þ z or nadir direction). The integral over 2pd can be phytoplankton biology, the relevant measure of the
evaluated as a double integral over y and j after a light field is the photosynthetic available radiation,
specific coordinate system is chosen. PAR, defined by
If the same instrument is oriented facing down- Z 700 nm
ward, so as to detect photons traveling upward, then lZ
PARðzÞ  Eo ðz; lÞdl
the quantity measured is the spectral upwelling sca- 350 nm hc
lar irradiance Eou(z, l). The spectral scalar irradiance ½photonss1 m2  ½6
Eo(z, l) is the sum of the downwelling and upwelling
components: where h ¼ 6.6255  1034 J s is the Planck constant
and c ¼ 3.0  1017 nm s1 is the speed of light. The
EO ðz; lÞ  Eod ðz; lÞ þ Eou ðz; lÞ factor l/hc converts the energy units of Eo to quan-
Z tum units (photons per second). Bio-optical literature
¼ Lðz; y; j; lÞdO ½4 often states PAR values in units of mol photons
4p
s1 m2 or einst s1 m2 (where one einstein is one
Eo(z, l) is proportional to the spectral radiant energy mole of photons).
density (J m3 nm1) and therefore quantifies how
much radiant energy is available for photosynthesis
Inherent Optical Properties
or heating the water.
Now consider a detector designed so that its sen- Consider a small volume DV of water, of thickness
sitivity is proportional to |cos y|, where y is the angle Dr as illuminated by a collimated beam of mono-
between the photon direction and the normal to the chromatic light of wavelength l and spectral radiant
surface of the detector. This is the ideal response of a power Fi(l) (W nm1), as schematically illustrated in
‘flat plate’ collector of area DA, which when viewed Figure 2. Some part Fa(l) of the incident power Fi(l)
at an angle y to its normal appears to have an area of is absorbed within the volume of water. Some part
DA|cos y|. If such a detector is located at depth z and Fi(c, l) is scattered out of the beam at an angle c,
is oriented facing upward, so as to detect photons and the remaining power Ft(l) is transmitted
traveling downward, then its output is proportional through the volume with no change in direction. Let
to the spectral downwelling plane irradiance Ed(z, l). Fs(l) be the total power that is scattered into all
This instrument is summing the downwelling radi- directions.
ance weighted by the cosine of the photon direction, The inherent optical properties usually employed
thus in radiative transfer theory are the absorption and
scattering coefficients. In the geometry of Figure 2,
R
Ed ðz; lÞ ¼ Lðz; y; j; lÞjcosyjdO the absorption coefficient a(l) is defined as the limit
2pd
½5 of the fraction of the incident power that is absorbed
½Wm2 nm1  within the volume, as the thickness becomes small
316 RADIATIVE TRANSFER IN THE OCEAN

ΔΩ Φs ( ) 10
4

ΔV 2
Tu
rb
10 id
ha
Co rbo

VSF,  (m sr )

_1
as r
tal
Cle oc

_1
ar ea
Φi Φa Φt 0 oc n
10 ea
n

_2
10
Δr
Pure sea water
_4

Figure 2 Geometry used to define inherent optical properties. 10

0.1 1.0 10.0 100.0
Scattering angle, (deg)

lim 1 Fa ðlÞ
aðlÞ  Dr-0 ½m1  ½7 Figure 3 Volume scattering functions (VSF) measured in three
Fi ðlÞ Dr different oceanic waters. The VSF of pure sea water is shown for
comparison.
The scattering coefficient b(l) has a corresponding
definition using Fs(l). The beam attenuation co-
efficient c(l) is defined as c(l) ¼ a(l) þ b(l). Integrating b(c, l) over all directions (solid angles)
Now take into account the angular distribution of gives the total scattered power per unit incident ir-
the scattered power, with Fs(c, l)/Fi(l) being the radiance and unit volume of water, in other words
fraction of incident power scattered out of the beam the spectral scattering coefficient:
through an angle c into a solid angle DO centered on Z Z p
c, as shown in Figure 2. Then the fraction of scat- bðlÞ ¼ bðc; lÞdO ¼ 2p bðc; lÞ sin c dc ½10
tered power per unit distance and unit solid angle, 4p 0
b(c, l), is
Eqn. [10] follows because scattering in natural
waters is azimuthally symmetric about the incident
lim lim Fs ðc; lÞ
bðc; lÞ  Dr-0 ½m1 sr1  ½8
DO-0 direction (for unpolarized light sources and ran-
Fi ðlÞDrDO
domly oriented scatterers). This integration is often
The spectral power scattered into the given solid divided into forward scattering, 0rcrp/2, and
angle DO is just the spectral radiant intensity scat- backward scattering, p/2rcrp, parts. Thus the
tered into direction c times the solid angle: Fs(c, backscatter coefficient is
l) ¼ Is(c, l) DO. Moreover, if the incident power Z p
Fi(l) falls on an area DA, then the corresponding bb ðlÞ  2p bðc; lÞ sin c dc ½11
incident irradiance is Ei(l) ¼ Fi(l)/DA. Noting that p=2
DV ¼ DrDA is the volume of water that is illuminated
by the incident beam gives The VSFs of Figure 3 have b values ranging from
0.037 to 1.824 m1 and backscatter fractions bb/b of
lim IS ðc; lÞ 0.013 to 0.044.
bðc; lÞ ¼DV-0 ½9 The preceding discussion assumed that no in-
Ei ðlÞDV
elastic-scattering processes are present. However,
This form of b(c, l) suggests the name volume inelastic scattering does occur owing to fluorescence
scattering function (VSF) and the physical interpret- by dissolved matter or chlorophyll, and to Raman
ation of scattered intensity per unit incident irradi- scattering by the water molecules themselves. Power
ance per unit volume of water. Figure 3 shows lost from wavelength l by scattering into wavelength
measured VSFs (at 514 nm) from three greatly dif- l0 al appears as an increase in the absorption a(l).
ferent water bodies; the VSF of pure water is shown The gain in power at l0 appears as a source term in
for comparison. VSFs of sea water typically increase the radiative transfer equation.
by five or six orders of magnitude in going from Two more inherent optical properties are com-
c ¼ 901 to c ¼ 0.11 for a given water sample, and monly used in optical oceanography. The single-scat-
scattering at a given angle c can vary by two orders tering albedo is oo(l) ¼ b(l)/c(l). The single-scattering
of magnitude among water samples. albedo is the probability that a photon will be
 RADIATIVE TRANSFER IN THE OCEAN 317

scattered (rather than absorbed) in any given inter- radiometric quantities. For example, it is not possible
action, hence oo(l) is also known as the probability of to write down an equation that can be solved directly
photon survival. The volume scattering phase func- for the irradiance Ed; one must first solve the RTE for
tion, b̃ðc; lÞ is defined by the radiance and then compute Ed by integrating the
radiance over direction.
bðc; lÞ
bðc; lÞ  ½sr1  ½12
bðlÞ
Apparent Optical Properties
Writing the volume scattering function b(c, l) as the Apparent optical properties are always a ratio of two
product of the scattering coefficient b(l) and the phase radiometric variables. This ratioing removes effects
function b̃ðc; lÞ partitions b(c, l) into a factor giving of the magnitude of the incident sky radiance onto
the strength of the scattering, b(l) with units of m1, the sea surface. For example, if the sun goes behind a
and a factor giving the angular distribution of the cloud, the downwelling and upwelling irradiances
scattered photons, b̃ðc; lÞ with units of sr1. A strik- within the water can change by an order of magni-
ing feature of the sea water VSFs of Figure 3 is that tude within a few seconds, but their ratio will be
their phase functions are all similar in shape, with the almost unchanged. (There will still be some change
main differences being in the detailed shape of the because the directional structure of the underwater
functions in the backscatter directions (c4901). radiance will change when the sun’s direct beam is
The IOPs are additive. This means, for example, removed from the radiance incident onto the sea
that the total absorption coefficient of a water body surface.)
is the sum of the absorption coefficients of water, The ratio just mentioned,
phytoplankton, dissolved substances, mineral par-
ticles, etc. This additivity allows the development of Eu ðz; lÞ
separate models for the absorption and scattering Rðz; lÞ  ½14
Ed ðz; lÞ
properties of the various constituents of sea water.
is called the irradiance reflectance (or irradiance
ratio). The remote-sensing reflectance Rrs(y, j, l) is
The Radiative Transfer Equation defined as
The equation that connects the IOPs and the radi-
ance is called the radiative transfer equation (RTE). Lw ðy; j; lÞ
Rrs ðy; j; lÞ  ½sr1  ½15
Even in the simplest situation of horizontally Ed ðlÞ
homogeneous water and time independence, the RTE
where Lw is the water-leaving radiance, i.e., the total
is a formidable integro-differential equation:
upward radiance minus the sky and solar radiance
that was reflected upward by the sea surface. Lw and
dLðz; y; j; lÞ Ed are evaluated just above the sea surface. Both
cos y ¼  cðz; lÞLðz; y; j; lÞ
dz Z Rrs(y, j, l) and R(z, l) just beneath the sea surface
þ Lðz; y0 ; j0 ; lÞ are of great importance in remote sensing, and both
4p can be regarded as a measure of ‘ocean color.’ R and
 bðz; y0; j0 -y; j; lÞdO0 Rrs are proportional (to a first-order approximation)
þ Sðz; y; j; lÞ ½13 to bb/(a þ bb), and measurements of Rrs above the
surface or of R within the water can be used to es-
The scattering angle c in the VSF is the angle be- timate water quality parameters such as the chloro-
tween the incident direction (y0 , j0 ) and the scattered phyll concentration.
direction (y, j). The source term S(z, y, j, l) can Under typical oceanic conditions, for which the
describe either an internal light source such as bio- incident lighting is provided by the sun and sky, the
luminescence, or inelastically scattered light from radiance and various irradiances all decrease ap-
other wavelengths. The physical environment of a proximately exponentially with depth, at least when
water body – waves on its surface, the character of its far enough below the surface (and far enough above
bottom, the incident radiance from the sky – enters the bottom, in shallow water) to be free of boundary
the theory via the boundary conditions necessary to effects. It is therefore convenient to write the depth
solve the RTE. Given the IOPs and suitable boundary dependence of, say, Ed(z, l) as
conditions, the RTE can be solved numerically for Z z
the radiance distribution L(z, y, j, l). Unfortunately, Ed ðz; lÞ  Ed ð0; lÞexpleft½ Kd ðz0 ; lÞdz0  ½16
there are no shortcuts to computing other 0
318 RADIATIVE TRANSFER IN THE OCEAN

where Kd(z, l) is the spectral diffuse attenuation be the dominant absorber at the blue end of the
coefficient for spectral downwelling plane irradiance. spectrum, especially in coastal waters influenced by
Solving for Kd(z, l) gives river runoff.

dlnEd ðz; lÞ Organic Particles

Kd ðz; lÞ ¼ 
dz Biogenic particles occur in many forms.
1 dEd ðz; lÞ 1
¼ ½m  ½17
Ed ðz; lÞ dz Bacteria Living bacteria in the size range 0.2–
1.0 mm can be significant scatterers and absorbers of
The beam attenuation coefficient c(l) is defined in
light, especially at blue wavelengths and in clean
terms of the radiant power lost from a collimated
oceanic waters, where the larger phytoplankton are
beam of photons. The diffuse attenuation coefficient
relatively scarce.
Kd(z, l) is defined in terms of the decrease with depth
of the ambient downwelling irradiance Ed(z, l),
which comprises photons heading in all downward Phytoplankton These ubiquitous microscopic
directions (a diffuse, or uncollimated, light field). plants occur with incredible diversity of species, size
Kd(z, l) clearly depends on the directional structure (from less than 1 mm to more than 200 mm), shape,
of the ambient light field, hence its classification as and concentration. Phytoplankton are responsible
an apparent optical property. Other diffuse attenu- for determining the optical properties of most
ation coefficients, e.g., Ku, Kod, or KPAR, are defined oceanic waters. Their chlorophyll and related
in an analogous manner, using the corresponding pigments strongly absorb light in the blue and red
radiometric quantities. In most waters, these K and thus, when concentrations are high, determine
functions are strongly correlated with the absorption the spectral absorption of sea water. Phytoplankton
coefficient a and therefore can serve as convenient, if are generally much larger than the wavelength of
imperfect, descriptors of a water body. However, visible light and can scatter light strongly.
AOPs are not additive, which complicates their in-
terpretation in terms of water constituents. Detritus Nonliving organic particles of various
sizes are produced, for example, when
phytoplankton die and their cells break apart, and
Optical Constituents of Seawater when zooplankton graze on phytoplankton and
leave cell fragments and fecal pellets. Detritus can
Oceanic waters are a witch’s brew of dissolved and
be rapidly photooxidized and lose the characteristic
particulate matter whose concentrations and optical
absorption spectrum of living phytoplankton,
properties vary by many orders of magnitude, so that
leaving significant absorption only at blue
ocean waters vary in color from the deep blue of the
wavelengths. However, detritus can contribute
open ocean, where sunlight can penetrate to depths
significantly to scattering, especially in the open
of several hundred meters, to yellowish-brown in a
ocean.
turbid estuary, where sunlight may penetrate less
than a meter. The most important optical constitu- Inorganic Particles
ents of sea water can be briefly described as follows.
Particles created by weathering of terrestrial rocks
Sea Water can enter the water as wind-blown dust settles on the
sea surface, as rivers carry eroded soil to the sea, or
Water itself is highly absorbing at wavelengths below
as currents resuspend bottom sediments. Such par-
250 nm and above 700 nm, which limits the wave-
ticles range in size from less than 0.1 mm to tens of
length range of interest in optical oceanography to
micrometers and can dominate water optical prop-
the near-ultraviolet to the near infrared.
erties when present in sufficient concentrations.
Particulate matter is usually the major determinant
Dissolved Organic Compounds
of the absorption and scattering properties of sea
These compounds are produced during the decay of water and is responsible for most of the temporal
plant matter. In sufficient concentrations these com- and spatial variability in these optical properties.
pounds can color the water yellowish brown; they A central goal of research in optical oceanography
are therefore generally called yellow matter or col- is to understand how the absorption and scattering
ored dissolved organic matter (CDOM). CDOM properties of these various constituents relate to the
absorbs very little in the red, but absorption increases particle type (e.g., microbial species or mineral com-
rapidly with decreasing wavelength, and CDOM can position), present conditions (e.g., the physiological
 RADIATIVE TRANSFER IN THE OCEAN 319

state of a living microbe, which in turn depends on chlorophyll profile measured in the Atlantic Ocean
nutrient supply and ambient lighting), and history north of the Azores in winter. The water was well
(e.g., photo-oxidation of pigments in dead cells). Bio- mixed to a depth of over 100 m. The chlorophyll
geo-optical models have been developed that attempt concentration Chl varied between 0.2 and
(with varying degrees of success) to predict the IOPs 0.3 mg m3 between the surface and 116 m depth; it
in terms of the chlorophyll concentration or other then dropped to less than 0.05 mg m3 below 150 m
simplified measures of the composition of a water depth. The water was oligotrophic, case 1 water, and
body. commonly used bio-optical models for case 1 water
were used to convert the chlorophyll concentration
to absorption and scattering coefficients (which were
Examples of Underwater Light
not measured). A scattering phase function similar in
Fields shape to those seen in Figure 3 was used for the
Solving the radiative transfer equation requires particles; this phase function had a backscatter
mathematically sophisticated and computationally fraction of bb/b ¼ 0.018. The second simulation was
intensive numerical methods. Hydrolight is a widely for an idealized, case 2 coastal water body con-
used software package for numerical solution of taining 5 mg m3 of chlorophyll and 2 g m3 of
oceanographic radiative transfer problems. The input brown-colored mineral particles representing resus-
to Hydrolight consists of the absorption and scattering pended sediments. Bio-optical models and measured
coefficients of each constituent of the water body mass-specific absorption and scattering coefficients
(microbial particles, dissolved substances, mineral were used to convert the chlorophyll and mineral
particles, etc.) as functions of depth and wavelength, concentrations to absorption and scattering co-
the corresponding scattering phase functions, the sea efficients. The large microbial particles of low index
state, the sky radiance incident onto the sea surface, of refraction were assumed to have a phase function
and the reflectance properties of the bottom boundary with bb/b ¼ 0.005, and the small mineral particles of
(if the water is not assumed infinitely deep). Hydro- high index of refraction had bb/b ¼ 0.03. The water
light solves the one-dimensional, time-independent was assumed to be well mixed and to have a brown
radiative transfer equation, including inelastic scat- mud bottom at a depth of 10 m. Both simulations
tering effects, to obtain the radiance distribution L(z, used a clear sky radiance distribution appropriate for
y, j, l). Other quantities of interest such as irradiances midday in January at the Azores location. The sea
or reflectances are then computed using their defin- surface was covered by capillary waves corres-
itions and the solution radiance distribution. ponding to a 5 m s1 wind speed.
To illustrate the range of behavior of underwater Figure 4 shows the component and total ab-
light fields, Hydrolight was run for two greatly dif- sorption coefficients just beneath the sea surface for
ferent water bodies. The first simulation used a these two hypothetical water bodies, and Figure 5

Case 1 Case 2
0.08 0.8
Absorption coefficient, a (m )
_1

0.06 0.6

Total

0.04 0.4
Total
Min
Water
Chl
0.02 0.2 Water
Chl

CDOM CDOM
0 0
400 500 600 700 400 500 600 700
Wavelength,  (nm) Wavelength,  (nm)

Figure 4 Absorption coefficients for the case 1 and case 2 water bodies. The contributions by the various components are labeled.
320 RADIATIVE TRANSFER IN THE OCEAN

Case 1 Case 2
0.30 3.0

Total
0.25

Scattering coefficient, b (m )
_1
0.20 2.0
Total Minerals
0.15
Chl Chl
0.10 1.0

0.05

Water Water
0 0
400 500 600 700 400 500 600 700
Wavelength,  (nm) Wavelength,  (nm)

Figure 5 Scattering coefficients for the case 1 and case 2 water bodies. The contributions by the various components are labeled
(CDOM is nonscattering).

00 m
th 1
th 0 Dep
log [radiance, L (W m sr nm )]

Dep
_1

_4
log [radiance, L (W m sr nm )]
_1

0
_1
_1

_5
_2

_1
_2

_6
_2
_7
_3
_8
_4 0
0
Vi _ 90
Vi

ew
ew

in _ 90
in

0
g 18
g

di 700
di

re 80
re

ct 1 90 6 00
700
ct

io 500 nm)
io

n, 90 600 ,(
n,

 50 0 (nm
) 0 400 g t h
elen
v

(d 400 th,  Wav

v
0
(d

eg e n g
eg

el
) Wav
)

Figure 6 The case 1 water radiance distribution in the azimuthal plane of the sun at depth 0 (just below the sea surface) and at
100 m.

shows the corresponding scattering coefficients. For Figure 6 and 7 show the radiance in the azimuthal
the case 1 water, the total absorption is dominated by plane of the sun as a function of polar viewing dir-
chlorophyll at blue wavelengths and by the water ection and wavelength, for selected depths. For the
itself at wavelengths greater than 500 nm. However, case 1 simulation (Figure 6), the depths shown are
the water makes only a small contribution to the zero, just beneath the sea surface, and 100 m; for the
total scattering. In the case 2 water, absorption by case 2 simulation (Figure 7), the depths are zero and
the mineral particles is comparable to or greater than 10 m, which is at the bottom. Note that the radiance
that by the chlorophyll-bearing particles, and water axis is logarithmic. A viewing direction of yv ¼ 0
dominates only in the red. The mineral particles are corresponds to looking straight down and seeing the
the primary scatterers. upwelling radiance (photons traveling straight up).
 RADIATIVE TRANSFER IN THE OCEAN 321

m
th 0 th 10
Dep Dep

log [radiance, L (W m sr nm )]

log [radiance, L (W m sr nm )]
_2

_1

_1
1
_1

_1
_3
_2

_2
0
_4
_1
_5
_2
_6
_3 0
0 Vi _ 90
ew
Vi

_ 90
ew

in 180
g
in

180 di 700
g

re 90 600 m)
di

700 ct 0
re

io 5 0  ( n
90 600 m) 400 length,
c

n, 0
tio

50 0  (n  e
400 length, Wav
n

0
,

e
v (d
av eg
v

W
(d

)
eg
)

Figure 7 The case 2 water radiance distribution in the azimuthal plane of the sun at depth 0 (just below the sea surface) and at 10 m.

Case 1 1
Case 2
2
10 10

Depth = 0 Depth = 0
Scalar irradance Eo (W m 2 nm )
_1

0 0
10 10
5m
_

_2 _1
10 100 m 10
10 m
_4 200 m _2
10 10

_6 _3
10 10

_8 _4
10 10

_ 10 _5
10 10
400 500 600 700 400 500 600 700
Wavelength,  (nm) Wavelength,  (nm)

Figure 8 The scalar irradiance Eo at selected depths for the case 1 and case 2 waters.

Near the sea surface, the angular dependence of the (  901ryvr901) at the bottom is isotropic; this is a
radiance distribution is complicated because of consequence of having assumed the mud bottom to
boundary effects such as internal reflection (the be a Lambertian reflecting surface. As the depth in-
bumps near yv ¼ 901, which is radiance traveling creases, the color of the radiance becomes blue for
horizontally) and refraction of the sun’s direct beam the case 1 water and greenish-yellow for the case 2
(the large spike near yv ¼ 1401). As the depth in- water. In the case 1 simulation at 100 m, there is a
creases, the angular shape of the radiance distri- prominent peak in the radiance near 685 nm, even
bution smooths out as a result of multiple scattering. though the solar radiance has been filtered out by the
By 100 m in the case 1 simulation, the shape of the strong absorption by water at red wavelengths. This
radiance distribution is approaching its asymptotic peak is due to chlorophyll fluorescence, which is
shape, which is determined only by the IOPs. In transferring energy from blue to red wavelengths,
the case 2 simulation, the upwelling radiance where it is emitted isotropically.
322 RADIATIVE TRANSFER IN THE OCEAN

0.015

Remote-sensing reflectance, Rrs (sr )

_1
Case 2
Case 1
0.010

0.005

0
400 500 600 700
Wavelength,  (nm)

Figure 9 The remote-sensing reflectance Rrs for the case 1 and case 2 waters.

As already noted, the extensive information con- 2 waters, in particular, is a research topic for the next
tained in the full radiance distribution is seldom decades.
needed. A biologist would probably be interested only
in the scalar irradiance Eo, which is shown at selected
depths in Figure 8. This irradiance was computed by
integrating the radiance over all directions. Although See also
the irradiances near the surface are almost identical, Bio-Optical Models.
the decay of these irradiances with depth is much
different in the case 1 and case 2 waters.
The remote-sensing reflectance Rrs, the quantity of
interest for ‘ocean color’ remote sensing, is shown in Further Reading
Figure 9 for the two water bodies. The shaded bars at
the bottom of the figure show the nominal SeaWiFS Bukata RP, Jerome JH, Kondratyev KY, and Pozdnyakov
sensor bands. The SeaWiFS algorithm for retrieval of DV (1995) Optical Properties and Remote Sensing of
the chlorophyll concentration uses a function of the Inland and Coastal Waters. New York: CRC Press.
ratio Rrs(490 nm)/Rrs (555 nm). When applied to these Caimi FM (ed.) (1995) Selected Papers on Underwater
Rrs spectra, the SeaWiFS algorithm retrieves a value of Optics. SPIE Milestone Series, vol. MS 118.
Bellingham, WA: SPIE Optical Engineering Press.
Chl ¼ 0.24 mg m3 for the case 1 water, which is close
Jerlov NG (1976) Marine Optics. Amsterdam: Elsevier.
to the average value of the measured profile over the Kirk JTO (1994) Light and Photosynthesis in Aquatic
upper few tens of meters of the water column. How- Ecosystems 2nd. New York: Cambridge University
ever, when applied to the case 2 spectrum, the Sea- Press.
WiFS algorithm gives Chl ¼ 8.88 mg m3, which is Mobley CD (1994) Light and Water Radiative Transfer in
almost twice the value of 5.0 mg m3 used in the Natural Waters. San Diego: Academic Press.
simulation. This error results from the presence of the Mobley CD (1995) The optical properties of water. In Bass
mineral particles, which are not accounted for in the M (ed.) Handbook of Optics, 2nd edn, vol. I. New
SeaWiFS chlorophyll retrieval algorithm. York: McGraw Hill.
These Hydrolight simulations highlight the fact Mobley CD and Sundman LK (2000) Hydrolight 4.1
that it is now possible to compute accurate under- Users’ Guide. Redmond, WA: Sequoia Scientific. [See
also www.sequoiasci.com/hydrolight.html].
water radiance distributions given the IOPs and
Shifrin KS (1988) Physical Optics of Ocean Water. AIP
boundary conditions. The difficult science lies in Translation Series. New York: American Institute of
learning how to predict the IOPs for the incredible Physics.
variety of water constituents and environmental Spinrad RW, Carder KL, and Perry MJ (1994) Ocean
conditions found in the world’s oceans, and in Optics. New York: Oxford University Press.
learning how to interpret measurements such as Rrs. Walker RE (1994) Marine Light Field Statistics. New York:
The development of bio-geo-optical models for case Wiley.
 TRANSMISSOMETRY AND NEPHELOMETRY
C. Moore, WET Labs Inc., Philomath, Oregon, USA between the values most certainly will differ depending
Copyright & 2001 Elsevier Ltd.
upon the composition of the materials in the water.
Using a transmissometer one can derive an at-
tenuation coefficient that mathematically describes
the ability of the water to transmit light. This co-
efficient is a fundamental optical characteristic and
an absolute quantity for a given medium. The scat-
tering sensor, on the other hand, collects a very small
Introduction portion of the scattered light and is usually calibrated
Transmissometry and nephelometry are two of the to some secondary standard. The units of measure-
most common optical metrics used in research and ment are themselves relative to that standard. Other
monitoring of the Earth’s oceans, lakes, and streams. differences also prove crucial in defining these meas-
Both of these measurements relate to what we per- urements. Limitations imposed by the instruments
ceive as the clarity of the water, and both provide themselves, application-specific requirements, sensor
vital information in numerous studies of natural sizes, and cost all play roles in determining the pos-
processes and human activities’ impact upon water sible suitability of one measurement versus another.
bodies. Applications involving these measurements Thus, in order to best fit these two methods to po-
range from monitoring drinking water suitability to tential applications, it is necessary to understand the
understanding how carbon is transferred into and measurements, the design of the sensors performing
transported within ocean waters. them, and the products that the sensors provide.
Transmissometry refers to measurements made by
transmissometers or beam attenuation meters. These
sensors infer the total light lost from a beam of light Measurements and Fundamental
passing through the water. These losses are caused by
Values
two primary mechanisms. Suspended particles and
the molecules of the water itself scatter the light In the realm of water sciences, transparency and tur-
away from its original path; the water, and dissolved bidity are two of the most commonly used terms in
and particulate matter contained within, absorb the describing optical clarity. These are general terms and
light and convert it into heat, photosynthetic activity, typically not tied to absolute physical quantities other
fluorescence, and other forms of energy. Larger than through the use of secondary standards. How-
concentrations of scattering and absorbing sub- ever, the set of underlying optical processes that de-
stances therefore result in greater losses in signal. scribe the impact of water-based media upon light
Nephelometry refers to measurements made by propagating through them are well defined, if not
optical scattering sensors, often referred to as tur- completely understood. In the study of the transmis-
bidity sensors or nephelometers. These sensors pro- sion of light energy through water, the inherent optical
ject a beam of light into the water and measure the properties (IOPs) refer to the set of intrinsic optical
radiant flux of light scattered into the direction of a characteristics of the water and components contained
receiver. Since the receiver signal increases with therein. The IOPs define how light propagates through
greater numbers of particles, the device infers the the water. In comparison to apparent optical prop-
concentration of suspended particles in the water. erties (AOPs), the other general class of in-water op-
Scattering sensors are used more commonly in en- tical measurements, the IOPs are not affected by
vironmental monitoring applications, especially in changes in the radiance distribution from sunlight or
highly turbid waters with large concentrations of par- other sources. The IOPs include coefficients for the
ticles; transmissometers see more use in general scien- attenuation, absorption, and scattering of light as well
tific studies. However, the uses for which they are as the volume scattering function.
employed broadly overlap. Nevertheless, transmiss- The coefficients of attenuation (c), absorption (a),
ometers perform quite different measurements from and scattering (b) determine radiance losses of a ray
those of scattering sensors and the quantities they of light propagating through the water. Light is either
measure are independent of one another and typically lost to absorption by the water and material con-
offer no direct comparison. In fact, while the data tained within or it is scattered by the same. The at-
products they provide may covary, the relationship tenuation coefficient accounts for losses attributed to

323
324 TRANSMISSOMETRY AND NEPHELOMETRY

both the absorption and the scattering and is equal to 3

the sum of these coefficients eqn [1]. Water

Dissolved organic matter

c¼aþb ½1 Particulate absorption

Particulate scattering
One determines the beam attenuation coefficient by
comparing the radiant flux of a collimated beam of Total non-water attenuation
2
light at source (Fs) with the radiant flux of the beam Total attenuation

Attenuation (m 1 )
at a receiver detector (Fd), a finite distance (r) away.

_
This ratio is known as the beam transmittance (T),
Wavelength of
given by eqn[2] or equivalently by eqn[3]. most common
transmissometers

Fd =Fs ¼ T ¼ ecr ½2

1
c ¼ lnðT Þ=r ½3

Here r is the path length between the source and the

receiver. This coefficient is the value ultimately de-
termined by a transmissometer. The attenuation co-
efficient is expressed in units of inverse meters (m1).
Thus, when one refers to water with an attenuation 0
coefficient of 1 m1, the implication is that within a 1 400 450 500 550 600 650 700
Wavelength (nm)
m path the available light within a collimated beam
is reduced to 1/e or approximately 37% of its ori- Figure 1 Relative contributions of water and non-water
ginal energy. scattering and absorbing components are seen in formulation of
Within the visible light spectrum the scattering and the attenuation coefficient within ‘typical’ waters.
absorption losses from the water itself remain ef-
fectively constant, and thus variability found in field VSF coefficient, bðyÞ, defines the radiant energy lost
measurements results from non-water particulate into a given angular region of the light scattering and
and dissolved matter. The extent of absorption-based is expressed in terms of inverse meters per steradian.
losses compared to scattering-based losses depend The VSF integrated over the entire spherical volume
both on the materials being measured and on the into which light is scattered provides b, the total
spectral configuration of the meters. Both the scat- scattering coefficient (eqn [4]).
tering and absorbing properties of water-based Zp
components are prone to variation with the wave- b ¼ 2p bðyÞsinðyÞdy ½4
length of light at which measurements are con- 0
ducted. Variations in the absorption depend heavily
upon the amount of colored dissolved organic matter The actual shape of the VSF depends upon the
(CDOM) and chlorophyll content. CDOM absorbs particle field being measured. Specific properties
very strongly in the blue wavelengths; chlorophyll that define this shape include the particle size and
absorbs heavily in the blue and in addition has a shape and the index of refraction. Particle size is
pronounced absorption peak in the deep red portion probably the single most pronounced factor in de-
of the spectrum (676 nm). Absorption by these ma- fining the VSF in that it dictates the regime of light
terials provides the appearance of color to the water. interaction with the particles themselves. Very small
Visually, CDOM laden waters tend to appear brown, particles that fall within the wavelength of the light
and chlorophyll-rich waters appear green. A deep impinging upon the particles are subject to mo-
blue cast to the water indicates very low levels of lecular or Rayleigh scattering. This interaction is
both of these substances. The spectral dependency of relatively weak, and creates a VSF that is relatively
the scattering signals is largely due to the size of the constant with angle. While Rayleigh scatterers are
particles from which the light is scattered (Figure 1). by far the most prevalent in most waters, most of
In addition to the optical loss coefficients, the the scattering signal seen by sensors is attributed to
volume scattering function (VSF) forms another im- particles ranging from 1 mm to 450 mm. The scat-
portant component of the IOPs in describing the fate tering behavior of these particles is typically mod-
of light in water. The VSF describes optical scattering eled using Mie theory. Mie theory uses Maxwell’s
as a function of the angle, y, away from the direction equations to predict perturbations of an incident
of propagation of the incident beam of light. The planar wave by spherical particles in its path. In
 TRANSMISSOMETRY AND NEPHELOMETRY 325

general, larger particles will create a greater degree intercalibrating with transmissometers (or by fil-
of near-forward scattering. tering and weighing) in natural waters.
Most scattering sensors are not considered tools
for determination of in-water optical properties, but
all scattering sensors including turbidity sensors Sensors
measure the VSF within a given angular region, Transmissometers
typically somewhere in the region of 90–1601 with
respect to the incident direction of the light. It is A basic transmissometer consists of a collimated light
perhaps ironic that while these sensors are among source projected through an in-water beam path and
the most ubiquitous of in-water optical tools, the then refocused upon a receiver detector. Typically
VSF is one of the least-characterized of all the IOPs. single-wavelength transmissometers employ a light-
This is because no single angle measurement can emitting diode coupled with an optical bandpass filter
account for the shape of the entire function. This in as the source. Source light is often split so that a por-
turn points to a major source of error in all tur- tion of the beam impinges upon a reference or com-
bidity-based measurements. Different materials dic- pensation detector that is either used in numerical
tate different VSFs and a single angle measurement processing of the data or integrated into a source sta-
will vary with concentration from one type of ma- bilization feedback circuit. The source output is often
terial to the next. In actual fact a diverse amalgam modulated and the lamp and receiver detector samples
of organic and inorganic particulates reside within are in phase with the source modulation. This greatly
most waters. This ultimately tends to homogenize reduces ambient light detection by the receiver from
the VSFs such that the variability in the VSF of the the sun or other unwanted sources. Path lengths are
composite is less than the variability of individual fixed with distances typically ranging from 5 cm to
components (Figure 2). 25 cm depending upon the waters in which the sensors
Most scattering measurements are based upon are used (Figure 3).
some standard such as formazin, diatomaceous The receiver detector converts radiant flux into
earth, or more recently spherical styrene bead sus- current and its output is thus proportional to the
pensions. These standards are used because they tend radiant energy passed through the water. Electronics
to be reproducible and easy to mix into various subsequent to the detector amplify and rectify the
concentrations for calibrations. Units of quantity are signal for digitization or direct output as a DC
expressed in form of turbidity units such as NTU voltage level. This signal is known as the instrument
(nephelometric turbidity units). Because of the dis- transmittance (Ti) (eqn [5]).
parate VSFs of these standards and natural waters,
total attenuation (or particle concentration) cannot Ti ¼ S  T ½5
be obtained from turbidity measurements without
S represents the instrument transmittance scaling
constant. This constant is a combined term that
_3
10
Open Ocean
1 23 11 12 13
4 5 6
San Diego Harbor 14
_4 California Coast 8 9 10
10
Normalized VSF

This is a common scattering

angle for many nephelometers
_5
10
7 15

_6 Figure 3 Cutaway view showing the primary optical

10 components found in a modern transmissometer. A transmitter
assembly and receiver assembly are mounted and aligned within
_7 a rigid frame. The transmitter assembly consists of (1) a source
10 lamp; (2) a pinhole aperture; (3) a collimating lens; (4) field
0 50 100 150 200 aperture; (5) an interference filter; (6) a beam splitter; (7) a
Angle (deg) reference detector; and (8) a pressure window. The beam (9)
then passes through a fixed-path volume of water and enters the
Figure 2 Normalized VSF data for three representative ocean receiver assembly. The receiver consists of (10) a pressure
water types. Note that at 901, the most common nephelometer window, (11) field aperture, (12) a refocus lens, (13) a pinhole
scattering angle, significant differences exist for the respective aperture, and (14) the receiver detector. Signals from the
coefficients. Data collected by Theodore Petzold and Seibert detector are then fed to the electronics for processing and
Duntley of Scripps Institute of Oceanography. output (15).
326 TRANSMISSOMETRY AND NEPHELOMETRY

includes signal amplification, losses through win- reduce utility of the longer path instruments. On the
dows and lenses, and other sensor gain factors. From other hand, shorter paths impose more demands
eqn[5] and assuming a 25 cm pathlength, we obtain than just high levels of precision in measurement.
eqn[6] or equivalently eqn[7]. Cleaning of optical surfaces also becomes a major
issue in maintaining sensor reproducibility and ac-
Ti =S ¼ ecð0:25Þ ½6 curacy. Again using the 25 cm path length instrument
as an example, maintaining signal reproducibility of
c ¼ 4 ln Ti  Q ½7 0.01 m1 over time requires a cleaning technique
that gives results that repeat within 0.25% trans-
The constant Q ¼ 4 ln S is a general scaling term that mittance. For a 10 cm path length instrument, re-
is removed, or compensated for, during the cali- peatability would need to be within 0.10%
bration process. transmittance. Likewise, internal correction mech-
An ideal transmissometer would reject all but the anisms such as compensation of temperature-related
parallel incident light into its receiver. This implies drift impose stringent requirements upon the sensor’s
that there is no error associated with near-forward electronics as well as the subsequent characterization
scattered light getting into the receiver. However, process. Long-term drift and general mechanical
limitations in real-world optics make this a near stability also must be tightly constrained for the in-
impossibility. Transmissometers thus provide a value strument to provide accurate results over time. The
for a system attenuation coefficient that has a finite requirements prove challenging in light of the forty
scattering error and is defined primarily by the ac- degree (centigrade) temperature swings and the 6000
ceptance angle of the receiver optics. These values meter depth excursions to which the instruments
range from around 0.51 to 11 in water for most potentially get exposed.
commercial instruments. Because that VSF for in- While the calculation of the attenuation coefficient
water particles is highly peaked at these angles, this from raw transmittance is independent of the cross-
can result in underestimation of the attenuation co- sectional area of the beam, the beam size does play
efficient and can also lead to sensor-to-sensor dis- an important role in the transmissometer’s ability to
crepancies in measurement. It thus becomes measure. Accurate transmittance measurements rely
important to know this angle in treating data care- upon the water and the materials it contains acting as
fully. While it is possible to build sensors with nar- a homogenous medium. This model starts to break
rower acceptance angles than 0.51, scattering in the down in two important cases: when the number
very near-forward direction becomes dominated by concentration of particulates becomes significantly
turbulent fluctuations in the density of the water it- low compared to the total volume of the illuminated
self. This turbulence-induced scattering is irrelevant sample area; and when the particle sizes become
to particulate studies and, depending upon the dis- significantly large in comparison to the cross-sec-
tances and receiver sizes involved, to most signal tional area of the beam. Taken in the extreme, one
transmission applications. can easily imagine a very narrow beam providing a
The conceptual framework for the transmiss- binary response at the receiver depending upon
ometer measurement involves starting with a full whether a particle occludes its path. Practically
signal and monitoring small negative deviations from speaking, most transmissometers need to show
it. The sensitivity of the instrument thus depends minimal spiking for particle sizes up to 100 mm
upon its ability to resolve these changes. In many diameter. Particles more than a few micrometers in
oceanic and other clear water investigations, signal diameter are ‘seen’ by the receiver at about two times
changes as small as 0.001 m1 become significant. In their actual size as a result of diffraction. This means
a 25 cm instrument this implies a requirement for for a beam of 5 mm nominal width that a single
transmittance resolution on the order of 0.025%. At 100 mm particle could reduce signal at the receiver by
the other end of the environmental spectrum, many approximately 0.08% or on the order of 0.0032 m1
inland waterways and some harbor areas would in a 25 cm path (or 0.008 m1 in 10 cm path). This
render a 25 cm path instrument ineffective due to proves acceptable for most operational conditions.
loss of all signal. Therefore, range and resolution On the other hand, a 1 mm particle could create an
become the two critical factors in determining a 8% deviation in sensor output, creating a noticeable
transmissometer’s effectiveness in a given appli- spike. Fortunately, 1 mm particles are extremely rare
cation. While it is easy to imagine using arbitrarily except in active erosion zones.
long path lengths to obtain increased sensitivity, the There are presently two primary methods used in
instrument path begins to impose other limitations calibrating transmissometers. The first uses funda-
upon its utility. Size and mechanical stability both mental principles of beam optics and knowledge of
 TRANSMISSOMETRY AND NEPHELOMETRY 327

the index of refraction difference between air and Ultimately, reproducibility of results becomes
water to directly estimate the sensor output. Electro- the best check for calibration. That said, this level
optical linearity in response to signal changes is as- of accuracy is really only required in conditions
sumed or verified. The sensor’s gain level is set near where particle concentrations are approaching
full scale for transmission in air and the sensor is minimal levels. Relative changes of transmittance
checked to ensure that if the source output is com- will still be precisely reflected in the instrument’s
pletely blocked it provides a real zero output. Ac- measurements.
counting for the differences in reflection and
transmission of the air–glass interfaces compared to
the water–glass interfaces, one can then assume that, Scattering Sensors
upon immersion, any further deviations in signal are
due to the attenuation of the water and materials A simple scattering sensor consists of a source
contained therein. This measurement is then verified element projecting a beam of light in the water and a
by immersion in clean water and subsequent com- receiver detector positioned at a fixed angle with
parison to clean water values. Error terms in this respect to the source. The source beam is sometimes
method usually include deviations of the modeled stabilized by inclusion of a second receiver that re-
optics from the real world. These errors include lens- ceives a portion of the light coming directly out of
induced focusing aberrations, alignment issues, the lamp. This signal is then fed back into the lamp
spectral content of the source, and any dust or film driver circuitry to compensate for fluctuations in the
on any of the optical components. The primary ad- source with time and temperature. The source beam
vantages of this method are that the calibration has a defined primary projection angle and a distri-
process relies only upon the air value measured by bution of light about that angle. Conversely, the re-
the meter, and that the attenuation due to the water ceiver is placed at a specific angle and maintains a
is included in the water-based measurements. defined field of view about that angle. These factors
The second method involves blanking the meter combine to form the distribution of angular response
directly with clean water. More akin to calibration for the scattered light (Figure 4).
approaches used in spectrophotometry, this method As with transmissometers, it is necessary to reject
involves immersion of the instrument into optically ambient light from the sun and other non-sensor
clean water, measuring the value, and setting that sources during measurement. With scattering sensors
value as full-scale transmittance or, conversely, this is achieved both through the use of synchron-
0.000 m1 attenuation (clean water values for the ously modulated light and detector amplification and
attenuation can then be added back in accordance also through the use of direct optical rejection. Direct
with published values). The chief disadvantage of optical rejection is employed at the source through
this method lies in the difficulty of creating and the use of relatively narrow spectral band sources
verifying optically clean water. While various levels that emit light in the infrared away from the water-
of filtering can remove most of the particulates from penetrating wavelengths of sunlight. Accordingly the
the water, filters can also introduce bubbles. These receiver incorporates narrowband optical filters that
bubbles are seen as particles by the sensor. Assuming reject wavelengths away from the primary emission
that one achieves filtration without introducing any bands of the source.
bubbles, bubble creation is still a concern in that any
partial pressure imbalances between the gases con-
tained within the water and the surrounding en-
vironment will result in subsequent bubble
formation. Added to that is the possibility that the
containers and the sensors themselves may also act as
sources of particulate contamination. The chief ad-
vantage of this method is that it accommodates for
small deviations in the real instrument with respect
to the ideal.
The overriding issue with calibration of trans-
missometers is the same as in the discussion of the
need for and difficulty of proper cleaning. In order
to calibrate an instrument to operate accurately in
cleaner waters, the calibrations must achieve ac-
curacy to within 0.25% of full-scale measurement. Figure 4 Typical scatter sensors and transmissometers.
328 TRANSMISSOMETRY AND NEPHELOMETRY

Specific angular configurations used in modern effective volume of interaction is strictly a function
scattering sensors vary widely. Some sensors are de- the attenuation coefficient in that it is infinite other
signed to operate within a highly constrained, nar- than for induced losses of light. As with transmiss-
rowband, angular relationship, and some are ometers, the volume of interaction also affects a
designed to collect as much scattered light as possible scattering sensor’s sensitivity and the effect of larger
and thus encompass a very wide angular range. In particles upon the signal. Small volumes show less
general two truths hold for all the designs: they will sensitivity and measure larger particles as signal
all provide a roughly linear response that is pro- spikes. The combined issues of long-path attenuation
portional to the particle concentration (at least in coupling and volumetric sensitivity point to the
low to moderate concentrations); and different op- preference of designs incorporating larger beams
tical configurations will demonstrate different abso- with greater interaction volumes for measurements
lute response curves with respect to each other even of cleaner waters and narrower beams with inter-
when calibrated with the same standard (Figure 5). action volumes close to the sensor surface for use in
A scattering sensor works by the simple principle highly turbid waters.
that when particles are present they will scatter light The response of a given scattering sensor is very
and the receiver will collect some of that light. Using highly dependent upon its specific optical con-
Beer’s law, which states that increasing concen- figuration. Angle of interaction, angular distri-
trations will result in a linear increase in output bution, wavelength at which the source emits, and
signal, the sensor’s output varies from a zero value in the relative path distance from the source and back
clean water to a full-scale value at the upper end of to the receiver are all factors in how a sensor will
its range. While it is convenient to assume a linear behave. As mentioned earlier, it should be expected
response with concentration, this is not strictly true. that two different designs will provide two different
Light reaching the volume of interaction and the responses. In studies in which researchers require
light scattered back into the detector is subject to only relative responses with space or time, this is
secondary losses due to attenuation. As the concen- not a major issue. A twofold change in a given
tration of scattering components in the water in- concentration of particles will generate an associ-
creases, so does the attenuation. This produces a ated response in the instrument output. However,
nonlinearity in the output signal. In sensors with many studies require some form of reproducible
large interaction volumes and a wide angular re- results. It is not enough that two sensors are cali-
sponse, this becomes a particularly messy analytical brated to the same medium. They must also re-
problem in that the light is subject to a large range of spond in the same way to any other medium that
effective path lengths in propagation from the source they might mutually measure. Standards such as
and back to the receiver. In the extreme case, sensors ISO 7027 have been published. These standards
exist that position a near-isotropic source next to a impose constraints on the angle of interaction be-
wide-angle detector such that they both project out, tween the source and the receiver (901), the angular
perpendicular to the same plane. In these sensors the distribution of the source, and its wavelength of
operation, as well as other design parameters. The
goal is to ensure that all sensors built within the
constraints imposed by the standard will provide
similar results in similar waters. This is a very
important step toward achieving consistent results
amenable to intercomparison.
Straightforward in concept, sensor calibration
employing a standard suspension, provides several
pitfalls in practice. First and foremost, no calibration
can be achieved to better accuracy than the standard
solutions themselves. Secondly, it is critical to ensure
that the container in which the calibration takes
place is not a cause of secondary reflections of light
that can get back to the receiver. Care must also be
taken to ensure that the suspension is not settling or
Figure 5 One of many possible optical configurations for a
flocculating during the measurement. Finally, one
scattering sensor. A source assembly consisting of a LED lamp,
reference detector, lens, and right angle prism projects light into variation of this technique is to use arbitrary con-
the water. The receiver is placed to receive light at 901 with centration of the calibration media and calibrate
respect to projected source beam. against another ‘standard’ precalibrated sensor.
 TRANSMISSOMETRY AND NEPHELOMETRY 329

Great care should be applied when using this agents with well-defined VSFs. These two techniques
method. Standard sensors often already incorporate address different elements of a sensor calibration and
compensation schemes for linearizing the data. These may well find optimum effectiveness when used in
schemes in turn are developed for use with a specific conjunction with one another (Figure 5).
type of suspension. This can create dramatic and
surprising results when using another suspension.
While scattering sensors are predominantly used Applications
to determine relative concentrations of particulates,
Domains of Use
another very important set of applications involve
characterization of the volume scattering function The use of transmissometers and nephelometers falls
itself. One of the important goals in observational broadly into two categories. We want to study the
oceanography involves the use of remotely sensed water’s optical properties and how they might relate
data from satellites and other airborne platforms to to ongoing processes occurring in the water, and we
rapidly characterize large areas of surface and near want to determine how much foreign matter is in it.
surface waters. Of particular interest are the emer- While, ultimately, both thrusts of study lead to
ging methodologies associated with using ocean measurement of the same media within a given body
color data captured from airborne and space-borne of water, the products that the instruments provide
platforms to provide information about the biology differ, and the requirements surrounding the given
and chemistry of waters. In the United States, NASA areas of study tend to drive the development of the
projects such as the Coastal Zone Color Scanner and different technologies. The factors ultimately deter-
the more recent SeaWiFs satellite program stimu- mining the appropriateness of one sensor versus an-
lated this interest, and in the case of SeaWiFs con- other do not always pertain to the data products
tinue to contribute a growing body of information. provided. Size, cost, ease of deployment, ease of
The light that these platforms receive is a function of maintenance, and researcher’s experiences all con-
the sea surface state and the resultant reflections and tribute to decisions on which type of sensor is the
the water-leaving radiance. This radiance in turn is best to use.
defined by the absorption and scattering character- Optical oceanographic research motivated much
istics of the water. Scattering in the region of of the development of modern transmissometers.
90–1801 is specifically important because it repre- This arena also stimulated development of scattering
sents incoming light from the sun that is scattered sensors that are specifically designed and calibrated
back into the atmosphere. To quantify this, a class of for providing coefficients related to the VSF. Much of
sensors called optical backscattering sensors have this work in the United States revolves around Naval
been developed and calibrated specifically for this research needs, and primary development of sensors
purpose. In many respects these sensors are very now available commercially was in large part funded
similar to other scattering sensors in that they use the through Naval research dollars. Naval applications
same basic optical configurations and respond simi- include mine hunting, underwater tactical assessment
larly to variations in the particle field. The major for diving operations, and sea truthing for laser
differences involve design constraints upon the communications and imaging research. The US Na-
wavelengths of the source emitters and the angles of tional Aeronautics and Space Administration
interaction. Equally importantly, the calibration of (NASA) has also played a major role in developing
these sensors involves tying the sensor response dir- underwater tools for optical characterization. These
ectly to the volume scattering function. tools help calibrate the airborne sensors. Similarly,
Calibration of scattering sensors for radiometric numerous other governments foster the development
measurements involves detailed knowledge of the and use of these tools through their respective Naval,
sensor optics geometry and some known scattering space and other scientific agencies. While not en-
agent. The prevalent method for single-angle meas- gaged in the study of ocean optical properties per se,
urements incorporates a sheet of highly reflective many other ocean scientists working under aegis of
diffuse material and maps the sensor response as a funds supplied by these agencies use transmiss-
function of the distance between the target and the ometers and optical backscattering sensors in on-
sensor. This information is then applied to derive the going efforts to understand physical, biological, and
angular weighting function of the interaction vol- chemical distributions and processes in the water.
ume. Finally, this weighting function is applied to a Scattering sensors remain the dominant optical
typical ocean water VSF. More recently, researchers tools used by environmental researchers. These sen-
have begun to apply a calibration technique that sors’ size and cost make them widely affordable and
incorporates known concentrations of scattering easily used, and the newer sensors incorporate
330 TRANSMISSOMETRY AND NEPHELOMETRY

fouling-retardant features such as shutters and bio- flavors of these various platforms are now emerging.
cidal exposed surfaces. As such they are becoming Some will find important niches for acquisition of
increasingly subscribed to as the sensor of choice in data over space and time.
compliance-driven monitoring applications de-
veloped by various governmental agencies through-
Some Current Applications
out the world. Naturally, the more attractive size and
costs of scattering sensors also make them favorable There are many different applications engaging the
choices in many larger-scale applications. use of transmissometers and scattering sensors.
It is likely that remote sensing will to some degree Table 1 represents only a sampling across numerous
change preferences for sensors among fresh water disciplines.
researchers over the next ten years. Presently there is
relatively little airborne color data available for fresh
Extending Capabilities
water bodies, and thus many limnology researchers
have not yet been compelled to measure optical As mankind’s need to understand and monitor the
properties of lakes directly. With the next generation Earth’s waters has increased, they have driven the
color airborne sensors and new governmental man- development of more rugged, more reliable, smaller,
dates driving more effective broader-scale sampling and cost-effective technologies for transmissometry
strategies, the need and desire for transmissometer and nephelometry as measurement techniques. These
measurements and scattering measurements for VSF resultant technologies have not only carved greater
determination will undoubtedly grow. roles for optical measurement methods but have also
proved seminal in the development of entirely new
sensors. Recently, a new generation of IOP tools has
How Sensors are Deployed
been made available to the oceanographic com-
One major constraint in an underwater sampling is munity. They include sensors for the determination
how to use the instrumentation effectively in the of the in-water absorption coefficient, multiangle
environment for which it is intended. Researchers scattering sensors, and a set of IOP tools with spec-
often want to measure the water in places they can- tral capabilities. Transmissometers and simple scat-
not easily get to, or over timescales that make per- tering sensors have laid the foundation for the optical
sonal attendance of equipment an unappealing techniques and data methods of these new devices. In
proposition. To these considerations must be added turn, these new sensors promise to significantly en-
the requirement that the data gathered must truly hance the role of IOP measurements in modern ob-
reflect changes at the time and space scales of the serving platforms.
governing processes within the water column, and One of the more significant recent breakthroughs
the constraint imposed by doing this sampling at a in optical measurement techniques lies in the devel-
reasonable cost. The sampling challenge becomes opment of the absorption meter. This sensor uses a
formidable. As a result, the development of effective measurement method and optical geometry similar
sampling platforms has become as challenging and to a transmissometer except that it encompasses the
competitive a discipline of research as instru- sensor’s beam path with a reflective tube and in-
mentation design itself. corporates a large-area detector at the receiver end of
Transmissometers and scattering sensors are typi- the path. The reflective tube and large-area detector
cally integrated into multiparameter sampling pack- combine to collect the bulk of the light scattered
ages for acquiring and storing data (CTDs, data from the source beam. Thus the light not detected is
sondes, loggers). The packages are then deployed primarily due to absorption by the water and its
from boats or other platforms and lowered through constituents.
the water column, travel on or are towed by a vessel, The wide-band spectral nature of sunlight coupled
or are placed on buoys or mooring lines in order to with the selective filtering capabilities of water and
log measurements over an extended period. Many the absorption characteristics of phytoplankton and
variations of these basic methods exist but virtually dissolved organic material make spectral optical
all entail these basic concepts. characterization of the water highly desirable. Like-
A new class of autonomous deployment platforms wise, the spectral information from the scattering of
will serve to revolutionize underwater sampling. particles provides more direct correlation with re-
These range from miniature programmable under- mote color data as well as a more complete de-
water vehicles, to freely drifting ocean profilers that scription of the type of particles scattering. New
can continuously move through the water column, tools encompassing spectral attenuation, absorption,
and to rapidly deployable profiling moorings. Many and scattering are now commercially available.
 TRANSMISSOMETRY AND NEPHELOMETRY 331

Table 1 Applications of transmissometry and nephelometry

Application Description

Monitoring terrestrial runoff Scattering sensors stationed in rivers and streams allow researchers to determine impacts of inflows
and impact of industrial upon water quality. Inflows might be created by logging, agriculture, mining, land development,
inflows on water quality controlled and uncontrolled outflows from water treatment plants, natural runoff and other events
that introduce new matter into the monitored bodies.
Compliance monitoring United States’ compliance monitoring of fresh water bodies is soon likely to include turbidity as a
required parameter for ongoing measurement.
Determining biological Both transmissometers and scattering sensors are deployed in viewing the biological variability in
distribution in the water space and time through the water column.
Radiative transfer studies – In verifying the optical relationships between the inherent and apparent optical properties,
optical closure researchers seek to test the relationships through direct measure and comparison of values from
the disparate instrument types. Scientists also seek to reconcile measurements of the inherent
properties among themselves in validation of IOP theory.
Remote sensing validation Satellite and other airborne remote imaging systems require in-water transmissometry, scattering,
and absorption measurements to calibrate these sensors to water-borne optical properties.
Studying the benthic layer In understanding the processes effecting the settling and re-suspension of particles near the bottom
processes of the water column both scattering sensors and transmissometers can provide relative
indications of particle flux.
Frazil ice formation Transmissometers have been shown to ‘see’ signal fluctuations associated with the formation of
frazil or supercooled ice. These studies are imperative in understanding how polar ice sheets are
formed.
Diver visibility Navies require better tactical assessment of waters for determining operational risk for divers and
other visibility-related operations.
Small-scale structure in the In coastal regimes many physical and ecological processes take place on smaller time and space
water column scales than previously thought. The speed of acquisition and sensitivity of modern scattering
sensors and transmissometers allow accurate particulate mapping within the water column, which
in turn serves as a tracer for these processes.
Tracking particulate Data from transmissometers has been shown to accurately reflect total particulate organic carbon
organic carbon within the water column. Understanding in-water carbon transport processes is, in turn, vital to
understanding carbon flux between the water and atmosphere through the uptake and output of
CO2.
Tracking bloom cycles Transmissometers on moorings located both in open ocean and in coastal areas track seasonal
bloom cycles as well as event-driven changes from major storms or other potential system
disturbances.
Monitoring activity around Scattering sensors on moorings and underwater vehicles track plumes from underwater vents and
thermal vents and eruptions.
underwater volcanoes

These tools are playing increasingly important roles a new set of multiangle scattering sensors is now
in various applications. coming into commercial availability.
Despite the plethora of scattering sensor data Other development efforts and new instru-
available, very little information exists concerning mentation incorporate scattering and transmittance
the range of variability of the VSF, and how it relates measurements in unique ways to obtain specific
to different water masses and the processes within underwater chemical and biological components.
them. One of the chief constraints in fully charac- One example of these includes an underwater
terizing the VSF is that it requires a multiangle transmissometer that uses polarized light to deter-
scattering measurement encompassing in excess of 4 mine concentrations of particulate inorganic carbon.
orders of magnitude of scattered light intensity. After These instruments promise to fill a vital niche in
some seminal work performed by researchers at the understanding the fate of carbon in the seas. Another
Scripps Institute of Oceanography during the late example in development are underwater flow cyt-
1960s and early 1970s, very little has since been ometers. While the prevalence of IOP measurements
done to add to this body of data. In fact, VSF func- look at bulk phase phenomena, new instruments are
tions measured then remain de facto calibration now available as ship-board and dock mounted units
standards for instruments being built today. In recent that couple scattering and fluorescence measure-
years researchers in Europe and the United States ments of individual cells and organisms to provide
have refocused attention upon this issue. As a result, identifying signatures. Patterned after laboratory
332 TRANSMISSOMETRY AND NEPHELOMETRY

flow-cytometers, the in water devices will offer sensors that extend IOP capabilities. These new
break-through capability in typing specific organisms sensors hold the ability to determine absorption co-
in their natural environment. efficients, to determine coefficients as a function of
One of the most exciting aspects of the recent wavelength, and to characterize the volume scat-
advancements in IOP-related technologies lies in the tering function at more than one angle. These im-
opportunities offered by their combined use. One provements not only allow more complete
marked example lies in the characterization of par- characterization of natural waters but also provide a
ticle aggregations in the water. While the attenuation tangible means of relating remotely sensed data from
or scattering at one wavelength will provide data air and space to in-water processes.
about relative concentrations of particles within the
water column, spectral data from these sensors
combined with absorption measurements can move See also
us a long way toward characterizing the aggregation
into various biological and inorganic components. Optical Particle Characterization. Radiative
Transfer in the Ocean. Satellite Remote Sensing
Microwave Scatterometers. Turbulence Sensors.
Summary
Transmissometry and nephelometry provide in-
creasingly valuable information relating to the light- Further Reading
transmitting characteristics of water as well as an
Bogucki DJ, Domaradzki JA, Stramski D, and Zaneveld
idea of the relative concentration of suspended ma-
JRV (1998) Comparison of near-forward light
terial within lakes and oceans. While sometimes scattering on oceanic turbulence and particles. Applied
viewed as near-synonymous techniques, these meth- Optics 37: 4669--4677.
ods use different measurement methods, provide Bricaud A, Morel A, and Prieur L (1981) Absorption by
different products, and have different strengths and dissolved organic matter of the sea (yellow substance) in
weaknesses in considering the applications to which the UV and visible domains. Limnology and
they are applied. Applications vary widely and across Oceanography 26: 43--53.
numerous disciplines, but tend to be divided into two Greenberg AE, Clescerl LS, and Eaton AD (eds.) (1992)
major classes: those that attempt to characterize the Standard Methods for the Examination of Water and
fundamental optical properties of the water; and Wastewater 18th edn. Washington, DC: American
Public Health Association, AWWA, WEF.
those that seek the relative concentrations of foreign
Jerlov NG (1976) Marine Optics. Amsterdam: Elsevier.
particulate matter in the water. In general, nephel- Kirk JTO (1994) Light and Photosynthesis in Aquatic
ometry is the preferred technique in environmental Ecosystems. Cambridge: Cambridge University Press.
and fresh water applications and transmissometry is Mobley CD (1994) Light and Water: Radiative Transfer in
more common in oceanographic research. Although Natural Waters. New York: Academic Press.
transmissometry and nephelometry differ as meas- Pegau WS, Paulson CA, and Zaneveld JRV (1996) Optical
urement techniques, in their application domains, measurements of frazil concentration. Cold Regions
and in subsequent calibration and handling, all of Science and Technology 24: 341--353.
these sensors are capable of providing outputs in Petzold TJ (1972) Volume Scattering Functions for Selected
terms of absolute coefficients that describe the fate of Ocean Waters. Reference Publication 72–28. La Jolla,
light passing through water. These coefficients of CA: Scripps Institute of Oceanography.
Tyler JE, Austin RW, and Petzold TJ (1974) Beam
light transfer are collectively known as the inherent
transmissometers for oceanographic measurements.
optical properties or IOPs. Their values are related In: Gibbs RJ (ed.) Suspended Solids in Water. New York:
through the volume scattering function that describes Plenum Press.
scattering as a function of angle into which light is Zaneveld JRV, Bartz R, and Kitchen JC (1990) A reflective-
deflected. While these sensors play an increasing role tube absorption meter. Ocean Optics X, Proceeding of
in observing in water processes, they also provide a the Society for Photo-Optical Instrumentation and
technological foundation for a new generation of Engineering 1302: 124--136.
SENSORS: CHEMICAL
 WET CHEMICAL ANALYZERS
A. R. J. David, Bere Alston, Devon, UK discrete samples into a continuous flowing carrier
Copyright & 2001 Elsevier Ltd.
stream by a pumping system. Reagents are added by
continuous pumping and merging of the sample
carrier and reagent streams. The sample carrier
stream is segmented with air before reagent addition,
which typically allows between 20 and 80 samples to
Introduction be processed in an hour. The insertion of standards in
the sample carrier stream provides regular datum
Since the early 1960s there has been a requirement points during a particular analysis. There is usually
for seawater laboratories to carry out increasing no problem with distinguishing between the samples
numbers of routine analyses, many of which were at the detection stage as the regular timing between
performed by traditional manual methods. The per- stages is controlled. However, unless precautions are
formance of standard manual methods was generally taken to prevent carryover, interaction can occur in a
variable due to human error and the efficiency was continuous system causing loss in discrimination
poor. Method automation has since enabled in- between successive samples at the detection stage.
creased numbers of samples to be analyzed with Figure 1 shows a typical air-segmented CFA
improved efficiency and reduced the risk of human manifold arrangement for total oxidized nitrogen
error. Air-segmented continuous flow analyzers (TON) determinations. It can be seen that the
(CFA) and flow injection analyzers (FIA) have han- manifold is relatively complex, with eight separate
dled the bulk of this automation. Instrument manu- streams pumped at fast flow rates. The sample
facturers have continued to improve both hardware stream is air-segmented prior to merging with the
and software over the years, which has resulted in ammonium chloride (10 g l1) carrier stream and the
better reliability and analytical performance. bubbles are removed by a debubbler before entering
the cadmium wire reduction coil. Air-segmentation is
Air-Segmented Continuous Flow then re-introduced into the sample stream before
merging with the separate sulfanilamide and
Analyzers N1NED streams. A second debubbler finally re-
These instruments are in widespread use for the de- moves the bubbles before entering the flow cell.
termination of nutrient concentrations in natural Sophisticated laboratory CFA systems are used in
waters. The technique is based on the fundamental shipborne laboratories for the routine determination
principles developed in 1957 and converts a series of of nitrate, nitrite, silicate, phosphate, and ammonia

1
Flow rate: ml min

0.10 Debubbler
N1NED
Flow cell
0.32
Air
Re-sample 0.80
0.10
Sulfanilamide
Debubbler

0.10 Cu/Cd reduction column

Air
0.32
Sample in
Ammonium 0.20
chloride
0.60
Waste
Return

Figure 1 Schematic diagram of a typical air-segmented continuous flow analyzer manifold (for total oxidized nitrogen).

335
336 WET CHEMICAL ANALYZERS

in sea water simultaneously. Although the technol- case of reverse flow injection) into a liquid carrier
ogy is now very mature and the instruments perform stream flowing to the detector, where the analyte
extremely well in true laboratory conditions, the forms a colored species in the reaction zone, which
same cannot be said for shipborne applications. contains both sample and reagent. The reaction does
The main drawback with air-segmented systems is not have to go to completion, and as the flow is
the need for a reproducible bubble pattern, which incompressible, the extent of the reaction will be
can be difficult to achieve when at sea in rough similar in all samples. This gives significant advan-
conditions. The air bubbles are compressible and tages over air-segmented CFA, including faster an-
therefore will create pulsations in the system and alysis rates and less complicated equipment, which
for most detector designs the bubbles have to be has resulted in FIA being readily adapted to the an-
removed to avoid flow problems within the cell. alysis of seawater nutrients. In the late 1970s,
Multichannel systems are complex and con- methods were described for the simultaneous de-
sequently require specialist knowledge of the system termination of nitrite and nitrate by FIA. This
in order to achieve optimum performance in adverse allowed up to 30 samples per hour to be analyzed
conditions, which is not always possible when op- with a relative precision of 1% in the range
erating a watch system on a research cruise. How- 0–0.05 mM for Nitrite-N and 0–0.1 mM for Nitrate-
ever, these instruments have been in use for many N. Similar work was carried out later that year,
years with standard validated methods and generally where up to 90 samples were analyzed with a relative
they are the standard by which other techniques are precision of 0.5 and 1.5% for nitrite and nitrate
judged. respectively; in the range of 0.1–0.5 mg l1
In addition to the standard methods applied to for Nitrite-N and 1–5 mg l1 for Nitrate-N. Further
CFA, various techniques have been developed over developments of this method, which utilized
the years in order to eliminate some of the problems copperized cadmium wire in a pre-valve in-valve
associated with air-segmented continuous flow tech- reduction technique, allowed synchronous de-
niques, e.g. the use of EDTA (ethylene diamine tetra- terminations of nitrite and (nitrite þ nitrate) using
acetic acid) to segment the carrier stream instead of one manifold and detector. In the early 1980s,
air. The technique is also very flexible, allowing methods were described which were modifications to
customized methods to be developed for a variety of previous methods that utilized reverse flow injection
additional chemical species and more sensitive analysis, whereby the sample was the carrier stream
methods for the normal nutrient species. For ex- and reagents were injected into it. This method gave
ample, methods for the determination of nanomolar a limit of detection (LOD) of 0.1 mM and allowed up
concentrations of nitrate and nitrite in sea water to 70 samples per hour to be analyzed with a relative
using an air-segmented CFA system have been precision of 1%. FIA is finding increased use in the
developed. water industry, where laboratories that had previ-
Air-segmented CFA has been widely used at sea for ously used air-segmented CFA have introduced FIA
the analysis of all major nutrient species and until the to complement their working practices. FIA is the
mid-1980s was the only technique that was available preferred technique for small batch sizes and low-
at a reasonable cost for automated analysis. level concentrations, where speed of analysis is es-
sential to eliminate the risk of airborne contamin-
ation. FIA techniques are also readily adaptable to
online monitoring of a watercourse.
Flow Injection Analysis
Over the past 20 years commercially available FIA
Flow injection analysis (FIA) techniques were de- instruments have established flow injection analysis
veloped to overcome some of the practical problems as a reliable technique with the level of sensitivity for
associated with air-segmented CFA that were per- monitoring micronutrient species in the environ-
ceived by some workers. Flow injection analysis ment. Like CFA, it has also been accepted as a
differs from air-segmented CFA in that the sample is standard method for the examination of waters and
injected directly into a moving liquid carrier stream associated materials.
without the addition of air. The main distinction FIA is also finding increasing applications in re-
between air-segmented CFA and FIA is that the search, routine analysis, teaching of analytical
continuous mixing of sample and reagents in a tur- chemistry, monitoring of chemical processes, sensor
bulent stream segmented by bubbles is replaced by testing and development, and enhancing the per-
the periodic mixing in an unsegmented laminar formance of various instruments. It is also used for
stream. The periodic mixing in a FIA system is the measurement of diffusion coefficients, reaction
achieved by injecting the sample (or reagent in the rates, stability constants, composition of complexes
 WET CHEMICAL ANALYZERS 337

and extraction constants and solubility products. which is sensed by a flow-through detector and
The versatility of FIA has allowed the technique to be recorded.
adapted to different detection systems such as elec- The height, width, and areas of a typical peak
trochemistry, molecular spectroscopy, and atomic output from a simple FIA system are all related to the
spectroscopy, using numerous manifold configur- concentration of the analyte. The time span between
ations. FIA systems can also be designed to dilute or the sample injection and the peak height is the resi-
to preconcentrate the analyte; to perform separations dence time during which the chemical reaction takes
based on solvent extraction, ion exchange, gas dif- place. With rapid response times, typically in the
fusion or dialysis; and to prepare unstable reagents range of 5–20 s, a sampling frequency of two samples
in situ. per minute can be achieved. The injected sample
Figure 2 shows a typical FIA manifold arrange- volumes may be between 1 and 300 ml, which in turn
ment for the determination of total oxidized nitrogen requires typically between 0.5 and 5 ml of reagent
(TON). per sampling cycle. This makes FIA a simple, auto-
The basic principle of FIA is the injection of a mated micro-chemical technique, which is capable of
liquid sample into a moving, unsegmented carrier a high sampling rate and a low sample and reagent
stream of a suitable liquid. The injected sample consumption.
forms a zone, which is transported towards a de- FIA is based on the combination of three prin-
tector that continuously records the absorbance, ciples: sample injection, controlled dispersion of the
electrode potential or other parameter as the sample injected sample zone, and reproducible timing of its
passes through the detector. movement from the injection point to the detector.
Optimization and design of the flow channels to The chemical reactions take place whilst the sample
achieve maximum sampling frequency, best reagent material is dispersing within the reagent. The phys-
and sample economies, and proper exploitation of ical dispersion processes form the concentration
the chemistries is possible through understanding of gradient of the sample zone. The sample zone
the physical and chemical processes taking place broadens as it moves downstream and changes from
during the movement of the fluids through the FIA the original asymmetrical shape to a more sym-
manifold. metrical and eventually Gaussian form. For standard
The simplest FIA analyzer, shown schematically in
Figure 3, consists of a pump, which is used to propel
the carrier stream through a narrow tube; an in- Pump
Injection valve
jection port, by means of which a well-defined vol- Reaction zone Detector
ume of a sample solution is injected into the carrier
stream in a reproducible manner; and a micro-
reactor in which the sample zone reacts with the Waste
components of the sample stream, forming a species Figure 3 Schematic diagram of a simple FIA system.

1
Flow rate: ml min

Sample

Injection Reduction column

valve
Ammonium chloride Flow cell
0.32
Reaction coil

Sample loop

Sulfanilamide
0.16
N1NED
0.16 Waste

Figure 2 Schematic diagram showing a typical FI manifold (for total oxidized nitrogen).
338 WET CHEMICAL ANALYZERS

conditions, the procedure is totally reproducible in acid at pHo2 for 28 days. However, this has been
that one injected sample behaves in the same way as shown to have an effect on environmental samples
all other subsequently injected samples. where nitrite is converted to nitrate by further mi-
crobial activity, whereas pH 12 is thought to elim-
inate the conversion-causing bacteria. Consequently,
significant errors may be introduced into preserved
In Situ Monitoring
samples and in both cases the cost and logistics of
As greater pressure is placed upon the environment continuously monitoring a particular environment
through anthropogenic activities and climatic chan- would generally be prohibitive.
ges, effective monitoring and control of nutrient en-
richment is vital to protect what is becoming a very
In Situ Instruments
delicately balanced marine environment. For ex-
ample, quantitative knowledge of nutrients and pri- Advances in analytical chemistry have made it feas-
mary production is essential for investigating the ible to perform a wide range of chemical determin-
ecology and biogeochemistry of aquatic ecosystems. ations in situ. The development of field automated
Until recently, the only way to monitor nutrient methods, e.g. flow injection, has been particularly
levels has been to collect discrete samples or use re- important in this respect.
search vessels with onboard laboratory facilities to The application of in situ automated FIA techni-
steam through a particular area of study. Collected ques can produce low-cost, rapid analysis with high
samples are either analyzed at the collection site if sampling frequency analytical systems that are sim-
convenient or transported to a central laboratory to ple and easily maintained. Early examples of FIA
be analyzed later. Monitoring schemes such as these based field instruments successfully completed field
may not detect short-term changes such as algal trials on the River Frome in Dorset. In situ FIA
bloom conditions, storm events, or point discharges techniques have also been developed for environ-
between sampling events. In addition, weather con- mental monitoring of phosphate, ammonia, and
ditions and the high cost and logistics of using re- aluminum, all using solid-state LED/photodiode de-
search vessels for routine studies may not allow tectors. These early systems were based on mains-
complete data sets to be compiled. Therefore, the powered microcomputers which are unsuitable for
current problems associated with compiling long- use in portable battery-powered systems. Since then,
term or continuous data can be summarized as advances in microchip technology have resulted in
follows: the availability of specialized microcontroller devices
for control and automation of a variety of everyday
• High cost and logistics of using research vessels
uses. This technology has been exploited in the de-
for long-term routine surveys.
velopment of in situ FIA monitoring systems for the
• Samples collected for analysis at a later time rely
analysis of nitrate and phosphate in natural waters.
on good sample preservation techniques. This can
All functions required for field-based operation, i.e.
lead to erroneous results with no means of re-
control of peristaltic pumps, injection and switching
placing samples.
valves, data acquisition, processing, and logging
• Existing methods require some degree of human
were controlled automatically. The early develop-
input to perform tasks, which can introduce ex-
ment systems were powered by 12 V sealed lead-acid
perimental error.
batteries, which were capable of 2–3 week’s oper-
• Research cruise weather conditions may not per-
ation depending on the mode of operation.
mit complete data sets to be compiled.
However, there are additional constraints for the
• Post-cruise processing of data can take several
in situ monitoring of sea water, i.e. systems must be
months.
made more rugged and capable of being submersed
There is also considerable evidence to suggest that to facilitate measurements at a particular point in the
many of the sample preservation techniques intro- water column. In the late 1980s, the use of a sub-
duce some level of variance into nutrient determin- mersible FIA system was reported to monitor silicate,
ations. For example, freezing of coastal and estuarine sulfide, and nitrate concentrations in sea water that
water at  101C for nitrate determination was found gave good correlation with laboratory techniques.
to give variance on samples tested, whereas freezing The effects of extremes of temperature, pressure, and
at  201C was found to be acceptable. The US EPA salinity on flow analysis and chemistries used in these
methodology for analysis for anions in water states systems have all been studied. During 1996 a sub-
that unpreserved samples must be analyzed within 48 mersible nitrate sensor based on FIA techniques was
hours otherwise they can be preserved with sulfuric successfully tested in estuarine and coastal waters,
 WET CHEMICAL ANALYZERS 339

over complete tidal cycles, and to depths of 40 m. Grasshoff K, Ehrhardt M, and Kremling K (1976) Methods
Commercially available submersible FIA instruments of Seawater Analysis. New York: Verlag Chemie.
permit laboratory FIA methods to be used in situ. HMSO (1981) Oxidised Nitrogen in Waters: Methods of
Other commercially available wet chemistry field Examination of Waters and Associated Materials.
instruments, which utilize the same basic proven London: HMSO.
HMSO (1988) Discrete and Air Segmented Automated
chemistries as the CFA and FIA instruments, are
Methods of Analysis including Robots, An Essay
available. For example, some instruments utilize Review, 2nd edn: Methods for the Examination of
rugged microprocessor-controlled syringe systems Waters and Associated Materials. London: HMSO.
and a unique design of manifold for the collection of HMSO (1990) Flow Injection Analysis, An Essay Review
the sample, reagent addition, and colorimetric de- and Analytical Methods: Methods for the Examination
termination of the resultant colored species. of Waters and Associated Materials. London: HMSO.
Field instruments enable a wide range of chemical Karlberg B and Pacey GE (1989) Flow Injection Analysis –
determinations to be performed in situ and as the A Practical Guide. Amsterdam: Elsevier.
technology matures greater use of these will be made Ruzicka J and Hansen EH (1988) Flow Injection Analysis,
in the years to come. 2nd edn. Chichester: Wiley Interscience.
Strickland JDH and Parsons TRA (1972) Practical
Handbook of Seawater Analysis, 2nd edn.
US EPA No. 353.2 (1979) Methods for the Chemical
Further Reading Analysis of Water and Wastes. Washington:
Crompton TR (1989) Analysis of Seawater. Sevenoaks: Valcarcel MD and Luque de Castro MD (1987) Flow
Butterworths. Injection Analysis – Principles and Applications.
David ARJ, McCormack T, Morris AW, and Worsfold PJ Chichester: Ellis Horwood.
(1998) Anal. Chim. Acta 361: 63.
ACOUSTIC METHODS
 ACOUSTIC MEASUREMENT OF NEAR-BED
SEDIMENT TRANSPORT PROCESSES
P. D. Thorne and P. S. Bell, Proudman thought of as dynamic interactions between: (1) the
Oceanographic Laboratory, Liverpool, UK seabed morphology, (2) the sediment field, and (3) the
& 2009 Elsevier Ltd. All rights reserved. hydrodynamics. These three components interrelate
with each other in complex ways, being mutually
interactive and interdependent as illustrated sche-
matically as a triad in Figure 1. The objective there-
fore is to measure this interacting triad with sufficient
Introduction: Sediments and Why resolution to study the dominate mechanisms.
Sound Is Used Acoustics uniquely offers the prospect of being able to
nonintrusively provide profiles of the flow, the sus-
Marine sediment systems are complex, frequently
pension field, and the bed topography. This exclusive
comprising mixtures of different particles with non-
combination of being able to measure all three com-
cohesive (sands) and cohesive (clays and muds)
ponents of the sediment triad, co-located and simul-
properties. The movement of sediments in coastal
taneously, has been and is the driving force for
waters impacts on many marine processes. Through
applying acoustics to sediment transport processes.
the actions of accretion, erosion, and transport, sedi-
The idea of using sound to study fundamental
ments define most of our coastline. Their deposition
sediment processes in the marine environment is
and resuspension by waves and tidal currents in es-
attractive, and, in concept, straightforward. A pulse
tuarine and nearshore environments control seabed
of high-frequency sound, typically in the range
morphology. Fine sediments, which act as reservoirs
0.5–5 MHz in frequency, and centimetric in length, is
for nutrients and contaminants and as regulators of
transmitted downward from a directional sound
light transmission through the water column, have
source usually mounted a meter or two above the
significant impact on water chemistry and on primary
bed. As the pulse propagates down toward the bed,
production. Therefore, an improved understanding of
sediment in suspension backscatters a proportion of
sediment dynamics in coastal waters has relevance to
the sound and the bed generally returns a strong
a broad spectrum of marine science ranging from
echo. The amplitude of the signal backscattered from
physical and chemical processes, to the complex bio-
the suspended sediments can be used to obtain ver-
logical and ecological structures supported by sedi-
tical profiles of the suspended concentration and
mentary environments. However, it is commonly
particle size. Utilizing the rate of change of phase of
acknowledged that our capability to describe the
the backscattered signal provides profiles of the three
coupled system of the bed, the hydrodynamics, and
orthogonal components of flow. The strong echo
the sediments themselves is still relatively primitive
from the bed can be used to measure the bed forms.
and often based on empiricism. Recent advances in
observational technologies now allow sediment pro-
cesses to be investigated with greater detail and pre- Sediments
cision than has previously been the case. The Bed load−suspended load
combined use of acoustics, laser and radar, both at
large and small scales, is facilitating exciting meas-
urement opportunities. The enhancement of com-
puting capabilities also allows us to make use of more
complex coupled sediment–hydrodynamic models,
which, linked with the emerging observations, provide Sediment
triad
new openings for model development.
It is readily acknowledged by sedimentologists that
the presently available commercial instrumentation
does not satisfy a number of requirements for near-
bed sediment transport processes studies. Here we Flow Bed forms
Waves−current Flat−rippled
focus on the development of acoustics to fulfill some
of these needs. The question could be asked: Why use Figure 1 Illustration of the sediment processes triad and their
acoustics for such studies? Sediment transport can be interactions.

343
344 ACOUSTIC MEASUREMENT OF NEAR-BED SEDIMENT TRANSPORT PROCESSES

Acoustics therefore has the potentiality to provide multifrequency acoustic backscatter system (ABS),
profile measurements of near-bed sediment pro- consisting, in this case, of three downward-looking
cesses, with sufficient spatial and temporal resolution narrow beam transceivers. The differential scattering
to allow turbulence and intrawave processes to be characteristic of the suspended particles with fre-
probed; this coupled with the bedform morphology quency is used to obtain profiles of suspended sedi-
observations provides sedimentologists and coastal ment particle size and concentration profiles. ‘B’ is a
engineers with an extremely powerful tool to ad- three-axis coherent Doppler velocity profiler (CDVP)
vance understanding of sediment entrainment and for measuring co-located profiles of the three orth-
transport. All of this is delivered with almost no in- ogonal components of flow velocity; two horizontal
fluence on the processes being measured, because and one vertical. It consists of an active narrow beam
sound is the instrument of measurement. transceiver pointing vertically downward and two
passive receivers having a wide beam width in the
vertical and a narrow beam width in the horizontal.
This system uses the rate of change of phase from the
Some Historical Background
backscattered signal to obtain the three velocity
For over two decades the vision of a number of components. ‘C’ is a pencil beam transceiver which
people involved in studying small-scale sediment rotates about a horizontal axis and functions as
processes in the coastal zone has been to attempt to an acoustic ripple profiler (ARP). This is used to
utilize the potential of acoustics to simultaneously extract the bed echo and provide profiles of the bed
and nonintrusively measure seabed morphology, morphology along a transect. These measurements
suspended sediment particle size and concentration are used to obtain, for example, ripple height and
profiles, and profiles of the three components of flow, wavelength, and assess bed roughness. ‘D’ is a high-
with the required resolution to observe the perceived resolution acoustic bed (ripple) sector scanner (ARS)
dominant sediment processes. The capabilities to for imaging the local bed features. Although the ARS
measure the three components have developed at does not provide quantitative measurements of bed
different rates, and it is only in the past few years form height, it does provide the spatial distribution
that the potentiality of an integrated acoustic ap- and this can be very useful when used in conjunction
proach for measuring the triad has become realiz- with the ARP. ‘E’ is a rapid backscatter ripple profiler
able. A schematic of the vision is shown in Figure 2. system (BSARP) for measuring the instantaneous
The figure shows a visualization of the application relationship between bed forms and the suspended
of acoustics to sediment transport processes. ‘A’ is a sediments above it.

A B C D E

B Waves

Current
Suspension

Bed forms

Figure 2 A vision of the application of acoustics to sediment transport processes. A, multifrequency acoustic backscatter for
measuring suspended sediment particle size and concentration profiles. B, coherent Doppler velocity profiler for measuring the three
orthogonal components of flow velocity. C, bed ripple profiler for measuring the bed morphology along a transect. D, high-resolution
sector scanner for imaging the local bed features. E, backscatter scanning system for measuring the relationship between bed form
morphology and suspended sediments.
 ACOUSTIC MEASUREMENT OF NEAR-BED SEDIMENT TRANSPORT PROCESSES 345

What Can Be Measured? Acoustic ripple scanners, ARSs, are based on sec-
tor-scanning technology, which has been specifically
The Bed adapted for high-resolution images of bed form
Whether the bed is rippled or flat has a profound morphology. They typically have a frequency of
influence on the mechanism of sediment entrainment around 1–2 MHz with beam widths of about 11 in the
into the water column. Steep ripples are associated horizontal and 301 in the vertical. As the pulse is
with vortices lifting sediment well away from the backscattered from the bed, the envelope of the signal
bed, while for flat beds sediments primarily move in is measured and usually displayed as image intensity.
a confined thin layer within several grain diameters An example of the data collected by an ARS is
of the bed. Therefore knowing the form of the bed is shown in Figure 4. As can be seen, this provides an
a central component in understanding sediment aerial image of the bed, clearly showing the main bed
transport processes. The development of the ARP features. The advantage of the ARS is the area cov-
and the ARS has had a significant impact on how we erage that is obtained, as opposed to a single line
interpret sediment transport observations. These profile with the ARP; however, direct information on
specifically designed systems typically either provide the height of features within the image cannot readily
quantitative measurements of the evolution of a bed be extracted.
profile with time, the ARP, or generate an image of Ideally one would like to combine the two
the local bed features over an area, the ARS. instruments and recently such systems have become
Figure 3 shows data collected with a 2-MHz nar- available. This is essentially an ARP which also ro-
row pencil beam ARP in a marine setting. The figure tates horizontally through 1801 and therefore allows
shows the variability of a bed form profile, over a three-dimensional (3-D) measurements of the bed.
nominally a 3-m transect, covering a 24-h period. The sonar gathers a single swath of data in the ver-
Over this period the bed was subject to both tidal tical plane and then rotates the transducer around
currents and waves and the figure shows the complex the vertical axis and repeats the process until a cir-
evolution of the bed with periods of ripples and less cular area underneath the sonar has been scanned in
regular bed forms. The figure clearly shows the de- a sequence of radial spokes. An example of data
tailed quantitative measurements of the bed that can collected by a 3-D ARP operating at a frequency of
be obtained with the ARP. 1.1 MHz is shown in Figure 5.

0.2
Height (m)

−0.2
1.5

1 25
0.5 20
Di
sta 0 15
nc
e(
m) −0.5 10 )
e (h
Tim
−1 5

−1.5 0

Figure 3 ARP measurements of the temporal evolution of a 3-m profile on a rippled bed, covering a 24-h period.
346 ACOUSTIC MEASUREMENT OF NEAR-BED SEDIMENT TRANSPORT PROCESSES

1
y (m)

−1

−2
−2 −1 0 1 2
x (m)

Figure 4 Image of ripples on the bed in a large-scale flume facility, the Delta Flume, collected using an acoustic ripple sector
scanner, ARS. The 0.5-m-diameter circle at the center right of the image is one of the feet of the instrumented tripod used to collect the
measurements, while the dark area at the center of the image is a blind spot directly beneath the sonar.

−0.25
z (m)

−0.5
−2

−1

0
x(
m
)

2
1 1
0
−1 y (m)
2
−2

Figure 5 Three-dimensional measurements of a rippled bed collected in the marine environment using a 3-D ripple scanner, 3-D
ARP. The artifacts in the image are associated with reflections from the instrumented tripod used to collect the measurements.

The data shown in the figure were collected during samples at 0.91 intervals and spanning 1801. The plot
a recent marine deployment. The bed surface shown clearly shows a rippled bed. There are one or two
is based on a 3-D scan, with a 100 vertical swaths at artifacts in the plot; these are associated with re-
1.81 intervals, each of which comprised 200 acoustic flection from the rig on which the 3-D ARP was
 ACOUSTIC MEASUREMENT OF NEAR-BED SEDIMENT TRANSPORT PROCESSES 347

mounted. However, the figure plainly contains in- To examine the capability of such a system,
formation on both the horizontal and vertical di- measurements from a three-axis CDVP have been
mensions of the ripples and can therefore be used to compared with a commercially available ADV. The
precisely define quantitatively the features of the bed system had a spatial resolution of 0.04 m, operated
over an area. The 3-D ARP is a substantial advance over a range of 1.28 m, and provided 16-Hz velocity
on both the ARP and the ARS. measurements of the vertical and two horizontal
components of the flow. Figure 7 shows a typical
example of CDVP and ADV time series, power
The Flow spectral density, and probability density function
The success of acoustic Doppler current profilers plots for u, the streamwise flow. The velocities pre-
(ADCPs), which typically provide mean current sented in Figure 7 show CDVP results that compare
profiles with decimeter spatial resolution, and more very favorably with the ADV measurements; having
recently the acoustic Doppler velocimeter (ADV), time series, spectra, and probability distributions
which measures, subcentimetric, subsecond, three in general agreement. There are differences in the
velocity components at a single height, has stimu- spectrum; the CDVP spectra begins to depart from
lated interest in using acoustics to measure near-bed the ADV above about 4 Hz, with the CDVP mea-
velocity profiles. The objective is to use the same suring larger spectral components at the higher fre-
backscattered signal as used by the ABSs, but process quencies. This trend was common to all the records
the rate of change of phase of the signal (rather than and is a limitation of the CDVP system used to ob-
the amplitude as used by ABSs) to obtain velocity tain the data, rather than an intrinsic limitation to
profiles with comparable spatial and temporal reso- the technique. Figure 8 illustrates the capability of
lution to ABSs. The phase technique is utilized in the CDVP for flow visualization in the marine en-
CDVPs and the phase approach has been the pre- vironment. The figure shows mean zeroed velocity
ferred method for obtaining high spatial and tem- vectors, u–w, v–w, and u–v, plotted over a 5-s time
poral resolution velocity profiles. An illustration of a period, between 0.05 and 0.7 m above the bed. The
three-axis CDVP is shown in Figure 6. A narrow- length of the velocity vectors is indicated in the fig-
beam, downwardly pointing transceiver, Tz, trans- ure. A single-point measurement instrument such as
mits a pulse of sound. The scattered signal is picked an ADV can provide the time-varying velocity vec-
up by Tz, and two passive receivers Rx and Ry which tors at a single height above the bed; however, the
are orthogonal to each other and have a wide beam spatial profiling which is achievable with the three-
in the vertical and narrow in the horizontal. axis CDVP provides a capability to visualize struc-
tures in the flow. The structures seen in Figure 8 are
associated with combined turbulent and wave flows.
Receiver Receiver This type of plot exemplifies the value of developing
Rx Ry
a three-axis CDVP with co-located measurement
z
Transmit volumes, since it clearly illustrates the fine-scale
transceiver
Tz temporal and spatial flow structures which can be
measured in the near-bed flow regime. Linking such
measurements with ABS profiles of particle size and
concentration will provide a very powerful tool for
studying near-bed fluxes and sediment transport
processes.
y
The Suspended Sediments
Multifrequency acoustic backscattering, ABS, can
Range
be used to obtain profiles of mean particle size
bin and concentration. The ABS is the only system
available that profiles both parameters rapidly and
simultaneously. Also the bed echo references the
x profile to the local bed position. This is important
because all sediment transport formulas use the bed
as the reference point and predict profiles of sus-
Figure 6 Schematic of the transducer arrangement for a three- pension parameters relative to the bed location. Ex-
axis CDVP. amples of the results that can be obtained are shown
348 ACOUSTIC MEASUREMENT OF NEAR-BED SEDIMENT TRANSPORT PROCESSES

(a)
1.5

u (m s−1) 1

0.5

−0.5
0 Time (s) 100

(b) (c)
3

10−1 2.5
PSDu (m2 s−2 Hz−1)

2
10−2

PDFu
1.5

10−3 1

0.5
10−4
0
10−1 100 101 −1 −0.5 0 0.5 1
Frequency (Hz) Velocity (m s−1)

Figure 7 Comparison of the streamwise flow, u, measured by an ADV (red) and a CDVP (black). (a) The velocities measured at
16 Hz over 100 s. (b) The power spectra of the zero-mean velocities. (c) The probability density functions of the zero-mean velocities
for the ADV (open circles) and the CDVP (crosses).

(a)
Height above bed (m)

0.8
0.6
0.4
0.2
0
155 155.5 156 156.5 157 157.5 158 158.5 159 159.5 160
(b) Time (s)
Height above bed (m)

0.8
0.6
0.4
0.2
0
155 155.5 156 156.5 157 157.5 158 158.5 159 159.5 160
Time (s) −1
(c) 1ms
0.8
Height above bed (m)

0.6

0.4

0.2

0
155 155.5 156 156.5 157 157.5 158 158.5 159 159.5 160
Time (s)

Figure 8 Demonstration of the capability of the triple-axis CDVP to provide visualizations of intrawave and turbulent flow. Plots in
(a)–(c) show a time series over a 5 s period of the zero-mean velocities displayed as vectors u–w, v–w, and u–v, respectively.
 ACOUSTIC MEASUREMENT OF NEAR-BED SEDIMENT TRANSPORT PROCESSES 349

(a) (b)
0.8 0.8

0.7 0.7

0.6 0.6

0.5 0.5
z (m)

z (m)
0.4 0.4

0.3 0.3

0.2 0.2

0.1 0.1

0 0
0 100 200 300 10−2 10−1 100
d (μm) C (kg m−3)

Figure 9 Profiles of suspended sediments: (a) particle size and (b) concentration.

in Figures 9 and 10. These observations were col- concentration with height above the bed. It can be
lected off Santa Cruz Pier, California, as a seasonal clearly seen that some of the periods of increased
storm passed through the area over the period of a particle size are associated with substantial suspended
couple of days. Figure 9 shows profiles of the sus- sediment events, as one might expect, however, there
pended sediment particle size and concentration. The are one or two events where the correlation is not as
profiles of particle size are seen to be relatively clear. Figures 9 and 10 clearly illustrate the capability
consistent with a mean diameter of c. 180 mm near of ABS to simultaneously measure profiles of con-
the bed and reducing to c. 120 mm at 0.7 m above the centration and particle size and the combination of
bed. The variability in size is relatively limited and both significantly adds to the assessment and devel-
due to changing bed and hydrodynamic conditions as opment of sediment transport formulas.
the storm passed by.
In Figure 9(b) the suspended concentrations are
comparable in their form, although they have abso- A Case Study of Waves over a Rippled
lute values that vary by greater than an order of
magnitude. This variation in suspended concen-
Bed
tration is associated with the changing conditions Here the use of acoustics is illustrated by application
as the storm passed through the observational area. to a specific experimental study. Over large areas of
It is interesting to note that over the period even the continental shelf outside the surf zone, sandy
though there is a large variation in concentration, the seabeds are covered with wave-formed ripples. If the
particle size remains nominally consistent. Figure 10 ripples are steep, the entrainment of sediments into
shows the temporal variation in particle size and the water column, due to the waves, is considered
350 ACOUSTIC MEASUREMENT OF NEAR-BED SEDIMENT TRANSPORT PROCESSES

(a)

0.6 180

160
0.4
z (m)

μm
140

0.2
120

0 100
14.8 15 15.2 15.4 15.6 15.8 16 16.2

(b)

0.6
−1

−1.5

log(kg m−3)
0.4
z (m)

−2

0.2 −2.5

−3

0
14.8 15 15.2 15.4 15.6 15.8 16 16.2
Date Mar. 2003

Figure 10 Measurements of the temporal variations with height above the bed, of: (a) particle size with the color bar scaled in
microns and (b) logarithmic concentration with the color bar scaled relative to 1.0 kg m  3.

(a) (b) (c)

v1 v1 v1

(d) (e) (f)

v1 v1 v1
v2
v2 v2

Figure 11 A schematic of vortex sediment entrainment by waves over a steeply rippled bed. The arrows show the direction and
relative magnitude of the near-bed wave velocity. v1 and v2 are the lee slope-generated vortices.

to be primarily associated with the generation cycle. This sediment-rich vortex is then thrown up
of vortices. This process is illustrated in Figure 11. As into the water column at around flow reversal
shown in Figures 11(a) and 11(b), a spinning parcel (Figures 11(c) and 11(d)), carrying sediment well
of sediment-laden water, v1, is formed on the leeside away from the bed and allowing it to be transported
of the ripple at the peak positive velocity in the wave by the flow. At the same time, a sediment-rich vortex,
 ACOUSTIC MEASUREMENT OF NEAR-BED SEDIMENT TRANSPORT PROCESSES 351

v2, is being formed on the opposite side of the ripple 330 mm, and this was located approximately halfway
due to the reversed flow. As shown in Figures 11(d)– along the flume in a layer of thickness 0.5 m and
11(f), v2 grows, entrains sediment, becomes de- length 30 m. In order to make the acoustic and
tached, and moves over the crest at the next flow other auxiliary measurements, an instrumented tri-
reversal carrying sediments into suspension. The pod platform was developed and is shown in
main feature of the vortex mechanism is that sedi- Figure 12(b). The tripod, STABLE II (Sediment
ment is carried up into the water column twice per Transport and Boundary Layer Equipment II), used
wave cycle at flow reversal. This mechanism is an ABS to measure profiles of particle size and con-
completely different to the flat bed case where centration, a pencil beam ARP to measure the bed
maximum near-bed concentration is at about the forms, and, in this case, electromagnetic current
time of maximum flow velocity. meters (ECMs) to measure the horizontal and ver-
To study this fundamental process of sediment tical flow components. Figure 12(c) shows a wave
entrainment, experiments were conducted in one of propagating along the flume with STABLE II sub-
the world’s largest man-made channels, specifically merged in water of depth 4.5 m, typical of coastal
constructed for such sediment transport studies, the zone conditions.
Delta Flume; this is located at the De Voorst Lab- To investigate and then model the vortex entrain-
oratory of the Delft Hydraulics in the north of the ment process it was necessary to establish at the
Netherlands. The flume is shown in Figure 12(a); it is outset whether or not the surface waves were gen-
230 m in length, 5 m in width, and 7 m in depth, and erating ripples on the bed in the Delta Flume. Using
it allows waves and sediment transport to be studied an ARP a 3-m transect of the bed was measured over
at full scale. A large wave generator at one end of the time. The results of the observations over a 90-min
flume produced waves that propagated along the recording period are shown in Figure 13. Clearly
flume, over a sandy bed, and dissipated on a beach at ripples were formed on the bed and the ripples were
the opposite end. The bed was comprised of coarse mobile. To obtain flow separation and hence vortex
sand, with a mean grain diameter by weight of formation requires a ripple steepness (ripple height/

(a) (b) (c)

Figure 12 (a) Photograph of the Delta Flume showing the sand bed at approximately midway along the flume and the wave
generator at the far end of the flume. (b) The instrumented tripod platform, STABLE II, used to make the measurements. (c) A surface
wave propagating along the flume.
352 ACOUSTIC MEASUREMENT OF NEAR-BED SEDIMENT TRANSPORT PROCESSES

−0.4

−0.6
Height (m)

−0.8

−1

−1.2

20

40
Ti
m
e
(m
in

60
)

−1.5
−1
−0.5
80 0
0.5
ce (m)
1 Distan
1.5

Figure 13 Acoustic ripple profiler measurements of a transect of the bed, over time, in the Delta Flume.

ripple wavelength) of the order of 0.1 or greater; an away from the bed and a new sediment-laden lee
analysis of the observations showed that this was vortex, v2, is initiated on the lee slope of the ripple.
indeed the case. In order to capture the essential features of these
Using the ABS, some of the most detailed full scale data within a relatively simple, and hence practical,
measurements of sediment transport over a rippled 1-DV (one-dimensional in the vertical) model, the data
bed under waves were captured. These measure- has first been horizontally averaged over one ripple
ments from the Delta Flume are shown in Figure 14. wavelength at each phase instant during the wave
The images shown were constructed over a 20-min cycle. The resulting pattern of sediment suspension
period as a ripple passed beneath the ABS. The sus- contours is shown in the central panel of Figure 15,
pended concentrations over a ripple, at the same while the upper panel shows the oscillating velocity
velocity instants during the wave cycle, were com- field measured at a height of 0.3 m above the bed. The
bined to generate a sequence of images of the con- concentration contours shown here are relative to the
centration over the ripple with the phase of the ripple crest level, the mean (undisturbed) bed level
wave. Four images from the sequence have been being at height z ¼ 0. The measured concentration
shown to illustrate the measured vortex entrain- contours presented in Figure 15 show two high con-
ment. The length and direction of the arrows in centration peaks near the bed that propagate rapidly
the figure give the magnitude and direction of the upward through a layer of thickness corresponding to
wave velocity, respectively. Comparison of Figure 14 several ripple heights. The first, and the strongest, of
with Figure 11 shows substantial similarities. In these peaks occurs slightly ahead of flow reversal,
Figure 14(a), there can be observed the development while the second, weaker and more dispersed peak, is
of a high-concentration event at high flow velocity centered on flow reversal. The difference in the
above the lee slope of the ripple, v1. In Figure 14(b), strengths of the two peaks reflects the greater positive
as the flow reduced in strength, the near-bed velocity that can be seen to occur beneath the wave
sediment-laden parcel of fluid travels up the leeside crest (time ¼ 0 s) than beneath the wave trough
of the ripple toward the crest. As the flow reverses, (time ¼ 2.5 s). Between the two concentration peaks
this sediment-laden fluid parcel, v1, travels over the the sediment settles rapidly to the bed. Maybe rather
crest and expands. As the reverse flow increases in unexpectedly this settling effect occurs at the times of
strength (Figure 14(d)), the parcel v1 begins to lift strong forward and backward velocity at measurement
 ACOUSTIC MEASUREMENT OF NEAR-BED SEDIMENT TRANSPORT PROCESSES 353

Concentration (kg m−3)

v1
6 cm

v1

42 cm

v1
v2

Figure 14 Acoustic imaging of suspended sand entrainment over a rippled bed due to waves, at four phases of the wave velocity.
The length of the white arrow in each plot gives the magnitude and direction of the near-bed wave velocity.

levels well above the bed. The underlying mechanism concentration peak being dominant. The decay rate
of sediment entrainment by vortices shed at or near of the concentration peaks as they go upward is also
flow reversal is clearly evident in the spatially averaged represented quite well, though a phase lag develops
measurements shown in Figure 15. with height that is not seen to the same extent in the
Any conventional ‘flat rough bed’ model that at- data. Essentially, the detailed acoustic observations
tempts to represent the above sequence of events in of sediment entrainment under waves over ripples of
the suspension layer runs into immediate and severe moderate steepness have begun to establish a new
difficulties, since such models predict maximum type of 1-DV modeling, thereby allowing the model
near-bed concentration at about the time of maxi- to go on to be used for practical prediction purposes
mum flow velocity, and not at flow reversal. Here in the rippled regime, which is the bed form regime
therefore, for the first time in a 1-DV model, it has of most importance over wide offshore areas in the
been attempted to capture these effects realistically coastal seas.
through the use of a strongly time-varying eddy vis-
cosity that represents the timing and strength of the
upward mixing events due to vortex shedding. The
Discussion and Conclusions
model initially predicts the size of the wave-induced
ripples and the size of the grains found in suspension, The aim of this article has been to illustrate the
and then goes on to solve numerically the equations application of acoustics in the study of near-bed
governing the upward diffusion and downward sediment processes. It was not to detail the theoretical
settling of the suspended sediment. The resulting aspects of the work, which can be found elsewhere.
concentration contours in the present case are shown To this end, measurements of bed forms, the hydro-
on the lower panel of Figure 15. The essential two- dynamics, and the movement of sediments have
peak structure of the eddy shedding process can be been used. These results show that acoustics
seen to be represented rather well, with the initial is progressively approaching the stage where it
354 ACOUSTIC MEASUREMENT OF NEAR-BED SEDIMENT TRANSPORT PROCESSES

0.5

Velocity (m s−1)
0

−0.5

Observed
0.25

0.2
z (m)

0.15

0.1

0.05

Modeled
0.25

0.2
z (m)

0.15

0.1

0.05

0 1 2 3 4 5
Time (s)

−0.5 0.25 1
log10[Concentration] (re: kg m−3)

Figure 15 Measurement and modeling of suspended sediments with height z above a rippled bed under a 5-s period wave.
 ACOUSTIC MEASUREMENT OF NEAR-BED SEDIMENT TRANSPORT PROCESSES 355

can measure nonintrusively, co-located and simul- proven and its use with a vengeance in sediment
taneously, with high temporal and spatial resolution, studies will begin to make an important impact in the
all three components of the interacting sediment next couple of years. It was with the concept of using
triad. Figure 16 shows the instrumentation used in a acoustics to measure suspended sediment concen-
further recent deployment in the Delta Flume. This tration that its application to the other components
shows the convergence of the instrumentation and of the small-scale sedimentation processes triad fol-
also its use in conjunctions with instruments such as lowed. To date the use of sound to measure sus-
laser in situ scattering transmissometry (LISST). pended sediment concentration and particle size has
Although substantial advances have been made in been successful when systems have been deployed
the past two decades, the application of acoustics to over nominally homogeneous sandy beds. However,
sediment transport processes is still in an ongoing all who use acoustics recognize that the marine
developmental phase and there are limitations and sedimentary environment is frequently much more
shortcomings that need to be overcome, and further complicated, and suspensions of cohesive sediments
applications explored. Although there have been few and combined cohesive and noncohesive sediments
reports to date on data collected using 3-D ARP, such are common. To employ acoustics quantitatively in
systems are now becoming available. The 3-D ARP is mixed and cohesive environments requires the de-
a substantial development over the single-line ARP velopment of a description of the scattering prop-
and the ARS, and should make a considerable con- erties of suspensions of cohesive sediments and
tribution to the measurement and understanding of sediment mixtures. This would be interesting and
the formation and development of bed forms. There very valuable work, and should significantly extend
have been a number of reports on single-axis CDVP; the deployment regime over which acoustic back-
however, it is the three-axis CDVP which is the way scattering can be employed quantitatively.
ahead. Again, these instruments are coming online, In conclusion, the objective of this article has been
though they are still very much a research tool. to describe the role of acoustics, in near-bed sedi-
However, the concept has now essentially been ment transport studies. It is clearly acknowledged

ARP LISST

ARS
PT

PS

ADV

ABS Three-axis CDVP

Figure 16 The instrumentation used in a recent Delta Flume experiment. LISST, laser in situ scattering transmissometry, PS,
pumped samples, PT, pressure transducer.
356 ACOUSTIC MEASUREMENT OF NEAR-BED SEDIMENT TRANSPORT PROCESSES

that acoustics is one of a number of technologies Hay AE and Mudge T (2005) Principal bed states during
advancing our capabilities to probe sediment pro- SandyDuck97: Occurrence, spectral anisotropy, and the
cesses. However, its nonintrusive profiling ability, bed storm cycle. Journal of Geophysical Research 110:
coupled with its capability to measure all three C03013 (doi:10.1029/2004JC002451).
components of the sediment dynamics triad, make it Thorne PD and Hanes DM (2002) A review of acoustic
measurements of small-scale sediment processes.
a unique and very powerful tool for studying the
Continental Shelf Research 22: 603--632.
fundamental mechanisms of sediment transport. Vincent CE and Hanes DM (2002) The accumulation and
decay of near-bed suspended sand concentration due to
waves and wave groups. Continental Shelf Research 22:
See also 1987--2000.
Zedel L and Hay AE (1999) A coherent Doppler profiler
Acoustic Scattering by Marine Organisms. for high resolution particle velocimetry in the ocean:
Acoustics in Marine Sediments. Laboratory measurements of turbulence and particle
flux. Journal of Atmosphere and Ocean Technology 16:
Further Reading 1102--1117.

Crawford AM and Hay AE (1993) Determining suspended Relevant Websites

sand size and concentration from multifrequency
acoustic backscatter. Journal of the Acoustical Society http://www.aquatecgroup.com
of America 94(6): 3312--3324. – Aquatec Group Ltd.
Davies AG and Thorne PD (2005) Modelling and meas- http://www.marine-electronics.co.uk
urement of sediment transport by waves in the vortex – Marine Electronics Ltd.
ripple regime. Journal of Geophysical Research 110: http://www.pol.ac.uk
C05017 (doi:1029/2004JC002468). – POL Research, Proudman Oceanographic Laboratory.
 ACOUSTIC NOISE
I. Dyer, Marblehead, MA, USA sounds from marine earthquakes, from marine ani-
Copyright & 2001 Elsevier Ltd.
mals, from nearby ships or other nearby commercial
activities in the ocean, from rain/hail/snow striking
the sea surface, and from fractures of ice in the north
or south polar oceans. With such a large number of
prevailing and other noise sources, the band desig-
nations given in the previous paragraph are unlikely
Introduction to be associated unequivocally with just one noise
Some ocean scientists consider ambient noise to be a source or, for that matter, adopted fully by most am-
fairly simple and well-behaved property of the ocean. bient noise researchers or practitioners. They are of
Ambient noise, after all, is often reported and sum- use, however, to help present the material to follow.
marized in highly averaged form, its naturally large The spectral summaries used in this Introduction
variance mostly unstated. Other ocean scientists are based on Wenz, and although still useful, modi-
consider the variational complexity of ambient noise fications and additions are needed in the light of new
a richly colored portrait carrying images of basic knowledge. Urick published an excellent summary of
ocean processes, including the physics of various ambient noise data acquired in various measurement
noise sources and the acoustics of multiple noise programs through about 1980. Practitioners com-
propagation paths. Space and incomplete knowledge monly use these data, plus the Wenz results, for
precludes a description here that can fully satisfy all prediction. Nevertheless, basic understanding of
ocean scientists or technologists. Instead, the ob- many ambient noise mechanisms through about
jective is to summarize those aspects of ocean am- 1980 was meager and, indeed, some suggested
bient noise that convey the more important recent mechanisms were considered speculative. Fortu-
research results and the more significant remaining nately, mechanisms for prevailing ambient noises
research questions. have received considerable research attention since
A 1962 summary of ambient noise measurements then, particularly from 1985 or so. The other noises
in the ocean (see Figure 1) is still useful today, at least have also been researched, in general to a lesser de-
to classify the various noise sources and their average gree. Two volumes edited by Kerman and one by
levels and smooth frequency spectra. Prevailing Buckingham and Potter are conference proceedings
noises (those observed almost always) are caused by of recent ambient noise research, and are extra-
wave–wave interactions at the sea surface, by dis- ordinary seminal contributions to the understanding
tributed seismic activity in the earth, by atmospheric of ambient noise mechanisms in the ocean. The
or oceanic turbulence, by distant shipping, by wind- continuing flow of research results in archival jour-
induced sea surface agitation, and by thermally in- nals and books, and the aforementioned volumes,
duced molecular agitation. According to Wenz, provide important modifications and additions to the
wave–wave interaction effects, seismic background, classical summary of ambient noise by Wenz. In
and/or turbulence dominates the noise at VLF (very what follows the more important new knowledge, or
low frequency band: 1ofo20 Hz), with power lack thereof, is summarized.
spectral density of the pressure field Sðf Þpf 4 . Dis-
tant shipping noise dominates at LF (low frequency
ULF Band: Wave–Wave Interaction
band: 20ofo200 Hz), has a broad spectral peak
around 50 Hz, and falls off sharply for f > 200 Hz as Noise
f 6 . At MF (midfrequency band: 200 Hzo Measurements within 0.1ofo2 Hz, which has come
fo50 kHz), noise caused by sea surface agitation to be called the ultralow frequency band (ULF), ex-
typically dominates, with a broad peak within tended the Wenzian picture one decade lower in
200 Hzofo2 kHz and, beyond f E2 kHz, with frequency1, and showed ULF noise to be a function
Sðf Þpf 1:7 . Finally, molecular agitation typically of wind speed. The data (Figure 2), have a strong
dominates the noise at HF (high frequency band: peak f0 located between about 0.2 and 0.7 Hz, with
f4100 kHz), with Sðf Þpf 2 .
Other noise sources are classified as temporally
intermittent or spatially discrete, rather than pre- 1
To accommodate the ULF band, the band scheme described in
vailing, and can often dominate. These include the Introduction is redefined to be VLF2–20 Hz.

357
358 ACOUSTIC NOISE

120
Intermittent and local effects
Earthquakes
and explosions
Biologics
Precipitation
Ships, industrial activity
100 Sea ice
26
KEY
Limits of prevailing noise
Wind-dependent bubble and spray noise
Low-frequency very-shallow-water
wind dependence

Ln ′ dB re 1 dyn cm 2 and 1Hz (add 100 dB for ref. of 1 μPa and 1 Hz)
Heavy precipitation
80 6
Heavy traffic noise
Usual traffic noise _ shallow water
Ln ′ Spectrum level, dB re 0.0002 dyn cm 2 and 1 Hz

Usual traffic noise _ deep water

Thermal noise
General pattern of noise from
_

earthquakes and explosions

Extrapolations _14
60

Wind force
(Beaufort)

40 _ 34

_
8

20 _ 54
3

Prevailing noises
1
Turbulent-pressure
fluctuations _ 74
0
Oceanic traffic

Bubbles and spray

(Surface agitation)
Surface waves _ Second-order pressure effects Molecular
(Seismic background) agitation
_ 20
1 10 2 3 4 5
10 10 10 10
Frequency (Hz)

Figure 1 Ambient noise spectra summarized by Wenz. (The ordinates are Ln ¼ 10 log10 Sðf Þ, with respect to the reference value.
Add 100 dB to the right-hand scale to obtain Ln in dB re 1 mPa and 1 Hz.) The Beaufort Force translates to wind speeds, in ms 1, as
follows: 1, 0.5–1.5; 2, 2–3; 3, 3.5–5; 5, 8.5–10.5; 8, 17–20. (Reproduced from Wenz, 1962.)

Sðf Þ at higher f proportional to about f 3 2f 5 , Pressure Spectral Density

dependent upon wind speed. A long history of
measurements, as well as theoretical surface wave Wave–wave interaction noise is caused by opposing
interaction studies, presaged this result. The ap- wind-driven surface gravity waves, each at frequency
pearance of systematic data such as in Figure 2 ap- fw , that to second-order create a pressure field in the
parently sparked even more research efforts that water at f ¼ 2fw . (Orders higher than the second
ultimately confirmed the basic aspects of ULF noise. have been shown to be negligible.) In its simplest
 ACOUSTIC NOISE 359

160 The noise spectral density Sðf Þ is proportional to

[Sw ðf =2Þ2 , the surface elevation density squared and
shifted in frequency. Since the frequency dependence
of the last three terms in eqn [1] is relatively weak
150 around the peak frequency fwo of Sw , the peak of S is
essentially fo E2fwo. Both fwo and Sw are functions
of wind speed U, as affected by other sea conditions
(fetch, sea age, etc.), and similarly lead to the de-
Ambient noise spectrum level (dB re μPa2/Hz)

140 pendence of fo and S on U.

Microseism Spectral Density

130 Displacement, rather than pressure, is often meas-

ured on the seafloor. Termed microseism noise, it can
be obtained from eqn [1] with the substitution of a
modified transfer function Td ðf ; zÞ for Tp ðf ; zÞ, each
120 in general given for all z, to yield the displacement
spectral density Sd ðf ; zÞ rather than Sðf ; zÞ. These
transfer functions incorporate the acoustics of the
medium, including the seismoacoustics of the bot-
110 tom. It is no surprise, therefore, that noise pressure
_
Time (z) Wind speed (m s 1) or displacement data may show subsidiary peaks in
17 OCTOBER (0340) 30
17 OCTOBER (1940) 25 the frequency domain, additional to the major peak
18 OCTOBER (1940) 15 at fo , associated with acoustic modes of the oceanic
100 19 OCTOBER (1940) 7.5
waveguide.
20 OCTOBER (1140) 5
20 OCTOBER (1540) <5 With modifications as outlined in the foregoing
20 OCTOBER (1940) <5 and as detailed in the literature, and with site-specific
24 OCTOBER (0340) 35
seismic–acoustic bottom properties and time- and
90
0.05 0.1 0.5 1.0 2 site-specific sea surface elevations, data such as in
Frequency (Hz) Figure 2 are remarkably well predicted with use of
the theory symbolized by eqn [1]. The theory may
Figure 2 Ambient noise pressure derived from vertical also be extended to cover interaction of swells with
displacement (seismometer) data at the water/bottom interface
(depth about 100 m). (Reproduced from Kibblewhite and Evans,
wind waves, with use of Sw Ss instead of S2w , where Ss
1985.) is the swell spectral density, and with F similarly
including the swell azimuthal directivity.

form, the spectral density S of the pressure field is

VLF Band: Atmospheric Turbulence
 Forcing
3 2 2

as a Noise Source
2 2 3
Sðf ; zÞ ¼ p r g =2c ½Sw ðf =2Þ f Fðf =2ÞTp ðf ; zÞ ½1
The heading of this section is in reality a question. In
where r is water density, c is water sound speed, g is the VLF band (2–20 Hz), and to somewhat higher
acceleration of gravity, Sw is the spectral density of frequencies in seas with very low shipping density,
the surface wave elevation, F is an integral over all noise measurements do not match direct extrapo-
azimuth angles f of the product of the normalized lation of wave–wave interaction data as in Figure 2.
azimuthal directivity of the opposing surface waves Other mechanisms may become important in this
(for f and f þ p), and Tp is a function normalized by band, and one has attracted considerable attention,
its value at depth z ¼ 0 that relates the pressure namely turbulent pressure fluctuations in the at-
spectral density at the surface to that at z, which mosphere that drive the ocean surface. Is the latter
importantly includes acoustic coupling to the sea the responsible mechanism? And if not, what is? A
bottom. For simplicity, eqn [1] includes contri- robust answer is not yet available.
butions to the noise field only for horizontal phase In one of the more extensive data summaries dat-
speeds given by cx  c. Because of its exponential ing from the 1960s, Crouch and Burt showed that in
decay in z, the remaining regime cx oc is important the absence of shipping noise, and at low wind
mostly for depths approaching z ¼ 0, especially for speeds, S is p f 1 U0 between about 10–50 Hz. This
lower f, and is included in fuller analyses. holds until a crossover wind speed Uc is reached
360 ACOUSTIC NOISE

(E16 m s1 at 11 Hz), beyond which S is roughly functional form of the foregoing data, but does come
pf 3 U4 . The crossover speed Uc tends to increase close to Nichols and others (at the higher wind
with increasing frequency up to about 50 Hz. speeds) for the 5–25 Hz range. It seems inappropri-
Nichols reported some data that also included ate, however, to choose among available data sets for
frequencies lower than those of Crouch/Burt. These the ones that confirm a model. The difference be-
data show that for a small spread of wind speeds tween the data sets may well be caused by some
around UE9 m s1 , Spf 22:5 from 3 to 8 Hz, and mechanism that we are collectively ignorant of.
Spf 0 from 8 to 20 Hz. These results compare well Finally, it is possible that available data are at least
with those of Crouch/Burt at the overlapping fre- partially contaminated by hydrophone flow noise,
quencies, and add a functional form below 8 Hz not whose functional form goes as f 4 . None of the data
covered by Crouch/Burt. Nichols also summarized sets matches this form. Thus, it can be concluded
unpublished data for U varying from about 3 to 9 m that the hydrophone flow noise mechanism is an
s1, that show Spf 25 U5 from 2 to 5 Hz, and unlikely cause of VLF noise. The identification of the
Spf 0 U5 from 5 to 10 Hz. (The f 0 U5 functional form mechanism responsible for VLF noise can thus not be
could be argued from the data shown to be f 0 U0 at made with confidence.
lower U, and f 0 U>5 at higher U.) This form at the In searching for candidate VLF noise mechanisms,
lowest frequencies has a much sharper falloff with one is inclined to look toward appropriate extensions
frequency than the Nichols’ form, but preference or modifications of mechanisms in the adjacent ULF
should be given to it because the frequency reso- and LF bands, mainly because wave–wave inter-
lution was much finer. Accordingly, the overall result, actions and distant shipping, respectively, are well
labeled ‘Nichols’, is stated here as Spf 25 U5 from 2 established. Nonprevailing mechanisms should also
to 5 Hz, and f 0 U5 from 5 to 20 Hz. be considered. For example, whale vocalizations are
More recently, as part of LF noise measurement observed for 15of o35 Hz and can affect the VLF
programs, data in the upper part of the VLF band band.
were published. Deep ocean noise versus U at f ¼ 13 Noise data sets beyond those referred to here,
and 50 Hz has been reported. For 5rUr15 m s1, supported by environmental data as suggested by
these data have a functional form approximately candidate mechanisms, may well be needed. The
Spf 0:8 U1:3 (perhaps more like f 0 U1:6 at the highest Crouch/Burt data set incorporated a plausible but
U). A report of shallow water noise versus U in convoluted data analysis path to extract the VLF
third-octave bands for 10rf r20 Hz, with form noise. The reported database of Nichols is not large.
Spf 3:2 U3:4 for U ¼ 3 and 5 m s1 in late fall, had The VLF data of other workers could have been af-
data that differed so widely for U ¼ 5 and 9 m s1 in fected, as the authors acknowledged, by distant
late spring that a functional form could not be stated. shipping noise. Perhaps because many of these re-
One report of deep ocean noise at f ¼ 15 and 25 Hz, search efforts were aimed at other objectives, en-
for 2rUr12 m s1, suggests Spf 0 U0:5 for Uo Uc, vironmental data provided with the noise data are
and Spf 0 U3 for U > Uc, with the crossover speed generally too fragmentary to aid the search for VLF
Uc E8 m s1. mechanisms.
What noise mechanism could account for all the
foregoing observations? An extrapolation of wave–
wave interaction noise to the VLF band, from data LF Band: Distant Shipping Noise
such as in Figure 2, suggests that Spf 4 U1=2 , give or Evolving technology has altered the view of distant
take one integer in the exponent of f , and one-half shipping noise. Increasing use of large aperture
integer in the exponent of U. However, this is un- acoustic arrays, with attendant high-resolution beam-
acceptably far from the data. The Crouch/Burt data widths and focused scanning in range and use of high-
suggest f 1 U0 and f 3 U4 for low and high U, re- resolution frequency filters, blurs the distinction once
spectively. The overall Nichols result is f 5 U5 from 2 sharp between distant and local ships. That is, ambient
to 5 Hz, and f 0 U5 from 5 to 20 Hz. Other data give noise at LF can be observed with high-resolution
f 0:8 U1:3 , f 3:2 U3:4 or an indefinite form, and technology as a countable number of discrete ship
f 0 U0:5 and f 0 U3 for low and high U, respectively. noise sources, rather than as a sum of noise from a
Without significant modifications applicable to the very large number of widely distributed ships.
VLF band, it seems that the wave–wave interaction
possibility must be set aside.
Frequency Spectra
Next, consider the atmospheric turbulence model.
It has evolved as most theories do, but is contentious. Figure 3 shows the noise radiated by a contemporary
It predicts Spf 0 U4 . This, too, is mostly far from the cargo ship. The radiation is largely tonal, as has long
 ACOUSTIC NOISE 361

190
B B
3 4

0.5 Hz Bandwidth levels (dB re. 1 μPa at 1 m)

G F B
180 4 2 B B
9
5 6
B F F F
B B
1 1 F 4 6
2 G B 8 B
5 3 10 B
7
170 11

F
5
G
G
8
160 G 9
140 rpm
6 G G
G 10 11
3
G
150 2
G
G G G G
7 G
12 G 15 G 20 22 30
14 58 rpm
G 13 16 G G
1 17 18 25 32
G
140 19
27
31
24 26 33
21 23 28
29

130

20 40 60 80 100 120 140 160 180 200

Frequency (Hz)

Figure 3 On-axis source level spectra of a cargo ship at 8 and 16 knots (4 and 8 m s1) measured directly below the ship. Noise
levels at distances beyond 1 m may be obtained by subtracting the transmission loss. Levels for bandwidths other than 0.5 Hz can not
be determined from this figure because the bandwidths of the tones are not given. (B, F, and G in the figure identify, respectively, the
harmonics of the (propeller) blade rate, the (diesel engine) firing rate, and the (ship’s service) generator rate.) (Reproduced from
Arveson and Vendittis, 2000.)

been known. However, this is not obvious from the cosyw Ecz =cb , where cz and cb are the sound speed at
spectra shown in Figure 1, because they entail sums the observation depth and at the bottom, respect-
over many ships. The tonal envelopes in Figure 3 ively. Distant shipping noise in deep water thus ar-
maximize between about 20 and 80 Hz, in good rives mostly from source radiation at the surface near
agreement with the summation spectra shown in 7ys 7 ¼ 0, and then propagates to the observer via
Figure 1. Acoustic propagation losses in the ocean refraction and surface reflection paths in the water.
change the shape of the source spectrum shown in Bottom reflection or transmission losses are relatively
Figure 3; above about 80 Hz, the spectrum observed high, so that these paths are less important. Other
distantly is increasingly reduced with increasing f effects influence the pedestal including, but not lim-
and with increasing distance from the ship. ited to, surface waves that modulate the source
amplitude, scattering rather than specular reflection
Directional Spectra from the rough sea surface, and scattering from a
Noise radiated by a ship is a function of azimuth fs , seamount or continental margin. Because of these
and vertical angle ys , in a cylindrical coordinate oceanographic and topographical complexities, the
system attached to the ship. The azimuthal spectral shape of the noise pedestal in Figure 4 is not general
shape Sf ðf; f ; zÞ observed for a single ship at longer but instead suggestive of the main features of Sy in
ranges is close to that measured near the ship. the deep ocean.
However, propagation of the noise to large ranges In shallow waters, if cz =cb > 1 (downward re-
fundamentally affects the vertical directional spec- fracting profile), then Sy is governed by path or mode
trum Sy ðy; f ; zÞ. In these spectra, f and y are in a losses, including those attributable to the bottom. If
coordinate system attached to the observer (y ¼ 0 is cz =cb o1 (upward refracting profile), then Sy would
the local horizontal plane). Figure 4 shows Sy as have a pedestal, but with yw typically an order of
measured in deep water by a vertical line array. It magnitude smaller than that observed in the deep
sums over the directional spectrum in azimuth Sf , ocean.
and therefore over the areal distribution of ships. A
Summation Issues
prominent feature of Sy is a pedestal of high noise
around the horizontal, which is weakly dependent With several thousand ships underway in each of the
on f and z, and varies in half-width yw from about heavily traveled oceans, taking account of all at the
715 to 201. These values are consistent with same time to determine distant shipping noise is
362 ACOUSTIC NOISE

from limits that may be imposed by the focusing,

Wind speed
High
filtering, or beam-forming processor, the spectral
70 Mid
valleys are set by the ship’s radiation of continuous
Low
Low rather than tonal noise, and the spatial valleys by
60
the sum of noise from more distant or less powerful
50 ships. In addition, the valleys can be influenced
40 by nonprevailing sources, the most prominent of
(A) which at LF is whale vocalization from about 15 to
35 Hz.
High
Power (dB / μPa/√Hz deg)

70 Mid VLF Implications

60 With respect to the VLF band, the tonal envelope of
Low
50 a ship for f o20 Hz, at all of its higher speeds is
40
about f 0 . Because very low frequency sounds can be
(B) detected at transoceanic distances, distant ships
could cause the measured result to be Spf 0 U0 found
in some VLF experiments, in the apparent but not
High
real absence of distant ships.
70
Mid
60

50
Low MF Band: Wind-driven Sea Surface
40
Noise
_ _
_
80 60 40 _ 20 0 20 40 60 80 Bubbles created by wind-driven surface waves have
(C) Angle (deg) long been thought to be the dominant source of
Figure 4 Vertical angle directional spectrum SY summed in
prevailing noise in the MF band (0.2–50 kHz). Many
azimuth in deep water for f ¼ 75 Hz and U ¼ 3; 7, and 11 m s1. basic physical details, however, have only recently
Positive angles are upward looking. (A) Shallow, (B) mid, and (C) become better understood, and some relevant add-
deep depth refers to the vertical line array (VLA) centered at itional questions only recently posed. Various wave-
z ¼ 850, 1750, and 2650 m. The half-power beamwidth of the VLA breaking processes of wind-driven surface waves
at this frequency is about 1.31 when steered to y ¼ 0.
(Reproduced from Sotrin and Hodgkiss, 1990.)
entrain air in the upper part of the ocean. Air-filled
bubbles in the water are pinched off from the en-
trained air, which in turn oscillate and radiate noise
neither feasible nor necessary. At one extreme, low- as acoustic monopoles. Such noise thus entails wave-
resolution data in the LF band (20–200 Hz) are breaking and bubble hydrodynamics, both of which
relatively insensitive to the detailed noise source are addressed elsewhere in this encyclopedia. The
characteristics of individual ships. Because sum- acoustical aspects are addressed here.
mation from a large number of ships merges the
details, only the broad and slowly evolving trends in
Vertical Directional Spectrum
shipping lane location, in shipping density, and in
shipping composition will affect the level and hori- In its simplest form, the theory for noise generated by
zontal directionality of the noise. At the other ex- a uniform distribution of sources on the surface is,
treme, high-resolution data resolve the frequency and from ray acoustics,
spatial spectra associated with distant ships, thus
giving the experimenter the in situ noise field in Sþ 1
y ðy; f ; zÞ ¼ ðcs =cz ÞSDfsinys ð1  Rb Rs Þg
relevant detail. The experimenter can avoid such ½2
yw ryrp=2
noise in the spectral valleys between tones or the
spatial valleys between high-noise beams2. Aside
S þ
y ðy; f ; zÞ ¼ Rb Sy ; p=2ryr  yw ½3

2
Many ocean processes broaden tonal bandwidths and spatial where
beamwidths as the sound propagates from source to receiver. Such
broadening is typically large enough to be a candidate ocean cosyw ¼ cz =cs for cz =cs o1;
monitoring tool, but not so much that it completely fills the ½4
valleys. or yw ¼ 0 for cz =cs 41
 ACOUSTIC NOISE 363

and where the directional spectrum in vertical angle surface, and the other assumes a similarly situated
Sy , per unit solid angle, is a function of f (at least monopole distribution concentrated on a submerged
through S, the pressure spectral density of the source plane. Then, respectively,
per unit surface area) and of observation depth z
(through sound speed cz at z). The superscripts þ D ¼ 2f1  sincð2ks dsinys Þg ½5
and – refer to positive (upward looking) and negative 2
D ¼ 4sin ðks dsinys Þ ½6
(downward looking) y obtained from cos cosy ¼
ðcz =cs Þcos ys (the grazing angle ys and sound speed cs where ks is the acoustic wavenumber at the surface,
pertain to z ¼ 0). Dðys ; f ) is the directivity of the d is the effective depth (the e-folding depth and the
elemental bubble noise sources distributed just below d-function depth, respectively), and sinc (x)(sin
the surface. For present purposes, the surface source xÞ=x. In the limit ks d sin ys 51, these functions have
distribution S, and the acoustic waveguide, is taken the same shape, and close to the same magnitude
as uniform and independent of range and azimuth, (E1 dB different). Data, however, show that the two
contributions from propagation in the bottom are are distinct. For eqn [5], the data suggest ks dEp,
neglected, and volumetric absorption in the seawater whereas for eqn [6] ks dEp In either case, the ideal-
is neglected. (These simplifications are adopted to ized states assumed in eqn [6] might not represent the
keep the main ideas clear but, for more precise needs, relevant complexity of the radiating bubbles beneath
can readily be replaced by assumptions that are more a breaking wave. For example, the exponential decay
realistic.) of bubble density with depth may well be a good
For a downward refracting sound speed profile to model for horizontally isotropic bubbles quasistati-
depth zðcz =cs o1), the ray theory of [3][4] predicts a cally present as a result of previous wave breaking
refractive shadow zone or notch of width 2yw events, but a poor model for radiating bubbles im-
around y ¼ 0, the horizontal plane. Wave theory mediately caused by a new event.
must be used to properly predict the field in the Integration of eqns [2]–[4] over y to obtain the
notch, which also can be partially filled by scattering noise spectral density Sðf ; z) also depends sensitively
of the noise from midwater depths by fish schools on Rb and Rs (and on possible bottom propagation
and by ocean inhomogeneities. For an upward re- paths). This emphasizes the need to compare ex-
fracting profile (cz =cs 41), the field around y ¼ 0 is perimentally derived values of Sðf ; z) with appro-
directly due to the surface-generated noise, plus priate knowledge of the acoustic waveguide.
possible scattering contributions. Alternatively, with use of eqns [2]–[4],S may be ex-
In eqns [2] and [3], Rb and Rs are, respectively, tracted from vertical line array (VLA) data. When a
the coefficients of bottom and surface specular re- VLA is steered to y ¼ p=2, the specular reflection and
flection. Terms involving these parameters can be the bottom propagation paths will contribute at most
important in the directional spectrum (but since
perfect reflection is not likely for an acoustic wave-
guide in the ocean, they do not lead to singularities as 65
eqns [2]–[4] might appear to suggest). For example,
consider that Rb and Rs approach unity (but do not
Noise level (dB// 1 μPa2/sr-Hz)

reach it) as the grazing angles at the bottom and the 60

surface, respectively, approach zero. Then, for y
within about 7p=4; Sy can be increased in typical
situations by about 10 dB. In addition, the bottom 55
propagation paths neglected here can actually con-
tribute, especially at the lower MF frequencies. Thus
details of the acoustic waveguide affect Sþ y and Sy
 50
and, along with the sound speed profile cðzÞ), could
account for the plethora of somewhat dissimilar
measured MF vertical directional spectra in the 45
literature.
Eqn [2] contains the bubble source directivity
Dðys ; f ) that, unfortunately, is not known with con- 5 10 15 20 25 30
20 Log (Wind speed) (kn)
fidence. At least two models for directive radiation
from aggregated bubbles have been considered. One Figure 5 Level of the source spectral density S, in dB re mPa2
assumes an exponential decrease of uncorrelated per m2 Hz, for f ¼ 110 Hz. The 10 m wind speed U is in kn (1
monopoles below a horizontal perfectly reflecting knE1/2 m s1). (Reproduced from Chapman and Cornish, 1993.)
364 ACOUSTIC NOISE

weakly. Such a measurement is thus dominated by is rough. Kennedy analyzed VLA data in a deep,
local surface sources, so that S may be compared acoustically isolated bay (40of o4000 Hz,
among measurements with less concern for wave- 2oUo15 m s1), with unlimited wind fetch but
guide properties. limited wave fetch. It was found that u =uc 0.9 de-
fined a rough surface regime. (It may therefore be
Source Spectral Density presumed that the crossover speed discussed in the
foregoing section is Uc E0:9uc ) Figure 6 shows that
Chapman and Cornish measured S in deep water the spectral data for u =uc 40.9 aggregate to an al-
with an upward-looking VLA. They apparently as- most universal scalable spectrum. What garners the
sumed eqn [6] for D, with ks dEp=2 . Their data at caveat of ‘almost’ is that frequency is scaled by fp, the
f ¼ 110 Hz, and for the wind speed interval observed peak frequency. Both Kerman and Kennedy
2oUo15 m s1, are reproduced in Figure 5, and point out that fp does not vary strongly. It ranges
show a crossover wind speed Uc E4.3 m s1. The from about 300 to 800 Hz in the Kennedy data, and
frequency interval for their measurements is is not unlike that sketched by Wenz (Figure 1). But
13of o300 Hz, within which they found that Uc experimental interest does not always include meas-
is about 4.3 m s1 for f r110 Hz, and that Uc is urement of fp , in which case a user of Figure 6 must
somewhat smaller (E3.5 m s1) for f 4110 Hz. slide the frequency scale without benefit of Kennedy’s
Chapman/Cornish attribute the crossover speed to a judgement. Neither, however, can properly be ac-
transition in source mechanism physics. Further- cused of intellectual sloth. Kerman provides a model
more, by regression analyses, their data show that for for fp, which contains wave-breaking parameters that
UoUc ; Spf 2:1 U0:6 , and for U > Uc ; Spf 2:1 U2:7. unfortunately are poorly known. Kennedy’s col-
These results hold on average within the speed and lapsed spectral spread although acceptably small,
frequency intervals measured. is large enough, and the frequency dependence for
Kewley, Browning, and Carey reviewed and com- 1/3of =fp o10 is weak enough, to forego fine atten-
pared several data sets, mostly deep water VLA tion to fp . Although apparently not used, fp is related
measurements, to extract S. They also used eqn [6] semi-empirically to breaking-wave whitecap size
and ks dEp=2, and concluded that for 30of o fp (in Hz)E1400/(OLW), where L and W are, re-
1300 Hz and 1oUo15 m s1, Uc E6 m s1 with spectively, the whitecap along-crest length and cross-
SpU1 for UoUc and Sp U3 for U4Uc. The crest width, both in meters].
Kewley et al. wind speed exponents of 1 and 3 are The source spectral density S in Figure 6 is ob-
not too different from those of Chapman/Cornish. tained by Kennedy from a dipole directivity model.
When one considers that the former tilted their In effect, eqn [6] was used with ks d51, in this limit
exponent choices somewhat to agree with extant known as a compact dipole. With this assumption,
physical models proposed for the below and above the integral of Dðf ; ys Þ over a hemispherical surface
Uc regimes, the agreement can be considered quite
satisfactory.3 What is more relevant, however, is that
a universal spectral shape is not evident for either _7
10
regime in the Kewley et al. comparisons. More likely u*
> 0.9
than not S, f , and D need to be scaled by hydro- uc
_8
dynamic parameters other than or additional to U, as 10
shown below.
pawcu *3
ADSD (f)B

_9
10
Basic Wave-Breaking Correlates
Research results on hydrodynamically based scaling _ 10

of noise from breaking waves have been reported. 10

Kerman has proposed that at u =uc E1, where u is

the friction velocity and uc is the minimum phase 10
_ 11

speed of gravity/capillary surface waves, the wave- 0.01 0.10 1 10

breaking process transitions from one that has an _f
fp
aerodynamically smooth sea surface to another that
Figure 6 Source spectral density S versus f , both
nondimensional as described in the text. The source directivity
3
When compared at the same U and f , S is about 3 dB higher in model used is presumably eqn [6] with ks d 51 (the compact
the Chapman/Cornish data set than in the data reviewed by dipole model). Data are for the aerodynamically rough regime
Kewley et al. (u 4uc ). (Reproduced from Kennedy, 1992.)
 ACOUSTIC NOISE 365

yields AD ¼ 2p=3 and this appears as one of the be significantly better than that with U and, via
S scaling terms in Figure 6.4 Another term is physical arguments it was concluded that the source
B ¼ 1000 Hz, the nominally observed bandwidth of levels are well correlated with E. This measurement
the noise; its role is simply to create an integral technique is also important as it determined the
measure SB for dimensional clarity. In the term E ¼ probability density of the dipole source levels, and
5ra u3 ; ra is air density, and E is the major scaling the spatial density of discrete breaking wave events.
variable, the average rate of energy dissipated per It was also concluded, again via physical arguments,
unit surface area by the breaking waves. Finally, that the source spectral density for the frequencies
AD SB=ðrw cs ), where rw is water density, is the measured are on average pE0.83, which in view of
average rate of acoustic energy radiated per unit the lower frequencies observed might be taken as
surface area. reasonably consistent with the E0.74 obtained by
Virtually simultaneously and independently, others other workers. Thus the question remains on a pos-
have researched in greater depth some concepts that sible frequency-dependent bubble layer effect.
are related to the Kerman/Kennedy result. Noise The foregoing results clearly have not answered all
spectral density Sðf ; zÞ has been correlated with dis- questions on MF noise caused by breaking waves.
sipation E, in deep water under steady wind and They do, however, provide more general predictive
wave conditions, for the intervals 4.3of o14 kHz tools than those previously available, and identify at
and 2oUo12 m s1. The data on average show least some of the more important physical attributes
SpSpf 0:4 E0:74 , with the exponent of E varying of noise from breaking waves.
from 0.86 to 0.67 from the low to the high end of the
frequency interval. At constant E, the frequency de-
pendence agrees reasonably with an extrapolation of
Figure 6. But the dissipation dependence can not be HF Band: Molecular Noise
compared without scaling the peak frequency fp, Molecules impinging on the surface of a pressure
which was not observed. Thus, for a range of E one sensor cause noise, as estimated from physical prin-
can seek the range of fp to satisfy linear scaling in E. ciples and as plotted in Figure 1. Molecular motion,
The peak frequency fp would then need to decrease and thus momentum reversal on the sensor (i. e.,
about a factor of 4 from low to high E, a factor so force per unit area) is a function of molecular kinetic
large as to suggest that a major change in noise energy, and thus seawater temperature. On an ab-
physics occurs at these higher frequencies. Does the solute temperature scale, all oceans may be con-
quasistatic bubble layer below the sea surface in- sidered at a constant temperature. Hence, one line in
creasingly attenuate the noise, or increasingly inhibit Figure 1 is sufficient to estimate the noise.
its generation, at these frequencies? Bubbles are
known to attenuate sound as a function of frequency
and size distribution, but data analyses do not con- See also
sider this.
With use of the Fresnel field of an array of Acoustics, Arctic. Acoustics, Shallow Water. Ships.
hydrophones, sound radiated by individual breaking
waves has been measured in deep water
(0.35of o4 kHz, 4oUo15 m s1). The on-axis
source levels of individual breaking events, were
Further Reading
obtained and modeled as spatially and temporally Arveson PT and Vendittis DJ (2000) Radiated noise
discrete compact dipoles eqn ([6] with ks d51). The characteristics of a modern cargo ship. Journal of the
individual source levels were correlated with U and Acoustical Society of America 107: 118--129.
cb , the latter being the speed of a breaking wave Buckingham MJ and Potter JR (eds.) (1995) Sea Surface
event, a measure closely connected to breaking wave Sound ’94, vol. 3. Singapore: World Scientific, 494pp.
dissipation E. The correlation with cb was found to Chapman NR and Cornish JW (1993) Wind dependence of
deep ocean ambient noise at low frequencies. Journal of
the Acoustical Society of America 93: 782--789.
Crouch WW and Burt PJ (1972) The logarithmic
4 dependence of surface-generated ambient-sea-noise
A spherical surface for the integral would seem more appro-
priate, since the only way a monopole can become a dipole is by spectrum level on wind speed. Journal of the Acoustical
including the negative image above the free surface, in which case Society of America 51: 1066--1072.
AD ¼ 4p=3. Had a noncompact dipole been assumed with ks d ¼ Kennedy RM (1992) Sea surface sound source dependence
p=2 (eqn [6]), then AD ¼ 2p. There is as much as 5 dB difference in on wave-breaking variables. Journal of the Acoustical
these values compared to the one used by Kennedy. Society of America 91: 1974--1982.
366 ACOUSTIC NOISE

Kerman BR (1984) Underwater sound generation by in the ocean. Journal of the Acoustical Society of
breaking wind waves. Journal of the Acoustical Society America 78: 981--994.
of America 75: 149--165. Nichols RH (1981) Infrasonic ambient ocean noise
Kerman BR (ed.) (1988) Sea Surface Sound, vol. 1. measurements: Eleuthera. Journal of the Acoustical
Dordrecht: Kluwer Academic Publishers, 639pp. Society of America 69: 974--981.
Kerman BR (ed.) (1993) Sea Surface Sound, vol. 2. Sotrin BJ and Hodgkiss WS (1990) Fine-scale measurements
Dordrecht: Kluwer Academic Publishers, 750pp. of the vertical ambient noise field. Journal of the
Kewley DJ, Browning DG, and Carey WM (1990) Low- Acoustical Society of America 87: 2052--2063.
frequency wind-generated ambient noise source levels. Urick RJ (1986) Ambient Noise in the Sea. Los Altos, CA,
Journal of the Acoustical Society of America 88: Peninsula Publishing.
1894--1902. Wenz GM (1962) Acoustic ambient noise in the ocean:
Kibblewhite AC and Evans KC (1985) Wave–wave spectra and sources. Journal of the Acoustical Society of
interactions, microseisms, and infrasonic ambient noise America 34: 1936--1955.
 ACOUSTIC SCATTERING BY MARINE ORGANISMS
K. G. Foote, Woods Hole Oceanographic Institution, oriented directional echo sounder beams and the
Woods Hole, Massachusetts, USA possibility of fish avoidance reactions to the trans-
Copyright & 2001 Elsevier Ltd. ducer platform, typically a research vessel.
In another parallel development, the Doppler
principle was exploited to measure the rate of ap-
proach or recession of fish targets. Both horizontally
oriented echo sounder beams and sonar beams were
used. Early applications determined the swimming
Historical Overview speeds of schools of small pelagic fish and individual
Development of underwater sonar as a tool for salmon in rivers.
navigation and military operations, following sink- Applications of acoustics to fish in the 1970s were
ing of the Titanic in 1912, led inevitably to appli- accompanied by notable applications to zoo-
cations to marine organisms. By the 1930s, echoes plankton, if pursued less intensively owing to dif-
from fish schools had been detected. In the 1940s, ferences in commercial importance. Because of the
the deep sound-scattering layer was observed. Its enormous diversity of zooplankton species in size,
biological origin in mesopelagic fish was identified in shape, and composition, it was recognized early that
the 1950s. At the same time, applications to com- insonification over a band of frequencies is required,
mercial fish were pursued with vigor, and both sci- even for routine observation. This has usually been
entific echo sounders and fishery echo sounders achieved by the use of multiple resonant transducers,
began to be manufactured. but genuinely broadband sonars are also proving
Steady improvements in transduction enabled in- successful in yielding spectra of individual euphau-
dividual fish of certain species and sizes to be de- siids and copepods.
tected at ranges of hundreds of meters. The Recognition of the importance of bandwidth in
ultrasonic frequency of 38 kHz was becoming a scattering by zooplankton was accompanied by ap-
standard at this time; it was subsequently shown to preciation of the role of interpretive models. Acous-
be near the optimum for achieving detection of tic scattering models have been developed and
commercially important fish in the presence of at- applied to fish since the 1950s and to zooplankton
tenuation due to spherical spreading and absorption. since the 1970s.
Parallel to studies of single-fish scattering at ultra- The transition from analog to digital technologies
sonic frequencies were studies of scattering at sonic in the 1970s facilitated processing of echo data. This
frequencies, especially to determine the resonance has become steadily more automated and sophisti-
frequency in swimbladder-bearing fish, which is a cated, but always with operator control of important
measure of size. decisions through the man–machine interface.
Echo integration was introduced in 1965 as a tool Other developments in technology since the 1970s
for quantifying fish aggregations at essentially arbi- have extended the range of applications of acoustic
trary conditions of numerical density. This was rap- scattering by marine organisms. Multiple-element
idly developed, and it has been used routinely in transducers have been used to determine the three-
surveys of fish stock abundance since about 1975. dimensional locations and movement, as well as the
Introduction of standard-target calibration in the target strength, of individual animals. Compact,
early 1980s served the cause of quantification by high-frequency sonars have been mounted on fish
providing a rapid, high-accuracy method of enabling capture gear to observe the behavior of fish during
the results of echo integration to be expressed in catching operations. Steerable high-frequency sonars
absolute physical units. With few exceptions, stand- have been used to track fish schools during capture
ard-target calibration has become the method of and to map their three-dimensional shapes.
choice.
Sonar, with one or more obliquely oriented or
Physical Basis for Scattering
steerable beams, began to find common application
in the 1970s for counting fish schools that might be Acoustic scattering by a marine organism is, in
missed by a vertical echo sounder beam. This was a principle, no different from that of any other kind of
significant development for acknowledging the nar- scattering. Differences in the physical properties of
rowness of the sampling volume of vertically the causative bodies with respect to the surrounding

367
368 ACOUSTIC SCATTERING BY MARINE ORGANISMS

medium are accompanied by reflection and re- even diving birds. The first five groups are considered
fraction, or more generally diffraction, of incident in a separate section in the following.
waves. Organisms, with contrasts in mass density or
elasticity relative to sea water, are thus sources of
scattering. Dependences of Scattering
The processes of reflection, refraction, and dif- In general, scattering by marine organisms is affected
fraction occur at surfaces, both external and internal, by a number of factors. Some are listed here.
marking discrete changes in physical properties and
throughout the volume or inside embedded inho-
mogeneities, as characterized by continuous changes Intrinsic factors. Intrinsic to the scatterers are
in properties. The net result of the individual pro- size, shape, internal composition, and condition.
cesses is a redistribution in space of the incident en- Condition may be affected by the stage of
ergy field. Changes in direction and amplitude development, presence of reproductive products,
characterize the scattering. and degree of stomach filling. Behavior is another
intrinsic factor, if often directly affected or
determined by the external environment. It is
Classification of Marine Organisms as typically quantified through the attitude, or
Scatterers orientation, of the organism and its velocity of
movement.
Marine organisms are conveniently divided into
groups based on considerations of taxonomy and Extrinsic factors. Scattering is affected by the
anatomy. Two major groups are those of fish and insonification signal, hence by its spectral
zooplankton, but others are also treated. composition. For impulsive signals, the spectrum
Fish may be distinguished as cartilaginous or bony. may be broadly continuous. For a typical pulsed
Bony fish may be acoustically distinguished because sinusoid containing many wavelengths, the
the fish possesses or lacks a gas-filled swimbladder. spectrum will be narrow, and the signal can be
Swimbladders may be closed, with gas exchange ef- characterized by the center frequency, pulse
fected by the rete mirabile, or open, with gas ex- duration, and amplitude. Depth and history of
change effected by gulping air at the surface or by depth excursion may also influence the scattering, as
releasing a sphincter muscle on a duct leading to the in the case of rapid depth changes for physoclists.
exterior. The respective swimbladder types are called For swimbladdered fish lacking rete mirabile, depth
physoclists and physostomes. They are illustrated by excursions will necessarily affect the swimbladder
cod (Gadus morhua) and herring (Clupea harengus), form, with the volume changing in accordance with
respectively. Some mesopelagic fish possess gas-filled Boyle’s law, thus inversely with the ambient
swimbladders, including a number of myctophid pressure.
species. Some other myctophids, as well as the
deepwater fish orange roughy (Hoplostethus atlan-
ticus), possess swimbladders that are invested with Quantification of Scattering
wax esters. The whiptail (Coryphaenoides sub-
Nomenclature
serrulatus), a macrurid, possesses a swimbladder that
contains gas in a spongy matrix of tissue. Swim- Scattering properties of organisms are distinguished
bladderless fish are illustrated by mackerel (Scomber as belonging to individual organisms or to aggre-
scombrus). Cartilaginous fish lack a swimbladder, gations of organisms. The fundamental scattering
but their liver is large and presents a marked density property of a single organism is the scattering amp-
contrast with the surrounding fish flesh. litude. This is described through the idealization of a
Zooplankton come in many shapes and sizes, but plane harmonic wave incident on a finite scattering
acoustically their variable physical composition ad- body. At a great distance r from the body, the scat-
mits of a severe reduction. Three prominent classes tered pressure field or amplitude psc is related to the
have been identified: the liquidlike, the hard-shelled, incident pressure amplitude pinc by eqn [1].
and the gas-bearing. These are illustrated by, re-
spectively, euphausiids, pteropods, and siphono- psc ¼ pinc f expðikrÞ=r ½1
phores.
Other marine organisms have also been detected In eqn [1] f is the far-field scattering amplitude, r is
by scattering. These include squid, gelatinous zoo- the distance from the scatterer, k is the wavenumber
plankton, algae, benthos, marine mammals, and 2p/l, and l is the acoustic wavelength. The scattering
 ACOUSTIC SCATTERING BY MARINE ORGANISMS 369

amplitude f describes the angular characteristics of characteristic or mean backscattering cross section.
the scattered field. The differential or bistatic scat-
tering cross-section is jf j2 . In the backscattered dir- sA ¼ r A s ½8
ection f ¼ fb, and the backscattering cross-section is
given by eqn [2], where the dual convention of using Another measure of scattering is the extinction
both sbs and s is shown. cross-section. This measures the relative loss of en-
ergy due to scattering and internal absorption. It may
s be defined for an individual scatterer, but is generally
sbs ¼ jfb j2 ¼ ½2 applied to aggregations of organisms if they are
4p
sufficiently numerous.
The target strength TS is a logarithmic measure With few exceptions, the issue of calibration must
(eqn [3]; where r0 is the reference distance, typically be addressed when making measurements. Standard
1 m). methods are available for this, the aim being to de-
fine the system characteristics so that the result of a
sbs measurement, a voltage signal for instance, can be
TS ¼ 10 log ½3
r20 expressed as a pressure-wave amplitude in the water
medium.
When many scatterers are concentrated in a vol-
ume in which individual scatterers cannot be dis- Measurement
tinguished by their echoes, a collective standard
There are dozens of techniques for measuring the
measure of scattering is used. This is the volume
scattering properties of individual organisms and
scattering coefficient. In the backscattered direction,
aggregations of organisms. These are commonly
the volume backscattering coefficient sv is given by
distinguished as being in situ, without constraint in
eqn [4], where fb,i is the backscattering amplitude for
the natural environment of the organisms, or ex situ,
the ith scatterer of N, and V is the volume.
hence constrained in some way, wherever this might
be.
X
N   Target strength is a key quantity in many investi-
sv ¼ V 1 fb;i 2 ½4
i¼1
gations. It may be determined with a single-beam
echo sounder; for example, by repeated measurement
The volume backscattering strength is given by eqn of similar organisms that are acoustically resolved
[5]. and by appropriate statistical reduction of these
measurements. Alternatively, it may be measured
Sv ¼ 10 logðr0 sv Þ ½5 directly with a dual- or split-beam echo sounder, in
which the beam pattern can be determined in the
A quantity useful in echo integration is the area or direction of the organism, enabling the backscatter-
column backscattering coefficient sa, (eqn [6]), where ing cross-section to be extracted from each indi-
the integration is performed over the range interval vidual echo.
[r1, r2]. Similar measurements can be performed on single
organisms ex situ with greater control and hence
Zr2 knowledge of their state during measurement.
Measurements on tethered organisms, constrained to
sa ¼ sv dr ½6
maintain a given orientation during insonification,
r1
are popular.
Aggregations of organisms are frequently quanti-
In scattering by fish, a numerically more convenient
fied acoustically through the volume backscattering
measure of sa is eqn [7], which refers the back-
coefficient. If the number and occupied volume of the
scattering to the reference area of one square nautical
organisms are known, then the characteristic target
mile.
strength can be inferred through eqn [9].

sA ¼ 4p18522 sa ½7 Sv ¼ 10 logn þ TS ½9

This form is particularly useful, for the fundamental Here n is the numerical density of organisms, and TS
equation of echo integration is simply eqn [8], where is the so-called mean target strength corresponding
rA is the numerical density of fish referred to the to a single organism, but derived as the logarithmic
same area of one square nautical mile, and s is the measure of the mean backscattering cross-section.
370 ACOUSTIC SCATTERING BY MARINE ORGANISMS

Cages are often employed to confine a known or where g is the ratio of specific heats at constant
knowable number of organisms to a fixed volume. pressure and volume, P is the ambient pressure at
Measurement of Sv can then yield a value for TS. depth, r is the mass density of fish flesh, and a is the
equivalent spherical radius. For elongated bubbles or
Modeling swimbladder shapes, the resonance frequency is
The importance of target strength in many studies modified.
involving scattering by marine organisms is so great The backscattering cross-section s at frequency v
that recourse is frequently made to theoretical is given by eqn [11].
models. On the basis of assumptions about the shape
and internal composition of subject organisms, 4pa2
s¼ h i2 ½11
mathematical expressions may be derived that can be ½n0 =ðnH Þ2 þ ðn0 =nÞ2 1
evaluated for particular conditions of concentration
or frequency that might not be realistically explored H is the damping factor given by eqn [12], where c is
through measurement. Ultimately, measurements the speed of sound in water, and x is the viscosity of
may be used to refine models, and models to inter- fish flesh.
pret measurements.
2pan x
H 1 ¼ þ 2 ; ½12
n0 c pa n0 r
Fish as Scatterers
Swimbladder-bearing Fish Some numerical values for the various parameters
are r ¼ 1050 kg m3 and x ¼ 50 Pa s. The speed of
The swimbladder shape varies with species and with sound in sea water varies over the range 1450–
condition of the individual specimen. An example of 1550 m s1, depending on temperature, salinity, and
a swimbladder in corpus is shown in Figure 1. Here pressure.
the swimbladder of an Atlantic herring (Clupea For gadoids and clupeoids in the size range 8–
harengus) has been exposed by careful dissection. 30 cm, n0 varies over 2.2–0.3 kHz. Given the inverse
relationship of resonance frequency and size in eqn
Low frequencies At low frequencies, with acoustic [10], smaller fish will have higher resonance fre-
wavelengths much greater than characteristic quencies. Thus, mesopelagic fish with partially wax-
swimbladder dimensions, the effect of a pressure invested swimbladders may have resonance fre-
wave on the swimbladder is essentially that of quencies in the low ultrasonic range. Very large
uniform compression and rarefaction. Consequently, swimbladdered fish, say with a total fish length ex-
a spherical model can be used. In fact, some ceeding 1 m, will have resonance frequencies of the
swimbladder-bearing fishes have been modelled order of hundreds of hertz. The corresponding
successfully as a spherical gas bubble surrounded by a backscattering cross-section, hence target strength,
finite layer of fish flesh that acts as a viscous fluid can be computed from eqs [10]–[12]. It is important
medium supporting surface tension on the interface to note that the quality factor of the resonance
between the shell and fish flesh. The volume of a condition, eqn [13], where Dn describes the range in
bubble of radius a is equivalent to that of the frequency over which s decreases to one-half its
swimbladder. Equation [10] gives the resonance maximum value, may be of the order of 1.5–3.
frequency v0 of an immersed spherical gas bubble,

 Q ¼ n0 =Dn ½13
1 3gP 1=2
n0 ¼ ½10 Implicit in the low-frequency condition of the
2p ra2
model is that s is independent of orientation. Aver-
ages of s with respect to arbitrary orientation dis-
tributions will be identical to s itself.
When computing average values of s for aggre-
gations of swimbladdered fish of varying size, s must
be averaged with respect to the size distribution. The
characteristic target strength is determined from the
definition in eqn [3].
Figure 1 Drawing of a specimen of Atlantic herring (Clupea
harengus), female, 36.0 cm long, 453 g, with exposed Intermediate frequencies As the acoustic
swimbladder. (Drawing by H. T. Kinacigil, used with permission.) wavelength decreases toward characteristic
 ACOUSTIC SCATTERING BY MARINE ORGANISMS 371

swimbladder dimensions, the scattering becomes is the total fish length in centimeters).
markedly directional, and the backscattering begins
to depend sensitively on the orientation of the fish. TS ¼ 20 log l  72:7 ½20
From measurements made both in situ and ex situ,
the empirical relationship of eqn [14] between mean Some stocks of orange roughy (Hoplostethus
target strength TS at 38 kHz and total fish length l in atlanticus) are being surveyed about their seamount
centimeters has been derived for a number of habitats. Determination of the target strength of this
gadoids. deepwater fish with fat-invested swimbladder is ad-
mittedly problematical. Some work suggests con-
TS ¼ 20 log l  67:5 ½14 vergence of the mean target strength of a 35 cm long
orange roughy at 38 kHz to about  48 dB. If the
Equation [15] applies for clupeoids standard equation for mean target strength–length
were used, namely eqn [21],
TS ¼ 20 log l  71:9 ½15

The average backscattering cross-section s may be TS ¼ 20 log l þ b ½21

determined immediately from eqn [3]. For a cod of
the coefficient b would be  79 dB.
length l ¼ 50 cm, TS ¼  33.5 dB and s ¼ 56 cm2.
For modeling scattering by swimbladdered fish at
For a herring of length l ¼ 30 cm, TS ¼  42.4 dB
these frequencies, the Kirchhoff approximation
and s ¼ 7.2 cm2.
model can be used. This assumes that the fish is
Blue whiting is an important commercial stock in
represented by the swimbladder, which acts as a
both hemispheres, and it is routinely surveyed by
pressure-release surface where it is directly insoni-
acoustics. To convert measurements of acoustic
fied, and as a surface without response otherwise.
density at 38 kHz to numerical density in accordance
A more general scattering model is that of the
with the echo integration equation [8], eqn [16],
boundary-element method. The swimbladder is rep-
where l is the fork length in centimeters. is used for
resented by a mesh of points, called nodes, spanning
the northern-hemisphere blue whiting (Micro-
the surface, illustrated in Figure 2. The harmonic
mesistius poutassou):
wave equation is solved numerically, assuming con-
TS ¼ 21:7 log l  72:8 ½16 tinuity of pressure and normal component of velocity
at each node. It is thus possible to model the effects
Equation [17] applies for the southern-hemisphere of internal gas density and pressure.
southern blue whiting (Micromesistius australis), To convert modeled values for s as a function of
where l is again the fork length in centimeters. orientation to an average value, an orientation dis-
tribution is required. Ideally, this is done on the basis
TS ¼ 25:0 log l  81:4 ½17 of in situ observations, but often such data are
lacking and an orientation distribution must be as-
Coincidentally, perhaps, the target strength of sumed. Some orientation distributions are described
yellowfin tuna (Thunnus albacares) at 38 kHz is in the literature. In some special circumstances it has
nearly identical to that of Micromesistius australis
and is given by eqn [18].

TS ¼ 25:3 log l  80:6 ½18

The target strength of bigeye tuna (Thunnus obesus) z

under similar conditions is given by eqn [19].

TS ¼ 24:3 log l  73:3 ½19

x
These relations were established from specimens in
the approximate size range 50–130 cm and 3–50 kg.
The whiptail (Coryphaenoides subserrulatus),
with a swimbladder containing gas-filled spongy y
tissue, seems to have a mean in situ target strength at
Figure 2 Boundary element model of the swimbladder of a
38 kHz that is consistent with the equation de- specimen of pollack (Pollachius pollachius), 34.5 cm in length,
veloped for another macrurid, the blue grenadier or with anterior end to the lower right (y direction). (Model by D. T. I.
hoki (Macruronus novaezelandie) (eqn [20], where l Francis, used with permission.)
372 ACOUSTIC SCATTERING BY MARINE ORGANISMS

been possible to infer the orientation distribution a pelagic shark might be represented by a body
by a combination of acoustic measurement and with the size, shape, and physical properties of
modeling. the liver.
The relationship of maximum and average meas-
ures of s is given approximately by eqn [22].
Zooplankton as Scatterers
smax E7save ½22 Liquid-like Bodies

Alternatively, eqn [23] can be used. A number of prominent and abundant zooplankton
can be classified as liquidlike in their acoustic prop-
TSmax ETSave þ 5 dB ½23 erties. Extensive modeling and measurement have
demonstrated that internal shear waves have neg-
Measures of the extinction cross-section are rela- ligible influence in scattering by such organisms. The
tively rare, there being few occasions when it is ne- animals are thus generally fluidlike in their prop-
cessary to compensate for scattering losses. However, erties. If the same animals lack sizable organs or
measurement or inference suggest that the extinction other tissue presenting large contrasts in mass density
cross-section is very roughly 1–3 times the back- or compressibility relative to the sea water immer-
scattering cross-section at intermediate frequencies. sion medium, then the acoustic properties of the or-
Ultimately, the cross-sections and their ratio must ganisms are more particularly liquidlike, and their
depend on the behavior of the organism, as quanti- acoustic scattering is consequently relatively weak.
fied through the orientation distribution. Two examples of zooplankton with liquidlike prop-
erties are euphausiids and copepods. These are also
High frequencies When the acoustic wavelength representative of homogeneous and inhomogeneous
becomes very small compared to the swimbladder scatterers, respectively.
size, scattering by other tissues may become
important. The contributions of head structure, Homogeneous liquidlike bodies The expectation
vertebrae, and even scales at very high frequencies of relatively weak scattering by euphausiids has
have been established through ex situ measurement. been confirmed by measurement. For example, the
Modeling of scattering by such structures can be target strength of Antarctic krill (Euphausia
computationally excessive, suggesting the superba) of mean lengths 30–39 mm is in the range
advantages of stochastic modeling if direct from  88 to  83 dB at 38 kHz and from  81 to
measurement is not possible or convenient.  74 dB at 120 kHz. The respective acoustic
wavelengths are 39 and 12.5 mm.
For a scattering body that is relatively long com-
Swimbladderless Fish
pared to the wavelength, the scattering will be in-
The mackerel is a prominent example of a swim- herently directional. Laboratory measurement has
bladderless fish. Its target strength must be attributed demonstrated strong effects of orientation on scat-
to the non-swimbladder structures and hence is in- tering by euphausiids in the size range 30–42 mm at
trinsically complicated at nearly all frequencies. frequencies of 120 kHz and higher.
At intermediate frequencies, the mean target strength In modeling scattering by homogeneous liquidlike
is roughly 10 dB less than that of a gadoid of com- zooplankton, there are just two significant material
parable size (eqn [24]). properties, the mass density and compressibility, or
longitudinal-wave sound speed. A variety of models
TSmackerel ETSgadoid  10 dB ½24 can be used to represent shape. At low frequencies,
a single euphausiid can be represented by a finite
For cartilaginous fish, such as sharks, the liver may circular cylinder or even a sphere, with volume
be very large. In pelagic sharks, this may be of the equal to that of the animal. At higher frequencies, the
order of 7–23% by weight; in demersal sharks, same animal might be represented as a finite, bent,
3–6%. The specific gravity of lipids is of the order of tapered cylinder or, better, by the actual shape of the
0.87–0.92 in pelagic sharks and 0.93–0.94 in de- exoskeleton.
mersal sharks, further suggesting the role of the liver Scattering models for euphausiids have demon-
in buoyancy and its significance in acoustic scat- strated the sensitive dependence of target strength on
tering. At least for the pelagic sharks, the size and both the material properties and orientation of the
difference in mass density may explain much of organism. Given the rarity of measurements of ma-
the target strength. Were a model to be constructed, terial properties, their seasonal and individual
 ACOUSTIC SCATTERING BY MARINE ORGANISMS 373

variability, and the generally unknown orientation, z

there has been little systematization of measured
values of target strength.
Theoretical understanding of scattering by
euphausiids has succeeded in associating large lobes
with the echo spectrum at rather short acoustic
wavelengths. When these are combined with know-
ledge of the target strength to within about an order
of magnitude, it is possible to classify euphausiids by
their acoustic signature.

Inhomogeneous liquidlike bodies Copepods, like y

euphausiids, also display relatively weak acoustic
scattering. Unlike euphausiids, however, their x
z
internal structure is acoustically distinct, being
composed of two dominant scatterers, a prosome
and an embedded oil sac. Because of the low density
of lipids in the oil sac, of the order of 900 kg m3,
the prosome must be correspondingly more
massive. Because the copepod body as a whole is
close to neutral buoyancy in sea water, the target
strength is due to the internal contrast in mass
density and compressibility, or longitudinal-wave
y
sound speed, between the prosome and oil sac.
Measurement has shown that the target strength x
of a 2 mm long copepod, Calanus finmarchicus, is in
the approximate range from  95 to  90 dB over Figure 3 Boundary element models of the prosome and oil sac
of a specimen of Calanus finmarchicus, stage 6 female, 2.74 mm
the frequency range 1600–2400 kHz. in length, with anterior end to the lower left (x direction). (Models
Copepods have been modeled as composite two- by D. T. I. Francis, used with permission.)
liquid-body structures. Numerical values for the
mass density and longitudinal-wave sound speed
have been derived from measurements or have been measurements at wavelengths roughly comparable to
assumed. The shapes of embedded oil sac and en- the maximum shell dimension.
compassing prosome, illustrated in Figure 3, have
Gas-bearing Bodies
been determined from videomicroscopic cross-sec-
tions in dorsal and lateral views. Results of modeling Siphonophores are representatives of gas-bearing
of copepods have shown the expected weak de- zooplankton, with gas inclusions in the pneumato-
pendence on orientation at low or moderate fre- phores. These are generally small compared to
quencies, and an overall mean target strength that is overall dimensions of specimens, and the target
in line with measured values. strength varies widely over the frequency range 350–
750 kHz. In particular, the target strength varies over
Hard-shelled Bodies the range from  90 to  60 dB, but with no ap-
parent systematic dependence on frequency. This
An example of a hard-shelled zooplankton is the wide range is suggestive of interference between
pteropod Limacina retroversa, a marine snail with a echoes from the gas inclusions and the nongaseous
spiral shell, opercular opening, and wings that propel tissue, the basis of an acoustic model.
it through the mid-water column. The target strength
of specimens of shell length 2 mm has been measured
over the approximate frequency range from 350 to Other Organisms as Scatterers
750 kHz. The target strength varies between  80
Squid
and  60 dB, depending on both frequency and
orientation. A number of specimens of squid have been observed
The pteropod has been modeled as a rough by acoustics. These include Todarodes pacifica,
spherical shell with a circular opening. Predictions of Loligo opalescens, and Loligo vulgaris reynaudii. In
scattering have been in reasonable agreement with a survey of the second species, performed at
374 ACOUSTIC SCATTERING BY MARINE ORGANISMS

120 kHz, the target strength of specimens of mean and the masses 0.33–0.8 kg. Target strengths ex-
dorsal mantle length 11.6 cm and mean mass 23.7 g pressed relative to 1 kg of biomass vary from  35 to
was about  59 dB. Thus in the standard target  28 dB at 50 kHz, from  33 to  24 dB at 70 kHz,
strength–length equation [21], but with l repre- and from  29 to  22 dB at 200 kHz.
senting the mean dorsal mantle length, b is about Smaller algae, the phytoplankton Prorocentrum
 80 dB. For Todarodes pacifica of mean dorsal micans, Peridinium triquetrum, Olistodiscus luteus,
mantle length 16 cm and mean mass 95 g, the target Dunaliella salina, Platimous viridis, and Phaeo-
strength is about  51 dB at 28.5 kHz and  55 dB dactilum tricornutum, are also being studied by
at 96.2 kHz, corresponding to values of b of  75 acoustics. Measurements of reverberation, in par-
and  79 dB, respectively. For Todarodes pacifica of ticular, are being used in attempts to quantify the
mean dorsal mantle length 23.7 cm and mean mass volume of gas vacuoles.
340 g, the respective mean target strengths at 28.5,
50, 96.2, and 200 kHz are about  45.7,  46.5, Clams
 48.0, and  47.6 dB, with respective values of b of
Both the razor clam (Tagelus dombeii) and the surf
 75,  74,  76, and  76 dB. For Loligo vulgaris
clam (Mesodesma donacium) have been surveyed by
reynaudii, the target strength was measured at
acoustics. Beds of the razor clam have been surveyed
38 kHz for sufficiently dispersed animals of mean
in shallow water over a flat bottom. Echograms that
mass 300 g. The target strength when referred to 1 kg
show the bottom–surface–bottom reflection in add-
was  42.5 dB. This compares favorably with the
ition to the first bottom reflection show an enhanced
measurements on Loligo opalescens at 120 kHz and
registration above the so-called second bottom echo.
Todarodes pacifica at 28.5 kHz. When expressed
Counting of its characteristic serrations provides a
relative to 1 kg, the respective target strengths are
quantitative measure of clam density.
 42.3 and  41.1 dB.
Marine Mammals
Common Jellyfish
A few measurements have been reported on the tar-
In anticipation of acoustic surveying of the cteno- get strength of the sperm whale (Physeter catodon)
phore Mnemiopsis leidyi and other gelatinous zoo- and the humpback whale (Megaptera novaeangliae)
plankton, namely Aurelia aurita and Pleurobrachia in situ. Measurements have been made of the At-
pileus, in the Black Sea, measurements have been lantic bottlenose dolphin (Tursiops truncatus) in
made of the target strength of the common jellyfish captivity. Measurements made on a 2.2 m long
Aurelia aurita. Functional regression equations have 126 kg female dolphin in broadside aspect at the
related the mean target strength in decibels to the surface revealed a mean target strength that de-
disk diameter d in centimeters. At 120 kHz, the re- creased from about  10 dB at the lowest measure-
lation is eqn [25]. ment frequency of 23 kHz to about  24 dB at
45 kHz, rising to about  20 dB at 65 kHz, then
TS ¼ 14:7 log d  74:6 ½25 falling to  25 dB at 80 kHz. The observed degree of
variability about these nominal values due to re-
At 200 kHz it is eqn [26]. peated insonification was 4–11 dB to within the first
standard deviation to either side.
TS ¼ 39:6 log d  104:4 ½26

Thus for a specimen with mean diameter 10 cm, Challenges

TS ¼  59.9 and  64.8 dB at the respective
For all of the instances and applications of acoustic
frequencies.
scattering by marine organisms, there is an enormous
demand for enhanced imaging capability and more
Algae
quantitative understanding, including both improved
Algae, such as kelp, are being surveyed by acoustics. measurement methods and models. In addition to
For purposes of quantification, the acoustic prop- refinement of current measurement methods, in-
erties of the plants themselves are being studied, both cluding those for quantifying concentrations of
by experiment and by theoretical modeling. Meas- marine organisms, instruments are being developed
urements have been performed on leaves of Lami- or adapted for application. These include high-fre-
naria saccarina and L. digitata at three ultrasonic quency sonars, multibeam sonars, and continuously
frequencies. The lengths of these span the range 0.7– broadband echo sounders, operating at both low and
2 m; the widths 0.4–0.9 m; the thicknesses 1–5 mm; high frequencies.
 ACOUSTIC SCATTERING BY MARINE ORGANISMS 375

In general, the addition of bandwidth to acoustic See also

devices, whether achieved by multiple frequencies or
a continuous spectrum, is a firm objective of many Bioacoustics. Sonar Systems.
development efforts. Its usefulness in classification is
appreciated from certain studies in zooplankton
scattering, but it would aid studies of nekton scat- Further Reading
tering if successful.
Recognition of the importance of understanding Craig RE (ed.) (1984) Fisheries Acoustics A symposium
the acoustic properties of individual organisms is held in Bergen, 21–24 June 1982, Rapports et Proces-
similarly influential in promoting developments and Verbaux des Reunions, vol. 184. Copenhagen:
applications. Determining the properties of single International Council for the Exploration of the Sea.
Foote KG (1997) Target strength of fish. In: Crocker MJ
organisms when found en masse remains a challenge,
(ed.) Encyclopedia of Acoustics, vol. 1, pp. 493--500.
as does quantifying avoidance reactions or avoiding
New York: Wiley.
inducing them. While there are many techniques for Foote KG and Stanton TK (2000) Acoustical methods.
determining target strength, their application re- In: Harris RP, et al. (ed.) ICES Zooplankton Metho-
quires ingenuity to elucidate some of the principal dology Manual, pp. 223--258. London: Academic Press.
dependences. The general lack of information on the Freon P and Misund OA (1999) Dynamics of Pelagic Fish
depth dependence of target strength for gas-bearing Distribution and Behaviour: Effects on Fisheries and
organisms is a particular, prominent example. Stock Assessment. Oxford: Fishing New Books.
Modeling of scattering by marine organisms offers Karp WA (ed.) (1990) Developments in Fisheries
much potential for resolving physically intractable Acoustics, A symposium held in Seattle, 22–26 June
problems, such as those involving separation of 1987, Rapports et Proces-Verbaux des Reunions, vol.
189. Copenhagen: International Council for the
echoes from individual organisms in the midst of
Exploration of the Sea.
their social aggregations or inferring the acoustic
Margetts AR (ed.) (1977) Hydro-acoustics in Fisheries
properties of organisms that are very fragile or that Research, A symposium held in Bergen, 19–22 June
occur in extreme environments. Both analytical and 1973, Rapports et Proces-Verbaux des Reunions, vol.
numerical models, however, require knowledge of 170. Copenhagen: International Council for the
the physical properties, shape, and behavioral char- Exploration of the Sea.
acteristics, such as the orientation distribution, of the Medwin H and Clay CS (1998) Fundamentals of Acoustical
subject organisms. Acoustic inference of the in situ Oceanography. San Diego, CA: Academic Press.
properties of organisms, by special measurement Nakken O and Venema SC (eds.) (1983) Symposium on
techniques and aided by models, appears very at- Fisheries Acoustics, Selected papers of the ICES/FAO
tractive if generally difficult. Symposium on Fisheries Acoustics, Bergen, Norway,
21–24 June 1982, FAO Fisheries Report no. 300. Rome:
An enhanced imaging capability based on acoustic
Food and Agriculture Organization of the United
scattering is also valuable. If realized in a compact
Nations.
device, this could aid fishing practice, as in providing Ona E (1990) Physiological factors causing natural
fishers with information on the species and size of variations in acoustic target strength of fish. Journal of
organisms present in the water column or on the the Marine Biological Association of the United
bottom without actually having to capture the or- Kingdom 70: 107--127.
ganism to make the determination. For the re- Physics of Sound in the Sea (1969) Reprint of the 1946
searcher, being able to distinguish different edition. Washington, DC: Department of the Navy.
organisms with overlapping distributions would be Progress in Fisheries Acoustics (1989) Proceedings of the
invaluable in aiding the study of relationships, ul- Institute of Acoustics, vol. 11, no. 3. St. Albans:
timately to advance the goals of ecosystem analysis Institute of Acoustics.
Simmonds EJ and MacLennan DN (eds.) (1996) Fisheries
and understanding.
and Plankton Acoustics, Proceedings of an ICES
international symposium held in Aberdeen, Scotland,
Acknowledgments 12–16 June 1995. ICES Journal of Marine Science, vol.
53, no. 2. London: Academic Press.
This is Woods Hole Oceanographic Institution con- Urick RJ (1983) Principles of Underwater Sound, 3rd edn.
tribution number 10271. New York: McGraw-Hill.
 ACOUSTIC SCINTILLATION THERMOGRAPHY
P. A. Rona, Rutgers University, New Brunswick, scans of the seafloor to detect weak fluctuation in the
NJ, USA index of refraction of the water near the seafloor.
C. D. Jones, University of Washington, Seattle, Random index of refraction changes result from tem-
WA, USA poral variations in the water temperature caused by
& 2009 Elsevier Ltd. All rights reserved. turbulent mixing, which produce detectable changes in
travel time of an acoustic ray as the ray propagates
from a stationary acoustic transducer through the
diffuse flow to the seafloor and is scattered back
Introduction: Mapping Diffuse Flow at through the diffuse flow. These changes in travel time
Seafloor Hydrothermal Sites increase from the nearer to the farther portion of the
sonar footprint, causing the associated echo waveform
Acoustic scintillation thermography or AST is a to change in shape with each transmission and re-
method to detect and map diffuse flow at seafloor sulting in decorrelation of successive sound pulses.
hydrothermal sites. Diffuse flow is the discharge of In essence, the scintillation of the acoustic wave as it
low-temperature hydrothermal solutions (up to tens passes through a turbulent flow field and scatters off
of degrees Celsius) as seepage through areas of the the underlying seafloor makes the seafloor appear to
seafloor. It is considered widespread at low intensity shimmer, much as the eye would detect hot water or
(temperature less than 1 1C; flow rate less than hot air shimmering against a stationary backdrop. For
1 cm s 1) in ocean basins and at higher intensity certain types of seafloor hydrothermal flow (such as
(temperature less than 100 1C; flow rate less than the case with diffuse flow where buoyant turbulent
1 m s 1) in seafloor hydrothermal fields. Quantita- microplumes are concentrated near the seafloor), the
tive assessment of diffuse flow is important because AST method can provide a means of remotely de-
the cumulative thermal flux of diffuse flow through tecting areas of flow and a potential method of
areas of the seafloor may equal or exceed that measuring the scales of temperature and velocity
of focused flow from associated high-temperature fluctuations in the near-bottom boundary layer.
(up to 405 1C) higher-flow-velocity (flow rate tens to Consider the scattering geometry illustrated in
hundreds of centimeters per second) black smoker Figure 1, where a turbulent boundary layer overlays
vents. Chemical fluxes in diffuse flow may be se- a rough seafloor and the incident acoustic field
lectively significant. Measuring diffuse flow is dif- propagates forward and back between the transducer
ficult. Occurrence is patchy and temperatures and and the backscattering seafloor through the bound-
flow velocities are low. Because it is clear, lacking the ary layer. Assume that the seafloor is not changing in
suspended particulate matter of black smoker time and there is no motion of the acoustic trans-
plumes, diffuse flow cannot be detected by measur- ducer, but the turbulent flow is evolving temporally.
ing attenuation or backscatter of light and sound. Between consecutive and rapid sonar returns from
the same spot on the seafloor, turbulent mixing will
cause slight spatial and temporal fluctuations of the
Method
index of refraction of the water near the seafloor.
The AST method uses the phase-coherent correlation These fluctuations will cause weak random forward
of acoustic backscatter from consecutive sonar scattering of the acoustic field as it propagates

z
Stationary source receiver

Vertical Incident field

beam pattern
Turbulent flow

x
Seafloor

Figure1 Propagation of an acoustic pulse through a turbulent boundary layer near the seafloor produced by diffuse hydrothermal
flow and backscatter of the pulse from the seafloor.

376
 ACOUSTIC SCINTILLATION THERMOGRAPHY 377

through the turbulent flow, resulting in temporal Volume and heat fluxes from areas of diffuse flow can
fluctuations in the amplitude and phase of the be estimated when in situ measurements of water
acoustic field measured at the receiver. The temporal temperature and vertical flow velocity are made in the
changes in backscatter can be measured by esti- areas of diffuse flow. For example, in situ measure-
mating the phase-coherent temporal decorrelation of ments in diffuse flow of temperature using thermistors
backscatter between successive echoes, whereas the and of vertical flow velocity using video to record rise
backscatter intensity from the turbulence itself is too rate against a calibrated rod in an acoustically imaged
weak to measure. In this case, the rate and magni- area of diffuse flow and black smokers in the Main
tude of temporal decorrelation can be used to detect Endeavour Field indicated that heat flux of the diffuse
and potentially measure scales of mixing in the flow was a multiple of that of the associated black
boundary layer. smokers. Figure 3 shows areas of decorrelation in-
tensity (draped onto bathymetry) for a relatively large
area of seafloor surveyed at the Clam Bed hydro-
Application thermal field on the Juan de Fuca Ridge (near the
Main Endeavour Field site). The AST method was
The method was first applied to detect and map applied using a sonar mounted on a hovering ROV to
diffuse flow in seafloor hydrothermal fields using a map the diffuse flow over a 3500 m  900 m area of
human occupied vehicle (HOV). Several later ex- the axial valley of the Endeavour segment of the Juan
periments have shown that deep ocean diffuse flow de Fuca Ridge. Using in situ sensors to simultaneously
can be detected over relatively large areas of the measure temperature and flow velocity, a diffuse heat
seafloor using a remotely operated vehicle (ROV) flux of 150 MW was integrated over the areas of the
platform. In these experiments, backscatter was re- AST decorrelation anomalies, indicating that the heat
corded using a 200 kHz multibeam sonar mounted transferred by diffuse flow is a significant component
on an ROV as it hovered above the seafloor at of total heat flux in the study area.
intervals along a track line. Figure 2 shows an area of The temporal correlation of the backscattered field
detected diffuse flow (as verified by video) and the is found by correlating collocated returns from the
corresponding backscattered intensity image. The seafloor as a function of transmission time. Consider
two images are derived from the same sonar scans of the monostatic geometry for backscatter as illus-
the seafloor taken at 0.1 s intervals. Future appli- trated in Figure 1, where a source/receiver system is
cations will extend the AST method to seafloor ob- fixed in space and measures backscatter at discrete
servatories to record changes in area of diffuse flow pulse transmission times tn. The temporal corre-
at a seafloor site on timescales up to years. lation of backscatter signals between a transmission
An informative representation is to drape areas of at time tm and a later transmission at tn is defined as
diffuse flow detected by the AST method over a cor-
responding bathymetric map to show the relation
between diffuse flow and seafloor morphology. Cðr; tm ; tn Þ ¼ huðr; tm Þu ðr; tn Þi ½1

Backscatter intensity (dB) Decorrelation intensity (dB)

50 50

40 40

30 30

20 20

10 10

0 0
20 0 20 40 20 0 20 40
Meters Meters

Figure 2 An AST image of diffuse flow near Hulk vent in the Main Endeavour Field on the northern Juan de Fuca Ridge is shown in
the right panel. The areas of diffuse flow (light patches) are detected as an increase in the decorrelation intensity. A conventional sonar
image of the same area is shown in the left panel. Colors indicate level of backscatter intensity (red is highest intensity) related to
bottom roughness.
378 ACOUSTIC SCINTILLATION THERMOGRAPHY

30
20
10

50

100

150

Me 150
ter
s 200

100
250

50
300

Figure 3 Areas of diffuse hydrothermal flow from the seafloor (Clam Bed site on the northern Juan de Fuca Ridge) detected by
backscatter decorrelation (area in meters; Jones et al., 2000). Colors indicate level of decorrelation intensity (yellow is higher intensity)
corresponding to intensity of diffuse flow.

where u(r; tn) ¼ A(r; tn) exp[if(r; tn)] is the complex The correlation function is a measure of the
envelope of the backscattered field as a function of change in the scattering medium, but noise and
range r along the seafloor. Range is determined by motion will also cause signal decorrelation. To cor-
round-trip acoustic time-of-flight as is usual in sonar. rect for the effects of low signal-to-noise levels when
The complex envelope, also known as the baseband the return from the seafloor is weak, it is convenient
signal, incorporates the amplitude (A) and phase (f) to define the decorrelation intensity as
of the echo signal, and the asterisk in eqn [1] denotes
complex conjugation. For m ¼ n, the correlation Ir ðr; t0 ; tn Þ ¼ Iðr; t0 ; tn Þ½1  r̂ðr; t0 ; tn Þ ½3
function is real and proportional to the signal in-
tensity. With man, the correlation function will be where the normalized temporal correlation co-
complex due to signal fluctuations in amplitude and efficient is
phase, with phase dominating. If the area of seafloor
observed has not changed because of motion of the  .qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
source/receiver, fluctuations in the signal will be due  
r̂ðr; t0 ; tn Þ ¼ Ĉðr; t0 ; tn Þ Ĉðr; t0 ; t0 Þ Ĉðr; tn ; tn Þ
to changes in the water above the seafloor as a
function of time tn. With m ¼ 0 as the reference time ½4
and n ¼ 1, 2, 3,y increasing monotonically, eqn [1]
is the cumulative temporal correlation function. For The scalar intensity I is the geometric mean of the
a single realization and finite number of discrete data intensity in the range window, defined as
points, an estimate of the ideal correlation function is
found by windowing the backscattered signals in qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
range and summing over range bins, Iðr; t0 ; tn Þ ¼ Ĉðr; t0 ; t0 Þ Ĉðr; tn ; tn Þ ½5

1X M
Ĉðri ; t0 ; tn Þ ¼ uðri ; t0 Þu ðri ; tn Þ ½2 The decorrelation intensity [3] is useful as a relative
M i¼1
measure of change when the scattering strength
where ri is a set of M discrete range values that over an area of the seafloor is nearly uniform and
correspond to a finite area (or patch) of the seafloor helps avoid the comparison of areas where the scat-
centered at range r. tered signal level is relatively low. Motion of the
 ACOUSTIC SCINTILLATION THERMOGRAPHY 379

source/receiver is more difficult to compensate and See also

will depend on the randomness in the seafloor.
Angular motion of the source/receiver between re- Acoustic Measurement of Near-Bed Sediment
cordings should be much smaller than the angular Transport Processes. Acoustic Noise. Acoustic
correlation scale of the scattering from the seafloor. If Scattering by Marine Organisms. Acoustics, Deep
Water. Acoustics in Marine Sediments. Acoustics,
there is motion the rate of signal decorrelation due to
Shallow Water. Bioacoustics.
motion must be less than the rate of signal decorr-
elation due to the diffuse flow.
The size of the area of seafloor (or patch) used to
estimate the correlation is defined by the length of Further Reading
the transmitted pulse and the number of samples
needed to form an accurate estimate. For a band- Bendat JS and Piersol AG (2000) Random Data: Analysis
width-limited Gaussian signal with bandwidth B, it and Measurement Procedures, 3rd edn., pp. 291–296.
is well known that the normalized mean-square error New York: Wiley-Interscience.
of the correlation estimate is given as Johnson HP, Hautala SL, Tivey MA, et al. (2002) Survey
studies hydrothermal circulation on the northern Juan
h i 1 h i de Fuca Ridge. EOS, Transactions, American Geo-
e2 Ĉðr; t0 ; tn Þ ¼ 1 þ r2 ðr; t0 ; tn Þð1 þ N=SÞ2 physical Union 83(73): 78--79.
2BT Jones CD, Jackson DR, Rona PA, and Bemis KG (2000)
½6 Observations of hydrothermal flow (abstract). Journal
of the Acoustical Society of America 108(5): 2544--
where r is the desired true correlation coefficient and 2545.
T is the length of the record used to form the estimate Rona PA, Jackson DR, Wen T, Jones CD, Mitsuzawa K,
(corresponding to the number of points M or patch and Bemis KG (1997) Acoustic mapping of diffuse flow
size). The signal-to-noise ratio (S/N) is assumed at a seafloor hydrothermal site: Monolith Vent, Juan de
constant and uncorrelated between transmission Fuca Ridge. Geophysical Research Letters 24(19):
times t0 and tn. 2351--2354.
 ACOUSTICS IN MARINE SEDIMENTS
T. Akal, NATO SACLANT Undersea Research Centre, boulder, gravels, coarse and fine sand to silt and clay,
La Spezia, Italy including materials deposited from chemical and
Copyright & 2001 Elsevier Ltd. biological products of the ocean, all being saturated
with sea water. Marine sediments are generally a
combination of several components, most of them
coming from the particles eroded from the land and
the biological and chemical processes taking place in
Introduction sea water. Most of the mineral particles found in
shallow and deep-water areas, have been transported
Because of the ease with which sound can be trans- by runoff, wind, and ice and subsequently distributed
mitted in sea water, acoustic techniques have pro- by waves and currents.
vided a very powerful means for accumulating After these particles have been formed, trans-
knowledge of the environment below the ocean sur- ported, and transferred, they are deposited to form
face. Consequently, the fields of underwater acous- the marine sediments where the physical factors such
tics and marine seismology have both used sound as currents, dimensions and shapes of particles and
(seismo-acoustic) waves for research purposes. deposition rate influence the spatial arrangements
The ocean and its boundaries form a composite and especially sediment layering. Particles settle to
medium, which support the propagation of acoustic the ocean floor and remain in place when physical
energy. In the course of this propagation there is forces are not sufficiently strong to move them. In
often interaction with ocean bottom. As the lower areas with strong physical forces (tidal and ocean
boundary, the ocean bottom is a multilayered struc- currents, surf zones etc.) large particles dominate,
ture composed of sediments, where acoustic energy whereas in low motion energy areas (ocean basins,
can be reflected from the interface formed by the enclosed bays) small particles dominate.
bottom and subbottom layers or transmitted and During the sedimentation process these particles,
absorbed. At low grazing angles, wave guide phe- based on the physical and chemical interparticle
nomena become significant and the ocean bottom, forces between them, form the sedimentary acoustic
covered with sediments of different physical charac- medium: larger particles (e.g., sands) by direct con-
teristics, becomes effectively part of the wave guide. tact forces; small particles (e.g., clays and fine silts)
Depending on the frequency of the acoustic energy, by attractive electrochemical forces; and silts, re-
there is a need to know the acoustically relevant maining between sands and clays are formed by the
physical properties of the sediments from a few combination of these two forces. The amount of the
centimeters to hundreds of meters below the water/ space between these particles is the result of different
sediment interface. factors, mainly size, shape, mineral content and the
Underwater acousticians and civil engineers are packing of the particles determined by currents and
continuously searching for practical and economical the overburden pressure present on the ocean
means of determining the physical parameters of the bottom.
marine sediments for applications in environmental Figure 1 is an example of a core taken very care-
and geological research, engineering, and underwater fully by divers, to ensure an undisturbed internal
acoustics. Over the past three decades much effort structure of the sediment sample. Sediment structures
has been put into this field both theoretically and have been quantified by using X-ray computed
experimentally, to determine the physical properties tomography to obtain values of density with a
of the marine sediments. Experimental and forward/ millimeter resolution for the full three-dimensional
inverse modeling techniques indicate that the volume. The image shown in Figure 1 is a false color
acoustic wave field in the water column and seismo- 3D reconstruction of a core sample at a site where
acoustic wave field in the seafloor can be utilized for sediments consist of sandy silt (75% sand, 15% silt,
remote sensing of the physical characteristics of the and 10% clay) and shell pieces.
marine sediments. The results of the X-ray tomography of the same
core, can also be shown on an X-ray cross-section
Sediment Structure as an Acoustic medium
slice along the center of the core (Figure 2A) and the
Much of the floor of the oceans is covered with a corresponding two-dimensional spectral density
mixture of particles of sediments range in size from levels for that cross-section (Figure 2B). The complex

380
 ACOUSTICS IN MARINE SEDIMENTS 381

300

250

200

150

100
110
50 100
90
80
350 70
300 60
250 50
200 40
150 30
100 20
50 10

Figure 1 Three-dimensional reconstruction of a sediment core sample containing 75% sand, 15% silt and 10% clay. Scale in mm.

2
Spectral density level dB re m
CT number _ 100
2250 _ 55
X-ray CT Section _ 80
_ 20
_ 60 _ 60
Vertical frequency (m 1)

0 2200 _ 40
_
Vertical distance (mm)

_ 65
_ 20
20
_ 70
0
40 2150
20
_ 75
60 40
_ 80
2100 60
80
80 _ 85
100
100
20 40 60 80 100 120 140 160 _ 100 _ 80 _ 60 _ 40 _ 20 0 20 40 60 80 100
2050 _
(A) Horizontal distance (mm) (B) Horizontal frequency (m 1)

Figure 2 (A) X-ray cross-section slice along the center of the core shown in Figure 1. (B) 2-D spectral density levels for the cross-
section shown in (A).

structure of the sediments can be with seen strong sediments subsequently covered the shelves as the sea
heterogeneity (local density fluctuation) of the me- level are in postglacial times.
dium that controls the interaction of the seismo-
acoustic energy. In addition to the complex fine
structure described above, the seafloor can also show
Biot–Stoll Model
complex layering (Figure 3A) or a simpler structure Various theories have been developed to describe the
(Figure 3B). These structures result from the lowering geoacoustic response of marine sediments. The most
of sea levels during the glaciation of the Pleistocene comprehensive theory is based on the Biot model as
epoch during which sand was deposited over wide elaborated by Stoll. This model takes into account
areas of the continental shelves. Unconsolidated various loss mechanisms that affect the response of
382 ACOUSTICS IN MARINE SEDIMENTS

N 42 56.536 19-JUL-1999 Time UTC N 42 48.615

E10 13.836 E10 13.899
18:30 18:20 18:10 18:00 17:50 17:40 17:30 17:20 17:10 17:00

0.100 75

0.120 90

)
_1
Depth (m @ 1500 m s
0.140 105
Twitt (s)

0.160 120

0.180 135

0.200 150

1 2 3 4 5 6 7 8 9 10 11 12 13 14
(A) Range (km)

N 42 36.054 03-JUL-1999 Time UTC N 42 34.525

E10 47.726 E10 52.215
13:50 13:40 13:30 13:20 13:10 13:00 12:50
0.140 105

0.160 120

0.180 135
Depth (m)
Twtt (s)

0.200 150

0.220 165

0.240 180

6.50 6.00 5.50 5.00 4.50 4.00 3.50 3.00 2.50 2.00 1.50 1.00 0.50 0
(B) Range (km)

Figure 3 Seismic reflection profiles showing (A) complex and (B) simple structures of sediment layers.
 ACOUSTICS IN MARINE SEDIMENTS 383

porous sediments that are saturated with fluid. The parameters such as those shown in Table 1 are
Biot–Stoll theory shows that acoustic wave velocity available.
and attenuation in porous, fluid-saturated sediments
depend on a number of parameters including por-
osity, mean grain size, permeability, and the prop- Seismo-acoustic Waves in the Vicinity
erties of the skeletal frame. of the Water–Sediment Interface
According to the Biot–Stoll theory, in an un- As mentioned above, when acoustic energy interacts
bounded, fluid-saturated porous medium, there are with the seafloor, the energy creates two basic types
three types of body wave. Two of these are dilata- of deformation: translational (compressional) and
tional (compressional) and one is rotational (shear). rotational (shear). Solution of the equations of the
One of the compressional waves (‘first kind’) and the wave motion shows that each of these types of de-
shear wave are similar to body waves in an elastic formation travels outward from the source with its
medium. In a compressional wave of the ‘first kind’, own velocity. Wave type, velocity, and propagation
the skeletal frame and the sea water filling the pore direction vary in accordance with the physical
space move nearly in phase so that the attenuation properties and dimensions of the medium.
due to viscous losses becomes relatively small. In The ability of seafloor sediments to support the
contrast, due to out-of-phase movement of the frame seismo-acoustic energy depends on the elastic prop-
and the pore fluid, the compressional wave of the erties of the sediment, mainly the bulk modulus (in-
‘second kind’ becomes highly attenuated. The Biot compressibility, K) and the shear modulus (rigidity,
theory and its extensions by Stoll have been used by G). These parameters are related to the compres-
many researchers for detailed description of the sional and shear velocities CP and CS respectively, by:
acoustic wave–sediment interaction when basic input
CP ¼ ½ðK þ 4G=3Þr1=2
Table 1 Basic input parameters to Biot–Stoll model

Frequency-independent Variables CS ¼ ðG=rÞ1=2

Porosity (%) P
Mass density of grains rr where, r is the bulk density. Table 2 shows basic
Mass density of pore fluid rf seismic–acoustic wave types and their velocities re-
Bulk modulus of sediment grains Kr
lated to elastic parameters
Bulk modulus of pore fluid Kf
Variables affecting global fluid motion These two types of deformations (compressional
Permeability k and shear) belong to a group of waves (body waves)
Viscosity of pore fluid Z that propagate in an unbounded homogeneous me-
Pore-size parameter a dium. However, in nature the seafloor is bounded
Structure factor a
and stratified with layers of different physical prop-
Variables controlling frequency-dependent response of frame
Shear modulus of skeletal frame m̄ ¼ mr ðoÞ þ imi ðoÞ erties. Under these conditions, propagating energy
Bulk modulus of skeletal frame K̄b ¼ Kbr ðoÞ þ iKbi ðoÞ undergoes characteristic conversions every time it
interacts with an interface: propagation velocity,

Table 2 Basic seismo-acoustic wave types and elastic parameters

Basic wave type Wave velocity

Body wave Compressional CP ¼ ½ðK þ 4G=3Þ=r1=2

Shear CS ¼ ðG=rÞ1=2
Ducted wave Love CL ¼ ðG=rÞ1=2
Surface wave Scholte CSCH ¼ ðG=rÞ1=2
Elastic parameters in terms of wave velocities (C) and bulk density (q)
Bulk modulus (incompressibility) K ¼ rðCP2  4CS2 =3Þ
Compressibility b ¼ 1=K
Young’s modulus E ¼ 2CS2 rð1 þ sÞ
Poisson’s ratio (transv./long. strain) s ¼ ð3E  rCP2 Þ=ð3E þ 2rCP2 Þ
Shear modulus (rigidity) G ¼ rCS2
Lame’s constant l ¼ rðCP2  2CS2 Þ
384 ACOUSTICS IN MARINE SEDIMENTS

energy content, and spectral structure and propa- Shear waves Shear waves are those in which the
gation direction changes. In addition, other types of particle motion is perpendicular to the direction of
waves, i.e., ducted waves and surface waves, may be propagation. These waves can be generated at a
generated. These basic types of waves and their layer interface by the incidence of compressional
characteristics together with their arrival structure as waves at other than normal incidence. Shear wave
synthetic seismograms for orthogonal directions are energy is polarized in the vertical or horizontal
illustrated in Figure 4. planes, resulting in vertically polarized shear waves
(SV) and horizontally polarized shear waves (SH).
However, if the interfaces are close (relative to a
Body Waves wavelength), one cannot distinguish between body
and surface waves.
These waves propagate within the body of the ma-
terial, as opposed to surface waves. External forces Ducted Waves
can distort solids in two different ways. The first
involves the compression of the material without Love waves Love waves are seismic surface waves
changing its shape; the second implies a change in associated with layering; they are characterized by
shape without changing its volume (distortion). horizontal motion perpendicular to the direction of
From earthquake seismology, these compressional propagation (SH wave). These waves can be
and distortion waves are called primus (P) and considered as ducted shear waves traveling within
secoundus (S), respectively, for their arrival sequence the duct of the upper sedimentary layer where total
on earthquake records. However, the distortional reflection occurs at the boundaries; thus the waves
waves are very often called shear. represent energy traveling by multiple reflections.

Interface Waves

Compressional waves Compressional waves Seismic interface waves travel along or near an
involve compression of the material in such a interface. The existence of these waves demands the
manner that the particles move in the direction of combined action of compressional and shear waves.
propagation. Thus at least one of the media must be solid, whereas

Basic wave types Propagation Particle Synthetic

direction motion seismogram
z
y x y z

A. Body waves x

Compressional

Shear Sv
Sh

B. Ducted waves

Love

C. Surface waves

Scholte

Figure 4 Basic seismo-acoustic waves in the vicinity of a water–sediment interface.

 ACOUSTICS IN MARINE SEDIMENTS 385

the other may be a solid (Stoneley wave), a liquid received by a hydrophone and three orthogonal
(Scholte wave), or a vacuum (Rayleigh wave). For geophones placed on the seafloor, at a distance of 1.5
two homogeneous half-spaces, interface waves are km from the source. The broadband signal of a few
characterized by elliptical particle motion confined milliseconds duration generated by the explosive
to the radial/vertical plane and by a constant velocity source is dispersed over nearly 18 s demonstrating
that is always smaller than the shear wave. the arrival structure of different types of waves. The
When different types of seismic waves propagate characteristics of these waves are indicated in the
and interact with the layered sediments they are figure for four different sensors. They can be identi-
partly converted into each other and their coupling fied in order of their arrival time as: (a) head wave;
may create mixed wave types in the vicinity of the (b) water arrival (compressional wave); (c) interface
interface. Figure 4 shows the basic seismo-acoustic wave; (d) Love waves.
waves in the vicinity of a water/sediment interface.
Basic characteristics of these waves together with
their particle motion and synthetic seismograms at Seafloor Roughness
three orthogonal directions (x, y and z) are also il- The roughness of the water–sediment interface and
lustrated in the same Figure. Under realistic con- layers below is another important parameter that
ditions, in which the seafloor cannot be considered to needs to be considered in sediment acoustics. The
be homogeneous, isotropic, nor a half-space, some of seafloor contains a wide spectrum of topographic
these waves become highly attenuated or travel roughness, from features of the order of tens of
together, making identification very difficult. In fact, kilometers, to those of the order of millimeters. The
the interface waves shown in Figure 4 are for a shape of the seafloor and its scattering effect on
layered seafloor, where the dispersion of the signal is acoustic signals is be covered here.
evident (homogeneous half-space would not give any
dispersion).
Under realistic conditions, i.e., for an inhomo- Techniques to Measure Geoacoustic
geneous, bounded and anisotropic seafloor, some of Parameters of Marine Sediments
these waves convert from one to another. The dif-
ferent wave types may travel with different speeds or The geoacoustic properties of the seafloor defined by
together, and they generally have different attenu- the compressional and shear wave velocities, their
ation. As an example, Figure 5 shows signals from an attenuation, together with the knowledge of the
explosive source (0.5 kg trinitrotoluene (TNT)) material bulk density, and their variation as a func-
tion of depth, are the main parameters needed to
solve the acoustic wave equation. To be able to de-
termine these properties of the seabed, different
Hydrophone
techniques have been developed using samples taken
from the seafloor, instruments and divers conducting
measurements in situ, and remote techniques meas-
uring seismo-acoustic waves and inverting this in-
b c formation with realistic models into sediment
a Vertical
properties. Some of the current methods of obtaining
geoacoustic parameters of the marine sediments are
briefly described here.
d Transversal
Laboratory Measurements on Sediment Core
Samples
c
Most of our knowledge of the physical properties of
Radial
sediments is acquired through core sampling. A large
number of measurements on marine sediments have
been made in the past. In undisturbed sediment core
0 2 4 6 8 10 12 14 16 18 20 22 samples, under laboratory conditions, density and
Time (s) compressional velocity can be measured with accur-
acy, and having measured values of density and
Figure 5 Signals from an explosive source of 0.5 kg
trinitrotoluene received by a hydrophone and three orthogonal
compressional velocity, the bulk modulus can be se-
geophones at 1.5 km distance (a, Head wave; b, Compressional lected as the third parameter, where it is can be cal-
wave; c, Interface wave; d, Love wave). culated (Table 1).
386 ACOUSTICS IN MARINE SEDIMENTS

There are several laboratory techniques available than used here and shown to have a strong linear
to measure some of the sediment properties. How- correlation. Theoretically this linearity only exists if
ever, the reliability of such measurements can be the dry densities of the mineral particles are the
degraded by sample disturbance and temperature same for all marine sediments. The density of the
and pressure changes. In particular, the acoustic sediment would then be the same as the density of
properties are highly affected by the deterioration of the solid material at zero porosity, and the same as
the chemical and mechanical bindings caused by the the density of the water at 100% porosity. Figure 6
differences in temperature and pressure between the shows this relationship.
sampling and the laboratory measurements. Con-
trolling the relationships between various physical Porosity and relative compressional wave
parameters can be used to check the accuracy of velocity The relationship between porosity and
measured parameters of sediment properties. It has compressional wave velocity has received much
been shown that the density and porosity of sedi- attention in the literature because porosity can
ments have a relationship with compressional vel- be measured easily and accurately. Data from
ocity, and different empirical equations between the SACLANTCEN sediment cores giving the
them have been established. relationship between porosity and compressional
Over the past three decades, at the NATO wave velocity are shown in Figure 7. As shown in
Undersea Research Centre (SACLANTCEN), a large Figure 6 due to the linear relation between density
number of laboratory and in situ measurements have and porosity, the relationship between density and
been made of the physical properties of the seafloor compressional wave velocity is similar to the
sediments. These measurements have been conducted porosity compressional wave velocity relation.
on the samples with the same techniques as when the
same laboratory methods were applied. This data set
In situ Techniques
with a great consistency of the hardware and meas-
urement technique, has been used to demonstrate the There are several in situ techniques available that use
physical characteristics of the sediment that affect instruments lowered on to the seafloor mainly by
acoustic waves. means of submersibles, remotely operated vehicles
(ROVs), and autonomous underwater vehicles AUVs
The relationship between measured physical and divers. The first deep-water, in situ measure-
parameters From 300 available cores, 20 000 ments of sediment properties were made from the
measured data samples for the density and porosity, bathyscaph Trieste in 1962. These measurements
and 10 000 samples for the compressional velocity provided accurate results due to the minimum dis-
were obtained. To be able to handle this large data set turbance of temperature and pressure changes com-
taken from different oceans at different water depths, pared to bringing the sample to the surface and for
all bulk density and compressional velocity data were
converted into relative density ðrÞ and relative 2.6
velocity ðCÞ with respect to the in situ water values:  = 2.635_ 0.01675 P
Number of data points = 13885
2.4
r ¼ rS =rW
Relative density ( = s /w)

C ¼ CS =CW 2.2

where, rS is sediment bulk density, rW is water 2.0

density, CS is sediment compressional velocity and CW
is water compressional velocity. The data are not only 1.8
from the water–sediment interface but cover
sedimentary layers of up to 10 m deep. 1.6

Relative density and porosity The density and 1.4

porosity of the marine sediment are least affected
1.2
during coring and laboratory handling of the
samples. Porosity is given by the percentage volume
of the porous space and sediment bulk density by 1.0
0 20 40 60 80 100
the weight of the sample per unit volume. The Porosity (%)
relationship between porosity and bulk density has
been investigated by many authors with fewer data Figure 6 Relationship between relative density and porosity.
 ACOUSTICS IN MARINE SEDIMENTS 387

Figure 9 Examples of received compressional wave signals

from fine sand sediment.

Figure 7 Relationship between relative compressional wave

velocity and porosity.

Figure 10 Examples of signals recorded from two shear wave

receivers in hard-packed fine sand sediments.

Figure 8 In Situ Sediment Acoustic Measurement System

(ISSAMS) for near-surface in situ geoacoustic measurements.
Using the compressional and shear wave transdu-
cers measurements are made with a continuous wave
analysis in the laboratory. The most reliable direct (cw) pulse technique where the ratio of measured
geoacoustic measurement techniques for marine transducer separation and pulse arrival time yields
sediments are in situ techniques. Some of the ap- the wave velocity. Samples of compressional and
proaches used in recent years are discussed below. shear wave data are shown in Figure 9 and 10 re-
spectively. Table 3 gives a comparison of laboratory
Near-surface method A system has been and in situ values of compressional and shear wave
developed to measure sediment geoacoustic velocities from two different types of Adriatic Sea
parameters, including compressional and shear sediment.
wave velocities and their attenuation at tens of
centimeters below the sediment–water interface. Cross-hole method Measurements as a function of
Figure 8 shows the main features of the In situ depth in sediments can be made with boreholes,
Sediment Acoustic Measurement System (ISSAMS). using either single or cross-hole techniques.
Shear and compressional wave probes are attached Boreholes are made by divers using water–air jets to
to a triangular frame that uses weights to force the penetrate thin-walled plastic tubes for cross-hole
probes into the sediment. In very shallow water, measurements. Figure 11 shows the experimental
divers can be used to insert the probes into the set up for the cross-hole measurements. The source
sediment, whereas in deeper water, a sleeve system is in the form of an electromagnetic mallet securely
(not shown) allows the ISSAMS to penetrate into coupled to the inner wall of one of the plastic tubes
the seafloor. with a hydraulic clamping device.
388 ACOUSTICS IN MARINE SEDIMENTS

Table 3 Comparison of laboratory and in situ measurements

Wave velocity (ms1)

Sediment Porosity Mean grain In situ Laboratory In situ Laboratory

type (%) size (f) (CP) (CP) (CS) (CS)

Sand 37 3.5 1557–1568 1580–1604 78–82 50

Mud 68 8.6 1467–1488 1468–1487 27–31 15

Projector Pump
Clamp Receiver
S Geophones

Clamp
Electromagnetic
mallet

Figure 11 The experimental setup for cross-hole measurements.

0
8
Relative velocity amplitude (V)

4
Spectrum level (dB)

Time _ 20

0
_ 40
_4

0.1 s
_8 _ 60
5 10 50 100
(A) (B) Frequency (Hz)

Figure 12 Cross-hole shear wave signal (A) and its spectrum (B) for a silty-clay bottom in the Ligurian Sea.

With a separation range of 2.9 m, moving-coil analyzed for velocity and attenuation parameters.
geophone receivers are also coupled to the inner wall Examples of a time series and a frequency spectrum
of the second plastic tube with a hydraulic clamping for a shear wave signal received by a geophone
device. The electromagnetic mallet generates a with a natural frequency of 4.5 Hz are shown in
point source on the thin plastic tube wall, which Figure 12.
results in a multipolarized transient signal to the Figure 13 shows shear wave velocity as a function
sediment. Depending on the orientation of receiving of depth obtained in the Ligurian Sea. It can be seen
sensors, vector components of the propagating that the shear wave velocities are around 60 m1 s at
compressional and shear waves are received and the sediment interface and increase with depth.
 ACOUSTICS IN MARINE SEDIMENTS 389

Remote Sensing Techniques signal received by sensors on the seafloor and/or in

the water column. Figure 14 illustrates a character-
Even though in situ techniques provide the most re-
istic shallow water signal from an explosion received
liable data, they are usually more time consuming
by a hydrophone close to the sea bottom. Three
and expensive to make and they are limited to small
different techniques to extract information relative
areas. Remote sensing and inversion to obtain
to bottom parameters from these signals are de-
geoacoustic parameters can cover larger areas in less
scribed briefly below.
time and provide reliable information. These tech-
niques are based on the use of a seismo-acoustic
Reflected waves
The half-space seafloor. When the seafloor
0 consists of soft unconsolidated sediments, due to its
very low shear modulus it can be treated as fluid.

2
Depth (m)

w
Cw
3  r
Water
Sediment s
t
Cs

5
0 50 75 100 125
_
Shear wave velocity (m s 1)

Figure 13 Shear wave velocity profile obtained from cross-hole Figure 15 Geometry and notations for a simple half-space
measurements. water–sediment interface.

Refracted wave Interface wave

Waterborne wave

0.2 0.4 0.6 0.8 1.0 1.2 1.4

Time (ms)

Figure 14 A characteristic shallow-water signal.

390 ACOUSTICS IN MARINE SEDIMENTS

Reflection may occur whenever an acoustic wave is independent reflection loss as the ratio between the
incident on an interface between two media. The amplitude of the shock pulse and the peak of the
amount of energy reflected, and its phase relative to first reflection (after correcting for phase shift,
the incident wave, depend on the ratio between the absorption in the water column, and differences in
physical properties on the opposite sides of the spreading loss). If the relative density r ¼ rS =rW,
interface. It is possible to calculate frequency and the relative compressional wave velocity

0° 0°

20° 20°
angle

Grazing angle
40° 40°
Grazing

60° 60°

80° 80°

Time (ms) Time (ms)

0 12 24 0 12 24

0° Grazing angle 90° 0° Grazing angle 90°

10

20

dB Loss
Relative POR.(%) Relative POR.(%)
sound speed 60 100 sound speed 60 100

1.00 1.05 1.1 REL. DENSITY 1.9 1.00 1.05 1.1 REL. DENSITY 1.9
CLAY
SAND

1 1
CLAY

2 2

SAND
3 3

4
4m

5m
(A) (B)

Figure 16 Acoustic signals and reflection loss as a function of grazing angle and sediment core properties: (A) for critical angle; (B)
for angle of incidence case.
 ACOUSTICS IN MARINE SEDIMENTS 391

C ¼ CS =CW , are used to present the contrast However, the phase of the reflected wave is then
between the two media, reflection coefficient for shifted relative to the phase of the incidence wave
such a simple environmental condition can be by an angle varying from 01 to 1801 and is given as:
written as: p
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ðcos2 Y  1=C2 Þ
F ¼ 2 arctan
rsinY  ð1=C2  cos2 YÞ rsinY
R¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
rsinY  ð1=C2  cos2 YÞ Figure 16 shows measured and calculated reflection
Where Y is the grazing angle with respect to the losses (20 log R) for this simple condition together
interface as shown in Figure 15. For an incident with the basic physical properties of the core sample
path normal to a reflecting horizon, i.e., Y ¼ 901, taken in the same area for explosive signals at
the reflection coefficient is: different grazing angles.

R ¼ rC  1=rC þ 1 Angle of intromission case. Especially in deep-

water sediments the sound velocity in the top layer
of the bottom is generally less than in the water
Critical angle case. When the velocity of above (Co1). In such conditions there is an angle of
compressional wave velocity is greater in the incidence at which all of the incident energy is
sediment layer (C>1), as the grazing angle is transmitted into the sedimentary layer and the
decreased, a unique value is reached at which the reflection coefficient becomes zero:
acoustic energy totally reflects back to the water
column. This is known as the critical angle and is
p
r 2
 1=C2

given by: cosYi ¼
r2  1

Ycr ¼ arccosð1=CÞ The phase shift is 01 when the ray angle is greater than
the intromission angle and 1801 when it is smaller.
When the grazing angle is less than this critical Thus, acoustical characteristics of the bottom, such as
angle, all the incident acoustic energy is reflected. the critical angle, the angle of incidence, the phase shift,

Direct signal
Reflected signal

D(t ) R(t )

F F

D(f ) R(f )
R(f )
D(f )

H (f ) _1 Impulse
Transfer func. F response
0 20 40
i(f ) Time (ms)
A (f ) e

Reflection +180°
coefficient Phase shift
(f ) 0°
A (f ) _ 180°
0
Reflection loss 10
_ 20 log A dB
20
30 dB
0 1 2 3 4 5
kHz

Figure 17 Acoustic reflection data and analysis technique for a layered seafloor.
392 ACOUSTICS IN MARINE SEDIMENTS

Grazing

Loss
angle

dB
Computed Measured
0

36° 10

20

10

54° 20
30
10
72° 20

30

0 2 4 0 2 4
kHz kHz

Impulse responses

Computed Measured

18°

36°

54°
72°

90°

0 10 0 10
Time (ms) Time (ms)
For layered bottom

Figure 18 Measured and calculated reflection losses for a layered seafloor.

and the reflection coefficient, are primarily influenced

by the relative density and relative sound velocity of the
environment as a function of the ray angle.

The layered seafloor. Since the seafloor is generally

layered, a simple peak amplitude approximation
cannot be implemented because of the frequency
dependence. In this case one calculates the transfer
function (or reflection coefficient) from the convolution ray
one ar
of the direct reference signal with the reflected signal. Geoph Explosive
source
Examples of the phase shift and reflection loss as a
function of frequency are shown in Figure 17. The
reflectivity can also be described in the time domain by
the impulse response, which is the inverse Fourier Figure 19 Experimental setup to measure seismo-acoustic
waves on the seafloor.
transform of the transfer function as shown in the same
figure. This type of data can be utilized for inverse
modeling to obtain the unknown parameters of the
sediments. Figure 18 is an example of measured and Refracted waves Techniques developed for remote
calculated reflection losses and impulse responses for a sensing of the uppermost sediments (25–50 m below
layered seafloor. the seafloor) utilize broad-band sources (small
 ACOUSTICS IN MARINE SEDIMENTS 393

Offset (m)
0
25 50 75 100 125 150 175 200 225 250
0

500

1000

1500
Time (ms)

2000

2500

3000

3500

4000
6.81

8.71
8.21

10.00
8.94

10.26
0.68
0.37

6.96
0.23
0.06

0.57

0.77

3.44
4.88

3.64
7.26

7.54

8.95
8.83
3.75
0.47
3.52

3.86

8.59
0.19

0.27

7.40

8.84
3.92

8.08

8.46
5.02

4.24

8.39
3.76

8.50
7.82

9.07
6.49

7.95
7.68

4.87
3.09

7.22

8.40
Figure 20 Signals received by geophone array.

explosives) and an array of geophones deployed on The slope of the travel-time curve (fitted parabola)
the seafloor. To obtain estimates of the bottom gives the rate of change of the range with respect to
properties as a function of depth, both refracted time that is also the velocity of propagation of the
compressional and shear waves as well as interface diving compressional wave at the level of its deepest
waves are analyzed. Inversion of the data is carried penetration (turning point) into the sediment. At
out using modified versions of techniques developed each range D, the depth corresponding to the deepest
by earthquake seismologists to study dispersed penetration is then calculated using the following
Rayleigh-waves and refracted waves. Figure 19 integral
illustrates the basic experimental setup and Figure 20
Z D
the signals received by an array of 24 geophones that
permit studies of both interface and refracted waves. zðVÞ ¼ 1=p cosh1 ðVðdt=dxÞÞdx
0
These data are analyzed and inverted to obtain
both compressional and shear wave velocities as a where
function of depth in the seafloor. Studies of attenu-
ation and lateral variability are also possible using 1=V ¼ ðdt=dxÞx ¼ D
the same data set.
Figure 21 shows the expanded early portion of the
geophone data shown in Figure 20, where, the first
arriving energy out to a range of about 250 m has Offset (m)
been fitted with a curve showing that compressional 0 80 160 240 320
waves refracted through the sediments just beneath 15
05
the seafloor travel faster as they penetrate more m
50 s _
1
deeply into the sediment. At zero offset, the slope
indicates a velocity of 1505 m s1 whereas at a range 157
Time (ms)

100 3m
offset at 250 m the slope corresponds to a velocity of s _1
1573 m s1. At ranges over 250 m, a strong head
150 4577 m _1
wave becomes the first arrival and, the interpretation s
would be that there is an underlying rock layer with
200
compressional wave velocity of about 4577 m s1.
A compressional wave velocity–depth curve for
250
the upper part of the seafloor can be derived from the
first arrivals shown in Figure 21 using the classical Figure 21 Expanded early portion of the data shown in
Herglotz–Bateman–Wieckert integration method. Figure 20.
394 ACOUSTICS IN MARINE SEDIMENTS

The result is the solid velocity–depth curve shown in _

Compressional wave velocity (m s 1)
Figure 22. 1500 1600
0

Interface waves In order to obtain a shear wave

velocity–depth profile from the data, later arrivals
corresponding to dispersed interface waves may be
utilized (Figure 20). The portion of each individual
signal corresponding to the interface-wave arrival
can be processed using multiple filter analysis to
create a group velocity dispersion diagram (Gabor Data
diagram). The result of applying this technique to a
dispersed signal is a filtered time signal whose

Depth (m)
envelope reaches a maximum at the group velocity 10
arrival time for a selected frequency. The envelope is
computed by taking the quadrature components of
the inverse Fourier transform of the filtered signal.
Filtering is carried out at many discrete frequencies
over selected frequency bands. Once the arrival
times are converted in to velocity, the envelopes are
arranged in a matrix and contoured and dispersion
curves are obtained by connecting the maximum
values of the contour diagram (Figure 23).
Having obtained the dispersion characteristics of 20
the interface waves, the geoacoustic model, made of Predicted
a stack of homogeneous layers with different com-
pressional and shear wave velocities for each layer
that predict the measured dispersion curve is deter-
mined. Figure 24 illustrates a number of examples
from the Mediterranean sea covering data from soft Figure 22 Compressional wave velocity versus depth curves
clays to hard sands. derived from data and predicted (dashed line) by the model.

120

100
Velocity (m s 1)
_

80

60

40
2 6 10 14 2 6 10
Frequency (Hz) Frequency (Hz)

Figure 23 Dispersion diagram and measured and predicted dispersion curves.

 ACOUSTICS IN MARINE SEDIMENTS 395

Depth (m) 10

20

Mediterranean Sea
West Atlantic
30
0 200 400 600
_
Velocity (m s 1)

Figure 24 Summary of shear wave velocity profiles from the Mediterranean Sea.

60
SD: 50 m Water
50
RD: 50 m
Compressional
70 Shear
Depth (m)
TL (dB)

60
Data Sand

Calculate
d 80
70 Predicted
Sand model
Clay Clay model
80 90
0 2 4 6 8 10 12 0 500 1000 1500 2000 2500
_
Range (km) Velocity (m s 1)

Figure 25 Comparison of transmission loss (TL) data and SAFARI prediction. Effects of changing bottom parameters from sand to
clay is also shown together with the input profiles utilized for SAFARI predictions.

Transmission Loss Technique amount of energy lost along an acoustic propagation

path, carries the information relative to the en-
The seafloor is known to be the controlling factor in
vironment through which the wave is propagating.
low-frequency shallow water acoustic propagation.
Figure 25 shows a comparison of TL data and model
Forward modeling is performed with models giving
predictions together with the input parameters used
exact solutions to the wave equation, i.e., SAFARI
at 400 Hz. The effects of changes in bottom par-
where, compressional and shear wave velocities, the
ameters are also shown in the figure. This technique
attenuation factors associated with these waves, and
becomes extremely useful when seafloor information
the sediment density as a function of depth are the
is sparse.
main input parameters. Acoustic energy propagating
through a shallow water channel interacts with the
seafloor causing partitioning of waterborne energy
into different types of seismic and acoustic waves.
Conclusions
The propagation and attenuation of these waves Acoustic/seismic characteristics of the marine sedi-
observed in such an environment are strongly ments have been of interest to a wide range of ac-
dependent on the physical characteristics of the sea tivities covering commercial operations involving
bottom. Transmission loss (TL), representing the trenching and cable lying, construction of offshore
396 ACOUSTICS IN MARINE SEDIMENTS

foundations, studies of slope stability, dredging and Brekhovskikh LM (1980) Waves in Layered Media. New
military applications like mine and submarine de- York: Academic Press.
tection. Experimental and theoretical work over Cagniard L (1962) Reflection and Refraction of
the years has shown that it is possible to determine the Progressive Seismic Waves. New York: McGraw-Hill.
geoacoustic properties of sediments by different tech- Grant FS and West GF (1965) Interpretation and Theory in
Applied Geophysics. New York: McGraw Hill.
niques. The characteristics of the marine sediments
Hampton L (ed.) (1974) Physics of Sound in Marine
and techniques to obtain information about these Sediments. New York and London: Plenum Press.
characteristics have been briefly described. The studies Hovem JM, Richardson MD, and Stoll RD (eds.) (1992)
so far conducted indicate that direct and indirect Shear Waves in Marine Sediments. Dordrecht: Kluwer
methods developed over the last four decades may give Academic.
sufficient information to deduce some of the funda- Jensen FB, Kuperman WA, Porter MB, and Schmidt H
mental characteristics of the marine sediments. It is (1999) Computational Ocean Acoustics. New York:
evident that still more research needs to be done to Springer-Verlag.
develop these techniques for fast and reliable results. Kuperman WA and Jensen FB (eds.) (1980) Bottom
Interacting Ocean Acoustics. New York and London:
Plenum Press.
Lara-Saenz A, Ranz-Guerra C and Carbo-Fite C (eds)
See also
(1987) Acoustics and Ocean Bottom. II F.A.S.E.
Acoustics, Shallow Water. Specialized Conference. Inst. De Acustica, Madrid.
Pace NG (ed.) (1983) Acoustics and the Sea-Bed. Bath:
Bath University Press.
Further Reading Pouliquen E, Lyons AP, Pace NG et al. (2000)
Backscattering from unconsolidated sediments above
Akal T and Berkson JM (eds.) (1986) Ocean Seismo- 100 kHz. In: Chevret P and Zakhario ME (ed.)
Acoustics. New York and London: Plenum press. Proceedings of the fifth European Conference on Under
Biot MA (1962) Generalized theory of acoustic water Acoustics, ECUA 2000 Lyon, France.
propagation in porous dissipative media. Journal of Stoll RD (1989) Lecture Notes in Earth Sciences. New
Acoustical Society of America 34: 1254--1264. York: Springer-Verlag.
 ACOUSTICS, ARCTIC
P. N. Mikhalevsky, Science Applications and acoustic remote sensing (see Tomography). The
International Corporation, McLean, VA, USA Arctic Ocean is the world’s ‘air-conditioner’, main-
Copyright & 2001 Elsevier Ltd. taining the surface heat balance, and it provides fresh
water to the world’s oceans, principally in the form
of sea ice discharged from the Fram Strait. The latter
regulates convective overturning in the Greenland
and Norwegian Seas that in turn drives the global
thermohaline circulation with significant impact on
Introduction climate. Monitoring changes in the temperature and
The Arctic Ocean is an isolated mediterranean basin stratification of the Arctic Ocean and sea ice thick-
with only limited communication with the world’s ness using acoustics is an important capability that
oceans, principally the Atlantic Ocean via the Fram will improve our understanding of the Arctic Ocean
Strait and the Barents Sea, and the Pacific Ocean via and its role in global climate change.
the Bering Strait. The ubiquitous feature of the Arctic
Ocean is the sea ice that covers the entire Arctic
History
basin during the winter months and only retreats off
the shallow water shelf areas in the summer months, The first large program dedicated to acoustics re-
creating a permanent cap over most of the central search in the Arctic Ocean was undertaken by the US
Arctic basin (Figure 1). The presence of the year- Navy Underwater Sound Laboratory, as it was then
round sea ice cover determines the unique character of known, in 1958 in connection with the International
acoustic propagation and ambient noise in the Arctic Geophysical Year. Also in that year the US nuclear
Ocean. The sea ice nsulates the Arctic Ocean from submarine Nautilus (SSN 571) made its historic
solar heating in the summer months, creating a year- voyage to the North Pole, marking the beginning of
round upward refracting sound speed profile with the regular submarine operations in the Arctic Ocean.
sound speed minimum at the water–ice interface. From 1958 to 1975 the USA conducted much of its
Sound, therefore, is refracted upward and is con- acoustics research from manned ice islands (thick
tinuously reflected from the ice as it propagates, tabular sections of land-fast ice that occasionally
causing attenuation by scattering, mode conversion, break away from the Ellesmere ice shelf) that re-
and absorption that increases rapidly with frequency. mained within the polar pack ice (see Cryosphere:
The lack of solar forcing and the Arctic Ocean’s re- Sea Ice). The most famous of these was Fletcher’s Ice
stricted communication with the other oceans of the Island, also known as T3, which was discovered in
world creates a very stable acoustic channel with June 1950. T3, originally 14.5 km long, 6.4 km wide
significantly reduced fluctuations of acoustic signals in and 52 m thick, provided an ideal platform for Arctic
comparison with the temperate oceans. In contrast to research including acoustics. The first experiments in
the central basin, acoustic propagation on the Arctic 1958 were aimed at investigating the feasibility of
shelves and in the marginal ice zones (MIZs, those using the RAFOS (Ranging and Fixing of Sound) for
areas between the average ice minimum and max- submarine navigation in Arctic waters. Explosive
imum) (Figure 1), is quite complex and variable owing signals were deployed from T3 and other ice stations
to the seasonal retreat of the sea ice, river run-off, and while the USS Skate (SSN 578) attempted to receive
bottom interaction (see Acoustics, Shallow Water, the signals for acoustic crossfixing. While RAFOS
Acoustics in Marine Sediments). was never operationally deployed, a significant
Over the last half-century Arctic acoustics research amount was learned about the upward refracting
and development has largely supported submarine sound speed profile, propagation, and resulting
operations. The importance of the Northern Sea scattering losses, as well as ambient noise and
Route to the Soviet Union, and the prospect of Soviet reverberation.
nuclear ballistic missile submarines exploiting the The Soviet Union operated at least two manned
unique Arctic acoustic environment to remain un- year-round ice stations in the Arctic continuously
detected provided the need for this research. Since from 1937 through 1992. In 1971 they began
the end of the Cold War and the beginning of con- POLEX (Polar Experiment), which was an intense
cern about ‘global warming’ there has been a new set of oceanographic, ice, and atmospheric meas-
focus for Arctic acoustics on acoustic thermometry urements over the entire Arctic Ocean basin that

397
398 ACOUSTICS, ARCTIC

Figure 1 Map of the Arctic Ocean showing the average minimum and maximum sea ice extent, as well as the major bathymetric
features, and other significant geographic locations.

continued for a decade. From 1971 through 1994 the quantified sea ice properties and ice scattering pro-
US Office of Naval Research sponsored or co-spon- cesses, bottom and surface reverberation, Arctic
sored many international acoustic-related research plate tectonics with seismic reflection and refraction,
science programs based from seasonal ice camps in identified the mechanisms of ice generated ambient
the Arctic, staged out of facilities in Alaska, Canada, noise, discovered the exceptional phase stability of
Greenland, and Svalbard. These included the Arctic low frequency propagation (B20–30 Hz), and
Ice Dynamics Experiments (AIDJEX), 1971–76; the demonstrated the feasibility of basin scale acoustic
FRAM and MIZEX series in the central Arctic basin thermometry in the Arctic Ocean.
and marginal ice zones respectively, 1979–87; the The US, UK, and Soviet Union (now Russia) have
Coordinated East Arctic Experiment (CEAREX), operated submarines in the Arctic Ocean and con-
1988–89; the Sea Ice Mechanics Initiative (SIMI), ducted a significant amount of acoustics research. In
1993–94; and the Transarctic Acoustic Propagation particular, the US Navy’s Submarine Ice Exercises
Experiment (TAP) in 1994. These programs (SUBICEXs) were conducted over almost three
 ACOUSTICS, ARCTIC 399

decades by the Arctic Submarine Laboratory (ASL) Arctic Ocean Sound Speed Structure
under the leadership of Dr Waldo Lyon, often in
conjunction with seasonal ice camps. Dr Lyon was The structure of the typical sound speed profile in the
one of the first to investigate propagation in polar Arctic consists of components that correspond
stratified waters, beginning shortly after the end of closely to the primary stratified water masses of the
World War II, working with Canadian researchers. Arctic Ocean (Figure 2). The uppermost layer is
From the Naval Electronics Laboratory (NEL) in San Polar Water (PW) defined by temperatures o01C and
Diego, Dr Lyon led the development of upward- salinities o34.5 ppt and extends from the surface to
looking ice profiling sonar, and the forward-looking depths of 100–200 m. There is often a mixed layer of
ice avoidance sonar, that proved critical for safe Polar Water 30–60 m thick just below the ice with a
under-ice submarine operations. SUBICEXs were nearly constant temperature near the freezing point
designed to test all aspects of submarine operations and nearly constant salinity at around 33.6 ppt.
including surfacing through ice, and weapons ef- Below the mixed layer the temperature and salinity
fectiveness (see Under Ice in Further Reading). In increase uniformly, attaining 01C and 34.5 ppt re-
1993 the US Navy supported a dedicated 60 day spectively, at the base of the Polar Water layer. Arctic
science cruise in the Arctic with the USS Pargo (SSN Intermediate Water (AIW) – also known as the
650). This started the Submarine Science Ice Ex- Atlantic Layer reflecting its origin (and sometimes
peditions (SCICEX) that were conducted each year referred to as Atlantic Intermediate Water) – is de-
from 1995 to 1999. These multidisciplinary cruises fined by temperatures 401C and salinities 434.5
have confirmed the large changes in the Arctic Ocean ppt and extends from the base of the Polar Water to a
thermohaline structure, including the warming of the depth of 1000 m. Temperature increases with depth
Arctic Intermediate Water (AIW, see below) that was through the AIW layer to a maximum at approxi-
measured acoustically during the TAP experiment in mately 200–500 m depth and then decreases with
1994. Other data and modeling have shown strong depth to 01C at about 1000 m. AIW enters the Arctic
evidence of decadal Arctic oscillations, but longer Ocean via the West Spitzbergen Current from the
observational time series are needed. A trans-basin Fram Strait and via the Barents Sea east of Franz
acoustic section was started in October 1998 from a Josef Land (see Ocean Currents: Arctic Basin Cir-
source located in the Franz Victoria Strait to a re- culation). The shallower depths (200 m) of the
ceiving array in the Lincoln Sea. This 3-year US/ temperature maximum of the AIW occur in the
Russian collaborative effort is expected to be fol- eastern Arctic and approaches 21C. This temperature
lowed by more permanent long-term monitoring of maximum deepens to 500 m the western Arctic in the
the Arctic Ocean using acoustics. Canada Basin and is approximately 0.41C. Salinity

_
Sound speed profile (m s 1)
Mode Ra y
1440 1460 1480
functions trace
0 Polar water
Mode 1
9.3˚
Atlantic
500 Intermediate
Mode 2 Water
9.6˚
Sound
Depth (m)

speed
10.0˚
1000
Mode 3 10.5˚
Temperature
Mode 4
Deep Salinity
1500 Arctic Water

12.8˚
2000
0 20 40 60 80
Range (km) 33.5 34.5 35.5 _ 2 0 2
Salinity (ppt) Temperature (˚C)

Figure 2 The Arctic sound speed profile shown on the right was computed from the measured temperature and salinity, also shown,
from an ice camp in the eastern Arctic Ocean in April 1994. A ray trace for this sound speed profile is plotted for a source at a depth of
100 m. The mode shapes computed at 20 Hz using this profile are shown on the left. The major Arctic water masses (see text) are
indicated.
400 ACOUSTICS, ARCTIC

increases with depth in the AIW layer from 34.5 ppt

to 34.9 ppt at about the same depth of the tem-
perature maximum and remains constant at 34.9 ppt
below this depth. Deep Arctic Water (DW) is a
relatively homogeneous water mass from a depth of
1000 m to the bottom, with temperatures o01C and
a nearly constant salinity of 34.9 ppt.
The resulting sound speed profile (Figure 2) for
these typical deep Arctic basin conditions has a min-
imum at the ice–water interface of 1435–1440 m s1
and increases with depth to the bottom. In the near-
surface mixed layer of the PW the sound speed gra-
dient is þ 0.016 s1, due entirely to the increase of
pressure with depth, as temperature and salinity are
Figure 3 A photograph of the pack ice in the central Arctic
constant. Below the mixed layer the sound speed in- Ocean taken at the FRAM II Ice Camp in May 1980. A weathered
creases rapidly with the increasing temperature and pressure ridge is shown on the left bounding a relatively flat multi-
salinity (temperature having the far dominant effect) year ice floe on the right that is typical of the central arctic. The
to the depth of the maximum temperature in the AIW blocks of ice in the foreground are 3–4 m high. (Photo by P.
Mikhalevsky.).
layer with gradients of þ 0.1 s1 or more, reaching a
sound speed between 1455 and 1465 m s1. Below
the depth of the AIW temperature maximum the and AIW layers. Mesoscale and internal wave dy-
sound speed continues to increase, but with reduced namics are approximately one order of magnitude
gradients of þ 0.01 s1 or less as the temperature less energetic in the Arctic Ocean than in the tem-
decreases, but the pressure increases and the salinity is perate seas and so have a small influence on the
constant. Below 1000 m in the DW both temperature sound speed profile and this results in the exceptional
and salinity are nearly constant and the sound speed signal stability at low frequencies (discussed below).
continues to increase with the þ 0.016 s1 gradient to
the bottom associated with the increasing pressure.
In general the Arctic or polar profile can be char- Propagation
acterized (and is often approximated) roughly as a Acoustic propagation in the Arctic is dominated by
bilinear upward refracting profile with the large the repeated reflection of the sound from the sea ice
positive gradient at the near surface creating a strong as illustrated in Figure 2. The ray trace shown was
near-surface duct and a smaller positive gradient computed using the sound speed profile of Figure 2.
below this ‘knee’ to the bottom. This structure is The morphology of sea ice is quite complicated in
important to understanding the resulting propa- general. Figure 3 is a photograph taken in the pack
gation effects described below. There are both re- ice at an ice camp in the eastern Arctic. It shows a
gional and temporal variations of this typical sound typical pressure ridge on the left that forms between
speed structure associated with the corresponding two ice floes when winds and currents compress the
variations in the water masses described above. The pack. Pressure ridges can extend as much as several
strong positive gradient associated with the upper tens of meters below the ice, and consist of un-
AIW layer weakens as the ‘knee’ in the sound speed consolidated blocks in newly formed ridges as well as
profile (Figure 2) deepens from 200 m to about 500 refrozen consolidated structures in older ridges. In
m as one moves from the eastern Arctic Ocean near general the sea ice in the Arctic consists of relatively
Svalbard (where the profile in Figure 2 was meas- smooth floes (the right part of Figure 3) laced with
ured) into the western Arctic Ocean and the Beaufort pressure ridges and/or open leads as the pack works
and Chukchi Seas. In the Beaufort and north Chuk- between convergent and divergent conditions. Aver-
chi Seas the influx of warmer Bering Sea water can age ice roughness and thickness measurements de-
create a sound speed maximum just below the mixed rived from upward-looking submarine sonars have
layer of the PW at depths between 50 and 80 m, historically (1958–76) been in the range of 1–5 m
creating a double duct. Because of the year-round ice rms and 2–4 m respectively.1 Recent analysis of
cover, seasonal, interannual, and recently identified
decadal time scales dominate the temporal variations
of the sound speed profile. The seasonal and inter- 1
LeSchack LA (1980) Arctic Ocean Sea-ice Statistics Derived
annual variability is confined almost entirely to the from Upward-Looking Sonar Data Recorded during Five Nuclear
PW layer. The decadal variability is seen in the PW Submarine Cruises. Technical Report to ONR under Contract
 ACOUSTICS, ARCTIC 401

_ 50 becomes much greater than the ice thickness the re-

20 Hz
_ 60 50 Hz
100 Hz
flection occurs at the ice–air interface. Submarine ice
_ 70
+200 Hz
draft statistics from the eastern Arctic over the
Nansen Basin (Figure 1) were used for Figure 4. As
Transmission loss, dB

_ 80
Figure 4 shows the Arctic acoustic waveguide is a
_ 90 low pass filter. So much so in fact that in comparison
_ 100 with propagation in the temperate oceans which
_ 110
scatter from surface waves that have a similar
roughness spectrum, at a given range, equivalent
_ 120
propagation loss in the Arctic requires transmitting
_ 130 at a frequency as much a factor of ten less.
_ 140 On the left side of Figure 2 the acoustic mode
_ 150
shapes for modes 1–4 at 20 Hz are plotted for the
10
0 1
10
2
10 10
3 profile shown. The acoustic modes can be thought of
Range, km as the interference pattern of upward and downward
going acoustic rays (Figure 2) whose turning depths
Figure 4 The transmission loss plotted as a function of range (the depth where the rays are horizontal) correspond
for 20, 50, 100, and 200 Hz. The source and receiver were with the depth of the deepest peak of the mode
located at 60 m depth. The ice was modeled as a rough elastic
plate including statistical parameters for the upper and lower side
amplitude. At 20 Hz, mode 1 consists of those paths
of the ice representative of the ice floe and ridge structure typical that are trapped in the strong near-surface duct cre-
of the central Arctic. ated by the thermocline in the transition from the
PW to AIW. The higher modes correspond to rays
with greater launch angles that turn at successively
SCICEX data (1993–97), however, has shown that
deeper depths in the Arctic Ocean. Rough surface
since the late 1970s the ice has thinned by as much as
scattering theory tells us that the loss per bounce or
40%,2 which would also imply a reduction in ice
reflection increases with grazing angle and frequency
roughness as well. The thicker ice and correspond-
(i.e., the steeper rays, or higher modes have greater
ingly greater roughness is typically found in the
loss per bounce). However, even though the lower
eastern Arctic and just north of the eastern Canadian
modes (lower grazing angles) have lower per bounce
Archipelago and Greenland, where the transpolar
loss, those that are trapped in the upper duct ex-
drift pushes the ice.
perience many more interactions with the ice, as
The acoustic energy is reflected and scattered from
clearly seen in Figure 2, and consequently their net
the rough ice and converted to both shear waves and
loss per kilometer is higher. At shorter ranges for a
compressional waves within the ice, resulting in
source and receiver in the upper duct (Figure 4) the
significant frequency-dependent attenuation loss.
propagation is dominated by the lower modes (or
Because of this, trans-basin propagation in the Arctic
equivalently the lower grazing angle rays that are
is limited to frequencies typically below 30 Hz or
trapped), but these get stripped away as the range
wavelengths exceeding about 50 m. Figure 4 shows
increases and at longer ranges (>100–200 km) the
the propagation loss plotted as a function of range
higher modes (or equivalently the higher grazing
for several frequencies. The curves in Figure 4 were
angle rays that are not trapped) dominate. Because
generated using a model that includes a rough elastic
the source and receiver are shallow (at 60 m depth)
plate representing the ice cover, with a two scale
for the model run in Figure 4, there is a less efficient
roughness spectrum that models pressure ridges and
excitation of mode 1 at 20 Hz (i.e., destructive
the smoother intervening floes.3 The model includes
interference with the surface reflected energy) than at
the statistics for both the underside and surface of
the higher frequencies. Thus less of the energy is
the ice since at lower frequencies as the wavelength
trapped in the near-surface duct. This results in lower
energy being received on the shallow receiver at
closer ranges and hence the higher initial propa-
(footnote continued) gation loss at 20 Hz. As the frequency increases the
N00014-76-C-0757/NR 307-374. Maryland, LeSchack Associ-
mode shapes tend to compress toward the surface
ates, Ltd.
2 (because the wavelength is getting smaller). For a
Rothrock DA, Yu Y and Maykut GA (1999) Thinning of the
arctic sea-ice cover. Geophysical Research Letters 26(23): 3469– shallow source more of the energy is trapped in this
3472. upper duct at higher frequencies (see discussion of
3
Kudryashov VM (1996) Calculation of the acoustic field in an frequency dispersion and Figure 6 below). Given the
arctic waveguide. Physical Acoustics. 42: 386–389. higher per bounce losses at the higher frequencies,
402 ACOUSTICS, ARCTIC

0 scattering into the ice (particularly conversion to

shear waves and the attenuation in the ice) are in-
cluded, better agreement at the low frequencies has
been achieved. In particular the ice keels and their
spacing are important in the conversion of acoustic
compressional waves into flexural modes in the ice.
Le Page and Schmidt (see Further Reading) have
Time(s)

2 3
5 shown that the intensity of acoustic scattering into
1
the flexural modes of a rough ice plate strongly de-
pends on the width of the roughness spatial spectra.
In order to achieve consistent agreement with data
the propagation loss models have become more
complex, demanding additional input data, particu-
larly about the structure of the sea ice and its con-
10 stitutive properties.
0 10 20 30 40 50 60 70 80 90 100 Bottom interaction must also be included, for
Frequency (Hz) those paths which cross major features like the
Lomonosov Ridge, and propagation on the arctic
Figure 6 A composite of various measurements of Arctic
ambient noise. (1) Central Arctic pack ice Nansen Basin, first-
shelves (Figure 1). The Lomonosov Ridge can rise to
and multi-year floes, 2–3 m thick, with ridging, April 1982. 1500 m below the surface. Figure 2 shows that
(Reproduced with permission from Dyer, 1984.) (2) Beaufort modes 5 and higher (rays 411–121) will start to
Sea, summer and fall conditions: (a) highest levels observed, (b) interact with this feature and will begin to get
typical cold weather situation following rapid temperature drop stripped away for long-range propagation. As the
and thermal-induced cracking, (c) quietest conditions during
warmer stable temperature periods with low winds September–
frequency decreases below 20 Hz the modes tend to
October, 1961 and May–September 1962. (Reproduced with expand away from the surface and finally all modes
permission from DiNapoli et al., 1978, in Von Winkle, 1984.) (3) will begin to interact with the bottom except in the
Shore-fast ice in the Canadian Archipelago, winter conditions, very deepest parts of the Arctic basin. This leads to a
thermal cracking, February 1963. (Reproduced with permission lower frequency bound at 5–10 Hz for efficient long-
from Milne AR and Ganton JH (1964) Ambient noise under Arctic-
Sea ice. Journal of the Acoustic Society of America 36(5): 855–
range propagation. However, for high source levels
863.) (4) Shore-fast ice in the Canadian Archipelago, spring at these frequencies there is significant backscatter
conditions: (a) noisiest conditions observed during diurnal reverberation from these bathymetric features that
cooling, (b) quietest conditions observed during diurnal can be detected and mapped. Excellent correlation
warming, April 1961 (Milne and Ganton 1964). (5) Chukchi with known Arctic bathymetry as well as the dis-
Sea, spring conditions, cold stable temperatures, low winds, April
1999 (Mikhalevsky, APLIS Ice Camp).
covery of uncharted features at basin scale ranges
(1000 km) has been achieved using large explosive
sources deployed from the ice. Over the shelves the
and the greater number of bounces experienced by shallow water propagation dictates bottom and sur-
these trapped modes, it can be seen why attenuation face/ice interaction at all frequencies; however, there
in the Arctic increases rapidly with frequency. For are optimal propagation frequencies similar to deep
deeper sources the near-surface duct is not as im- water that depend upon the depth and boundary
portant and higher grazing angle (higher mode) properties.
propagation and loss are dominant at all ranges; The stable upward refracting Arctic sound speed
however, the higher loss per bounce at higher fre- profile causes a very predictable modal and fre-
quencies still results in more rapid attenuation as the quency dispersion. As can be seen in Figure 2, suc-
frequency increases. cessively higher modes are propagating at higher
There is a significant body of research on the sound speeds corresponding to higher group vel-
scattering and reverberation of acoustic energy from ocities. As a consequence at a given range, mode 1
the sea ice. The earliest models relied on rough free arrives last, preceded in order by modes 2, 3, 4, etc.
surface scattering theory with empirical fits to im- This can be observed by transmitting a wideband
portant parameters such as rms ice roughness and waveform such as an impulsive or explosive source
average ridge spacing. The free surface scattering and plotting the signal intensity as a function of
models that included ridge-like morphology per- frequency and time. Figure 5 shows the arrival time
formed reasonably well at frequencies from 250 Hz from the computed group velocities of modes 1–3
to 2000 Hz. At lower frequencies they tended to for an explosive shot at a range of approximately
underpredict the loss. When elastic coupling and 580 km in the Beaufort Sea. Although not plotted,
 ACOUSTICS, ARCTIC 403

110 again in 1994 and 1999 at 20 Hz and 2700 km.

100 This implies that coherent integration of up to 1 h
with optimal gain is achievable (and has been
90 1 demonstrated).
Spectrum level dB re IPa / Hz

80
4a
3
At very high frequencies (more than tens of kHz)
2a
Knudsen Sea
reflection occurs at the ice–water interface. Attenu-
70 2b
2c
State zero ation is rapid, due not only to the severe scattering
5
60 4b from the ice, but to volumetric absorption as well.
These frequencies are used for short-range appli-
50
cations including the upward-looking ice profiling
40 sonar, forward-looking ice avoidance sonar, and
30
downward-looking depth sounders and bottom-
mapping sonar. Torpedoes also operate at these fre-
20 quencies and ice capture can be a problem, but
10 Doppler processing can distinguish moving targets
1 10 10
2 3
10 10
4
from the stationary ice (see Sonar Systems).
Frequency (Hz)

Figure 6 The arrival time of a broadband acoustic pulse plotted

as a function of frequency showing the modal dispersion with Ambient Noise
mode 1 arriving last, preceded by modes 2 and 3 respectively.
The absolute travel time can be obtained by adding 396.5 s to the The ambient noise in the Arctic is highly variable,
time shown on the vertical axis. This modeled result agreed very exhibiting some of the quietest as well as the noisiest
closely with the received data. (Adapted with permission from ocean noise conditions of all the world’s oceans. A
DiNapoli et al., 1978, in Von Winkle, 1984.)
composite of various measurements of Arctic ambi-
ent noise is shown in Figure 6. In Figure 6, Knudsen
Sea State Zero refers to the ambient noise level in the
this calculation agreed very closely with the meas- temperate oceans at sea state zero (the quietest con-
ured data. Note also from Figure 5 that as the fre- ditions) for comparison (see Acoustic Noise). Ice-
quency increases (as described above), the modes generated ambient noise is the dominant mechanism
compress towards the surface and first mode 1, then contributing to the general character of ambient
mode 2 and finally mode 3 become trapped in the noise in the Arctic Ocean from a few tenths of Hz up
slower near-surface duct, as evidenced by the longer to 10 000 Hz. Episodic noise is also present in the
travel times. This modal dispersion and arrival pat- form of seismic events, such as earthquakes along the
tern is a classic Arctic acoustic result. The data upon Mid Arctic Ridge (Figure 1), biologics (mostly mar-
which Figure 5 is based were taken in 1962; 32 years ine mammals in the marginal ice zones), and man-
later in 1994 the TAP experiment measured the same made noise from ice breakers, and seismic
arrival pattern, attesting to the very long-term stable exploration.
water mass structure in the Arctic Ocean. Obser- Ice noise is generated when the sea ice deforms,
vations of the changes in the arrival times of these fractures, and breaks in response to environmental
modes are directly related to the temperature chan- forcing such as wind, current, thermal, and internal
ges of these water masses (since sound speed in- and surface wave-induced stresses. In general, during
creases approximately 5 m/s per 11C). In particular, stable or warming temperature conditions with low
at 20 Hz mode 2 is most sensitive to temperature winds the quietest conditions obtain, particularly
changes in the AIW, as can be seen from Figure 2. A under shore-fast ice. During periods of rapid cooling
decrease in the travel time of mode 2 by 2 s was the ice-fracturing events resulting from thermal-induced
indicator of the warming in the AIW observed in the tensile stresses can lead to higher noise levels. Higher
1994 TAP experiment. In addition to the long-term noise levels also occur in the pack ice when the ice is
stability of the Arctic sound speed profile, the short- in motion due to nonthermal forcing such as high
term stability results in exceptional amplitude winds causing bumping, grinding and rubbing of the
(B1 dB rms) and phase stability (B0.01 cycles rms) ice flows. In the marginal ice zone where the ice
of acoustic signals at low frequencies even over concentrations are typically less than in the central
ranges approaching 3000 km for up to 1 h (the Arctic pack, ice concentration and surface gravity
longest continuous observations made to date). This wave-induced flexural floe failure are the primary
applies to essentially fixed terminals. This was first correlates with the ambient noise. The actual noise-
observed in 1980 at 15–30 Hz and 300 km and then generating mechanics are complicated but fall into at
404 ACOUSTICS, ARCTIC

least two general categories. In the low frequency Conclusions

range (5–100 Hz) the important noise-generating
The presence of the year-round ice cap creates the
mechanism is the unloading motion of the ice im-
upward refracting sound speed profile in the Arctic
mediately following breaking. This has been shown
Ocean with the sound speed minimum at the ice–
to have a dipole radiation characteristic. The
water interface. Reflection, scattering, mode con-
breaking process itself is important at intermediate
version, and absorption by the rough elastic sea ice
frequencies (100–2000 Hz) and has an octopole ra-
cover causes high attenuation as frequency increases,
diation characteristic likely resulting from a slip–
limiting long-range propagation to very low fre-
dislocation process.
quencies. Bathymetric effects are important near the
Noise from earthquakes and other seismic events
major ocean ridges, basin margins, and on the shelf
occur with some regularity within the Arctic Ocean,
areas where significant mode coupling can occur. The
concentrated along the Mid Arctic Ridge (Figure 1).
Arctic sound channel is very stable and predictable in
These events have been recorded on acoustic arrays
the central Arctic basins and there is a close cor-
suspended from the ice within approximately 300 km
respondence of propagating acoustic modes with
of the Mid Arctic Ridge. The received frequencies are
the major water masses of the Arctic Ocean, espe-
typically 20 Hz and below. Earthborne acoustic
cially the important AIW. This latter fact makes the
pressure and shear waves emanate from the source
use of acoustic thermometry for monitoring long-
and couple through the ocean floor above the epi-
term Arctic Ocean temperature change particularly
center to compressional waterborne waves that
suitable. Ongoing research is exploring ways to
propagate vertically and are reflected from the ice
relate changes in acoustic travel time and intensity
canopy into the Arctic sound channel. Most of the
to monitor other important variables in the Arctic
energy from these events arrives via this path, but
Ocean including changes in the PW, DW, and the
there are weaker precursors associated with crustal
halocline, and average sea-ice thickness and rough-
propagating compressional and shear waves that
ness. The latter measurements, when combined
couple to the water directly below the receive array.
with sea ice extent from satellite remote sensing,
These arrivals are easily identified by their vertical
could provide an estimate of sea ice mass in the
arrival angle at the array as well as their arrival time.
Arctic.
Biological noise in the Arctic is concentrated in the
The role of the Arctic Ocean in shaping and re-
marginal ice zones and near the edge of the pack ice.
sponding to global climate change is only beginning
The bowhead is the most numerous of the baleen
to be explored. Cost-effective, long-term, year-round
whales in Arctic waters, migrating along the coast of
synoptic observations in the Arctic Ocean require
Alaska in the Chukchi and Beaufort Seas in the spring
new measurement strategies. The year-round ice
and then exiting the Arctic waters in the fall. They can
cover in the Arctic prevents the use of satellites for
vocalize from 25 to 3500 Hz, but their dominant
direct ocean observations common in the ice-free
frequencies are between 100 and 400 Hz. Of the
oceans. Shore-cabled mooring-based observations
toothed whales the beluga and narwhal are the most
using advanced biogeochemical sensors and acoustic
common, with calls ranging from a few hundred Hz
sources and hydrophone arrays, as well as instru-
to as high as 20 kHz. Many species of pinnipeds, in-
mented autonomous underwater vehicles (AUVs)
cluding hair seals such as the bearded, hooded, harp,
and under-ice drifters, represent new approaches for
ringed, and ribbon seal, and the walrus frequent the
observing the Arctic Ocean. Interestingly the RAFOS
marginal ice zones. As a group their calls typically
concept (using nonexplosive sources) is being evalu-
range from a few hundred Hz to 10 kHz.
ated anew as a way to track AUVs and drifters in the
Unlike the temperate oceans where shipping typi-
Arctic as well as for acoustic communication of data.
cally dominates the ambient noise spectrum, man-
It is clear that Arctic acoustics will have as large a
made noise in the Arctic is a small contributor except
role to play in this important new endeavor in the
for specific events that are isolated in time and space.
future, as it has had in the submarine and military
Such events include icebreakers, seismic exploration,
operations of the past.
and some military and experimental activities. Ice-
breaker noise peaks in the 50–100 Hz range, but
with broadband contributions up to 1000 Hz. Seis-
mic exploration occurs during the summer and fall
seasons when the ice extent is minimum (Figure 1),
See also
allowing easier access to the shelves. Most of the Acoustics, Arctic. Acoustics in Marine Sediments.
activity has been confined to the Beaufort Sea off the Acoustic Noise. Acoustics, Shallow Water.
North Slope of Alaska near Prudhoe Bay. Bioacoustics. Nepheloid Layers. Tomography.
 ACOUSTICS, ARCTIC 405

Further Reading climate monitoring in the Arctic. IEEE Journal of

Oceanic Engineering 24(2): 183--201.
Dyer I (1984) Song of sea ice and other Arctic Ocean Newton JL (1989) Sound speed structure of the Arctic
melodies. In: Dyer I and Chryssostomidis C (eds.) Arctic Ocean including some effects on acoustic propagation.
Technology and Policy, pp. 11--37. Washington, DC: US Navy Journal of Underwater Acoustics 39(4):
Hemisphere Publishing. 363--384.
Dyer I (1993) Source mechanisms of Arctic Ocean ambient Richardson WJ, Greene CR Jr, Malme CI, and Thomson
noise. In: Kerman BR (ed.) Natural Physical Sources of DH (1995) Marine Mammals and Noise. San Diego:
Underwater Sound, pp. 537--551. Netherlands: Kluwer Academic Press.
Academic. Urick RJ (1975) Principles of Underwater Sound. New
Leary WM (1999) Under Ice. College Station: Texas A&M York: McGraw-Hill.
University Press. Von Winkle WA (1984) Naval Underwater Systems Center
LePage K and Schmidt H (1994) Modeling of low- (NUSC) Scientific and Engineering Studies: Underwater
frequency transmission loss in the central Arctic. Acoustics in the Arctic. New London: Naval
Journal of the Acoustic Society of America 96(3): Underwater Systems Center Publisher.
1783--1795.
Mikhalevsky PN, Gavrilov AN, and Baggeroer AB (1999)
The transarctic acoustic propagation experiment and
 ACOUSTICS, DEEP OCEAN
W. A. Kuperman, Scripps Institution of with depth. Below the thermocline, the temperature
Oceanography, University of California, is constant and the sound speed increases because of
San Diego, CA, USA increasing ambient pressure. Therefore, between the
Copyright & 2001 Elsevier Ltd. deep isothermal region and the mixed layer, there is a
minimum sound speed; the depth at which this
minimum takes place is referred to as the axis of the
deep sound channel. However, in polar regions, the
water is coldest near the surface so that the minimum
Introduction sound speed is at the surface.
Figure 2 is a contour display of the sound speed
The acoustic properties of the ocean, such as the structure of the North and South Atlantic with the
paths along which sound from a localized source will deep sound channel axis indicated by the heavy dashed
travel, are mainly dependent on its sound speed line. Note that the deep sound channel becomes
structure. The sound speed structure is dependent on shallower toward the poles. Aside from sound speed
the oceanographic environment described by vari- effects, the ocean volume is absorbtive and will cause
ations in temperature, salinity, and density with attenuation that increases with acoustic frequency.
depth or horizontal position. This article will review The ocean surface and bottom also have a strong
the ocean acoustic environment, sound propagation, influence on sound propagation. The ocean surface,
ambient noise, scattering and reverberation, and the though a perfect reflector when flat, causes scattering
passive and active sonar equation. when its roughness becomes comparable in size with
the acoustic wavelength. The ocean bottom, de-
pending on its local structure will scatter and also
Ocean Acoustic Environment
attenuate the acoustic field.
Sound propagation in the ocean is governed by the
spatial structure of the sound speed and the sound
speed in the ocean is a function of temperature, sal- Units
inity, and ambient pressure. Since the ambient pres-
The decibel (dB) denotes a ratio of intensities (see
sure is a function of depth, it is customary to express
Section 3.3) expressed in terms of a logarithmic (base
the sound speed (c) in meters per second as an em-
pirical function of temperature (T) in degrees celsius, _1
Sound speed (m s )
salinity (S) in parts per thousand and depth (z) in
1440 1460 1480 1500 1520
meters, e.g. eqn [1].
Warmer surface
Surface water profile
2 3
c ¼ 14449:2 þ 4:6T  0:055T þ 0:00029T duct
Mixed layer
profile
þ ð1:34  0:01T ÞðS  35Þ þ 0:016z ½1 profile
Polar Thermocline
1000 region
There exist more accurate formulas, if needed. profile Deep sound
Figure 1 shows a typical set of sound speed pro- channel axis
Water depth (m)

files, indicating greatest variability near the surface.

In a warmer season (or warmer part of the day,
sometimes referred to as the ‘afternoon effect’), the 2000
Deep isothermal layer
temperature increases near the surface and hence the
sound speed increases toward the sea surface. In
nonpolar regions where mixing near the surface due
to wind and wave activity is important, a mixed layer 3000
of almost constant temperature is often created. In
this isothermal layer, sound speed increases with
depth because of the increasing ambient pressure, the
last term in eqn [1]. This is the surface duct region. 4000
Below the mixed layer is the thermocline where the
temperature and hence the sound speed decrease Figure 1 Generic sounds speed profiles.

406
 ACOUSTICS, DEEP OCEAN 407

0
1510

GREENLAND
1000 1475

ANTARCTICA
1500
1485
Depth (m)

2000 1495
1505
REYKJANES
3000 RIDGE
MID
1515
ATLANTIC
SCOTIA
RIDGE
4000 1525
ARGENTINE
RIDGE
CAPE MID RIO EAST
CANARY BRAZIL BASIN 1535 GRANDE BASIN
VERDE ATLANTIC SCOTIA
5000 BASIN
BASIN RIDGE RISE RIDGE WEDDELL
SEA
6000
70˚N 60˚N 50˚N 40˚N 30˚N 20˚N 10˚N 0˚ 10˚S 20˚S 30˚S 40˚S 50˚S 60˚S 70˚S 80˚S
Latitude

Figure 2 Sound speed contours of the North and South Atlantic along 30.501W. Dashed line indicates axis of deep sound channel.
(From Northrop and Colborn (1974).)

30
Ocean surface
0 Su
40 rfa
ce
Depth (m)

-re
Source fle
cte
Direct dp
path ath
50
Transmission loss (dB)

100
Receiver range

60

_4
70 ~r
_2
~r
80

90

100
0 1 2 3 4 5 6 7 8 9 10
Range (km)

Figure 3 The insert shows the geometry of the Lloyd mirror effect. The plots show a comparison of Lloyd mirror to spherical
spreading. Transmission losses are plotted in decibels corresponding to losses of 10 log r2 and 10 log r4, respectively, as explained in
the test.

10) scale. The ratio of two intensities, I1/I2 is 10 log these two paths interfere, they produce a spatial
(I1/I2) in dB units. Absolute intensities are expressed distribution of sound often referred to as a ‘Lloyd
using an accepted reference intensity of a plane wave mirror pattern’ as shown in the insert of Figure 3.
having an rms pressure equal to 105 dyn cm2 or,
equivalently, 1 mPa. Transmission loss is a decibel Basic Long-range Propagation Paths
measure of relative intensity, the latter being pro-
portional to the square of the acoustic amplitude. Figure 4 is a schematic of propagation paths in the
ocean resulting from the sound speed profiles (indicated
by the dashed line) described above in Figure 1. These
Sound Propagation paths can be understood from Snell’s law
eqn [2], which relates the ray angle y (z), with respect to
Very Short-range Propagation
the horizontal, to the local sound speed c(z) at depth z.
The pressure amplitude from a point source in free
space falls off with range r as r1; this geometric loss cosyðzÞ
¼ constant ½2
is called spherical spreading. Most sources of interest cðzÞ
in the deep ocean are nearer the surface than the
bottom. Hence, the two main short-range paths are The equation requires that the higher the sound
the direct path and the surface-reflected path. When speed, the smaller the angle with the horizontal,
408 ACOUSTICS, DEEP OCEAN

Ocean Continental Continental regions of high sound speed, larger launch angle
Arctic basin margin shelf
paths with longer path lengths arrive earlier than
Ice F
shorter paths launched at shallower angles. This is
B
C
just the opposite of boundary-limited propagation
A D such a bottom bounce (or shallow water) in which a
E Ocean bottom reflection occurs before refraction in a high-speed
region can take place. Hence, for deep water re-
fractive paths, those paths that penetrate to a deeper
A: Arctic D: Convergence zone
B: Surface duct E: Bottom bounce
depth have a greater group speed (the horizontal
C: Deep sound channel F: Shallow water speed of energy propagation) than those paths that
do not go as deep, with the axial, most horizontal
Figure 4 Schematic representation of sound propagations path along the deep sound channel axis having the
paths in the ocean.
slowest group speed. For pulse propagation, this
axial arrival is the last arrival.

meaning that sound bends away from regions of high Geometric Spreading Loss
sound speed or, put another way, sound bends
The energy per unit time emitted by a sound source
toward regions of low sound speed. Therefore, paths
flows through a larger area with increasing range.
A, B, and C are the simplest to explain since they are
Intensity is the power flux through a unit area, which
paths that oscillate about the local sound speed
translates to the energy flow per unit time through a
minima. For example, path C depicted by a ray
unit area. Hence, the simplest example of geometric
leaving a source near the deep sound channel axis at
loss is spherical spreading for a point source in free
a small horizontal angle propagates in the deep
space, for which the area increases as 4pr2, where r is
sound channel. This path, in temperate latitudes
the range from the point source. Thus spherical
where the sound speed minimum is far from the
spreading results in an intensity decay proportional
surface, permits propagation over distances of
to r2. Since intensity is proportional to the square of
thousands of kilometers. Path D, which is at slightly
the pressure amplitude, the fluctuations in pressure p
steeper angles and is usually excited by a near surface
induced by the sound decay as r1. For range-in-
source, is convergence zone propagation, a spatially
dependent ducted propagation, that is, where rays
periodic (35–65 km) refocusing phenomenon pro-
are refracted or reflected back toward the horizontal
ducing zones of high intensity near the surface due to
direction (which is the case for most long-range
the upward refracting nature of the deep sound speed
propagation), there is no loss associated with the
profile. Regions between these zones are referred to
vertical dimension. In this case, the spreading surface
as shadow regions. Referring back to Figure 1, there
is the area of a cylinder whose axis is in the vertical
may be a depth in the deep isothermal layer at which
direction passing through the source: 2prH where H is
the sound speed is the same as it is at the surface; this
the depth of the duct and is constant. Geometric loss
depth is called the critical depth and is the lower
in the near-field Lloyd mirror regime requires con-
limit of the deep sound channel. A positive critical
sideration of interfering beams from direct and surface
depth specifies that the environment supports long-
reflected paths. To summarize, the geometric spread-
distance propagation without bottom interaction,
ing laws for the pressure field (recall that intensity is
whereas a negative critical depth specifies that the
proportional to the square of the pressure) are
ocean bottom is the lower boundary of the deep
sound channel. The bottom bounce path E is also a • Spherical spreading loss: ppr1
periodic phenomenon but with a shorter cycle dis- • Cylindrical spreading loss: ppr1/2
tance and shorter propagation distance because of • Lloyd mirror loss: ppr2.
losses when sound is reflected from the ocean
bottom.
Volume Attenuation
An alternative way of describing paths is by de-
noting that they are composed of combinations of Volume attenuation increases with frequency. In
refraction (R), surface reflection (SR) and bottom Figure 3, the losses associated with path C include
reflection (BR) processes. Thus, Figure 4 also repre- only volume attenuation and scattering because this
sents some of these paths. Note from Snell’s law that path does not involve boundary interactions. The
refractive paths involve a path turning around at the volume scattering can be biological in origin or can
highest speed in its duct of confinement. Because arise from interaction with large internal wave ac-
such ray paths spend much of their propagation in tivity in the vicinity of the upper part of the deep
 ACOUSTICS, DEEP OCEAN 409

sound channel where paths are refracted before they water. A summary of the approximate frequency
would interact with the surface. Both of these effects dependence (f in kHz) of attenuation (in units of dB
are small for low frequencies. This same internal km1) is given in eqn [4] with the terms sequentially
wave region is also on the lower boundary of the associated with regions I–IV in Figure 5.
surface duct, allowing scattering out of the surface
duct and thereby also constituting a loss mechanism   0:11f 2 43f 2
a0 dBkm1 ¼ 3:3  103 þ þ
for the surface duct. This mechanism also leaks 1 þ f 2 4100 þ f 2
sound into the deep sound channel, a region that þ 2:98  104 f 2 ½4
without scattering would be a shadow zone for a
surface duct source. This type of scattering from in-
ternal waves is also thought to be a major source of Bottom Loss
fluctuation of the sound field.
The structure of the ocean bottom affects those
Attenuation is characterized by an exponential
acoustic paths that interact with the ocean bottom.
decay of the sound field. If A0 is the rms amplitude of
This bottom interaction is summarized by bottom
the sound field at unit distance from the source, then
reflectivity, the amplitude ratio of reflected and in-
the attenuation of the sound field causes the ampli-
cident plane waves at the ocean-bottom interface as a
tude to decay with distance r along the path ac-
function of grazing angle, y (see Figure 6A). For a
cording to eqn [3], where the unit of a is nepers/
simple bottom that can be represented by a semi-
distance (nepers is a unitless quantity).
infinite half-space with constant sound speed cb and
density rb, the reflectivity is given by eqn [5] with the
A ¼ A0 expðarÞ ½3 subscript w denoting water

This attenuation coefficient can be expressed in rb kwz  rw kbz

RðyÞ ¼ ; ½5
decibels per unit distance by the conversion rb kwz þ rw kbz
a0 ¼ 0.686a. The frequency dependence of attenu-
The wavenumbers are given by eqn [6].
ation can be roughly divided into four regimes as
displayed in Figure 5. In region I, leakage out of the
kiz ¼ ðo=ci Þ sin yi ¼ k sin yi ; i ¼ w; b ½6
sound channel is believed to be the main cause of
attenuation. The main mechanisms associated with The incident and transmitted grazing angles are re-
regions II and III are boric acid and magnesium lated by Snell’s law according to eqn [7] and the in-
sulfate chemical relaxation. Region IV is dominated cident grazing angle yw is also equal to the angle of
by the shear and bulk viscosity associated with fresh the reflected plane wave.

A' cb cos yw ¼ cw cos yb ½7

Region I Region II
B'
1000 Leakage Chemical
relaxation
For this simple water-bottom interface for which
B (OH)3 we take cb >cw, there exists a critical grazing angle yc
100
3:1 below which there is perfect reflection (eqn [8]).
Attenuation, (dB/km)

10 A cw
B cosyc ¼ ½8
30:1 cb
1
For a lossy bottom, there is no perfect reflection, as
_1
10 also indicated in a typical reflection curve in
10:1 Figure 6B. These results are approximately frequency
_2
10 independent. However, for a layered bottom, the
Region III Region IV
? MgSO4 Shear & reflectivity has a complicated frequency dependence
_3
10 ? volume as shown in the example in Figure 6C, where the
relaxation
viscosity contours are in decibels. This example shows the
_4
10 simple reflectivity result below 200 Hz and then a
10 100 1000 10 K 100K 1M 10 M
more complicated frequency dependence at higher
Frequency (Hz) frequencies. It should be pointed out that if the
Figure 5 Regions of different dominant processes of
density of the second medium vanishes, the reflect-
attenuation of sound in sea water. (From Urick (1979).) The ivity reduces to the pressure release case of
attenuation a is given in dB per 1000 yards. R(y) ¼  1.
410 ACOUSTICS, DEEP OCEAN

exp [i (k⊥ .r + k1z z)] acoustically relevant parameters of sound speed,

R exp [i (k⊥. r + k1z z)]
z density, and attenuation.
Sound propagation in the ocean is mathematically
r 1c1 i R
described by the wave equation whose coefficients
2c 2 T
and boundary conditions are derived from the ocean
(A) T exp [i (k⊥. r + k 2z z)] environment. There are essentially four types of
models (computer solutions to the wave equation) to
describe sound propagation in the sea: ray, spectral
 1 Nonlossy bottom
Lossy bottom or fast field program (FFP), normal mode (NM), and
Reflectivity R ()

Phase the parabolic equation (PE). Ray theory is an

Phase (°)

asymptotic high-frequency approximation to the

wave equation, whereas the latter three models are
more or less direct solutions to the wave equations
under an assortment of milder restrictions. The high-
0 0 frequency limit does not include diffraction
0 c 90 phenomena. All of these models can handle depth
(B) Grazing angle  (°) variation of the ocean acoustic environment. A
model that also takes into account horizontal vari-
0 ations in the environment (i.e., sloping bottom or
_
_43
spatially variable oceanography) is termed range-
Grazing angle  (°)

_ 1
_2

4
_

_3
5 _2 _ 4_ 5 dependent. For high frequencies (a few kHz or
above), ray theory is the most practical. The other
4

_4
_

10
three model types are more applicable and usable at
15 lower frequencies (below 1 kHz). The hierarchy of
3
_

underwater acoustic models is shown in schematic

3

_
_

_ 32 _3 _4
20
form in Figure 7. The output of these models is
0.05 0.1 0.2 0.4 0.8 1.6 3.2
typically propagation loss, which is the intensity
(C) Frequency (kHz)
relative to a unit source at unit distance, expressed in
Figure 6 The reflection and transmission processes. Grazing decibels. Transmission loss is the negative of propa-
angles are defined relative to the horizontal. (A) A plane wave is gation loss, and hence, a positive quantity.
incident on an interface separating two media with densities and An example of the output of propagation models
sound speeds rc. R(y) and T (y) are reflection and transmission is shown in Figure 8, indicating agreement between
coefficients. Snell’s law is a statement that k>, the horizontal
the models. However, we also see a difference among
component of the wavevector, is the same for all three waves. (B)
Rayleigh reflection curve (eqn [5]) as a function of the grazing the models in that ray theory predicts a sharper
angle (y in (A)) indicating critical yc. The dashed curve shows that if shadow zone than the wave theory model (i.e., the
the second medium is lossy, there is less than perfect reflection 10–30 km region in Figure 8B); this is an expected
below the critical angle. (C) Examples of contour of reflection loss result from the infinite-frequency ray approximation.
(20 log R) for a layered bottom, showing frequency and grazing
angle dependence. The simpler reflectivity curve for each
frequency is obtained from a vertical slice.
Scattering and Reverberation
Propagation Models
Scattering caused by rough boundaries or volume
An ocean acoustic environment is often very com- inhomogeneities is a mechanism for loss (attenu-
plex, with range- and depth-dependent properties. ation), reverberant interference, and fluctuation. At-
Such an environment does not in general lend itself tenuation from volume scattering was addressed
to simple analytic predictions of sound propagation. above. In most cases, it is the mean or coherent (or
Even in range-independent environments there are specular) part of the acoustic field that is of interest
many paths (multipaths) and these paths combine to for a sonar or communications application, and
form a complex interference pattern. For example, scattering causes part of the acoustic field to be
the convergence zones are an example of a more randomized. Rough surface scattering out of the
complex structure that cannot be described by a ‘specular direction’ can be thought of as an attenu-
monotonic geometric spreading law. Acoustic mod- ation of the mean acoustic field and typically in-
els play an important role in predicting sound creases with increasing frequency. A formula often
propagation; the inputs to these models are oceano- used to describe reflectivity from a rough boundary is
graphic quantities ultimately translated into the eqn [9], where R(y) is the reflection coefficient of the
 ACOUSTICS, DEEP OCEAN 411

Nonlinear Linear Frequency-domain

wave equation wave equation wave equation

Normal WKBJ Range-

TDFFP FFP approximation independent
modes

Coupled Adiabatic Ray Range-

NPE TDPE modes modes theory PE
dependent

Asymptotic Asymptotic

Spectral

Figure 7 Heirarchy of underwater acoustic models. TD refers to time domain. NPE is the nonlinear parabolic equation that describes
high-amplitude (e.g., shockwave) propagation. The arrows are directed toward the flow of derivation of the model.

smooth interface and G is the Rayleigh roughness 1520

_
1510 1530 m s 1 Source = 25 m, Receiver = 38 m
parameter defined as GR2ks sin y where k ¼ 2p/l,
0
l is the acoustic wavelength, and s is the rms Depth (m)
200
roughness (height). 400
600
G2
R0 ðyÞ ¼ RðyÞexp  ½9 800
2 1000
The scattered field is often referred to as re- 0 10 20 30
verberation. Surface, bottom, or volume scattering (A) Range (km)
strength, SS,B,V is a simple parametrization of the
60
production of reverberation and is defined as the
70
ratio in decibels of the sound scattered by a unit
surface area or volume referenced to a unit distance, 80
(Insert equqtion), to the incident plane wave inten-
sity, Iinc (eqn [10]).
60
Propagation loss (dB)

Iscat 70
SS;B;V ¼ 10 log ½10
Iinc 80
90
The Chapman–Harris curves predicts the ocean
surface scattering strength in the 400–6400 Hz 100 PE (250 Hz)
Ray (2000 Hz)
region; eqn [11], where y is the grazing angle in 110 Experiment
degrees, w the wind speed in m s1 and f is the 120
frequency in Hz. 0 10 20 30
(B) Range (km)
y
SS ¼ 3:3b log  42:4 log b þ 2:6; Figure 8 Model and data comparison for a range-dependent
30
 0:58 deep water case. (A) Sound speed profiles as a function of range
b ¼ 107 wf 1=3 ½11 together with a ray trace showing the breakdown of surface duct
propagation. (B) Parabolic equation comparison with data at
A more elaborate formula exists that is more accur- 250 Hz and ray theory comparison with data at 2000 Hz.
ate for lower wind speeds and lower frequencies.
The simple characterization of bottom back-
scattering strength utilizes Lambert’s rule for diffuse Under the assumption that all incident energy is
scattering, given by eqn [12] where the first term is scattered into the water column with no transmission
determined empirically. in to the bottom, A is  5 dB. Typical realistic values
for A that have been measured are  17 dB for big
Basalt Mid-Atlantic Ridge cliffs and  27 dB for
SB ¼ A þ 10 log sin2 y ½12 sediment ponds.
412 ACOUSTICS, DEEP OCEAN

Volume scattering strength is typically reduced Finally, near-surface bubbles and bubble clouds
to a surface scattering strength by taking (Insert can be thought of as either volume or surface scat-
equqtion) as an average volume scattering strength tering mechanisms acting in concert with the rough
within some layer at a particular depth; then the surface. Bubbles have resonances (typically greater
corresponding surface scattering strength is given by than 10 kHz) and at these resonances, scattering is
eqn [13], where H is the layer thickness. strongly enhanced. Bubble clouds have collective
properties; among these properties is that a bubbly
mixture, as specified by its void fraction (total bubble
SS ¼ Sv þ 10 log H ½13
gas volume divided by water volume) has a con-
siderable lower sound speed than water.
The column or integrated scattering strength is de-
fined as the case for which H is the total water depth.
Volume scattering usually decreases with depth
(about 5 dB per 300 m) with the exception of the
Ambient Noise
deep scattering layer. For frequencies less than There are essentially two types of ocean acoustic
10 kHz, fish with air-filled swimbladders are the noise: man-made and natural. Generally, shipping is
main scatterers. Above 20 kHz, zooplankton or the most important source of man-made noise,
smaller animals that feed upon phytoplankton and though noise from offshore oil rigs is becoming more
the associated biological chain are the scatterers. The and more prevalent. Typically, natural noise domin-
deep scattering layer (DSL) is deeper in the day than ates at low frequencies (below 10 Hz) and high fre-
in the night, changing most rapidly during sunset and quencies (above a few hundred hertz). Shipping fills
sunrise. This layer produces a strong scattering in- in the region between ten and a few hundred hertz.
crease of 5–15 dB within 100 m of the surface at A summary of the spectrum of noise is shown in
night and virtually no scattering in the daytime at the Figure 10. The higher-frequency noise is usually
surface since it migrates down to hundreds of meters. parametrized according to sea state (also Beaufort
Since higher pressure compresses the fish swim- number) and/or wind. Table 1 summarizes the de-
bladder, the backscattering acoustic resonance tend scription of sea state.
to be at a higher frequency during the day when the The sound speed profile affects the vertical and
DSL migrates to greater depths. Examples of day and angular distribution of noise in the deep ocean.
night scattering strengths are shown in Figure 9. When there is a positive critical depth, sound from

≤100 m
_ 45

≤ 850 m ≤ 340 m
_ 50 ≤ 780 m
10 log ∫0 MZ d Z

≤ 580 m
d

_ 55

≤ 140 m
_ 60

_ 65 Day Night

2 5 10 15 2 5 10 15
Frequency (kHz)
(A) (B)

Figure 9 Day and night scattering strength measurements using an explosive source as a function of frequency. The spectra
measured at various times after the explosion are labeled with the depth of the nearest scatterer that could have contributed to the
reverberation. The ordinate corresponds to Sv in eqn [13]. (From Chapman and Marchall (1966))
 ACOUSTICS, DEEP OCEAN 413

Intermittent and local effects

146 Earthquakes
and explosions
Biologics
Precipitation
Ships, industrial activity
Sea ice
KEY
126 Limits of prevailing noise
Wind-dependent bubble and spray noise

Low-frequency very-shallow-water
wind dependence
Heavy precipitation
Sound pressure spectrum level (dB re 1 Pa)

Heavy traffic noise

106 Usual traffic noise _ deep water
Usual traffic noise _ shallow water
Thermal noise
General pattern of noise from
earthquakes and explosions
Extrapolations
86

Wind force
(Beaufort)

66

46
3
2
Prevailing noises
Turbulent-pressure 1
fluctuations
26
Oceanic traffic
Bubbles and spray
(surface agitation)
Surface waves _ second-order pressure effects
(seismic background) Molecular
Agitation
6
2 3 4 5
1 10 10 10 10 10
Frequency (Hz)

Figure 10 Composite of ambient noise spectra. (From Wenz (1962).)

surface sources can travel long distances without In a range-independent ocean, Snell’s law predicts
interacting with the ocean bottom, but a receiver a horizontal noise notch at depths where the speed of
below this critical depth should sense less surface sound is less than the near-surface sound speed. Re-
noise because propagation involves interaction with turning to eqn [2] and reading off the sound speeds
lossy boundaries, surface and/or bottom. This is il- from Figure 11 at the surface (c ¼ 1530 m s1) and
lustrated in Figure 11, which shows a deep water say, 300 m (1500 m s1), a horizontal ray (y ¼ 0)
environment with measured ambient noise. Figure 12 launched from ocean surface would have an angle
is an example of vertical directivity of noise that also with respect to the horizontal of about 111 at 300 m
follows the propagation physics discussed above. depth. All other rays would arrive with greater ver-
The shallower depth is at the axis of the deep sound tical angles. Hence we expect this horizontal notch.
channel, while the other is at the critical depth. The However, the horizontal notch is often not seen at
pattern is narrower at the critical depth where the shipping noise frequencies. This is because shipping
sound paths tend to be horizontal since the rays are tends to be concentrated in continental shelf regions;
turning around at the lower boundary of the deep range-dependent propagation couples such noise
sound channel. sources to the deep ocean. Thus, for example,
414 ACOUSTICS, DEEP OCEAN

Table 1 Descriptions of the ocean sea surface. Approximate relation between scales of wind speed, wave height, and sea state

Sea criteria Wind speed 12-h wind Fully risen sea

Beaufort Range Mean Wave Wave Durationb, c

Fetchb, c
Seastate
scale (m s1) (m s1) height a, b
height a, b
(h) (km) scale
(m) (m)

Mirrorlike 0 o0.5 0
Ripples 1 0.5–1.7 1.1 1/2
Small wavelets 2 1.8–3.3 2.5 o0.30 o0.30 1
Large wavelets, scattered whitecaps 3 3.4–5.4 4.4 0.30–0.61 0.30–0.61 o2.5 o19 2
Small waves, frequent whitecaps 4 5.5–8.4 6.9 0.61–1.5 0.61–1.8 2.5–6.5 19–74 3
Moderate waves, many whitecaps 5 8.5–11.1 9.8 1.5–2.4 1.8–3.0 6.5–11 74–185 4
Large waves, whitecaps everywhere, spray 6 11.2–14.1 12.6 2.4–3.7 3.0–5.2 11–18 185–370 5
Heaped-up sea, blown spray, streaks 7 14.2–17.2 15.7 3.7–5.2 5.2–7.9 18–29 370–740 6
Moderately high, longwaves, spindrift 8 17.3–20.8 19.0 5.2–7.3 7.9–11.9 29–42 740–1300 7

a
The average height of the highest one-third of the waves (significant wave height).
b
Estimated from data given in US Hydrographic Office (Washington, DC) publications HO 604 (1951) and HO 603 (1955).
c
The minimum fetch and duration of the wind needed to generate a fully risen sea.
From Wenz (1962).

18 Hz
500 Hz

150 Hz
250 Hz

100 Hz
1000

25 Hz
50 Hz
2000
Depth (m)

3000

4000 CRITICAL
DEPTH = 4420 m

5000

BOTTOM 5322 m
BOTTOM 5322 m
6000
1480 1500 1520 1540 50 60 70 80 90
_1
Sound velocity (m s ) Spectrum level (dB relative to 1 μPa)
(A) (B)

Figure 11 (A) Sound speed profile and (B) noise level as a function of depth in the Pacific. (From Morris (1978))

propagation down a continental slope converts high- methodology of the sonar equation is analogous to
angle rays to lower angles at each bounce. There are an accounting procedure involving acoustic signal,
also deep sound channel shoaling effects that result interference, and system characteristics.
in the same trend in angle conversion. It is instructive, beyond the specific application to
conventional sonars, to understand this accounting
methodology and below is a simplified summary.
Sonar Equation
Passive Sonar Equation
A major application of underwater acoustics is sonar
technology. The performance of a sonar is often de- A passive sonar system uses the radiated sound from
scribed simply in terms of the sonar equation. The a target to detect and locate the target. A radiating
 ACOUSTICS, DEEP OCEAN 415

object of source level SL (all units are in decibels) is threshold. Conversely, since the FOM is a transmis-
received at a hydrophone of a sonar system at a sion loss, one can use the output of a propagation
lower signal level S because of the transmission loss model (or, if appropriate, a simple geometric loss plus
‘TL’ it suffers (e.g., cylindrical spreading plus at- attenuation) to estimate the minimum range at which
tenuation or a TL computed from one of the a 50% probability of detection can be expected. This
propagation models) (eqn [14]). range changes with oceanographic conditions and is
often referred to as the ‘range of the day’ in navy
S ¼ SL  TL ½14 sonar applications.

The noise, N, at a single hydrophone is subtracted

Active Sonar Equation
from eqn [14] to obtain the signal-to-noise ratio
(SNR) at a single hydrophone (eqn [15]). A monostatic active sonar transmits a pulse to a
target and its echo is detected at a receiver co-located
SNR ¼ SL  TL  N ½15 with the transmitter. A bistatic active sonar has the
receiver in a different location from the transmitter.
Typically a sonar system consists of an array or an- The main differences between the passive and active
tenna of hydrophones that provides signal-to-noise cases is that the source level is replaced by a target
enhancement through a beamforming process; this strength, TS; reverberation and hence reverberation
process is quantified in decibels by array gain AG level, RL, is usually the dominant source of inter-
that is therefore added to the single hydrophone SNR ference as opposed noise; and the transmission loss is
to give the SNR at the output of the beamformer over two paths: transmitter to target and target to
(eqn [16]). receiver. In the monostatic case the transmission loss
is 2TL, where TL is the one-way transmission loss;
SNRBF ¼ SL  TL  N þ AG ½16 and in the bistatic case the transmission loss is the
Because detection involves additional factors in- sum (in dB) over paths from the transmitter to the
cluding sonar operator ability, it is necessary to specify target and the target to the receiver, TL1 þ TL2. The
a detection threshold (DT) level above the SNRBFat concept of the detection threshold is useful for both
which there is a 50% (by convention) probability of passive and active sonars. Hence, for signal excess,
detection. The difference between these two quantities we have eqn [19].
is called signal excess (SE) (eqn [17]).
S ¼ SL  TL1 þ TS  TL2
SE ¼ SL  TL  N þ AG  DT ½17  ðRL þ N Þ þ AG  DT ½19
This decibel bookkeeping leads to an important sonar
The corresponding FOM for an active system is de-
engineering descriptor called the figure of merit,
fined for the maximum allowable two-way trans-
FOM, which is the transmission loss that gives a zero
mission loss with TS ¼ 0 dB.
signal excess (eqn [18]).

FOM ¼ SL  N þ AG  DT ½18
See also
The FOM encompasses the various parameters a
Acoustic Noise. Acoustic Scattering by Marine
sonar engineer must deal with: expected source level,
Organisms. Acoustics, Arctic. Acoustics, Shallow
the noise environment, array gain and the detection Water. Acoustics in Marine Sediments.
Bioacoustics. Deep-Sea Drilling Methodology.
Seismology Sensors. Sonar Systems. Tomography.
(dB / μPa / steradian)

80
Spectrum level

713 m
3781 m
70 Further Reading
Anderson VC (1979) Variations of the vertical directivity
_ 20 _ 10 0 10 20 of noise with depth in the North Pacific. Journal of the
DOWN Elevation angle (°) UP Acoustical Society of America 66: 1446--1452.
Brekhovskikh LM and Lysanov YP (1991) Fundamentals
Figure 12 The vertical directionality of noise at the axis of the of Ocean Acoustics. Berlin: Springer-Verlag.
deep sound channel and at the critical depth in the Pacific. (From Chapman RP and Harris HH (1962) Surface backscattering
Anderson (1979)) strengths measured with explosive sound sources.
416 ACOUSTICS, DEEP OCEAN

Journal of the Acoustical Society of America 34: 1592-- Munk W, Worcester P, and Wunsch C (1995) Acoustic
1597. Tomography. Cambridge: Cambridge University Press.
Chapman RP and Marchall JR (1966) Reverberation from Nicholas M, Ogden PM, and Erskine FT (1998) Improved
deep scattering layers in the Western North Atlantic. empirical descriptions for acoustic surface backscatter
Journal of the Acoustical Society of America 40: in the ocean. IEEE-JOE 23: 81--95.
405--411. Northrup J and Colborn JG (1974) Sofar channel axial
Collins MD and Siegmann WL (2001) Parabolic Wave sound speed and depth in the Atlantic Ocean. Journal of
Equations with Applications. New York: Springer-AIP. Geophysical Research 79: 5633--5641.
Dushaw BD, Worcester PF, Cornuelle BD, and Howe BM Ross D (1976) Mechanics of Underwater Noise. New
(1993) On equations for the speed of sound in seawater. York: Pergamon.
Journal of the Acoustical Society of America 93: 255--275. Ogilvy JA (1987) Wave scattering from rough surfaces.
Jensen FB, Kuperman WA, Porter MB, and Schmidt H Reports on Progress in Physics 50: 1553--1608.
(1994) Computational Ocean Acoustics. Woodbury: Urick RJ (1979) Sound Propagation in the Sea.
AIP Press. Washington, DC: US GPO.
Keller and Papadakis JS (eds.) (1977) Wave Propagation in Urick RJ (1983) Principles of Underwater Sound. New
Underwater Acoustics. New York: Springer-Verlag. York: McGraw Hill.
Makris NC, Chia SC, and Fialkowski LT (1999) The bi- Spiesberger JL and Metzger K (1991) A new algorithm for
azimuthal scattering distribution of an abyssal hill. sound speed in seawater. Journal of the Acoustical
Journal of the Acoustical Society of America 106: Society of America 89: 2677--2688.
2491--2512. Wenz GM (1962) Acoustics ambient noise in the ocean:
Medwin H and Clay CS (1997) Fundamentals of spectra and sources. Journal of the Acoustical Society of
Acoustical Oceanography. Boston: Academic Press. America 34: 1936--1956.
Morris GB (1978) Depth dependence of ambient noise in
the Northeastern Pacific Ocean. Journal of the
Acoustical Society of America 64: 581--590.
 ACOUSTICS, SHALLOW WATER
F. B. Jensen, SACLANT Undersea Research Centre, away from the bottom and therefore can propagate
La Spezia, Italy to long ranges with little attenuation. Moreover, the
Copyright & 2001 Elsevier Ltd. environmental variability is much higher in coastal
regions than in the deep ocean, with the result that
there is much more acoustic variability in shallow
water than in deep water.

Introduction The Ocean Acoustic Environment

Using the commonly accepted definition of shallow The ocean is an acoustic waveguide limited above by
water to mean coastal waters with depth up to the sea surface and below by the seafloor. The speed
200 m, the shallow-water regions of the world con- of sound in the waveguide plays the same role as the
stitute around 8% of all oceans and seas. These re- index of refraction does in optics. Sound speed is
gions are particularly important since they are normally related to density and compressibility. In
national economic zones and also more assessible. the ocean, density is related to static pressure, sal-
Sound waves in the sea play the role of light in the inity and temperature. The sound speed in the ocean
atmosphere, i.e. acoustics is the only means of ‘seeing’ is an increasing function of temperature, salinity, and
objects at distances beyond a few hundred meters in pressure, the latter being a function of depth. It is
seawater. All forms of electromagnetic waves (light, customary to express sound speed (c) as an empirical
radar) are rapidly attenuated in seawater. Low- function of three independent variables: temperature
frequency acoustic signals, on the other hand, (T) in degrees centigrade, salinity (S) in parts per
propagate with little attenuation and can be heard thousand (%), and depth (D) in meters. A simplified
over thousands of kilometers in the deep ocean. expression for this dependence is
The use of sound in the sea is ubiquitous. It is
employed by the military to detect mines and sub- c ¼ 1449:2 þ 4:6T  0:055T 2 þ 0:00029T 3
½1
marines, and ship-mounted sonars measure water þð1:34  0:010T ÞðS  35Þ þ 0:016D
depth, ship speed and the presence of fish shoals.
Side-scan systems are used to map bottom topo- In shallow water, where the depth effect on sound
graphy, sub-bottom profilers for getting information speed is small, the primary contributor to sound
about the deeper layering, and other sonar systems speed variations is the temperature. Thus, for a sal-
for locating pipelines and cables on and beneath the inity of 35%, the sound speed in seawater varies
seafloor. Sound is also used for navigating submerged between 1450 m s1 at 01C and 1545 m s1 at 301C.
vehicles, for underwater communications and for Seasonal and diurnal changes affect the oceano-
tracking marine mammals. In an inverse sense sound graphic parameters in the upper ocean. In addition,
is used for measuring physical parameters of the all of these parameters are a function of geography.
ocean environment and for monitoring oceanic pro- In a warmer season (or warmer part of the day) in
cesses through the techniques of acoustical ocean- shallow seas where tidal mixing is weak, the tem-
ography and ocean acoustic tomography. perature increases near the surface and hence the
Optimal sonar design for this great variety of ap- sound speed increases toward the sea surface. This
plications demands using a wide range of acoustic near-surface heating (and subsequent cooling) has a
frequencies. Practical shallow-water systems cover a profound effect on surface-ship sonars. Thus the di-
frequency range from 50 Hz to 500 kHz, which, with urnal heating causes poorer sonar performance in the
a mean sound speed of 1500 m s1, correspond to afternoon – a phenomenon known as the afternoon
acoustic wavelengths from 30 m down to 3 mm. effect. The seasonal variability, however, is much
The principal characteristic of shallow-water greater and therefore more important acoustically.
propagation is that the sound-speed profile is nearly A ray picture of propagation in a 100-m deep
constant over depth or downward refracting, mean- shallow water duct is shown in Figure 1. The sound-
ing that long-range propagation takes place ex- speed profile in the upper panel is typical of the
clusively via lossy bottom-interacting paths. This is Mediterranean in the summer. There is a warm sur-
very different from deep-water scenarios, where the face layer causing downward refraction and hence
sound-speed structure is such that sound is refracted repeated bottom interaction for all ray paths. Since

417
418 ACOUSTICS, SHALLOW WATER

SV SEA SURFACE
0
Depth

2.5˚ slope
100

Depth (m)
(A) Range 200
SV

300
Depth

400
(B) Range 0 1 2 3
Range (km)
Figure 1 Ray paths in shallow water for typical Mediterranean
summer and winter profiles. (A) In summer sound interacts Figure 3 Seismic profile of bottom layering in coastal-water
repeatedly with the seabed but not with the sea surface. (B) In area of the Mediterranean.
winter sound interacts with both the sea surface and the seabed,
except for shallow rays emitted near the horizontal.
propagation paths suffers increased bottom re-
flection loss) and of acoustic signal fluctuations with
the seafloor is a lossy boundary, propagation in time.
shallow water is dominated by bottom reflection loss Turning to the upper and lower boundaries of the
at low and intermediate frequencies (o1 kHz) and ocean waveguide, the sea surface is a simple hori-
scattering losses at high frequencies. The seasonal zontal boundary and a nearly perfect reflector. The
variation in sound-speed structure is significant with seafloor, on the other hand, is a lossy boundary with
winter conditions being nearly iso-speed (Figure 1B). varying topography. Both boundaries have small-
The result is that there is less bottom interaction in scale roughness associated with them which causes
winter than in summer, which again means that scattering and hence attenuation of sound due to the
propagation conditions are generally better in winter increased bottom reflection loss associated with
than in summer. steep-angle propagation paths. In terms of propa-
Of course, the ocean sound-speed structure is gation physics, the seafloor is definitely the most
neither frozen in time nor space. On the contrary, the complex boundary, exhibiting vastly different re-
ocean has its own weather system. There are cur- flectivity characteristics in different geographical
rents, internal waves and thermal microstructure locations.
present in most shallow-water areas. Figure 2 illus- The structure of the ocean bottom in shallow
trates the sound speed variability along a 15 km-long water generally consists of a thin stratification of
track in the Mediterranean Sea. The data were re- sediments overlying the continental crust. The nature
corded on a towed thermistor chain covering depths of the stratification is dependent on many factors,
between 5 and 90 m. In general, this type of time- including geological age and local geological activity.
varying oceanographic structure has an effect on Thus, relatively recent sediments will be character-
sound propagation, both as a source of attenuation ized by plane stratification parallel to the sea bed,
(acoustic energy being scattered into steeperangle whereas older sediments and sediments close to the
crustal plate boundaries may have undergone sig-
SV Summer nificant deformation. An example of a complicated
bottom layering is given in Figure 3, which displays a
seismic section from the coastal Mediterranean. The
upper stratification here is almost parallel to the
seafloor, whereas deeper layers are strongly inclined.
Depth

Transmission Loss
The decibel (dB) is the dominant unit in ocean
Range acoustics and denotes a ratio of intensities (not
Figure 2 Spatial variability of sound speed in shallow-water
pressures) expressed on a log10 scale.
area of the Mediterranean. The depth covered is around 100 m An acoustic signal traveling through the ocean
and the range 15 km. becomes distorted due to multipath effects and
 ACOUSTICS, SHALLOW WATER 419

weakened due to various loss mechanisms. The The cylindrical spreading loss is therefore given by
standard measure in underwater acoustics of the
change in signal strength with range is transmission TL ¼ 10 log r ½dB re 1 m ½4
loss defined as the ratio in decibels between the
acoustic intensity I(r,z) at a field point and the Note that for a point source in a waveguide, there is
intensity I0 at 1 m distance from the source, i.e. spherical spreading in the nearfield (rrD) followed
by a transition region toward cylindrical spreading
Iðr; zÞ which applies only at longer ranges (rbD).
TL ¼  10 log
I0 As an example consider propagation in a shallow-
jpðr; zÞj water waveguide to a range of 20 km with spherical
¼  20 log ½dB re 1 m: ½2 spreading applying on the first 100 m. The total
j p0 j
propagation loss (neglecting attenuation) then be-
Here use has been made of the fact that the intensity comes: 40 dB þ 23 dB ¼ 63 dB. This figure represents
of a plane wave is proportional to the square of the the minimum loss to be expected at 20 km. In prac-
pressure amplitude. The major contributors to tice, the total loss will be higher due both to the
transmission loss in shallow water are: geometrical attenuation of sound in seawater, and to various re-
spreading loss, water volume attenuation, bottom flection and scattering losses.
reflection loss, and various scattering losses.
Sound Attenuation in Seawater
Geometrical Spreading
When sound propagates in the ocean, part of the
The spreading loss is simply a measure of the signal acoustic energy is continuously absorbed, i.e. the
weakening as it propagates outward from the source. energy is transformed into heat. Moreover, sound is
Figure 4 shows the two geometries of importance in scattered by different kinds of inhomogeneities, also
underwater acoustics. First consider a point source in resulting in a decay of sound intensity with range. As
an unbounded homogeneous medium (Figure 4A). a rule, it is not possible in real ocean experiments to
For this simple case the power radiated by the source distinguish between absorption and scattering ef-
is equally distributed over the surface area of a sphere fects; they both contribute to sound attenuation in
surrounding the source. If the medium is assumed to seawater.
be lossless, the intensity is inversely
 proportional
 to A simplified expression for the frequency depend-
the surface of the sphere, i.e. Ip1= 4pR2 . Then from ence (f in kHz) of the attenuation is given by Thorp’s
eqn [2] the spherical spreading loss is given by formula,

TL ¼ 20 log r ½dB re 1 m ½3 0:11f 2 44f 2

a¼ þ ½dB km1 ; ½5
1 þ f 2 4100 þ f 2
where r is the horizontal range in meters.
When the medium has plane upper and lower where the two terms describe absorption due to
boundaries as in the waveguide case in Figure 4B, the chemical relaxations of boric acid, B(OH)3, and
farfield intensity change with horizontal range magnesium sulphate, MgSO4, respectively.
becomes inversely proportional to the surface of a According to eqn [5] the attenuation of low-fre-
cylinder of radius R and depth D, i.e. Ip1=ð2pRDÞ. quency sound in seawater is indeed very small. For
instance, at 100 Hz a tenfold reduction in sound
intensity (  10 dB) occurs over a distance of around
8300 km. Even though attenuation increases with
frequency (r10 dB C150 km at 1 kHz and C9 km at
R R 10 kHz), no other kind of radiation can compete
* D
* with sound waves for long-range propagation in the
ocean.

Bottom Reflection Loss

1 1
I∝ I∝ Reflectivity, the ratio of the amplitudes of a reflected
4πR 2 2πRD
plane wave to a plane wave incident on an interface
(A) (B)
separating two media, is an important measure of the
Figure 4 Geometrical spreading laws. (A) Spherical spreading; effect of the bottom on sound propagation. Ocean
(B) Cylindrical spreading. bottom sediments are often modeled as fluids which
420 ACOUSTICS, SHALLOW WATER

means that they support only one type of sound

wave – a compressional wave.
The expression for reflectivity at an interface sep- 10 Cp = 1550 m s−1
arating two homogeneous fluid media with density ri
and sound speed ci, i ¼ 1, 2, was first worked out by

Bottom loss (dB)

1600
Rayleigh as 1800
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 ffi
ðr2 =r1 Þsin y1  ðc1 =c2 Þ2 cos2 y1 5
Rðy1 Þ ¼ rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 ffi ½6
2 2
ðr2 =r1 Þsin y1 þ ðc1 =c2 Þ cos y1

θc
where y1 denotes the grazing angle of the incident 0
plane wave of unit amplitude. 0 30 60 90
The reflection coefficient has unit magnitude, Grazing angle (°)
meaning perfect reflection, when the numerator and
Figure 5 Bottom reflection loss curves for different bottom
denominator of eqn [6] are complex conjugates. This types. Note that low-speed bottoms (clay, silt) are more lossy
can only occur when the square root is purely than high-speed bottoms (sand, gravel). Cp is the compressional
imaginary, i.e. for cos y14c1/c2 (total internal re- wave speed.
flection). The associated critical grazing angle below
which there is perfect reflection is found to be
Real ocean bottoms are complex layered struc-

 tures of spatially varying material composition.
c1
yc ¼ arccos ½7 A geoacoustic model is defined as a model of the real
c2 seafloor with emphasis on measured, extrapolated,
and predicted values of those material properties
Note that a critical angle only exists when the sound important for the modeling of sound transmission. In
speed of the second medium is higher than that of the general, a geoacoustic model details the true thick-
first. nesses and properties of sediment and rock layers
A closer look at eqn [6] shows that the reflection within the seabed to a depth termed the effective
coefficient for lossless media is real for yc4yc, which acoustic penetration depth. Thus, at high frequencies
means that there is loss ðjRj ¼ 1Þ but no phase shift (41 kHz), details of the bottom composition are
associated with the reflection process. On the other required only in the upper few meters of sediment,
hand, for ycoyc we have perfect reflection ðjRj ¼ 1Þ whereas at low frequencies (o100 Hz) information
but with an angle-dependent phase shift. In the must be provided on the whole sediment column and
general case of lossy media (ci complex), the re- on properties of the underlying rocks.
flection coefficient is complex, and, consequently, The information required for a complete geoa-
there is both a loss and a phase shift associated with coustic model should include the following depth-
each reflection. dependent material properties: the compressional
The critical-angle concept is very important for wave speed, Cp; the shear wave speed, Cp; the com-
understanding the waveguide nature of shallow- pressional wave attenuation, Cp; the shear wave at-
water propagation. Figure 5 shows bottom loss tenuation, Cp; and the density, Cp. Moreover,
curves ðBL ¼  10 log jRj2 Þ for a few simple fluid information on the variation of all of these par-
bottoms with different compressional wave speeds ameters with geographical position is required.
(Cp), densities and attenuations. Note that for a lossy The amount of literature dealing with acoustic
bottom we never get perfect reflection. However, properties of seafloor materials is vast. Table 1 lists
there is in all cases an apparent critical angle the geoacoustic properties of some typical seafloor
(yCC331 for cp ¼ 1800 m s1 in Figure 5), below materials, as an indication of the many different
which the reflection loss is much smaller than for types of materials encountered just in continental
supercritical incidence. With paths involving many shelf and slope environments.
bottom bounces such as in shallow-water propa-
gation, bottom losses even as small as a few tenths of
Boundary and Volume Scattering Losses
a decibel per bounce accumulate to significant total
losses since the propagation path may involve many Scattering is a mechanism for loss, interference and
tens or even hundreds of bounces. fluctuation. A rough sea surface or seafloor causes
 ACOUSTICS, SHALLOW WATER 421

Table 1 Geoacoustic properties of continental shelf environments

Bottom type p rb/rw Cp/Cw Cp Cs ap as

(%) – – (m s1) (m s1) (aBl1
p ) (aBl1
s )

Clay 70 1.5 1.00 1500 o100 0.2 1.0

Silt 55 1.7 1.05 1575 Csa 1.0 1.5
Sand 45 1.9 1.1 1650 Csb 0.8 2.5
Gravel 35 2.0 1.2 1800 Csc 0.6 1.5
Moraine 25 2.1 1.3 1950 600 0.4 1.0
Chalk – 2.2 1.6 2400 1000 0.2 0.5
Limestone – 2.4 2.0 3000 1500 0.1 0.2
Basalt – 2.7 3.5 5250 2500 0.1 0.2

a
Cs ¼ 80 z̃0.3
b
cs ¼ 110 z̃0.3
c
cs ¼ 180 z̃0.3
Cw ¼ 1500 ms  1, rw ¼ 1000 kg m  3.

attenuation of the mean acoustic field propagating in appropriate geoacoustic parameters (see Table 1) are
the ocean waveguide. The attenuation increases with used for evaluating R0 (y), and the rough-bottom
increasing frequency. The field scattered away from reflection coefficient is then obtained from eqn [8].
the specular direction, and, in particular, the back- Volume scattering is thought to arise primarily
scattered field (called reverberation) acts as inter- from biological organisms. For lower frequencies
ference for active sonar systems. Because the ocean (less than 10 kHz), fish with air-filled swim bladders
surface moves, it will also generate acoustic fluctu- are the main scatterers whereas above 20 kHz, zoo-
ations. Bottom roughness can also generate fluctu- plankton or smaller animals that feed on the phyto-
ations when the sound source or receiver is moving. plankton, and the associated biological food chain,
The importance of boundary roughness depends on are the scatterers. Many of the organisms undergo a
the sound-speed profile which determines the degree diurnal migration rising towards the sea surface at
of interaction of sound with the rough boundaries. sunset and descending to depth at sunrise. Since the
Often the effect of scattering from a rough surface composition and density of the populations vary
is thought of as simply an additional loss to the with the environmental conditions, the scattering
specularly reflected (coherent) component resulting characteristics depend on geographical location, time
from the scattering of energy away from the specular of day, season and frequency. As an example, data
direction. If the ocean bottom or surface can be from the Mediterranean Sea for volume scattering
modeled as a randomly rough surface, and if the losses due to fish shoals show excess losses of
roughness is small with respect to the acoustic 10–15 dB for a propagation range of 12 km and
wavelength, the reflection loss can be considered to frequencies between 1 and 3 kHz.
be modified in a simple fashion by the scattering Finally, scattering off bubbles near the surface is
process. A formula often used to describe reflectivity sometimes referred to as either a volume- or surface-
from a rough boundary is: scattering mechanism. These bubbles arise not only
2
from sea surface action, but also from biological
R0 ðyÞ ¼ RðyÞe0:5G ½8 origins and from ship wakes. Furthermore, bubbles
are not the only scattering mechanism, but bubble
where R0 (y) is the new reflection coefficient, reduced clouds may have significantly different sound speed
because of scattering at the randomly rough interface. than plain seawater thereby altering local refraction
G is the Rayleigh roughness parameter defined as conditions. At the sea surface, the relative import-
ance of roughness versus bubble effects is not yet
G ¼ 2ks sin y ½9
resolved.
where k ¼ 2p/l is the acoustic wavenumber and s is
the rms roughness. Note that the reflection coefficient
Transmission-loss Data
for the smooth ocean surface is simply  1 (the
pressure-release condition is obtained from eqn [6] Figure 6 gives an example of transmission-loss vari-
by setting r2 ¼ 0) so that the rough-sea-surface ability in shallow water. The graph displays a
reflection coefficient
  for the coherent field is collection of experimental data from different shal-
R0 ðyÞ ¼ exp 0:5G2 . For the ocean bottom, the low-water areas (100–200 m deep) all over the
422 ACOUSTICS, SHALLOW WATER

40 SD = 50.0 m RD = 50.0 m
3200
1600
50

Frequency (Hz)
800
60 400
200
70
100
Transmission loss (dB)

50
80
25
(A)
90 SD = 50.0 m RD = 59.0 m
3200
100 1600

Frequency (Hz)
800
110 400
200
120
100

130 50
25
0 10 20 30 40
140
1 2 5 10 20 50 100 (B) Range (km)
Range (km)
Figure 7 Examples of frequency-dependent propagation
Figure 6 Transmission loss variability in shallow water. losses measured in two shallow-water areas: (A) Barents Sea,
(B) English Channel. Note the presence of an optimum frequency
of propagation between 200 and 400 Hz.
world. The data refer to downward-refracting sum-
mer conditions in the frequency band 0.5–1.5 kHz. over the entire frequency band. This second data set
Two features are of immediate interest. One is the represents typical propagation conditions for thick
spread of the data amounting to around 50 dB at sandy sediments with negligible shear-wave effects.
100 km and caused primarily by the varying bottom-
loss conditions in different areas of the world. The
Cutoff Frequency and Optimum Frequency
second feature is the fact that transmission is gener-
ally better than free-field propagation (20 log r) at A common feature of all acoustic ducts is the exist-
short and intermediate ranges but worse at longer ence of a low-frequency cutoff. Hence, there is a
ranges. This peculiarity is due to the trapping of critical frequency below which the shallow-water
energy in the shallow-water duct, which improves channel ceases to act as a waveguide, causing energy
transmission at shorter ranges (cylindrical versus radiated by the source to propagate directly into the
spherical spreading), but, at the same time, causes bottom. The cutoff frequency is given by,
increased boundary interaction, which degrades
transmission at longer ranges. cw
f0 ¼ rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 ½10
A second example of transmission-loss variability
in shallow water is given in Figure 7, where broad- 4D 1  ðcw =cb Þ2
band data from two different geographical areas are
compared. The data set in Figure 7A, was collected This expression is exact only for a homogeneous
in the Barents Sea in 60 m water depth. Note the high water column of depth D and sound speed cw over-
transmission losses recorded below 200 Hz, where lying a homogeneous bottom of sound speed cb. As an
energy levels fall off rapidly indicating that most of example, let us take D ¼ 100 m, cw ¼ 1500 m s1, and
the acoustic energy emitted by the source is lost to cb ¼ 1600 m s1 (sand–silt), which yields f0C11 Hz.
the seabed. It is believed that this excess attenuation Sound transmission in shallow water has the
is caused by the coupling of acoustic energy into characteristic frequency-dependent behavior shown
shear waves in the seabed. In contrast to the high- in Figure 7, i.e. there is an optimum frequency of
loss environment in the Barents Sea, Figure 7B shows propagation at longer ranges. Thus the 80 dB con-
a data set from the English Channel in 90 m water tour line extends farthest in range for frequencies
depth. Here propagation conditions are excellent around 400 Hz in Figure 7A and around 200 Hz in
 ACOUSTICS, SHALLOW WATER 423

Figure 7B, implying that transmission is best at these differences, and individual pulse shapes being
frequencies – the optimum frequencies of propa- modified due to frequency-dependent amplitude and
gation for the two sites. phase changes associated with each boundary re-
Optimum frequency is a general feature of ducted flection. From simple geometrical considerations, the
propagation in the ocean. It occurs as a result of time dispersion is found to be
competing propagation and attenuation mechanisms
at high and low frequencies. In the high-frequency

R 1
regime we have increasing volume and scattering loss Dt C 1 ½11
with increasing frequency. At lower frequencies the c̄ cosy
efficiency of the duct to confine sound decreases (the
cutoff phenomenon). Hence propagation and at- where R is the range between source and receiver, c̄ is
tenuation mechanisms outside the duct (in the sea- the mean sound speed in the channel, and y is the
bed) become important. In fact, the increased maximum propagation angle with respect to the
penetration of sound into a lossy seabed with de- horizontal. This angle will be determined either by
creasing frequency causes the overall attenuation of the source beamwidth or by the critical angle at the
waterborne sound to increase with decreasing fre- bottom (the smaller of the two). Since the dispersion
quency. Thus we get high attenuation at both high considered here is solely due to the geometry of the
and low frequencies, whereas intermediate frequen- waveguide, it is called geometrical dispersion.
cies have the lowest attenuation. It can be shown that An example of measured pulse arrivals over a 10 h
the optimum frequency for shallow-water propa- period in the Mediterranean is given in Figure 9.
gation is strongly dependent on water depth (foptp Note that the time-varying ocean (internal waves,
D1), has some dependence on the sound-speed currents, tides) causes strong signal fluctuations with
profile, but is only weakly dependent on the bottom time, particularly in the earlier part of the signal. The
type. Typically, the optimum frequency is in the time dispersion is 15–20 ms and at least four main
range 200–800 Hz for a water depth of 100 m. energy packets, each consisting of several ray arrivals
can be identified.

Signal Transmission in the Time

Domain
Even though underwater acousticians have tradi- 10
tionally favored spectral analysis techniques for
gaining information about the band-averaged energy 9
distribution within a shallow-water waveguide,
additional insight into the complication of multipath 8
propagation can be obtained by looking at signal
transmission in the time domain. 7
Geo-time (h)

Figure 8 indicates that the signal structure meas-

6
ured downrange will consist of a number of arrivals
with time delays determined by the pathlength
5

1
4
2 D′
S D
3 3
4
R
2

BOTTOM 1

Figure 8 Schematic of ray arrivals in shallow-water waveguide. 0

A series of multipath arrivals are expected, represented by 0 10 20 30 40
eigenrays connecting source (S) and receiver (R). The shortest Signal time (ms)
path is the direct arrival D followed by the surface-reflected arrival
D0 . Next comes a series of four arrivals all with a single bottom Figure 9 Measured pulse arrivals versus geo-time over a
bounce; then four rays with two bottom bounces as illustrated in 10 km shallow-water propagation track in the Mediterranean Sea.
the figure, followed by four rays with three bottom bounces, etc. The bandwidth is 200–800 Hz.
424 ACOUSTICS, SHALLOW WATER

Wave eq.

Range
WI NM independent

Range
Coupled Coupled Adiabatic dependent
Ray PE FD / FE
WI NM NM

Figure 10 Hierarchy of numerical models in underwater acoustics. WI, wavenumber integration; NM, normal modes; PE, parabolic
equation; FD, finite difference; FE, finite element.

Numerical Modeling Models that handle ocean variability in three spatial

dimensions have also been developed, but these
The advent of computers has resulted in an explosive
models are used less frequently than two-dimensional
growth in the development and use of numerical
versions because of the computational cost involved.
models since the mid-1970s. Numerical models have
become standard research tools in acoustic labora-
tories, and computational acoustics is becoming an Conclusions
evermore important branch of the ocean acoustic sci-
ence. Only the numerical approach permits an analysis The acoustics of shallow water has been thoroughly
of the full complexity of the acoustic problem. studied both experimentally and theoretically since
An assortment of models has been developed over World War II. Today the propagation physics is well
the past 25 years to compute the acoustic field in understood and sophisticated numerical models
shallow-water environments in both the frequency and permit accurate simulations of all processes (re-
time domains. Entire textbooks are dedicated to the flection, refraction, scattering) that contribute to the
development of theoretical and numerical formalisms complexity of the shallow-water problem. Sonar
which can provide quantitative acoustic predictions performance predictability, however, is limited by
for arbitrary ocean environments. Sound propagation knowledge of the controlling environmental inputs.
is mathematically described by the wave equation, The current challenge is therefore how best to collect
whose parameters and boundary conditions are relevant environmental data from the world’s enor-
descriptive of the ocean environment. As shown in mously variable shallow-water areas.
Figure 10, there are essentially five types of models
(computer solutions to the wave equation) to describe
See also
sound propagation in the sea: wavenumber integration
(WI); normal mode (NM); ray; parabolic equation Acoustic Scattering by Marine Organisms. Acous-
(PE) and direct finite-difference (FD) or finite-element tics in Marine Sediments. Acoustics, Arctic.
(FE) solutions of the full wave equation. All of these Acoustic Noise. Sonar Systems. Tomography.
models permit the ocean environment to vary with
depth. A model that also permits horizontal variations
in the environment, i.e. sloping bottom or spatially Further Reading
varying oceanography, is termed range dependent.
As shown in Figure 10, an a priori assumption Brekhovskikh LM and Lysanov YP (1990) Fundamentals
about the environment being range independent, of Ocean Acoustics, 2nd edn. New York: Springer-
leads to solutions based on spectral techniques (WI) Verlag.
or normal modes (NM); both of these techniques can, Etter PC (1996) Underwater Acoustic Modeling, 2nd edn.
London: E & FN Spon.
however, be extended to treat range dependence. Ray,
Jensen FB, Kuperman WA, Porter MB, and Schmidt H
PE and FD/FE solutions are applied directly to range- (2000) Computational Ocean Acoustics. New York:
varying environments. For high frequencies (a few Springer-Verlag.
kilohertz or above), ray theory, the infinite frequency Medwin H and Clay CS (1998) Fundamentals of
approximation, is still the most practical, whereas the Acoustical Oceanography. San Diego: Academic Press.
other five model types become more and more ap- Urick RJ (1996) Principles of Underwater Sound, 3rd edn.
plicable below, say, a kilohertz in shallow water. Los Altos: Peninsula Publishing.
 BIOACOUSTICS
P. L. Tyack, Woods Hole Oceanographic Institution, marine mammal. Marine mammal bioacoustics dur-
Woods Hole, USA ing this period was concerned primarily with iden-
Copyright & 2001 Elsevier Ltd. tifying which species produced which sounds heard
under water. Much of this research was funded by
naval research organizations because biological
sources of noise can interfere with military use of
sound in the sea.

Introduction
The term ‘bioacoustics’ has two different usages in
Elementary Acoustics
ocean sciences. Biological oceanographers use active Sound consists of mechanical vibrations that propa-
sonars to map organisms in the sea. Since they use gate through a medium. Sound induces movements
sound to detect marine life, they often call this ap- or displacements of the particles in the medium.
proach ‘bioacoustics’. The other sense of bioacous- Imagine a small sphere that expands to create a
tics involves studying how animals use sound denser area. This compression will propagate as
themselves in the ocean. This is the kind of bioa- particles are displaced in the direction of propa-
coustics covered in this article. gation. If the sphere then contracts, it can create an
Humans are visual animals, and we think of vision area of rarefaction, or lower density, and this also
as a primary distance sense because light carries so can propagate outward. These compressions or rar-
well in terrestrial environments. However, light is efactions can be expressed in terms of particle dis-
useful for vision under the sea only over ranges of placement or as a pressure differential.
tens of meters at best. Sound, on the other hand, Now imagine a sound source that creates a series
propagates extremely well in water – that is why of compressions and rarefactions that propagate
oceanographers so often select sound as a medium through the medium. A source with a purely sinus-
for exploring the sea or for communicating under the oidal pattern of compression and rarefaction would
sea. Sound propagates so well under water that a produce energy at only one frequency. The frequency
depth charge exploded off Australia can be heard in of this sound is measured in cycles per second. A
Bermuda. Just as we can hear well but emphasize sound that takes t seconds to make a full cycle has a
vision, so many marine mammal species see well but frequency f ¼ t1. Older references may refer to fre-
emphasize hearing. It is possible to gauge the relative quency in cycles per second, but the modern unit of
importance of audition versus vision in animals by frequency is the Hertz (Hz) and a frequency of 1000
comparing the number of nerve fibers in the auditory Hz is expressed as one kiloHertz (1 kHz). If a sound
versus the optic nerves. Of all marine mammals, the took 1 s for a full cycle, it would have a frequency of
cetaceans are the most specialized to use sound. 1 Hz. The wavelength of a tonal sound is the distance
Most cetaceans have auditory:optic ratios of fiber from one measurement of the maximum pressure to
counts that are 2–3 times those of land mammals, the next maximum. The speed of sound is approxi-
suggesting that audition is more important than vi- mately 1500 m s1 in water, roughly five times the
sion. Some cetaceans also use sound to echolocate. value in air, 340 m s1. The speed of a sound c is
Dolphins have a large repertoire of vocalizations related in a simple way to the frequency f and the
spanning frequencies from below 100 Hz to over 100 wavelength l by c ¼ lf. An under-water sound with
kHz, and dolphins have evolved high-frequency f ¼ 1 Hz would have l ¼ 1500 m; for f ¼ 1500 Hz,
echolocation similar to some human-made sonars l ¼ 1 m. Not all sounds have energy limited to one
and to the biosonar used by bats. frequency. Sounds that have energy in a range of
Marine mammals not only hear well, they are also frequencies, say in the frequency range between 2000
very vocal animals. The sounds of marine mammals and 3000 Hz (2 and 3 kHz), would be described as
are now well known, but the first recordings identi- having a bandwidth of 1 kHz.
fied from a marine mammal species were only made One can imagine a sound wave as a growing sphere
in the late 1940s. In the 1950s and 1960s, there was propagating outward from a compression or rar-
rapid growth in studies of how dolphins echolocate efaction generated by a point source. The initial
using high-frequency click sounds and of field studies movement of the source will have transmitted a cer-
associating different sounds with different species of tain amount of energy to the medium. If none of this

425
426 BIOACOUSTICS

energy is lost as the sound propagates, then it will be 36 Hz band and they last several tens of seconds. The
evenly diluted over the growing sphere. The acoustic pulses of finback whales, Balaenoptera physalus,
intensity is defined as the amount of energy flowing range roughly between 15 and 30 Hz and last on the
through an area over a unit of time. As the sphere order of 1 s. Particularly during the breeding season
increases in radius from 1 to r, the surface area in- in mid-latitudes, finbacks produce series of pulses in
creases to 4pr2. The intensity of a sound thus declines a regularly repeating pattern in bouts that may last
as the inverse of the square of the range from the many days.
source (r2). A sound in the middle of the ocean can These loud low-frequency sounds appear to be
be thought of as spreading in this way until it en- specialized for long-range propagation in the sea.
counters a boundary such as the surface or seafloor Absorption is negligible at the frequencies of these
that might cause reflection, or an inhomogeneity in sounds. While acoustic models predicted that these
the medium that might cause refraction. One fascin- sounds could be detected at ranges of hundreds of
ating acoustic feature of the deep ocean is that sound kilometers, it is only recently that this has been
rays propagating upward may refract downward as confirmed empirically. During the Cold War, the US
they encounter warmer water near the surface, and Navy developed bottom-mounted hydrophones to
downward-propagating rays will refract upward as locate ships and to track them. After the end of the
they encounter denser water at depth. When one is Cold War, these sophisticated systems were made
far from a sound source compared to the ocean available to biologists, who have worked with Navy
depth, the sound energy may be concentrated by re- personnel to locate and track whales over long ran-
fraction in the deep ocean sound channel. This sound ges, including one whale tracked for more than
can be thought of as spreading in a plane, to a first 1700 km over 43 days (Figure 1). These arrays have
approximation. In this case, sound intensity would proven capable of detecting whales at ranges of
decline as the inverse of the first power of the range, hundreds to thousands of kilometers, as was pre-
or r1. This involves much lower loss than the inverse dicted by the earlier acoustic models.
square spreading loss in an unbounded medium. The physics of sound can also help explain why
Sound spreading is a ‘dilution’ factor and is not a dolphins specialize in high-frequency sounds. Dol-
true loss of sound energy. Absorption, on the other phins can detect distant objects acoustically by pro-
hand, is conversion of acoustic energy to heat. The ducing loud clicks and then listening for echoes. The
attenuation of sound due to absorption is a constant clicks used by dolphins for echolocation have been
per unit distance, but this constant is dependent upon well described. The echolocation clicks of bottlenose
signal frequency. While absorption yields trivial ef- dolphins are very short (o100 ms), with a rapid rise-
fects at frequencies below 100 Hz, it can significantly time and a relatively broad bandwidth from several
limit the range of higher frequencies, particularly tens of kilohertz up to near 150 kHz (Figure 2A).
above 40 kHz or so. A 100 Hz sound can travel over
a whole ocean basin with little absorption loss, while
a 100 kHz sound would lose half its energy just 280° 290° 300°
traveling about 100 m.
40° 40°

Relating Acoustic Structure to

Biological Function of Marine
Mammal Calls
Understanding the physics of sound in the sea can 30° 30°
help us understand why animals make the kinds of
sound they do. For example, the calls of baleen
whales are low-frequency because they are adapted
for long-range propagation in the deep sea. Large
baleen whales have evolved abilities to produce and
20° 20°
to hear low-frequency calls well-suited for long-
280° 290° 300°
range communication. Blue whales and fin whales
produce the lowest-frequency signals of all marine
Figure 1 Track of a calling blue whale, Balaenoptera musculus,
mammals, so low that humans can barely hear as it swam 1700 km over 43 days. The whale was tracked using
them. The long moans of blue whales, Balaenoptera the Integrated Underwater Sound Surveillance System (IUSS) of
musculus, have fundamental frequencies in the 14– the US Navy. (From Figure 4.17 of Au et al. (2000).)
 BIOACOUSTICS 427

Kaneohe Bay
SL = 210 _227 dB re 1 μPa 22 kHz
58 kHz 30 kHz
100 kHz 40° 108 kHz
1 0 dB
0
Tank 250 μs _ 10 dB 30°
SL = 170 _185 dB re 1 μPa
_ 20 dB 20° 118 kHz
1 _ 30 dB 10°
0 250 μs
0°
1.0 Tank
_10°
Kaneohe Bay
|U (f )|

_ 20° 45 kHz
0.5
40 kHz
65 kHz _ 30°

0 0 150 μs
0 100 200
(A) Frequency (kHz) (B)

Figure 2 (A) Waveform and spectrum of echolocation clicks of bottlenose dolphins, Tursiops truncatus, in open ocean (Kaneohe
Bay) and in a tank. The spectrum of the click from the tank (indicated with a dashed line) has a lower frequency peak at 40 kHz. (B)
Beam pattern of Tursiops echolocation clicks. (mPa ¼ micropascal, reference for sound pressure measurements. SL ¼ source level.)
((A) from Figure 9.1, (B) from Figure 9.5 of Au et al. (2000).)

Captive dolphins in a reverberant pool make clicks 2pr ¼ 1500/(2p  0.005)B50kHz. This is within the
that are less loud and lower in frequency than dol- frequency range of dolphin echolocation clicks,
phins working on long-range echolocation in an which include energy up to about 150 kHz. This
open bay. The high-frequency components of these upper frequency is appropriate for detecting spher-
clicks are highly directional. If one moves 10 degrees ical targets with radii as small as 1.5 mm. The
off the axis of the beam, the click energy is halved hearing of dolphins is also most sensitive at fre-
and the click contains energy at lower frequencies quencies of roughly 50–100 kHz. If dolphins have a
(Figure 2B). The detection abilities of echolocating need to echolocate on rigid targets with sizes in the
dolphins are truly remarkable. For example, trained 1 cm range, that helps explain why their echo-
bottlenose dolphins can detect a 2.54 cm solid steel location system emphasizes these high frequencies.
sphere at 72 m, nearly a football field away.
The optimal frequency of a sound used for echo-
location depends upon the size of the expected tar-
get. Absorption imposes a penalty for higher
Marine Mammal Hearing
frequencies, but small targets can best be detected by In order to detect sound, animals require a receptor
short-wavelength, or high-frequency, signals. In the that can transduce the forces of particle motion or
nineteenth century, Lord Rayleigh solved the fre- pressure changes into neural signals. Most mech-
quency dependence of sound scattering from rigid anoreceptors in animals involve cells with hairlike
spherical targets; this is called Rayleigh scattering. A cilia on their surfaces. As these cilia move, the elec-
spherical target of radius r reflects maximum energy tric potential between the inside and the outside of
when the wavelength of the sound impinging on it the receptor cells changes, and this potential differ-
equals the circumference of the sphere, or when ence modifies the rate of nerve impulses that signal
l ¼ 2pr. The echo strength drops off rapidly from other parts of the nervous system.
signals with wavelength l42pr. Since l ¼ c/f, one Terrestrial mammals evolved an ear that is divided
can equate the two l terms to get c/f ¼ 2pr. The re- into three sections: the outer, middle, and inner ear.
lationship f ¼ c/2pr can be found by rearranging The outer ear and middle ear function in terrestrial
terms to calculate the optimal frequency for re- mammals to transduce airborne sound into vibra-
flecting sound energy off a spherical target of radius tions of a fluid the inner ear of mammals which
r. Higher frequencies than this would still be effective contains the cochlea, the organ in which sound en-
sonar signals, but frequencies below f would show a ergy is converted into neural signals. Sound enters
strong decrease in effectiveness with decreasing fre- the cochlea via the oval window and causes a
quency. A dolphin echolocating on rigid targets with membrane, called the basilar membrane, to vibrate.
a ‘radius’ of 0.5 cm should use a frequency fZc/ This membrane is mechanically tuned to vibrate at
428 BIOACOUSTICS

different frequencies. Near the oval window, the 140

basilar membrane is stiff and narrow, causing it to 130

Sound pressure level (dB)

vibrate when excited with high frequencies. Farther 120
into the cochlea, the basilar membrane becomes 110
wider and ‘floppier’, making it more sensitive to 100 Dolphin
lower frequencies. Sensory cells at different positions 90 Sea lion
along the basilar membrane are excited by different 80 Seal
frequencies, and their rate of firing is proportional to 70
the amount of sound energy in the frequency band to
60
which they are sensitive. Porpoise
50
Marine mammals share basic patterns of mam-
40
malian hearing but also have varying adaptations for 0.1 1 10 100 1000
listening under water as opposed to in air. All marine Frequency (kHz)
mammals other than sirenians, the sea otter, and
Figure 3 Audiograms from a variety of marine mammals:
cetaceans spend critical parts of their lives on land or
dolphin Tursiops truncatus; porpoise Phocoena phocoena; sea
ice and some phocid seals communicate both in air lion Zalophus californianus; seal Phoca vitulina.
and under water. The relative importance of hearing
in air and under water has been compared for three
pinniped species whose hearing has been tested in Mammalian hearing is designed to analyze the
both environments. The California sea lion (Zalo- frequency content of sound. Among mammals, dol-
phus californianus) is adapted to hear best in air; the phins have extraordinarily good abilities of dis-
harbor seal (Phoca vitulina) can hear equally well in criminating different frequencies. They can detect a
air and under water; and the northern elephant seal change of as little as 0.2% in frequency, which is
(Mirounga angustirostris) has an auditory system close to the resolution of human hearing.
adapted for under water sensitivity at the expense of
aerial hearing.
Vocalizations of Marine Mammals
The eardrum and middle ear in terrestrial mam-
mals functions to efficiently transmit airborne sound When terrestrial carnivores and ungulates invaded
to the inner ear where the sound is detected in a fluid. the sea, they encountered new constraints and op-
No such matching is required for an animal living in portunities for sensing signals. The sirenians, cet-
the water, and cetaceans, which are adapted ex- aceans, phocid seals, and the walrus (Odobenus
clusively for listening under water, do not have an rosmarus) evolved specializations for using sound to
air-filled external ear canal. The problem for cet- communicate under water and to explore the marine
aceans is isolating the ears acoustically, and the inner environment; other taxa, including the otariid
ear is surrounded by an extremely dense bone that is pinnipeds, sea otter (Enhydra lutra) and polar bear
isolated from the skull. High-frequency sound is (Ursus maritimus), vocalize mainly in air. As with
thought to enter the dolphin head through a thin hearing, cetaceans show the most elaborate and ex-
section of bone in the lower jaw and is conducted to treme specializations for acoustic communication
the inner ear via fatty tissue that acts as a waveguide. under water.
Hearing abilities have been tested for those species The best-known acoustic displays of marine
of marine mammals that can be held in captivity. mammals are the reproductive advertisement dis-
Figure 3 shows audiograms from a dolphin, por- plays called songs. The songs of humpback whales
poise, and several pinnipeds. As discussed above, are the best known advertisement display in the
dolphins have hearing specialized to hear very high cetaceans, but bowhead whales also sing. Male seals
frequencies up to ten times the upper limit of human of some species repeat acoustically complex songs
hearing. Seals have less acute hearing than do dol- during the breeding season. Songs are particularly
phins and they are less able to hear the highest fre- common among seals that inhabit polar waters and
quencies. The frequency range of hearing has never that haul out on ice. The songs of bearded seals,
been tested in baleen whales. Hearing is usually Erignathus barbatus, are produced by sexually ma-
tested by training an animal, and baleen whales are ture adult males and are heard most frequently dur-
so big that only a few have been kept for short per- ing the peak of the breeding season. Male walruses,
iods in captivity. However, both their low-frequency Odobenus rosmarus, also produce complex visual
vocalizations and the frequency tuning of their and acoustic displays near herds of females during
cochlea suggest they are specialized for low-fre- their breeding season. They use their lips to whistle,
quency hearing. and also produce loud sounds of breathing that are
 BIOACOUSTICS 429

audible in air when they surface during these dis- many ways of producing sounds. They produce
plays. When they dive, displaying males produce a gonglike impulse sounds using specialized pharyn-
series of pulses under water followed by bell-like geal sacs, and can even use their lips to whistle in air.
sounds. Antarctic Weddell seal males repeat under Odontocetes have well developed vocal folds in
water trills (rapid alternations of notes) during the the larynx, but most biologists argue that odonto-
breeding season. cetes produce sounds as air flows past the nasal plugs
Marine mammals also produce a broad variety of or phonic lips in the upper nasal passages
displays, including threat displays and recognition (Figure 4A). Mechanisms for sound production must
displays used for individual or group recognition. also match the acoustic impedance to the medium of
air or sea water, and they may function to direct
some sounds in a beam. The beam pattern of dolphin
clicks (shown in Figure 2B) stems from a complex
Mechanisms of Sound Production interaction of reflection from the skull and air sacs,
Most terrestrial mammals produce vocal sounds by coupled with refraction in soft tissues (Figure 4A).
vibrating vocal cords in the larynx. It is thought that There is a more detailed model of sound pro-
the polar bear and most pinnipeds make sounds duction for sperm whales (Physeter macrocephalus)
using similar mechanisms. Some adaptations for than for other cetacean species. Sperm whales have a
diving may affect vocalization mechanisms in large organ called the spermaceti organ, which lies
pinnipeds. Pinnipeds have a more flexible trachea dorsal and anterior to the skull (Figure 4B). Below
than do terrestrial mammals, so that air inside can the spermaceti organ is the ‘junk’, which is composed
compress during a dive, and they have a wider tra- of a series of fatty structures separated by dense
chea to allow higher rates of air flow. Most pinnipeds connective tissue. The primary vocalizations of
can vocalize under water without emitting bubbles; sperm whales are distinctive clicks comprising a
some species have sacs attached to the trachea or burst of pulses with equally spaced interpulse inter-
upper respiratory sac, but the role of these in vo- vals (IPIs). Bioacousticians suggest that these regular
calization has not been determined. Walruses have IPIs may result from reverberation within the

Figure 4 Functional anatomy of sound production in two odontocete cetaceans: (A) bottlenose dolphin Tursiops truncatus; (B)
sperm whale Physeter macrocephalus. (Adapted from Figures 1.4 and 3.1 of Au et al. (2000).)
430 BIOACOUSTICS

Dive #29
9 May 95
ES95-1 8 May 1995 2103:27 UT Dive #13 0002:32 UT
Frequency (Hz) Depth (m) 0
20
40
60
80

600
400
200
0

65 2 100
dB re 1 μPa /Hz
70
Heart rate (beats per min)

60

50

40

30
0 5 10 15 0 3
Time (min)

Figure 5 Data on dive profile, acoustic record, and acoustically determined heart rate from a tag on an elephant seal. The acoustic
record shows a vessel passing. The closest point of approach occurs at the minimum frequency of the ‘U’ shaped pattern in the
spectrogram at about 6 min. The heart rate differs little from the surrounding times or from a quiet period in a later dive from the same
seal. (mPa ¼ micropascal, reference for sound pressure measurements.) (Adapted from Figure 8 of Burgess et al. (1998).)

spermaceti organ. The frontal sac at the posterior which animal is vocalizing by using their own ears to
end of the spermaceti organ has been suggested as a locate the source of a sound and then looking for
potential reflector of sound and the distal sac as a movements associated with sound production.
partial reflector of sound at the anterior end Humans cannot locate sounds under water in the
(Figure 4B). The source of the sound energy in the same way that they locate airborne sounds. Fur-
click is thought to come from a strong valve (phonic thermore, marine mammals seldom produce visible
lips) in the right nasal passage at the anterior end of motions coordinated with sound production under
the spermaceti organ (Figure 4B). This sound pro- water. It is even more difficult to attempt behavioral
duction model suggests that some of the energy from observations on marine mammals during a dive
the first pulse within the click is transmitted directly when they are out of sight. The need for some
into the water. The remaining pulses are hypothe- technique to track behavior during a dive and to
sized to occur as some of the sound energy passes identify which cetacean produces which sound dur-
through the anterior reflector into the ocean at each ing normal social interaction has been discussed for
reflection there. over three decades. Two different approaches have
emerged: (1) passive acoustic location of sound
sources using an array of hydrophones; (2) recording
information about behavior and sound production
Methods for Bioacoustic Research by attaching a tag onto the animal.
It has been difficult to integrate visual observation of Acoustic location of vocalizing animals is a useful
social behavior with patterns of vocalization in method for identifying which animal is producing a
submerged mammals because it is difficult to identify sound. It involves no manipulation of the animals,
which animal within an interacting group produces a merely placement of hydrophones near them. In
sound under water. Biologists studying terrestrial some applications, animals may vocalize frequently
animals take it for granted that they can identify enough and be sufficiently separated that source
 BIOACOUSTICS 431

location data may suffice to indicate which animal the depth of the dive but also movement patterns
produces a sound. Tracks of continuously vocalizing such as the fluke beat and physiological parameters
finback and blue whales have been made using bot- such as heart rate. This information is useful to de-
tom-mounted hydrophones. Figure 1 shows a termine reactions of marine mammals to man-made
1700 km track of a blue whale that was tracked in noise, an issue of growing concern.
the early 1990s by US Navy personnel using arrays of
hydrophones initially developed to track submarines.
Bottom-mounted recording devices are proving See also
cheaper alternatives for biologists today.
Bioacousticians have also developed smaller, Acoustic Scattering by Marine Organisms. Sonar
portable hydrophone arrays that can be deployed Systems.
rapidly from a ship or from shore. These arrays have
been used to locate vocalizing finback whales, right
whales (Eubalaena glacialis), sperm whales, and Further Reading
several species of dolphins. Vertical hydrophone
arrays can in some settings be used to calculate the Au WWL (1993) The Sonar of Dolphins. New York:
range and depth of vocalizing whales. One classic Springer Verlag.
Au WWL, Popper AS and Fay R (eds.) (2000) Hearing by
configuration involves a linear horizontal array of
Whales and Dolphins. Springer Handbook of Auditory
hydrophones that is towed behind a ship. Signal Research Series. New York: Springer Verlag.
processing techniques allow one to determine what Burgess WC, Tyack PL, LeBoeuf BJ, and Costa DP (1998)
bearing a sound is coming from, and to reconstruct A programmable acoustic recording tag and first results
the signal from that bearing. Bioacousticians are only from free-ranging northern elephant seals. Deep-Sea
just beginning to explore how to use these techniques Research 45: 1327--1351.
in behavioral studies of whales. Kastak D and Schusterman RJ (1998) Low-frequency
The second technique does not require locating amphibious hearing in pinnipeds: Methods,
each animal within a group. If an animal carries a measurements, noise, and ecology. Journal of the
telemetry device that transmits acoustic data re- Acoustical Society of America 103: 2216--2228.
corded at the animal, then the device can record all Medwin H and Clay CS (1998) Fundamentals of
Acoustical Oceanography. New York: Academic Press.
vocalizations of the animals along with most every-
Miller P and Tyack PL (1998) A small towed beamforming
thing else it hears. This kind of tag can also record array to identify vocalizing resident killer whales
depth of dive, movement and orientation of the (Orcinus orca) concurrent with focal behavioral
tagged animal. However, it is difficult to telemeter observations. Deep-Sea Research 45: 1389--1405.
information through sea water, and marine mammals Rayleigh, Lord (1945) The Theory of Sound. New York:
might sense many of the signals one might want to Dover.
use for telemetry. These problems with telemetry Tyack P (1998) Acoustic communication under the sea.
have led biologists to develop recoverable tags that In: Hopp SL, Owren MJ, and Evans CS (eds.) Animal
record data while on an animal, but that need to be Acoustic Communication: Recent Technical Advances,
recovered from the animal in order for the data to be pp. 163--220. Heidelberg: Springer Verlag.
downloaded. Recently, biologists have had successful Tyack PL (2000) Functional aspects of cetacean
communication. In: Mann J, Connor R, Tyack PL, and
programs recovering such tags from many different
Whitehead H (eds.) Cetacean Societies: Field Studies of
kinds of marine mammal. Recoverable acoustic tags Dolphins and Whales, pp. 70--307. Chicago: University
may have scientific uses well beyond identifying vo- of Chicago Press.
calizations. Figure 5 shows acoustic and dive data Wartzok D and Ketten DR (1999) Marine mammal
sampled from an elephant seal. The tag was able to sensory systems. In: Reynolds JE III and Rommel SA
monitor both the acoustic stimuli heard by the (eds.) Biology of Marine Mammals, vol. 1,
whale, and orientation sensors monitored not just pp. 117--175. Washington, DC: Smithsonian Press.
 SEISMIC REFLECTION METHODS FOR STUDY OF
THE WATER COLUMN
I. Fer, University of Bergen, Bergen, Norway Primer on the Method
W. S. Holbrook, University of Wyoming, Laramie,
WY, USA Marine seismic reflection data are acquired by firing,
at constant distance intervals, a sound source of
& 2009 Elsevier Ltd. All rights reserved. relatively low-frequency (10–150 Hz) energy pro-
duced by an array of air guns towed several meters
beneath the sea surface. The data are recorded on
hydrophone cables, or streamers, which are typically
several kilometers long, contain 40–100 hydrophone
channels per kilometer, and are towed 3–8 m below
Introduction the sea surface. The pressure variations of the reflected
Recent work has shown that marine seismic re- sound signals are recorded by hydrophones installed
flection profiling, a technique commonly used by at regular intervals (usually 12.5 m) along the
geophysicists and geologists to image the Earth’s streamer. For a flat reflector, the point in the ocean’s
crust beneath the seafloor, can produce surprisingly interior at which a reflecting ray reaches its maximum
detailed images of thermohaline fine structure in the depth before reflecting (the ‘bottoming’ or ‘reflecting’
ocean. Seismic reflection profiling produces images point) occurs halfway between the source and the
by mapping the locations of ‘bounce points’ where receiver. Therefore, the subsurface sampling interval is
acoustic waves expanding out from the shot location half the hydrophone spacing (i.e., usually 6.25 m).
have reflected from subsurface interfaces where Reflections appear on the resulting records as
changes in material density and/or sound speed curved arrivals whose travel time increases hyper-
occur. Similar to sonar, the arrival time of a reflection bolically with distance from the source point. With
at a receiver, together with a measurement or esti- the exception of the direct arrival, refractions are not
mate of the sound-speed profile, is used to estimate returned to the streamer from within the water col-
the depth of the feature that generated the reflection. umn – only reflections. Shots are fired at intervals of
Within the ocean, density-compensating, fine-scale 50–150 m, creating a large volume of overlapping
(1–10-m thick) temperature-salinity contrasts (e.g., reflections on successive shot records. The vessel
temperature and salinity having opposing contri- maintains a constant speed along a straight track
butions to the density to keep the density constant) allowing for redundant sampling of a fixed point.
result in small changes in sound speed that produce Traces on the shot records are re-sorted into common
weak, but distinct, reflections. Recent studies image midpoint (CMP) records, which gather together all
returns that are specular reflections from laterally traces that have a common source–receiver midpoint.
continuous steps or sheets in the ocean’s tempera- The hyperbolic curvature of the reflections in each
ture-depth structure. The reflectors imaged in the CMP ensemble is then removed, and the resulting
ocean are 100–1000 times weaker than those from flattened reflections are summed (or stacked) together
the solid Earth below. However, by increasing the to create a single trace at each CMP. Summed traces
gain of the processing system, oceanic fine structure for each CMP are then joined into a stacked section.
can be made visible. Until recently, seismologists Each trace in the stack can be thought of as the re-
have been largely unaware of these weak reflections sponse that would be generated by a simple, single-
in their data and have not routinely processed returns channel geometry, that is, the wave field from a single
from the water column, since their focus is on the shot that travels downward, reflects off a boundary at
structure of the Earth. The discovery of seismic re- normal incidence, and is returned and recorded on a
flections from the water column and the ability to single hydrophone located at the source point. The
image large volumes of the ocean at full depth and at final processing step is migration, which accounts for
high lateral resolution opens up new possibilities for any distortions in the stacked section caused by the
probing the structure of the ocean with ‘seismic presence of dipping (nonhorizontal) reflectors. The
oceanography’. Spectacular images of thermohaline resulting migrated image resembles a snapshot cross
fine structure in the ocean have been produced from section of the acoustic impedance of the ocean and
features such as intrusions, internal waves, and underlying Earth at comparable horizontal and ver-
mesoscale eddies. tical sampling of about 5–10 m.

432
 SEISMIC REFLECTION METHODS FOR STUDY OF THE WATER COLUMN 433

Comparison with Other Marine however, high-frequency techniques are limited

Acoustic Technology to ranges of a few hundred meters. For the low-
frequency sound (10–150 Hz) of the seismic profiles,
Seismic reflection profiling differs substantially from, scattering theory is inappropriate, because the wave-
and thus complements, the high-frequency acoustic length of the sound is much larger than the length
imaging commonly used in oceanography. High- scale of common scatterers in the ocean.
frequency (typically above 10 kHz) ocean acoustics Unlike the long-range refracted sound typically
image scatterers in the water column – that is, dif- used in acoustic tomography in the ocean, the re-
fracting bodies with spatial dimensions on the order flections imaged with seismic reflection methods are
of the dominant wavelength of the sound source, such sensitive to the vertical derivative of the sound speed,
as zooplankton, bubble clouds or microstructure rather than the value of the sound speed itself. The
of temperature, salinity, or velocity. Scattering the- travel times of the reflections are sensitive to the
ories have been developed to quantify relationship average sound speed, but the reflection amplitudes
between microstructure intensity or zooplankton are sensitive to the vertical gradients. The frequencies
concentration to the backscatter strength. Many ob- generated by air gun sources are typically in the
servational programs have imaged internal waves in 10–150-Hz range, corresponding to vertical wave-
shallow water, often two-layer flows in straits that lengths of 10–150 m. The marine seismic reflection
show dramatic patterns of wave growth and breaking; method can resolve layers with thickness of about

0.0 25

Temp. (°C)
0.5 20
TWTT (s)

1.0 15

1.5 10
SE
NW 215 000 220 000 225 000 CMP 235 000 240 000 245 000
0

2
TWTT (s)

3
0.5
4

5
Se 1.0
aflo
or 20 km
6

1.5

Figure 1 (Top) Stacking sound speed, approximately equal to root-mean-square sound speed, in the ocean (color) and sea surface
temperature (SST, black) measured during a seismic survey in the Newfoundland Basin. Cold colors correspond to low sound speed
(minimum of B1440 m s  1); warm colors reflect higher sound speed (maximum of B1530 m s  1). Horizontal axis is the common-
mid-point (CMP) with 6.25-m spacing. The arrow marks the front between the Labrador current (NW of the front) and the North Atlantic
current (SE of the front), visible as an abrupt B5 1C increase in temperature at CMP 229 000. (Bottom) Stacked seismic section of
water column with a vertical exaggeration of 27. Vertical axis is two-way-travel time (TWTT) in seconds; the base of the section at 6 s
corresponds to a depth of B4500 m in the ocean. An intrusive feature is marked by an arrow. Box denotes the portion of the
profile depicted in the inset which shows slabs losing coherency at depths of B1000 m. Reproduced from Holbrook WS, Páramo P,
Pearse S, and Schmitt RW (2003) Thermohaline fine structure in an oceanographic front from seismic reflection profiling. Science 301:
821–824, with permission from AAAS.
434 SEISMIC REFLECTION METHODS FOR STUDY OF THE WATER COLUMN

one-quarter wavelength, that is, from several meters two-dimensional thermohaline structure of the ocean
to several tens of meter thick, provided that the in unprecedented detail (Figure 1). LC and NAC are
layers are separated by relatively sharp vertical gra- characterized by cold-fresh and warm-salty water
dients in sound speed as a result of predominant masses, respectively, separated by a front (marked by
temperature contrasts. For example, at surface an arrow in the upper panel of Figure 1, LC is on the
pressure, the sound speed increases by 4 m s  1 by NW side of the front) where strong thermohaline
increasing the temperature of seawater of 35 psu intrusions with 20- to 30-m-thick fine structure were
salinity from 5 to 6 1C. The oceanic fine structure is previously reported. Wedges of inclined reflections in
characterized by vertical scales of meters to tens of the vicinity of the front are representative of this
meters which are not directly affected by molecular phenomenon. A plausible mechanism that creates
diffusion. These scales are comparable to the dom- intrusive layers is double diffusion driven solely by
inant wavelength of the sound source. Hence, marine the factor 100 difference between the molecular dif-
seismic reflection data is ideally tuned to detect fusivity of heat and salt. Temperature-salinity struc-
oceanic fine structure and seismic theory developed ture in the intrusive layers is modified by the fluxes
for reflections from a laterally coherent layered induced by double diffusion when both salinity and
medium can be confidently applied. temperature increase or decrease with depth. The
strong tilt of the reflections matches the expected
inclination of density surfaces affected by double-
diffusive processes. Nearly horizontal, 0.5- to 5-km-
Observations long reflections in the warm waters of the NAC are
To date, published observations using seismic consistent with the weak tilt of lateral intrusions as a
oceanography come from three areas: the North result of double-diffusive mixing observed previously
Atlantic Front, the Kuroshio Front, and the Nor- in the NAC, whereas the isopycnal surfaces tilt more
wegian Sea. strongly in the vicinity of the front as a result of the
Seismic reflection data acquired in August 2000 geostrophic dynamical balance. Coherent slabs of
across the oceanographic front between the Labrador 100- to 300-m-thick reflections inclined at 1–41 from
current (LC) and the North Atlantic current (NAC) in the horizontal become progressively weaker and lose
Newfoundland basin yielded images showing the coherence with increasing depth, reaching down to

Distance (km)
W E
0 20 40 60 80 100 120 140 160
0 0

100 8 °C 100

200 200

300 300

400 400
Depth (m)

500 500

600 600
0 °C
700 700

800 800
Seafloor
900 900

1000 1000
°C
−2 0 2 4 6 8 10 12 14

Figure 2 Ocean temperature (color) overlain by seismic reflection data (black-and-white image). Thin solid lines are selected
isotherms contoured every 2 1C, derived from XBT and XCTD casts. Red stars and blue circles at top show locations of XBTs and
XCTDs, respectively. The warm (7–14 1C) AW is separated from the cold ( 0.5–2 1C) NSDW by a boundary layer delineated by
rapidly changing temperatures and strong seismic reflections. The top 140 m of the seismic profile, containing interference from the
direct arrival, has been muted. Reproduced from Nandi P, Holbrook WS, Pearse S, Páramo P, and Schmitt RW (2004) Seismic
reflection imaging of water mass boundaries in the Norwegian Sea. Geophysical Research Letters 31: L23311 (doi:10.1029/
2004GL021325). Copyright (2004) American Geophysical Union. Reproduced by permission of American Geophysical Union.
 SEISMIC REFLECTION METHODS FOR STUDY OF THE WATER COLUMN 435

about 1.5 s two-way-travel time (TWTT, approxi- vessel-mounted acoustic Doppler current profiler
mately 1100-m depth). (ADCP). Isotherms derived from XCTD measure-
Soon after the discovery of the ability to seismically ments across the Kuroshio current show the tem-
image oceanic thermohaline fine-structure in such perature characteristics of the frontal system with
detail, the first joint seismic reflection/physical ocean- intrusive features (Figure 5(a)). Two lines were ac-
ography study was conducted in the Norwegian Sea: a quired, one to the north (Line 1) and another to the
dense array of expendable bathythermographs (XBTs) south (Line 2) of the Kuroshio current core. The first-
and expendable conductivity-temperature depth order difference in the seismic images is that the
(XCTD) probes were deployed on several lines during reflectors become nearly horizontal in Line 2 with
the acquisition of seismic reflection data. The resulting increasing distance from the front, whereas more fine
data showed conclusively that the marine seismic re- structure and steeply inclined reflectors are common
flection method can map distinct water masses. At the in Line 1, consistent with the temperature section
survey site, the major water masses are the warm and (Figure 5(a)). With the purpose of determining the
saline Atlantic Water (AW) carried by the Norwegian best seismic source configuration to image the
Atlantic current and underlying cold and less-saline oceanic fine structure, the seismic lines of this survey
Norwegian Sea Deep Water (NSDW). Relatively were repeated with different air gun chamber vol-
nonreflective zones in the upper B400 m and below umes (20 l, 9 l, and a 3.4 l air gun of generator–
the 0 1C isotherm in Figure 2 correspond to AW and injector, GI, type). Upper portions of panels (b) to (c)
NSDW, respectively. Fine-scale temperature variability of Figure 5 show Line 1 sampled with decreasing air
measured by the XBT/XCTDs in the boundary be- gun source strength over a duration of 54 h. The
tween the two water masses affects the sound speed
and results in acoustic impedance contrasts manifested
Distance (km)
by the strong reflective zone in Figure 2. 121.0 121.5
Reflections generally follow isotherms derived 200 200

from the XBT/XCTD survey (Figure 2). A com-

parison of a temperature profile measured by an 300 300

XCTD to the coincident seismic reflection profile

reveals the sensitivity of the reflections to fine-scale 400 400
variability in temperature. Temperature fluctuations
derived by high-pass filtering the temperature profile
Depth (m)

500 500
for lengths smaller than 35 m, the dominant wave-
length of the range of the sound source used in this 600 600
2 4 6
study, match the seismic reflection signal (Figure 3). 450
Reflector amplitudes are enhanced where tempera- 700 700
ture anomalies are large. Even temperature fluctu- 500
ations as small as 0.04 1C, comparable to the 800 800
measurement accuracy of 70.02 1C, can be imaged
550
with seismic reflection techniques. 900 900
–1 0 1 2 3 4 5 6 7 8 9 –0.3 –0.2 –0.1 –0.0 0.1 0.2 0.3
Seismic profiles across the Kuroshio current, off Temp. (°C) Temp. (°C)
the Muroto peninsula southwest of Japan, revealed
fine structure continuous at a horizontal scale over °C
–2 0 2 4 6 8 10 12 14
40 km between 300- and 1000-m depth (Figure 4).
Successive seismic profiles at 2–3-day intervals, cor- Figure 3 (a) An unfiltered XCTD profile located at km 121.5
responding to about cross section snapshots every on the seismic profile (Figure 2) showing temperature from
180–450 km of the Kuroshio current moving with 200–900-m depth. (b) Short-wavelength temperature variations
1 m s  1, suggest that the lateral continuity of the fine (red), produced by removing wavelengths greater than 35 m from
the XCTD temperature profile, plotted with a 5-km-wide section of
structure at the Kuroshio current is not a transient
the reflection image surrounding the XCTD location (black and white
feature and can persist at least for 20 days. image). Background color scheme is ocean temperature, plotted as
The observations across the Kuroshio current in Figure 2. The seismic image has been shifted upward by 14 m to
motivated a joint seismic reflection and physical reflect the lag between the onset of energy and peak amplitude in
oceanographic survey in summer 2005 within the the seismic wavelet. Reproduced from Nandi P, Holbrook WS,
Pearse S, Páramo P, and Schmitt RW (2004) Seismic reflection
Kuroshio extension front east of Japan, where warm
imaging of water mass boundaries in the Norwegian Sea.
Kuroshio current meets cold Oyashio water. The Geophysical Research Letters 31: L23311 (doi:10.1029/
hydrography and ocean currents were sampled using 2004GL021325). Copyright (2004) American Geophysical Union.
XBT, XCTD, expendable current profiler (XCP), and Reproduced by permission of American Geophysical Union.
436 SEISMIC REFLECTION METHODS FOR STUDY OF THE WATER COLUMN

reflection patterns are consistent between consequent concert with the seismic reflection data show some
transects and correlate with hydrographic property similarities to the seismic images regarding the in-
changes in the water column. Depth-distance maps clined reflection patterns and their locations. Similar
of the ADCP backscatter intensity collected in to the observations in the Norwegian Sea (Figure 2),

(a) (d)
0.0 0 0.0 0
NW SE NW SE
300 300
0.5 0.5

Depth (m)
600 600
1.0 1.0
900 900

1.5 1.5
1200 1200
10 km 10 km
27 June 06 July
2.0 1500 2.0 1500
(e) 0.0 0
NW SE
Two-way time (s)

300
0.5

Depth (m)
3.0 600
1.0
900
Seafloor 1.5
1200
10 km
4.0 3000 09 July
2.0 1500
0.0 0
(f)
NW SE
300
0.5
5.0

Depth (m)
600
1.0
900

1.5
1200
6.0 4500
10 km
12 July
2.0 1500
(b) 0.0 0 (g) 0.0 0
NW SE NW SE
300 300
Two-way time (s)

0.5 0.5

Depth (m)
600 600
1.0 1.0
900 900

1.5 1.5
1200 1200
10 km 10 km
30 June 14 July
2.0 1500 2.0 1500
(c) 0.0 0 (h) 0.0 0
NW SE NW SE
300 300
Two-way time (s)

0.5 0.5
Depth (m)

600 600
1.0 1.0
900 900

1.5 1.5
1200 1200
10 km 10 km
03 July 17 July
2.0 1500 2.0 1500

Figure 4 Selected seismic profiles acquired across the Kuroshio current at 2–3-day intervals with date indicated on each panel. The
orientation of the line is northwest (left) to southeast (right). The vertical axis to the left is the TWTT and that on the right is the depth
calculated assuming an acoustic velocity of 1500 m s  1. The image for (a) 27 June is shown for TWTT 0–6 s, whereas other panels
are shown for 0–2 s. The reflection marked in (a) with arrows has a slope of B1/120. Reproduced from Tsuji T, Noguchi T, Niino H,
et al. (2005) Two-dimensional mapping of fine structures in the Kuroshio Current using seismic reflection data. Geophysical Research
Letters 32: L14609 (doi:10.1029/ 2005GL023095). Copyright (2005) American Geophysical Union. Reproduced by permission of
American Geophysical Union.
 SEISMIC REFLECTION METHODS FOR STUDY OF THE WATER COLUMN 437

N Latitude (degree) S
(a) 37.5 37.0 36.5 36.0 35.5 35.0
0

Depth (m) 200

400

600
Line 1
800
Line 2
1000
Temperature (°C)
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
XCTD-23
(b) N 2005/08/29 13:56 – 2005/08/30 09:00 XCP-03 S (e)
N 2005/09/01 03:15 – 2005/09/01 14:05 S
0 0

200 200

400 400
Depth (m)

600 600

Depth (m)
0

200 800 0°
0.5°
400 0° 1000 1.0°
V.E.~55 0
0.5°
600
1.0° 20 km
200
(c) N 2005/08/30 10:30 – 2005/08/30 23:30 S
0 400 V.E.~55
20 km
200 600
Temperature (°C)
(f) 2 4 6 8 10 0 5 10 15 20
400
Depth (m)

220 100
600 240
260 200
0
280
Depth (m)

300
200 300
320 400
400 0° 340
0.5° V.E.~55 500
360
1.0° 20 km
380
600 600
XCTD-26 400
XCP-05 2005/08/31 01:09 – 2005/08/31 20:00 Salinity (psu)
(d) 0N S 420 700
1.25 km 33.0 33.5 34.0 34.5 35.0
Temperature (°C)
200 4 6 8 10 0 5 10 15 20
(g) 100
160
400
Depth (m)

180 200
600 200
Depth (m)

0 220 300
240
200 400
260
280 500
400 0° V.E.~55 300
0.5° 320 600
1.0° 20 km
600 340 Salinity (psu)
700
1.25 km 33.0 33.5 34.0 34.5 35.0

Figure 5 Joint physical oceanography and marine seismic reflection survey within the Kuroshio extension front east of Japan. Two
seismic lines, indicated by boxes in (a) were sampled: Line 1 to the north and Line 2 to the south of the core of the Kuroshio current.
(a) Temperature section derived from XCTD measurements. Panels (b) to (e) show seismic reflection profile (gray scale, top) and
ADCP intensity (color, bottom) for (b–d) Line 1 and (e) Line 2. The air guns used are (b) 20 l, (c) 9 l, (d) 3.4 l, and (e) 20 l. Time period of
observation is indicated above each seismic profile. Red circles, green diamonds, and yellow diamonds are XCTD, XBT, and XCP
locations, respectively. Red waveforms on the seismic reflection profiles are synthetic seismograms calculated from XCTD profiles.
Expanded data from two blue rectangles marked in (b) and (d) are shown in (f) and (g), respectively. (f) Comparison of Line 1 seismic
profile (20 l air gun) and XCTD data marked with blue rectangle in (b). (left) Seismic profile overlain by a synthetic seismogram (red),
(right) temperature (blue) and salinity (red) profiles derived from XCTD measurements. (middle) Blow-up of temperature from the
rectangle to the right, which corresponds to the depth range of seismic profile to the left. (g) Same as (f) except using 3.4 l air gun at
location shown by blue rectangle in (d). Reproduced from Nakamura Y, Noguchi T, Tsuji T, Itoh S, Niino H, and Matsuoka T (2006)
Simultaneous seismic reflection and physical oceanographic observations of oceanic fine structure in the Kuroshio extension front.
Geophysical Research Letters 33: L23605 (doi:10.1029/ 2006GL027437). Copyright (2006) American Geophysical Union.
Reproduced by permission of American Geophysical Union.
438 SEISMIC REFLECTION METHODS FOR STUDY OF THE WATER COLUMN

Temperature (°C) Southward velocity (cm s−1) N S

0

200

400
Depth (m)

600

20 L
800
V.E.~38 0°
XCTD-23 XCP-03 0.5°
5 km
1.0°
1000
5 10 15 –20 0 20

Temperature (°C) Southward velocity (cm s−1) N S

0

200

400
Depth (m)

600

3.4 L
800
V.E.~38 0°
XCTD-26 XCP-05 0.5°
5 km
1.0°
1000
5 10 15 –40 –20 0 20

Figure 6 Comparison of (left) XCTD temperature, (middle) southward velocity recorded by XCP and (right) seismic reflection data.
XCTD and XCP station numbers indicated on panels are marked in Figures 5(b) and 5(d). The dashed blue horizontal lines indicate
abrupt temperature and velocity changes correlated with seismic reflections. Location of the XCP and XCTD measurements relative to
the seismic profile is shown by vertical black lines. All data are collected at Line 1 (see Figure 5(a)) with top panels using a 20 l air gun
and bottom panels using a 3.4 l air gun. These two measurements were separated in time by B19 h. Reproduced from Nakamura Y,
Noguchi T, Tsuji T, Itoh S, Niino H, and Matsuoka T (2006) Simultaneous seismic reflection and physical oceanographic observations
of oceanic fine structure in the Kuroshio extension front. Geophysical Research Letters 33: L23605 (doi:10.1029/ 2006GL027437).
Copyright (2006) American Geophysical Union. Reproduced by permission of American Geophysical Union.
 SEISMIC REFLECTION METHODS FOR STUDY OF THE WATER COLUMN 439

reflections are enhanced at the base of the thermo- seismograms constructed from the XCTD profiles.
cline (temperature range 4–8 1C). The synthetic profiles match the seismic observations
Given an observed CTD profile a synthetic re- well, lending confidence to the physical interpret-
flection pattern can be constructed to check whether ation of the reflectors (Figure 6).
the seismic interpretation of thermohaline fine Common in all seismic reflection images are un-
structure is credible or not. The sound-speed profile dulations of reflectors which cannot be attributed to
in seawater (calculated as a function of temperature, processing artifacts. It is well known that the velocity
salinity, and depth) is used to calculate acoustic im- shear of the internal wave field of the stratified ocean
pedance contrasts. Acoustic impedance is relatively displaces the isopycnals, isotherms, and irreversible
more sensitive to changes in sound speed than fine structure caused by mixing events. The undu-
changes in density. Reflection coefficients derived lating reflectors can be interpreted as the acoustic
from acoustic impedance are used to generate syn- snapshots of the fine-structure displacements due to
thetic seismograms. Red waveforms on the seismic the internal waves. This assertion was recently tested
reflection profiles of Figure 5 are such synthetic by digitizing selected reflectors where reflectors

(a)
300
Line 30
400

500
Depth (m)

600

700

800

900
95 100 105 110 115 120 125 130 135
Distance (km)
(b)
300
Line 32
400

500
Depth (m)

600

700

800

900
105 100 95 90 85 80 75 70 65
Distance (km)

Figure 7 Seismic sections for Lines (a) 30 and (b) 32 acquired across the isobaths in the Norwegian Sea. Data are displayed so that
peaks and troughs of reflections are red and blue, respectively, in the water column, and gray and black in the solid earth. Yellow lines
show reflectors picked and digitized by auto-tracking for spectral analysis shown in Figure 8. (a) Stacked seismic section of line 30
shows a clear progression toward the slope from smooth, continuous fine structure (inset left) to highly disrupted fine structure (inset
right). (b) Stacked seismic section of Line 32, similar to Line 30, shows fine structure changing from smoothly undulating to more
discontinuous pattern as the continental slope is approached. Reproduced from Holbrook WS and Fer I (2005) Ocean internal wave
spectra inferred from seismic reflection transects. Geophysical Research Letters 32: L15604 (doi:10.1029/ 2005GL023733). Copyright
(2005) American Geophysical Union. Reproduced by permission of American Geophysical Union.
440 SEISMIC REFLECTION METHODS FOR STUDY OF THE WATER COLUMN

roughly parallel isotherms (hence representative of marked change from smooth, continuous patterns to
isotherm displacements) and comparing the wave- more choppy reflections with high vertical displace-
number spectrum to the isopycnal displacement ex- ments as the continental slope is approached. The
pected from oceanic internal wave field. In the open spectra calculated away from the seafloor and hori-
ocean, frequency and wavenumber domain repre- zontally within 10 km of the continental slope are
sentations of internal wave energy show a remark- compared to the Garrett–Munk spectrum in
ably uniform energy spectrum that can be modeled Figure 8. For the open ocean reflectors, the agree-
by the so-called Garrett–Munk spectrum, which has ment with the oceanic internal wave horizontal wave
a frequency, o, and wavenumber, k, decay of energy number spectrum is within a factor of 2 down to
with a power law near o  2 and k  2. This spectrum wavelengths of about 30 m. Near the continental
is thought to be maintained by a cascade of energy slope reflector displacement spectrum suggests en-
from large scales and low frequencies to small scales hanced internal wave energy which could be caused
and high frequencies due to interaction of different by wave reflection and generation processes at the
wave packets. Two seismic lines presented in sloping boundary. At a length scale of about 300 m,
Figure 7 were acquired in the Norwegian Sea at there is an apparent change of slope from kx 2 of
6.25-m horizontal resolution. Reflectors roughly the internal wave spectrum to kx 5/3 of the inertial
following the isotherms derived from XBT meas- subrange of turbulence. Consistent with this infer-
urements in the vicinity are digitized (yellow traces in ence from the seismic profiles, recent measurements
Figure 7) to calculate the horizontal wavenumber of joint isopycnal displacement and turbulence-
(kx) power spectra of reflection displacements, z, dissipation rates indicate that turbulence subrange
obtained by removing a straight line fit over the ex- extends to surprisingly large horizontal wavelengths
tent of each reflector (typically 5–30 km). (4100 m). A fit of the spectral amplitude to a tur-
It is known that the internal wave characteristics bulence-model spectrum yields reasonable estimates
differ significantly in proximity to the sloping of the turbulence-dissipation rate in the water col-
boundaries where internal wave–boundary inter- umn, even when applied to horizontal scales of tens
actions lead to a variety of processes that can en- of meters, which could not be possible in vertical
hance the internal wave energy, shear, and mixing. profiles. The method is equally applicable to hori-
Consistently, the seismic reflection images show a zontal transects of seismic reflectors which follow
isopycnal displacements. A fit to the near-slope
spectrum of Figure 8 predicts diapycnal eddy diffu-
106
Open ocean sivity on the order B10  3 m2 s  1 (Figure 9).
Near slope
GM76-Tow
N Φ (kx) (m3 h−1)

104 101
–5/3

100
(N/N0)φ (cpm−1)

2
10
x

100
10–1 100 101 102  ( )
kx (cpkm)
( )
Figure 8 Horizontal wavenumber spectra of vertical
displacements inferred from digitized reflectors from open ocean
(squares), near slope (dots), all scaled by the average buoyancy kx (cpkm)
frequency, N, covering the representative depth range of the
chosen reflectors. Vertical bars are 95% confidence intervals. Figure 9 Reflection-slope spectrum, fzx, derived from the near-
Garrett–Munk tow spectrum is shown as a band for the observed slope spectrum shown in Figure 8. Spectrum is normalized by the
range of N ¼ (1.7 – 4.4)  10  3 s  1. The dashed line shows the buoyancy frequency, N ¼ 3.5  10  3 s  1, relative to N0 ¼ 5.3 
 5/3 slope of the inertial subrange of turbulence, for reference. 10  3 s  1. Dissipation rate of turbulent kinetic energy, e, is
Reproduced from Holbrook WS and Fer I (2005) Ocean internal estimated by a fit of the turbulent spectrum in the 4  10  3 –
wave spectra inferred from seismic reflection transects. 2  10  2 cpm wavenumber range. Eddy diffusivity is estimated
Geophysical Research Letters 32: L15604 (doi:10.1029/ using Kr ¼ 0.2e/N 2. Reproduced with permission from Klymak JM
2005GL023733). Copyright (2005) American Geophysical Union. and Moum JN (2007) Oceanic isopycnal slope spectra: Part II:
Reproduced by permission of American Geophysical Union. Turbulence. Journal of Physical Oceanography 37: 1232–1245.
 SEISMIC REFLECTION METHODS FOR STUDY OF THE WATER COLUMN 441

Conclusions N buoyancy frequency (s  1)

e dissipation of turbulent kinetic energy,
Standard multichannel marine seismic reflection pro- per unit mass (m2 s  3)
filing has the capability to image order 10-m vertical z vertical displacement (m)
scale sound speed and density fine structure in the zx horizontal derivative of z (  )
ocean, if processing gains are sufficiently high. Seismic o frequency (Hz)
reflection images visualize ocean fronts, water mass
boundaries, eddies, intrathermocline lenses, and ther-
mohaline intrusions. With the very dense lateral sam-
See also
pling of reflection profiling (order 10 m), exceptional
detail can be seen in the structure of oceanic reflectors Sonar Systems.
throughout portions of the water column where fine
structure is present. Interpreting marine seismic re-
flection data in terms of ocean fine structure and in- Further Reading
ternal motions is presently being investigated to
Holbrook WS and Fer I (2005) Ocean internal wave spectra
develop this technique into a tool that can produce
inferred from seismic reflection transects. Geophysical
useful and trusted information on properties of dy- Research Letters 32: L15604 (doi:10.1029/ 2005GL
namical interest to the physical oceanographers. A 023733).
basic physical understanding is achieved on the origin Holbrook WS, Páramo P, Pearse S, and Schmitt RW (2003)
of low-frequency reflections in the water column. Thermohaline fine structure in an oceanographic
Carefully analyzed horizontal transects of reflectors front from seismic reflection profiling. Science 301:
suggest that undulating patterns common in most 821--824.
seismic transects are a proxy to the isopycnal dis- Klymak JM and Moum JN (2007) Oceanic isopycnal slope
placements and can be used to remotely estimate the spectra: Part II: Turbulence. Journal of Physical
internal wave energy as well as turbulent kinetic en- Oceanography 37: 1232--1245.
ergy dissipation rate and mixing where ocean fine Nakamura Y, Noguchi T, Tsuji T, Itoh S, Niino H, and
Matsuoka T (2006) Simultaneous seismic reflection and
structure allows. In the near future, joint seismic-
physical oceanographic observations of oceanic fine
physical oceanography surveys are anticipated, which structure in the Kuroshio extension front. Geophysical
will provide ground-truth and further improvements Research Letters 33: L23605 (doi:10.1029/ 2006GL027
in interpreting and analyzing seismic reflection sections 437).
and finally allow for exploitation of the extensive Nandi P, Holbrook WS, Pearse S, Páramo P, and Schmitt
global archive of marine seismic reflection data for RW (2004) Seismic reflection imaging of water mass
interpreting the ocean dynamics. boundaries in the Norwegian Sea. Geophysical Research
Letters 31: L23311 (doi:10.1029/ 2004GL021325).
Páramo P and Holbrook WS (2005) Temperature contrasts
Nomenclature in the water column inferred from amplitude-versus-offset
analysis of acoustic reflections. Geophysical Research
k wavenumber (cycles per meter, cpm) Letters 32: L24611 (doi:10.1029/ 2005GL024533).
kx horizontal wavenumber (cycles per Tsuji T, Noguchi T, Niino H, et al. (2005) Two-dimensional
meter, cpm; or cycles per kilometer, mapping of fine structures in the Kuroshio current using
cpkm) seismic reflection data. Geophysical Research Letters 32:
Kr diapycnal eddy diffusivity (m2 s  1) L14609 (doi:10.1029/ 2005GL023095).
 SONAR SYSTEMS
A. B. Baggeroer, Massachusetts Institute of development of the Fessenden oscillator in 1914
Technology, Cambridge, MA, USA which eventually led to the development of fath-
Copyright & 2001 Elsevier Ltd. ometers, an acoustic system for measuring the
depth to the seabed. The French physicist/chemist
Paul Langevin was the first to detect a submarine
using sonar in 1918, motivated by the extensive
damage of German U-boats. Between World Wars I
and II both Britain and the US sponsored sonar
Introduction and Short History research, especially on transducers. The former was
Sonar (Sound Navigation and Ranging) systems conducted under the Antisubmarine Detection In-
are the primary method of imaging and communi- vestigation Committee, or ASDIC as sonar is still
cating within the ocean. Electromagnetic energy does often referred to within the British military, and the
not propagate very far since it is attenuated by either latter was performed at the Naval Research
absorption or scattering – visibility beyond 100 m is Laboratory.
exceptional. Conversely, sound propagates very well The re-emergence of the German U-boat stimu-
in the ocean especially at low frequencies; con- lated the modern era of sonars and the physics of
sequently, sonars are by far the most important sys- sound propagation in the ocean where major research
tems used by both man and marine life within the programs were chartered in the USA (Columbia,
ocean for imaging and communication. Harvard, Scripps Institution of Oceanography,
Sonars are classified as being either active or pas- Woods Hole Oceanographic Institution), UK, and
sive. In active systems an acoustic pulse, or more Russia. A very comprehensive summary was com-
typically a sequence of pulses, is transmitted and a piled by the US National Defense Research Council
receiver processes them to form an ‘image’ or to after World War II, which still remains a valuable
decode a data message if operating as a communi- reference (see Further Reading Section).
cation system. The image can be as simple as the The development of the nuclear submarine, both as
presence of a discrete echo or as complex as a visual an attack boat (SSN) or as a missile carrier (SSBN)
picture. The receiver may be coincident with the provided a major emphasis for sonar throughout the
transmitter – a monostatic system, or separate – a cold war. The USA, UK, Russia, and France all had
bistatic system. Both the waveform of the acoustic substantial research programs on sonar for many
pulse and the beamwidths of both the transmitter applications, but ASW certainly had a major priority.
and receiver are important and determine the per- The nuclear submarine could deny use of the oceans
formance of an active system. One typically associ- but could also unleash massive destruction with
ates an active sonar with the popular perception of nuclear missiles. With the end of the cold war, ASW
sonar systems. Many marine mammals use active now has a lower priority; however, the submarine still
sonar for navigation and prey localization, as well as remains the platform of choice for many countries
communication in ways which we are still at- since modern diesel/electric submarines operating on
tempting to understand. Many of the signals used by batteries are extremely hard to detect and localize.
modern sonars have some of the same features as Undoubtedly, the most extensively used reference was
those of marine mammals. compiled by Urick (1975), which is frequently refer-
Passive systems only receive. They sense ambient enced as a handbook for sonar engineers.
sound made by a myriad of sources in the ocean such While military operations have dominated the
as ships, submarines, marine mammals, volcanoes. development of sonars, they are now used exten-
These systems have been, and still are, especially sively for both scientific and commercial appli-
important in anti-submarine warfare (ASW) where cations. The use of fathometers and closely related
stealth is an important issue, and an active ping seismic methods provided much of the important
would reveal the location of the source. data validating plate tectonics. There is also a lot
The use of sound for detecting underwater of overlap between geophysical exploration for
objects was first introduced in a patent by
Richardson in June 1912 for the ‘sonic detection
of icebergs,’ 2 months after the sinking of the 1
Much of this material in the history has been extracted from
Titanic.1 This was soon followed by the Beyer, 1999.

442
 SONAR SYSTEMS 443

hydrocarbons and modern sonars. High resolution over which strong echos (called ‘returns’) from the
and multibeam systems are extensively used for previously transmitted pulse can be expected, so that
charting the seabed and its sub-bottom character- one return does not overlap and become confused
istics, fish finding, current measurements exploiting with another. With some systems with well confined
Doppler, as well as archaeological investigations. response durations, several pulses may be in transit
at the same time.
Active Sonar systems The ocean introduces three important components
before it is detected by a receiver.
The major components of an active system are indi-
cated in Figure 1. A waveform generator forms a pulse • There is the desired echo from the target itself.
or ‘ping’, which is then modulated, or frequency This may be a simple echo, especially if the target
shifted, to an operating frequency, fo which may be as is close, but it may also include many multipaths
low as tens of Hertz for very long-range systems, or as and/or modes as a result of reflections of the ocean
high as 1 MHz, for high resolution short-range im- surface and bottom as well as paths refracted
aging sonars. Next, the signal is often ‘beamformed’ completely within the ocean itself.
by an array of transducers, that focuses the signal in • The ocean is filled with spurious, or unwanted
specific directions either by mechanically rotating the reflectors which produce reverberation. The
array or by introducing appropriate time delays or dominant source of this is the sea bottom, but the
phase shifts. The signals are amplified and then con- sea surface and objects (e.g. fish) can be important
verted from an electrical signal to a sound wave by the as well. Typically, the bottom is characterized in
transmit transducers. Efficient transduction, the con- terms of a scattering strength per unit area
version of electric power to sound power, and even a insonified.
modest amount of directivity of the transmitter re- • Finally, the ocean is filled with ambient noise
quires that the transducer have dimensions on the which is created by both natural and man-made
scale of the wavelength of the operating frequency; sources. At low frequencies, 50–500 Hz, shipping
hence, low frequency transmitters are typically large tends to dominate the noise in the Northern
and not very efficient, whereas high frequency trans- Hemisphere, especially near shipping lanes. Wind
mitters are smaller and very efficient. and wave processes as well as rain can also be
The pinging rate, usually termed the pulse repe- important. In specific areas, marine life may be a
tition frequency (PRF) is determined by the duration very important component.

Figure 1 Active sonar system components.

444 SONAR SYSTEMS

The sonar receiver implements operations similar function of aspect angle; NL, is the level of the am-
to the transmitter. Hydrophones convert the acoustic bient noise at a single hydrophone; RL, is the re-
signal to an electric one whereupon it usually verberation which is determined by the area
undergoes some ‘signal conditioning’ to amplify it to insonified, the scattering strength, and the signal
an appropriate level. In modern sonars the signal is level; DIt, is the directivity index of the transmitter,
digitized, since most of the subsequent operations are which is a measure of the gain compared omnidir-
more easily implemented by digital signal processors. ectional radiation; AGr, is the array gain of the re-
Next, a receiver beamformer, which may be quite ceiver in the direction of the target (often this is
different from the transmitter, focuses energy arriv- described as a receiver); directivity index, DIr; DT, is
ing from specific directions for spatial imaging. This the direction threshold for a target to be seen on the
is also done either by mechanically steering the array output display (this can be a complicated function of
or by introducing time delays or phase shifts (if the the complexity of the environment and the sophisti-
processing is done in the frequency domain). The cation of an operator); TL, is the transmission loss,
signal is then usually demodulated to a low fre- i.e. the loss in signal energy as it transmits to the
quency band which simplifies further electronics and target and returns. If the SE40, then a target is
signal processing steps. Finally it is ‘pulse com- discernible on a display.
pressed,’ or ‘matched filtered,’ which is a process that The notation max(NL, RL) distinguishes the two
maximizes energy arriving at the travel time that important regimes for an active sonar. When
corresponds to the range to a target. The matched NL4RL, the sonar is operating in a noise-limited
filter is simply a correlation operation which seeks environment; conversely, when RL4NL the en-
the best replica of the transmitted signal among all vironment is reverberation-limited which is the case
the signal components introduced by the ocean. In in virtually all applications.
the simplest form of processing a sequence of ‘pings’ Reverberation is the result of unwanted echoes
is rastered, i.e. the echo time-series are displayed one from the sea surface, seafloor, and volume. In the
after the other, to construct an image. This is typical simplest formulation its level for surfaces both bot-
of a sidescan sonar system. In more sophisticated tom and top is usually characterized by a scattering
systems, especially those operating at low frequen- strength per unit area, so the level is given by the
cies where phase coherence can be preserved or ex- product of the resolved area multiplied by the scat-
tracted, a sequence of outputs from the pulse tering strength (or sum if using a decibel formulation).
compression filter is processed to form images. This Often, the Rayleigh parameter 2ps l sin(f) is used,
s

is typical of synthetic aperture sonars. In both types where ss is the rms surface roughness and f is the
of systems display algorithms, sometimes termed incident angle as a measure of when surface roughness
‘normalizers,’ are important for emphasizing certain becomes important, i.e. when the Rayleigh parameter
features by improving contrast and controlling the is greater than 1. Similarly with volume scattering, a
dynamic range of the output. scattering strength per unit volume is used.
The performance of an active sonar system is The operating frequency of a sonar is an important
captured in the active sonar equation; while im- parameter in its design and performance. The im-
perfect in its details, it is very useful in assessing gross portant design issues are:
performance. It is expressed in logarithmic units, or
decibels which are referenced to a standard level. For • Resolution. There are two aspects of resolution –
‘cross-range’ resolution and ‘in range’ resolution.
a monostatic system it is:
‘Cross-range’ resolution is determined by the di-
mensions of the transmit and receive apertures
SE ¼ SL  2  TL þ TS  maxðNL; RLÞ þ DIt relative to the wavelength, l, of the acoustic sig-
þ AGr  DT nal.4 It is given by Rl/L where R is the range and
L is the transmitter and/or receiving aperture.
where2,3 SE, is the signal excess at the receiver out- Higher resolution requires higher operating fre-
put; SL, is the source level referenced to a pressure quencies since these result in smaller Rl/L ratios.
level of 1 mPa; TS, is the target strength, which is a
‘In range’ resolution is determined by the band-
width of the signal and is given approximately by
2
There are many versions of the sonar equation and the no-
menclature differs among them (see Urick, 1975)
3 4
If the system is bistatic wherein the transmitter and receiver are The wavelength in uncomplicated media is given by l ¼ c=fo,
not colocated, then the sonar equation is significantly more com- where c is the sound speed, nominally 1500 m s1 and fo is the
plicated (Cox, 1989) operating frequency.
 SONAR SYSTEMS 445

c=2W, where W is the available bandwidth and c is

the sonar speed. Since W is usually proportional to
the operating frequency, fo, one tends to try to use
higher frequencies, which limits practical ranges.
Also the sonar channel is often very band-limited. As
a result in most sonar systems the ‘in range’ reso-
lution is typically significantly smaller than the
‘cross-range’ resolution, so care needs to be taken in
interpreting images.
• Maximum operating range. Acoustical signals can
propagate over very long ranges, but there are a
number of phenomena which can both enhance
and attenuate the signal power. These include the
geometrical spreading, the stratification of the
sound speed versus depth, absorption, and scat-
tering processes. The first two are essentially in- Figure 2 Attenuation of sound vs frequency.
dependent of frequency while the latter two have
strong dependencies. Figure 2 indicates the ab- number is larger than unity, the target strength
sorption loss in dB km1 of sound at 201 and 35 normalized by the presented area is typically be-
apt salinity.5 Essentially, the absorption loss factor tween 1 and 10. There are two additional features
increases quadratically with frequency. The two to consider: (i) large, flat surfaces lead to large
‘knees’ in the figure relate to the onset of losses returns, often termed specular, and (ii) in a bistatic
introduced by ionic relaxation phenomena. The sonar, the forward scattering, essentially the sha-
net implication is that efforts are made to min- dow, is determined almost solely by the inter-
imize the operating frequency so that it contrib- cepted target shape, often called Babinet’s
utes o10 dB of loss for the desired range, i.e principle.
aðfo ÞRo10 dBwhere aðfo ÞR is attenuation per unit
distance in dB. The net effect is a desire for higher operating fre-
quencies in order to stay in the Rayleigh region
The penalty of this, however, is a less directive where there is significant target strength.
signal, so it is more difficult to avoid contact with the Overall high frequencies produce better resolution
ocean boundaries and consequent scattering losses, and higher target strengths. However, at high fre-
especially the bottom is shallow-water environments. quencies acoustic propagation is more complicated
As a result, the actual operating frequency of a sonar and absorption limits the range.
is a compromise based on the sound speed profile, Table 1 indicates the operating frequencies of
the directivity of transmitter, receiver beamformers, some typical active sonars.
and desired operating range and isolation.
• Target cross-section. The physics of sound re- Sonar System Components
flecting from a target can be very complicated and
The components in Figure 1 all have significant im-
there are few exact solutions, most of which in-
pact on the performance of an active sonar. The signal
volve long, complicated mathematical functions.
processing issues are complex and there is a large
In addition, the geometry of the target can intro-
sonar-related literature as well as radar where the
duce a significant aspect dependence. The im-
issues are similar. The essential problem is to separate
portant scaling number is 2pa/l where a is a
a target amidst the reverberation and ambient noise in
characteristic length scale presented to the in-
either of two operating realms – ‘noise-limited’ and
coming sound wave. Typically, if the number is
‘reverberation’-limited environments. In a ‘noise-
less than unity, the reflected target strength de-
limited’ environment the ambient background noise
pends upon the fourth power of frequency, so the
limits the performance of a system, so increasing the
target strength is quite small; this is the so-called
transmitter output power improves performance.
Rayleigh region of scattering. Conversely, if this
A ‘reverberation-limited’ environment is one where
the noise is composed of mostly unwanted reflections
from objects other than the target, so increasing the
5
db km1 represents 10 log of the fractional loss in power per transmitted power simply increases both the target
kilometer representing an exponential decay versus range. and reverberation returns simultaneously with no
446 SONAR SYSTEMS

Table 1 Features of some typical active sonars

Sonar system Operating frequency range Wavelength Nominal range

Long-range, low frequency 50–500 Hz 30–3 m 1000 km

Military ASW sonars 3–4 kHz 0.5–0.75 m 100 km
Bottom-mapping echosounders 3–4 kHz 0.5–0.75 m vertical
High-resolution fathometers 10–15 kHz 15–10 cm vertical
Acoustic communications 10–30 kHz 15–5 cm 10 km
Sidescan sonar (long-range) 50–100 kHz 6–1.5 cm 5 km
Sidescan sonar (short range) 500–1000 kHz 3–1.5 mm 100 m
Acoustic localization nets 10–20 kHz 15–5 cm vertical
Fish-finding sonars 25–200 kHz 6–1 cm 1–5 km
Recreational 100–250 km 15–6 mm vertical

net gain in signal to noise ping. Most active sonars can distinguish it against a fixed reverberant back-
operate in a reverberation limited environment. ground. The ability to resolve Doppler frequency
Effective design of an active sonar depends upon depends upon the duration of a signal, with a de-
controlling reverberation through a combination of pendence of 1/duration, so good ‘Doppler gating’
waveform design and beamforming. waveforms are long.
There has been a lot of research on the topic of
Waveform design There are two basic approaches optimal waveform design. Ideally, one wants a
to waveform design for resolving targets – ‘range waveform which can resolve range and Doppler
gating’ and ‘Doppler gating.’ The simplest approach simultaneously which implies long duration, and
to ‘range gating’ is a short, high powered pulse. This wide bandwidth. These requirements are difficult to
essentially resolves every reflector and an image is satisfy simultaneously.
constructed by successive pulses and then the
returns are rastered. While this is the simplest Beamforming Both the transmitter and the
waveform it has limitations when operating in receiver beamformers provide spatial resolution for
environments with high noise and reverberation the sonar system. The angular resolution in degrees
levels, since the peak power of most sonars, both is approximately DyE60l=L, where l is the
man-made and marine mammal, is limited. This wavelength and L is the aperture length. Since
shortcoming can often be mitigated by exploiting acoustic wavelengths are large when compared with
bandwidth (resolution a/b). This has led to a large optical wavelengths, the angular resolutions tend to
literature on waveform design with the most be large especially at low frequencies.7 Beam-
popular being frequency modulated (FM) and coded formers, often termed array processors in receivers,
(PRN) signals. With these signals6 the center have been an important research topic for several
frequency is swept, or ‘chriped’ across a frequency decades with the advent of digital signal processing
band at the transmitter and correlated, or which permitted increasingly more sophistication,
‘compressed’ at the receiver. This class of signals is especially in the realm of adaptive methods. One of
commonly used by marine mammals including the simplest transmit beamformers consists of a line
whales and dolphins for target localization. array of transducers each radiating the same signal.
‘Doppler gating’ is based on differences in target The simple receiver and also a line of transducers
motion. A moving target imparts a Doppler shift to adds all the signals together. This resolves the paths
the reflected signal which is proportional to oper- perpendicular or broadside to the array. If one
ating frequency and the ratio v=c, where v is the wants to ‘steer’ the array, or resolve another
target speed and c is the sound speed. ‘Doppler gat- direction, the array must be mechanically rotated.
ing’ is particularly useful in some ASW contexts since This method of beamforming is still used by many
it is difficult to keep a submarine stationary, thus a systems since it is quite robust. Another simple
properly designed signal, one that resolves Doppler, beamformer is a planar array of transducers.

6 7
Pseudo Random Noise (PRN) are coded signals which appear Sonars with angular resolutions of 11 are generally considered
to be random noise. Well designed signals have useful mathemat- to have high resolution. Compare this with that of the human eye
ical constructs which led to good outputs at the output of the pulse with a nominal diameter of 4 mm and the wavelength of light in
compression, or matched filter, processor. the visible region is 0.4 mm leading to a resolution scale of 0.1 ms.
 SONAR SYSTEMS 447

Digital signal processing has led to more sophis- for deep water operate at a 12 kHz center frequency.
ticated array processing, especially for receivers. The transmit beam is produced by a linear array
Beamformers which steer beams electronically by running fore to aft along the bottom of the ship,
introducing delays, or phase shifts, shape beams to thwartships beam, which produces a swath which
control sidelobes, place nulls to control strong re- resolves the seafloor along track (Figure 4). The re-
flectors, and reduce jamming are now practical be- ceiver array is oriented port to starboard. The signals
cause these features can be practical electronically from this array are beamformed electronically, so the
rather than mechanically. seafloor is resolved port to starboard within the
transmitted swath, since the patch is the product of
the transmit and receive beamwidth. This configur-
Examples of Active Sonar Images ation allows two-dimensional resolution with two
This section describes two examples of sonars used linear arrays instead of a full planar array. The
for mapping seafloor bathymetry. In the first the depths from each of the multibeams are measured by
sonar is carried on an unmanned underwater vehicle combining the travel time and the ray refraction
(UUV) close to the seafloor. The operating frequency from the sound speed profile to obtain a depth.
is 675 kHz and the beam is mechanically steered Subsequent processing edits anomalous returns and
from port to starboard as well as fore and aft as interpolates all the data to generate the contour map.
the UUV proceeds along its track, so the beams Active sonar systems with additional features have
are steered forward and directly below the UUV been developed for special applications, but they all
(Figure 3). The onset time of the first echo return is use the basic principles described above.
the parameter of interest. It is converted to the depth
of the seafloor after including the vehicle position,
Passive Sonars
the direction of the beam and possibly refraction
effects in the water itself. Usually straight-line Passive sonars that only listen and do not transmit
acoustic propagation is assumed. are used in a variety of applications including the
The signals are combined to generate a high military for antisubmarine warfare (ASW), tracking
resolution map of the seafloor. The processing to and classification of marine mammals, earthquake
achieve this includes editing for spurious responses, detection, and nuclear test ban monitoring.
registration of the rasters or images from successive Since the signals are passive there is no pulse com-
transmission using the navigation sensors on the pression, or matched filtering, so a passive sonar
UUV (or more generally any vehicle) and normal- design primarily focuses upon the ‘short-term’ fre-
ization to improve the contrast so that weak features quency wavenumber spectrum, or the directional
can be detected amidst strong ones. spectrum and the power density spectrum and how it
The second example of an active sonar is a mul- evolves in time. The data are nonstationary and in-
tibeam bathymetric mapper. Most of these systems homogeneous, but many of the processing algorithms

Figure 3 Image with forward-looking and down-looking sonar. (Figure courtesy of Dr Dana Yoerger, Woods Hole Oceanographic
Institution.)
448 SONAR SYSTEMS

Depth (m)

Figure 4 High resolution bathymetric map of the seafloor near the Mid-Atlantic Ridge: (A) contour map; (B) isometric projection
(from top-right). (Figures courtesy of Dr Brian Tucholke, Woods Hole Oceanographic Institution.)

are based upon stationary and homogeneous as- important, and it is necessary to separate the array
sumptions; hence the term ‘short-term.’ The per- gain, AGr into two terms, or:
formance of a passive system is characterized by the
passive sonar equation:
AGr ¼ AGr;n  SGDs
SE ¼ SL  TL  NL þ AGr  DT
where AGr,n is the array gain against the ambient
where the terms are essentially the same as for an noise and SGDs is the signal gain degradation due to
active sonar. In some applications the arrays are so lack of coherence. SGDs ¼ 0 for a signal that is co-
large that the coherence of the received signal is herent across the entire array.
 SONAR SYSTEMS 449

Passive Sonar Beamforming parameter for signal processing schemes and display
formats includes time (the epoch for data processing,
The signals received by the sonar’s hydrophone are
T), angle (azimuth and elevation), and frequency (the
preconditioned, which might include editing bad
spectral content of the data) (Figure 5).
data channels, calibration, and filtering. They are
then beamformed, either in the time domain by
Bearing-time Recording
introducing delays to compensate for the travel time
across the array, or in the frequency domain. With Bearing time processing takes the beam outputs over
digital signals the former usually requires upsam- a specified frequency band and plots the output
pling or interpolation of the data to avoid distortion. versus time. Two modes of processing are often used:
The latter is accomplished by FFT (fast Fourier (i) energy detection, which forms an average of the
transforms), phase shifting to compensate for the beam outputs, or (ii) cross-correlation detection,
delays, and then IFFT (inverse fast Fourier trans- where the array is split at the beamformer and then
forming). Frequency domain beamforming allows the two outputs are cross-correlated versus the dir-
simpler implementation of adaptive techniques that ection. The processing is often classified according to
are useful in cases where the ambient field has many the width of the band used, passive broadband (PBB)
discrete components. Adaptive algorithms form or passive narrowband (PNB).
beams with notches, i.e. poor response in the dir- The data for each time epoch are normalized to
ection of interferers, thereby suppressing them. improve the contrast for signals of interest and each
Many algorithms have been designed to accomplish raster is plotted. Over a sequence of epochs, the
this, but the MVDR (minimum variance distortion directional components in the ambient field which
filter – first introduced by Capon) and related algo- are associated with shipping are observed. By man-
rithms have been used most extensively in practice. euvering the array one can triangulate to obtain a
range to each source as well.
Passive Sonar Display Formats
Low Frequency Acoustic Recording and Analysis
The output of the beamformer is a time-series for (LOFAR) grams Once a bearing or direction of
each beam. In certain applications the time-series interest has been determined, spectral analysis of
itself may be of interest, however, in most cases the the selected beam is used to produce a LOFAR
time-series is further processed to assist in extracting gram, which is a plot of the signal spectrum for each
weak signals from the background noise. The analysis epoch, T, versus time. By examining the

Figure 5 Passive sonar with modes of displaying output.

450 SONAR SYSTEMS

features of the LOFAR gram, such as frequency, it at specific range (rather than at infinite range
peaks, harmonic signals, and their changes as a which is the case for plane waves). When focused at
function of time, the source of the signal and some short ranges, long-range targets are attenuated. This
of its characteristics, such as speed, can be deduced. introduces the focal range as another parameter for
As in the case of the bearing-time display, the displaying the sonar output.
LOFAR gram is normalized to enhance features of At long ranges and low frequencies or in shallow
interest. water acoustic signals have complex multipath or
multimode propagation which leads to coherent
FRAZ displays Frequency-azimuth, or FRAZ interference along a vertical array or a very long
displays plot the spectral content as a function of horizontal array. The appropriate array processing is
frequency and azimuth for each epoch, T. Often a to determine the full field Green’s function for the
number of FRAZ outputs are averaged to improve signal and match the beamforming to it, a technique
signal to noise. FRAZ displays allow connection of known as matched field processing (MFP). MFP re-
the spectral content of a single source along a given quires knowledge of the sound speed profile along
bearing since the display contains a number of lines the propagation path, so its performance depends
for each source at a given azimuth. As with the upon the accuracy of environmental data. It is a
previous display, normalization algorithms to computationally intensive process, but has the
improve contrast are usually employed. powerful advantage of being able to resolve both
target depth and range, as well as azimuth. MFP is an
Trackers The objective of an ASW passive sonar is active subject of passive sonar research.
to detect, classify and track sources of radiating
sound. Trackers are used to follow sources through
direction and frequency space. There are a number
of tracking algorithms build upon various signal See also
models. Some separate the direction and frequency Acoustic Scattering by Marine Organisms. Acous-
dimensions while others are coupled models. Since tics, Shallow Water.
the ambient field can often have a number of
sources, some targets and some interferers and since
there are complicated propagation effects, the
design of trackers is difficult. Most involve some Further Reading
form of Kalman filtering and some have integral
propagation models. Baggeroer AB (1978) Sonar signal processing. In: Oppenheim
AV (ed.) Applications of Digital Signal Processing.
Englewood Cliffs NJ: Prentice-Hall.
Advanced beamforming Baggeroer A, Kuperman WA, and Mikhalevsky PN (1993)
An overview of Matched Field Processing. IEEE
Most passive sonars are based upon a plane wave Journal of Oceanic Engineering 18: 401--424.
model for the ambient field components. Plane wave Beyer RT (1999) Sounds of Our Times: 200 Years of
beamforming is robust, has several computational Acoustics. New York: Springer-Verlag.
advantages, and is well understood. However, in Capon J (1969) High resolution frequency-wavenumber
many sonars this model is not adequate because the spectrum analysis. Proceedings of the IEEE 57:
arrays are so long that wavefront curvature becomes 1408--1418.
an issue or the arrays are used vertical where acoustic Clay C and Medwin H (1998) Fundamentals of Acoustical
propagation introduces multipaths. Advanced Oceanography. Chestnut Hill, MA: Academic Press.
Cox H (1989) Fundamentals of bistatic active sonar.
beamforming concepts can address these issues.
In: Chan YT (ed.) Underwater Acoustic Data
Wavefront curative becomes important when the Processing. Kluwer Academic Publishers.
target is in the near field of the array (usually given National Defense Research Council (1947) Physics of
by the Fresnel number 2L2 =l) which is a con- Sound on the Sea: Parts 1–6. National Defense Research
sequence of either very long arrays or high frequen- Council, Division 6, Summary Technical Report.
cies. A quadratic approximation to the curvature is Urick RJ (1975) Principles of Underwater Sound for Engi-
often used and the array is actually designed to focus neers, 2nd edn. McGraw-Hill Book Co.
 TOMOGRAPHY
P. F. Worcester, University of California The short spatial scales mean that moored sensors
at San Diego, La Jolla, CA, USA must be too closely spaced to be practical. Munk and
Copyright & 2001 Elsevier Ltd. Wunsch proposed that the travel times of acoustic
signals propagating between a relatively small num-
ber of sources and receivers could be used to map the
evolving temperature field in the intervening ocean.
Their work led directly to the first 3D ocean acoustic
tomography experiment, conducted in 1981. In spite
Introduction of the marginal acoustic sources that were available at
Ocean acoustic tomography is a method for acoustic the time, the experiment showed that it was possible
remote sensing of the ocean interior that takes ad- to use acoustic methods to map the evolving mesos-
vantage of the facts that the propagation of sound cale field in a 300 km by 300 km region (Figure 1).
through the ocean is sensitive to quantities of It was quickly realized, however, that the integral
oceanographic interest, such as temperature and measures provided by acoustic methods are powerful
water velocity, and that the ocean is nearly transpar- tools for addressing certain types of problems, in-
ent to low-frequency sound so that signals can be cluding the measurement of integral quantities such
transmitted over long distances. The procedure is (i) as heat content, mass transport, and circulation.
to transmit acoustic signals through the ocean, (ii) to Acoustic measurements of the integrated water vel-
make precise measurements of the properties of the ocity around a closed contour, for example, provide
received signals, e.g., travel times, and (iii) to use in- the circulation, which is directly related to the areal-
verse methods to infer the state of the ocean traversed average vorticity in the interior by Stokes’ theorem.
by the sound field from the measured properties. The Vorticity is difficult to measure in other ways. The
characteristics of the ocean between the sources and suppression of small-scale variability in the spatially
receivers are determined, rather than the character- integrating acoustic measurements also makes them
istics of the ocean at the instruments as is the case for well suited to measure large-scale phenomena, such
conventional thermometers and current meters. as the barotropic and baroclinic tides. Finally, the
Ocean acoustic tomography has a number of at- integral measurements provided by the acoustic data
tractive attributes. It makes possible the rapid and can be used to test the skill of dynamic models and to
repeated measurement of ocean properties over large provide strong model constraints.
areas, taking advantage of the speed with which Acoustic scattering due to small-scale oceanic
sound travels in water (B1500 m s1). It permits the variability (e.g., internal waves) causes the properties
monitoring of regions in which it is difficult to install of the received acoustic signals to fluctuate. Although
instruments to make direct measurements, such as these fluctuations limit the precision with which the
the Gulf Stream or the Strait of Gibraltar, using signal characteristics can be measured and with
sources and receivers on the periphery of the region. which oceanic parameters such as temperature and
Acoustic measurements are inherently spatially inte- water velocity can be inferred, it was soon realized
grating, suppressing the small-scale variability that that measurements of the statistics of the fluctuations
can contaminate point measurements and providing can be used to infer the statistical properties of the
direct measurements of horizontal and vertical small-scale oceanic variability, such as internal-wave
averages over large ranges. Finally, the amount of energy level, as a function of space and time.
data grows as the product (S  R) of the number of Summarizing, tomographic methods can be used
acoustic sources S and receivers R, rather than lin- to map the evolving ocean, to provide integral
early as the sum of the number of instruments (S þ R) measures of its properties, and to characterize the
as is the case for point measurements. statistical behavior of small-scale oceanic variability.
Ocean acoustic tomography was originally intro-
duced by Munk and Wunsch in 1979 to address the Ocean Acoustics: The Forward
difficult problem of observing the evolving ocean
Problem
mesoscale. Mesoscale variability has spatial scales of
order 100 km and timescales of order one month. The ‘forward’ problem in ocean acoustics is to
The short timescales mean that ships move too compute the properties of the received signal given
slowly for ship-based measurements to be practical. the sound-speed C(x, y, z) and current v(x, y, z) fields

451
452 TOMOGRAPHY

R CTD I 66 _ 85 76 79 82 85
S R -6
-5 -6 -4 -2
-2 -2 -8 -6
S
300 km

-8
R -7 -10 -12
-10 -9
-8 -4 -7
R -4
S -4 -5 -8 -6
-5 -5
R -5
S
300 km
88 91 94 97 100 103
-3 -1 -1
-2 0 -3 -1
-4 -1 -2 -1
-6 -3 -4 -5 -3
-5 -3
-5 -1
-8 -7 -6 -7 -5

-3 -3
-3 -2
-5
-4 -1

106 109 112 115 118 CTD II 120 _ 139

-2 1
0 -2
-4 0 -2 -6 -4
-2 -6 -4 -2 -4 -2 -2 -1 1
-6 -6
-4
-3

-5
-4 -5
-2 -4
-4 -4 -7 -3

Figure 1 The 1981 tomography experiment. The first panel shows the geometry, with four source (S) and five receiver (R) moorings
on the periphery of a 300 km by 300 km region in the north-west Atlantic Ocean. Subsequent panels show the sound-speed
perturbations at 700 m depth derived from the acoustic data at 3-day intervals, with regions of high uncertainty shaded. The initial and
final panels are derived from two ship-borne conductivity-temperature-depth (CTD) surveys, each of which required about 20 days to
complete. The label on each panel is the year day in 1981. The contour interval is 1 m s1 (0.21C). Adapted from Cornuelle B, Wunsch
C, Behringer D, et al. (1985) Tomographic maps of the ocean mesoscale. Part I: Pure acoustics. Journal of Physical Oceanography
15: 133–152.

between the source and receiver. Acoustic remote perform full-field inversions that use the time series
sensing of the ocean interior requires first a full of intensity and phase for the entire received signal as
understanding of the forward problem, i.e., of observables. Unfortunately, neither normal modes
methods for finding solutions to the wave equation. nor the intensities and phases of the received signal
A variety of approaches are available to do this, in- are robust in the presence of internal-wave-induced
cluding geometric optics, normal mode, and para- scattering, and so tend to be useful only at short
bolic equation methods. The appropriate method ranges and/or low frequencies where internal-wave-
depends in part on the character of the sound-speed induced scattering is less important. A number of
and current fields (e.g., range-independent or range- other possible observables have been proposed as
dependent) and in part on the choice of the observ- well. In what follows the use of ray travel times as
ables in the received signal to use in the inverse observables will be emphasized, in part because they
problem. have been the observable most commonly used to
The approach most commonly used in ocean date and in part because they are robust to internal-
acoustic tomography has been to transmit broad- wave-induced scattering.
band signals designed to measure the impulse re-
sponse of the ocean channel and to interpret the
peaks in the impulse response in terms of geometric Ocean Sound Channel
rays. Ray travel times are robust observables in the The speed with which sound travels in the ocean
presence of internal-wave-induced scattering because increases with increasing temperature, salinity, and
of Fermat’s principle, which states that ray travel pressure. As a result, over much of the temperate
times are not sensitive to first-order changes in the world ocean there is a subsurface minimum in sound
ray path. Other observables are possible, however. speed at depths of roughly 1000 m. Sound speed in-
The peaks in the impulse response are in some cases creases toward the surface above the minimum be-
more appropriately interpreted in terms of normal- cause of increasing temperature and toward the
mode arrivals, for example, and the observables are bottom below the minimum because of increasing
then modal group delays. Another possibility is to pressure. Salinity does not play a major role because
 TOMOGRAPHY 453

its effect on sound speed is normally less than that of and reflected from the sea surface are referred to as
either temperature or pressure. The depth of the refracted-surface-reflected (RSR) rays. Both RR and
sound-speed minimum is called the sound channel RSR rays can propagate to long distances and are
axis. The axis shoals towards high latitudes where commonly used in ocean acoustic tomography. Rays
the surface waters are colder, actually reaching the that interact with the seafloor tend to be strongly
surface during winter at sufficiently high latitudes. scattered, however. Rays with multiple bottom
The sound speed gradients above and below the interactions therefore tend not to propagate to long
sound channel axis refract acoustic rays toward the ranges.
axis in accord with Snell’s law. Near-horizontal rays A receiver at a specified range from an omnidir-
propagating outward from an omnidirectional ectional acoustic source will detect a discrete set of
source on the axis will therefore tend to be trapped, ray arrivals (Figure 2), corresponding to the rays that
cycling first above and then below the axis (Figure 2). are at the depth of the receiver at the appropriate
Such rays are referred to as refracted–refracted (RR) range. These rays are called eigenrays and are des-
rays. Steeper rays will interact with the surface and/ ignated 7p, where 7 indicates an upward/down-
or seafloor. Rays can reflect from the sea surface with ward launch direction and p is the total number of
relatively low loss. Rays that are refracted at depth ray turning points (including reflections). The ray

_1
C (km s )
1.50 1.55
0

1
W N
+20
2
Depth (km)

4 +11

5
0 100 200 300
(A) Range (km)
+20

Measured +11

Predicted
+11

_11

_ 10

_ 12

_ 13
_8

_ 10

_ 14

_ 15

_16

_ 17

_ 18
+8

+10

+12

+13

+11
+14
+10
_8

_ 19
+8

+9

+9
_7
_7

+15

+16

+17

+18

+19

+20
_9

203 204
(B) τ (s)

Figure 2 (A) Sound-speed profile in the western North Atlantic and the corresponding ray paths for source and receiver near the depth
of the sound channel axis and about 300 km apart. The geometry is that of a reciprocal acoustic transmission experiment conducted in
1983 with transceivers designated W(est) and N(orth). (B) Measured and predicted acoustic amplitudes as a function of time for the 1983
experiment. The arrivals are labeled with their ray identifier. The earliest arrivals are from steep ray paths that cycle through nearly the
entire water column. The latest arrivals are from flat ray paths that remain near the sound-channel axis. The differences between the
measured and predicted arrival times are the data used in tomographic inversions. Adapted from Howe BM, Worcester PF and Spindel
RC (1987) Ocean acoustic tomography: mesoscale velocity. Journal of Geophysical Research 92: 3785–3805.
454 TOMOGRAPHY

geometry controls the vertical sampling properties of because the fluctuations in current at a fixed location
tomographic measurements. One can obtain signifi- in the ocean are typically large compared to the time-
cant vertical resolution even for the case of source mean current.
and receiver both on the sound channel axis, because Setting vðr; Þ  0; DvðrÞ  vðrÞ, forming perturb-
the eigenrays in general have a range of turning ation travel times, and linearizing to first order in
depths. DC=C and jvj=C gives
Dtþ þ
i ¼ ti  ti ðÞ ¼
Ray Travel Time Z
½DCðrÞ þ vðrÞ  r0 ðÞ
To first order in jvj=C, the travel time ti of ray i is  ds ½7
C2 ðr;Þ
GiðÞ
Z
ds Dt 
ti ¼ ½1 i ¼ ti  ti ðÞ ¼
C ð r Þ þ vð r Þ  r 0 Z
Gi ½DCðrÞ  vðrÞ  r0 ðÞ
 ds ½8
C2 ðr;Þ
where Gi is the ray path for ray i along which dis- GiðÞ
tance s is measured and r0 is the tangent to the ray at
position r. The sign of v . r0 depends on the direction where GiðÞ ; r0 ðÞ are the ray path and tangent vector
of propagation, and the travel times and ray paths in for the reference state. The superscript plus (minus)
opposite directions differ because of the effects of refers to propagation in the þ ðÞ direction. The
currents. (Sound travels faster with a current than reference travel time is
Z
against a current.) The eigenrays Gi are obtained ds
ti ðÞ ¼ ½9
using a numerical eigenray code. Cðr;Þ
GiðÞ

The Inverse Problem The sum of the travel time perturbations

Z
The ‘inverse’ problem is to compute the sound-speed 1  ds
Dsi ¼ Dtþ i þ Dt 
i ¼  DCðrÞ ½10
CðrÞ and current vðrÞ fields given the measured travel 2 C2 ðr;Þ
GiðÞ
times. In fact, a great deal is normally known about
CðrÞ in the ocean from climatological or other data. depends only on the sound-speed perturbation
The interesting problem is therefore to compute the DCðrÞ. The difference
perturbations from an assumed reference state, using Z
the measured perturbations from the travel times 1  ds
Ddi ¼ Dtþ i þ Dt 
i ¼  2
vðrÞ  r0 ðÞ ½11
computed for the reference state. 2 C ðr;Þ
Gi ð  Þ

Data depends only on the water velocity v . r0 along the ray

Travel times are in general a nonlinear function of the path. Forming sum and difference travel times sep-
sound-speed and current fields, because the ray path arates the effects of DC and v. This separation is
Gi depends on CðrÞ and vðrÞ. Linearize by setting crucial for measuring v, because jvj is usually much
smaller than DC. It is not crucial for measuring DC,
CðrÞ ¼ Cðr;Þ þ DCðrÞ ½2 however, and one-way, rather than sum, travel time
perturbations are often used for this purpose.
vðrÞ ¼ vðr;Þ þ DvðrÞ ½3 The data used in the inverse problem can therefore
either be the one-way travel time perturbations, e.g.,
where Cðr; Þ; vðr; Þ are the known reference states. Dtþi , or the sum and difference travel time perturb-
The argument ðÞ denotes the dependence of the ations, Dsi and Ddi . The use of one-way travel time
variables only on the reference state, independent of measurements to estimate DC is sometimes given the
the measurements. Normally, special name of acoustic thermometry, reflecting the
fact that sound-speed perturbations depend mostly
jDCðrÞj5Cðr;Þ ½4
on temperature.
jDvðrÞj5vðr;Þ ½5
Reference States and Perturbation Models
In general, however, The perturbation field DC and therefore the data
depend on the choice of reference state, Cðr; Þ. Al-
jDvðrÞj > jvðr;Þj ½6 though there is some freedom in the choice, reference
 TOMOGRAPHY 455

states that include the range and time dependence of gives

the sound-speed field available from ocean climatol- X  Z 
Fm ðzÞ
ogies, for example, usually yield reference ray paths Dti ¼ am  ds þ dti ;
C2 ðr;Þ
that are acceptably close to the true ones. Accurate m
Gi ð  Þ
prior estimates of the ray paths help ensure that the
i ¼ 1; y; M ½15
sampling properties of the rays are included properly
in the inverse procedure and that the nonlinearities X
associated with ray-path mismatches are minimized. ¼ Eim am þ dti ; i ¼ 1; y; M ½16
The continuous perturbation fields DC and v are m

parametrized with a finite number of discrete par- The elements Eim depend only on prior infor-
ameters using a model. Because the tomographic mation. This equation can be written in compact
inverse problem is always underdetermined, it is matrix notation as
important to use perturbation models that are
oceanographically meaningful and that provide an y ¼ Exþn ½17
efficient representation of ocean variability. Separ-
able models using a linear combination of basis where
functions describing the horizontal (x, y) and vertical y ¼ ½Dti ; E ¼fEim g;
½18
(z) structures, x ¼½am ; n ¼ ½dti 
X The inverse problem consists (i) of finding a par-
DCðx; y; zÞ ¼ aklm Xk ðxÞYl ðyÞFm ðzÞ ½12
ticular solution x̂ and (ii) of determining the un-
k;l;m
certainty and resolution of the particular solution.
have frequently been used for simplicity, although Writing the estimate x̂ as a weighted linear sum of
more general models are of course possible. The the observations,
coefficients aklm are the model parameters to be de- x̂ ¼ By ¼ BðEx þ nÞ ½19
termined, and the Xk, Yl, Fm are the basis functions.
For zero-mean noise, /nS ¼ 0, the expected value is
Inverse Methods: Vertical Slice
/x̂S ¼ BE/xS ½20
The inverse problem is most simply described for the
case of a single acoustic source–receiver pair. Neg- The matrix BE is called the resolution matrix. It
lecting currents and assuming that the sound-speed gives the particular solution as a weighted average of
perturbation is a function of depth only, the true solution x, with weights given by the row
vectors of BE. If the resolution matrix is the identity
Z
DCðzÞ matrix I, then the particular solution is the true so-
Dti ¼  ds þ dti ; i ¼ 1; y; M ½13 lution. If the row vectors of BE are peaked on the
C2 ðr;Þ
Gi ðÞ diagonal with low values elsewhere, the particular
solution is a smoothed version of the true solution.
where there are M rays. (Note that although the The solution uncertainty is described by the covar-
sound-speed perturbation has been assumed to be iance matrix
independent of range, the reference state can be a D  T E
more general function of position.) The quantity dti P ¼ x̂  x x̂  x ½21
has been introduced to represent the noise contri-
bution that is inevitably present. The noise term where superscript T denotes transpose.
arises not only from observational errors but also There is an immense literature on inverse methods,
from modeling errors associated with the represen- and a variety of approaches are available to con-
tation of DC using a finite number of parameters and struct the inverse operator B, including least squares,
from nonlinearity errors associated with the use of singular-value decomposition (SVD), and Gauss–
the ray paths for the reference state rather than the Markov estimation. To provide an example of one
true ray paths. (In the absence of currents the prob- approach that has been widely used, the Gauss–
lem can be restated somewhat more simply in terms Markov estimate is discussed briefly here. (The
of sound slowness, S ¼ C1, if desired.) Gauss–Markov estimate is sometimes known as
Substituting the ‘stochastic inverse’ or as ‘objective mapping’.)
X The Gauss–Markov estimate is derived by min-
DCðzÞ ¼ am Fm ðzÞ ½14 imizing the expected uncertainty between the true
m value xj and the estimate x̂j , i.e., by individually
456 TOMOGRAPHY

minimizing the diagonal elements of the uncertainty differences imply that the reference state is in-
covariance matrix P. The result is the Gauss–Markov adequate or the model is inadequately formulated.
theorem, When nonlinearities are important, iterative or other
methods are needed to find a solution consistent with
B ¼ Uxy U1
xy ½22 the original data.
The linear inverse methods used in ocean acoustic
where tomography are well known and widely used in a
    variety of fields. The crucial problem in the appli-
Uxy  xyT ; Uyy  yyT ½23 cation to tomography is the construction of the
model used to describe oceanic variability, including
are the model–data and data–data covariance
the choice of parametrization and the specification of
matrices, respectively. These covariances can be
the (co)variances of the model parameters and noise.
rewritten using y ¼ Ex þ n. The model–data covar-
iance matrix becomes
Sampling Properties of Acoustic Rays
 
Uxy ¼ xxT ET ¼ Uxy ET ½24 Vertical slice: range-independent The vertical
sampling properties of acoustic rays are most easily
where it has been assumed that the model x and understood for the range-independent case, in
noise n are uncorrelated. The data–data covariance which sound speed is a function of depth z only. In
matrix becomes that case, eqn [13] can be converted to an integral
D E over depth
Uyy ¼ ðEx þ nÞðEx þ nÞT ¼ EFxx ET þ Unn ½25
z̃ þZ
ðÞ
Finally, the inverse estimate, x̂ ¼ By, can be writ- dz
Dti ¼  qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
  DCðzÞ
ten in the familiar form C 2 ðz; Þ 1  C  ðz; Þ=C̃ 2

z̃ ðÞ
 1 þ dti ; i ¼ 1; y; M ½28
x̂ ¼ By ¼ Uxx ET EFxx ET þ Unn y ½26

The Gauss–Markov estimator requires that the per- using Snell’s law, CðzÞ=cosðyÞ ¼ C̃, where y is the ray
turbation model discussed above include the a priori angle relative to the horizontal and C̃ is the sound
specification of the statistics of the model par- speed at the ray turning points z̃7. The function
ameters, i.e., of the covariance matrix Uxx .
The solution uncertainty P includes contributions 1
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2 ½29
due to data error and due to a lack of resolution. In C2 ðz; Þ 1  Cðz; Þ=C̃
most realistic cases the lack of resolution dominates
the solution uncertainty estimate. A key, and un- gives the weighting with which DCðzÞ contributes to
familiar, feature of the acoustic methods is that the Dti . There are (integrable) singularities in the
solution uncertainty matrix is not diagonal, i.e., the weighting function at both the upper and lower
uncertainties in the model parameters are correlated turning point depths, where ( z̃7; Þ ¼ C̃: The ray
in a way that depends on the ray sampling prop- travel times are therefore most sensitive to sound-
erties. These correlated uncertainties often cancel in speed perturbations at the ray turning points
the computation of integral properties of the solu- (Figure 3).
tion, such as the vertically averaged heat content. The value of the weighting function is the same for
Once a solution and its uncertainty have been depths z7 above and below the sound-channel axis
found, the solution must be evaluated for consistency at which
with the various assumptions made in its con-
struction before it can be accepted. The statistics of Cðzþ ; Þ ¼ Cðz ; Þ ½30
the residuals ŷyÞ, where ŷ¼ E x̂, need to be exam-
ined for consistency with the assumed noise statistics There is a fundamental up–down ambiguity for
Unn , for example. Further, ray trajectories should be acoustic measurements in mid-latitudes. It is in
recomputed for the field principle impossible to distinguish from the acoustic
data alone whether the observed travel-time per-
CðrÞ ¼ Cðr;Þ þ DCðrÞ ½27 turbations are due to sound-speed perturbations lo-
cated above or below the sound channel axis. This
and the resulting ray travel times compared with the ambiguity has to be resolved from a priori infor-
original data to test for consistency. Significant mation or from other data. This up–down ambiguity
 TOMOGRAPHY 457

0 than physical space, using a truncated Fourier series

in x as the model for the sound-speed perturbations,

XX  
2pi
0.1 DCðx; zÞ ¼ akm exp ðkxÞ Fm ðzÞ;
k m
L
k ¼ 0; 71; y; 7N ½31
0.2 The Fourier series is periodic over a domain of
length L and is truncated at harmonic N. The
Upper turning depth z (km)

domain is normally chosen sufficiently large (say

~+

0.3 twice the size of the source–receiver range) to avoid

artifacts within the area of interest that might be
caused by the periodicity.
The travel-time perturbations are then
0.4
( Z
XX ds
Dti ¼ akm 
k m
C2 ðr;Þ
0.5 Gi ð  Þ
  )
2pi
exp ðkxÞ Fm ðzÞ þ dti ½32
L
0.6
where the integrals depend only on prior infor-
mation. The problem has again reduced to the form
y ¼ Ex þ n, with the solution vector x containing an
0 1 2 3 4 5 ordered set of the complex Fourier coefficients akm.
Δ (ms) The sensitivity of the travel-time inverse to various
wavenumbers can be quantified by plotting the di-
Figure 3 Travel-time perturbations computed at about 1000 km agonal of the resolution matrix BE defined in eqn
range in the North-east Pacific for a sound-speed perturbation with
an amplitude of 1 m s1 at 100 m depth, linearly decreasing to
[20]. For the specific case of two moorings separated
zero at 90 m and 110 m. The travel-time perturbations are zero for by 600 km, with a source and five widely-separated
rays with upper turning depths below 110 m, because they do not receivers on each mooring, the resolution matrix
sample the perturbed region. The perturbations are sharply shows the sensitivity of tomographic measurements
peaked for rays with upper turning depths between 90 m and to the features that match the ray periodicity (i.e.,
110 m. Rays that have upper turning depths above 90 m have
nonzero perturbations because they traverse the perturbed
have the same wavelength as the ray double loops)
region, but the perturbations are relatively small because the ray (Figure 4). Further, because the ray paths are some-
weighting function falls off rapidly with distance from the turning what distorted sinusoids in midlatitudes, the reso-
point. Adapted from Cornuelle BD, Worcester PF, Hildebrand JA, lution matrix displays sensitivity to harmonics of the
et al. (1993) Ocean acoustic tomography at 1000-km range using basic ray double loop length. Finally, as expected, the
wavefronts measured with a large-aperture vertical array. Journal
of Geophysical Research 98: 16365–16377.
measurements are sensitive to the mean. There are
obvious spectral gaps for wavenumbers between the
mean and first harmonics of the ray paths, and again
is not present in polar regions when the temperature between the first and second harmonics. The har-
profile is close to adiabatic, so that sound speed is a monics extend over bands of wavenumbers because
minimum at the surface and increases monotonically the eigenrays connecting the source and receiver have
with depth (pressure). a range of double-loop lengths.

Vertical slice: range-dependent A ray trapped in Horizontal slice The sampling issues present when
the sound channel, cycling between upper and lower the goal is to map the evolving ocean using integral
turning points at regular intervals, samples the data are most easily understood by considering the
ocean periodically in space, so that its travel time is two-dimensional horizontal slice problem. In this
sensitive to some spatial frequencies but is case sound speed is assumed to be constant in the
unaffected by others. The key to understanding the vertical, so that ray paths travel in straight lines in
horizontal sampling properties of acoustic travel the horizontal plane containing the sources and
times is to consider the wavenumber domain, rather receivers. Neglecting currents,
458 TOMOGRAPHY

1.0 two ships start in the left and right bottom corners of
a 1000-km square and steam northward in parallel,
0.8 transmitting from west to east through an isotropic
mesoscale field constructed to have a 1/e decay scale
Resolution

0.6
of 120 km (Figure 5). Inversion of the resulting tra-
0.4
vel-time data leads to an estimate that consists only
of east–west contours, as all the ray paths measure
0.2 only zonal averages. To interpret this result in
wavenumber space, note that
0.0
Z L
0 10 20 30 40 50 60
dxDCðx; yÞ ¼ 0 for ka0 ½36
k (cycles / 600 km)
0
Figure 4 Diagonal elements of the resolution matrix (‘transfer
function’) for tomographic measurements over a 600-km path in a East–west transmissions therefore give information
range-dependent ocean. The plot is for the lowest baroclinic only on the parameters a0l, which are nearly perfectly
mode. Adapted from Cornuelle BD and Howe BM (1987) High determined for the assumptions made in this simple
spatial resolution in vertical slice ocean acoustic tomography.
Journal of Geophysical Research 92: 11680–11692.
scenario. Similarly, for north–south transmissions
between two ships traveling from east to west, only
the parameters ak0 are determined. Combining east–
Z
DCðx; yÞ west and north–south transmissions determines both
Dti ¼  ds þ dti ;
C2 ðx; y; Þ a0l and ak0, but nonetheless fails to give useful maps
Gi ð  Þ because the majority of the wavenumbers are still
i ¼ 1; y; M ½33 undetermined. Adding scans at 451 determines
wavenumbers for which k ¼ l, giving improved, but
where there is one ray path per source–receiver pair still imperfect, maps.
and there are a total of M ray paths connecting the The conclusion is that generating accurate maps
sources and receivers. from integral data requires sampling geometries with
As was the case for the range-dependent vertical ray paths at many different angles to provide ad-
slice problem, the key to understanding the horizontal equate resolution in wavenumber space. This re-
sampling properties of acoustic travel times is to quirement must be independently satisfied in any
consider the wavenumber domain, rather than phys- region with dimensions comparable to the ocean
ical space, using a truncated Fourier series in x and y correlation scale. These results are a direct con-
as the model for the sound-speed perturbations, sequence of the projection-slice theorem.
XX  
2pi Time-dependent Inverse Methods
DCðx; yÞ ¼ akl exp ðkx þ lyÞ ;
k l
L
The discussion of inverse methods to this point has
k; l ¼ 0; 71; y; 7N ½34 implicitly assumed that data from a single instant in
time are used to estimate the state of the ocean at
The Fourier series is doubly periodic over the square that instant. Observations from different times can
domain of size L and is truncated at harmonic N. be combined to generate improved estimates of the
The travel-time perturbations are then evolving ocean, however, using a time-dependent
( Z ocean model to connect the oceanic states at those
XX ds
Dti ¼ akl  2
times. One seeks to minimize the misfit between the
k l C ðx; y; Þ estimate x̂(t) and the true state x(t) over some finite
GiðÞ
  ) time span, instead of at a single instant.
2pi The practice of combining data with time-evolving
exp ðkx þ lyÞ þ dti ½35
L ocean circulation models, referred to as ‘assimi-
lation’ or ‘state estimation’, simultaneously tests and
where the integrals depend only on prior infor- constrains the models. A variety of approaches are
mation. The problem has again reduced to the form available to solve this problem, including, for ex-
y ¼ Ex þ n, with the solution vector x containing an ample, Kalman filtering and the use of adjoint
ordered set of the complex Fourier coefficients akl. methods. Although the problem of combining inte-
To explore the horizontal sampling properties of gral tomographic data with time-evolving models
integral data, consider a simple scenario in which does not differ in any fundamental way from the
 TOMOGRAPHY 459

Figure 5 The top center panel is the ‘true ocean,’ constructed assuming a horizontally homogeneous and isotropic wavenumber
spectrum, to be mapped using tomographic data. (A) W-E transmissions between two northward-traveling ships (left panel).
Inversion of the travel time perturbations produces east–west contours in DC (middle) with only a faint relation to the ‘true ocean.’
Expected predicted variances in wavenumber space (right) are 0% (no skill) except for (k,l) ¼ (0,1),(0,2),y,(0,7), which account for
s2 ¼ 16% of the a priori DC variance. (B) S-N transmissions between two eastward traveling ships. (C) Combined W-E and S-N
transmissions, accounting for 32% of the DC variance. (D) Combined W-E, S-N, SW-NE, and SE-NW transmissions,
accounting for 67% of the variance and giving some resemblance to the true ocean. Distances are shown in magameters (Mm) and
wavenumbers are shown in cycles per megameter (cpMm). Adapted from Cornuelle BD, Munk WH and Worcester PF (1989) Ocean
acoustic tomography from ships. Journal of Geophysical Research 94: 6232–6250.
460 TOMOGRAPHY

problem of using other data types, tomographic data Oceanic Convection

do differ from most other oceanographic data be-
Oceanic convection to great depths occurs at only a
cause their sampling and information content tend to
few locations in the world (see )Nonetheless, it is be-
be localized in spectral space rather than in physical
lieved to be the process by which the properties of the
space, as discussed above. It is therefore important to
surface ocean and deep ocean are connected, with
use methods that directly assimilate the tomographic
important consequences for the global thermohaline
measurements and preserve the integral information
circulation and climate. The deep convective process is
they contain. Approximate data assimilation meth-
temporally intermittent and spatially compact, con-
ods optimized for measurements localized in physical
sisting of convective plumes with scales of about 1 km
space are generally inappropriate because they do
clustered in chimneys with scales of tens of kilometers.
not preserve the nonlocal tomographic information.
Observing the evolution of the deep convective process
and quantifying the amount of deep water formed
Selected Tomographic Results presents a difficult sampling problem. Tomographic
Tomographic methods have been used to study a measurements have been key components in programs
wide range of ocean processes, at diverse locations. to study deep convection in the Greenland Sea (1988–
Measurements have been made at scales ranging 1989) and the Mediterranean Sea (1991–1992), as well
from a few tens of kilometers (e.g., to measure the as in an ongoing program in the Labrador Sea (1996 to
transport through the Strait of Gibraltar) to thou- present). In all of these regions the tomographic data
sands of kilometers (e.g., to measure the heat content provide the spatial coverage and temporal resolution
in the north-east Pacific Ocean). This review con- necessary for observing the convective process.
cludes by presenting results from a few selected ex- In the Greenland Sea, for example, six acoustic
periments to provide some indication of the breadth transceivers were deployed from summer 1988 to
of possible applications and to illustrate the strengths summer 1989 in an array approximately 210 km in
and weaknesses of tomographic measurements. diameter (Figure 6), as part of the intensive field

Figure 6 (A) Geometry of the tomographic transceiver array deployed in the Greenland Sea during 1988–1989. Mooring 2 failed
about one month after deployment. A deep convective chimney was observed near the center of the array during March 1989 (shaded
region). (B) Time-depth evolution of potential temperature averaged over the chimney region. Contour interval is 0.21C. Typical rms
uncertainty (1C) as a function of depth is shown to the right. Total heat flux (from the British Meteorological Office) and daily averaged
ice cover (derived from satellite SSM/I measurements) are shown above. Adapated from Morawitz WML, Cornuelle BD and Worcester
PF (1996) A case study in three-dimensional inverse methods: combining hydrographic, acoustic, and moored thermistor data in the
Greenland Sea. Journal of Atmospheric and Oceanic Techniques 13: 659–679.
 TOMOGRAPHY 461

phase of the International Greenland Sea Project. extracted from the three-dimensional inverse esti-
The acoustic data were combined with moored mates were averaged over the chimney region to
thermistor data and hydrographic data to estimate show the time-evolution of the chimney (Figure 6). A
the evolution of the three-dimensional temperature one-dimensional vertical heat balance adequately
field T(x, y, z) in the Greenland Sea during winter. described changes in total heat content in the chim-
(During the convective period, the hydrographic data ney region from autumn 1988 until the time of
were found to be contaminated by small-scale vari- chimney break-up, when horizontal advection be-
ability and were not useful for determining the came important and warmer waters moved into the
chimney and gyre-scale structure.) A convective region. The average annual deep-water production
chimney reaching depths of about 1500 m was ob- rate in the Greenland Sea for 1988–1989 was esti-
served to the south west of the gyre center during mated from the average temperature change over the
March 1989. The chimney had a spatial scale of region occupied by the tomographic array to be
about 50 km and a timescale of about 10 days about 0.1  106m3 s1.
(Figure 7). The location of the chimney seemed to be
sensitively linked to the distribution of the relatively
Barotropic and Baroclinic Tides
warm, salty Arctic Intermediate Water found at
intermediate depths. Potential temperature profiles Sum and difference travel times from long-range re-
ciprocal transmissions provide precise measurements
of the sound-speed (temperature) changes associated
with baroclinic (internal) tidal displacements and of
barotropic tidal currents, respectively.
The availability of global sea-surface elevation
data from satellite altimeter measurements has made
0 possible the development of improved global tidal
models. Tomographic measurements of tidal currents
made in both the central North Pacific and western
North Atlantic Oceans have shown that the har-
monic constants for current derived from a recent
global tidal model (TPXO.2) are accurate to a frac-
Depth (km)

tion of a millimeter per second in amplitude and a

few degrees in phase in open ocean regions
1
(Figure 8). Small, spatially coherent differences be-
tween the modeled and measured harmonic con-
stants are found in the western North Atlantic near
complicated topography that is unresolved in the
model. These differences are almost certainly due to
˚
errors in the TPXO.2 currents. The integrating na-
76 ture of the tomographic measurements suppresses
2 short-scale internal waves and internal tides, pro-
6˚ 75
˚ de viding tidal current measurements that are sub-
itu
5˚
4˚ lat
3˚ r th stantially more accurate than those derived from
Wes
t lo ˚ No
ngitu
de
2˚ 74 current-meter data.
Tomographic measurements of sound-speed fluc-
Figure 7 Mixed layer depth in the central Greenland Sea on 19 tuations at tidal frequencies from the same experi-
March 1989, as defined by the minimum depth of the 1.21C ments revealed large-scale internal tides that are
isotherm. The ocean is colder than 1.21C above this depth as a phase-locked to the barotropic tides. Prior to these
result of surface cooling and is warmer below. The main chimney
measurements it had commonly been assumed that
reaches a maximum depth of about 1500 m in an area about
50 km in diameter centered on 74.751N, 3.51W, south west of the midocean internal tides are not phase-locked to the
gyre center. A secondary chimney with a maximum depth of barotropic tides (except for locally forced internal
about 1000 m is evident to the north east of the gyre center, tides) and have correlation length scales of order
separated from the primary chimney by a ridge of warmer water. only 100–200 km. The measurements in the North
Contours of mixed layer depth are shown below. Adapted from
Pacific were consistent with a large-scale, phase-
Morawitz WML, Sutton PJ, Worcester PF et al. (1996) Three-
dimensional observations of a deep convective chimney in the locked internal tide that had been generated at the
Greenland Sea during winter 1988/1989. Journal of Physical Hawaiian Ridge and then propagated to the tomo-
Oceanography 26: 2316–2343. graphic array over 2000 km to the north (Figure 9).
462 TOMOGRAPHY

50°
N

40°

30°

20°

10°

0°
80° W 70° 60° 50° 40° 30° 20° 10° 0°
2.5
400
2
300
Model (cm/s)

Model (deg)

1.5
200
1
100

0.5 0
Zonal Zonal
0
2.5
400
2
300
Model (cm/s)

Model (deg)

1.5
200

1
100

0.5 0
Meridional Meridional
0
0 0.5 1 1.5 2 2.5 0 200 400
Measured (cm / s) Measured (deg)

Figure 8 (Bottom) Comparison of the M2 current harmonic constants (amplitude and phase) in the North Atlantic Ocean derived from
reciprocal acoustic transmissions (filled circles) and from current meter data (dots) with those predicted by a global tidal model derived
from satellite altimeter measurements (TPXO.2). (Top) The acoustic data are from the pentagonal tomographic transceiver array
deployed in the western North Atlantic between Puerto Rico and Bermuda during 1991–1992. The current-meter mooring locations are
indicated by crosses. Adapted from Dushaw BD, Egbert GD, Worcester PF et al. (1997) A TOPEX/POSEIDON global tidal model
(TPXO.2) and barotropic tidal currents determined from long-range acoustic transmissions. Progress in Oceanography 40: 337–367.

These observations were subsequently confirmed diurnal K1 internal tide, 1100 km distant at 30.01N
from satellite altimeter data. The measurements in (Figure 10). In both cases the peak-to-peak tem-
the western North Atlantic revealed a diurnal in- perature variations associated with the maximum
ternal wave resonantly trapped between the shelf just displacement of the first baroclinic modes were only
north of Puerto Rico and the turning latitude for the about 0.041C. Once again, the acoustic observations
 TOMOGRAPHY 463

160 km

40°
N RTE87

30°

Midway

20°
Hawaii

10 cm
10°
km
0 500
100

125

112
49

87

74
11

36

23

61

180° 170° W 160° 150°

Figure 9 Schematic diagram showing the phase-locked internal tide generated at the Hawaiian Ridge and the triangular
tomographic array deployed north of Hawaii during 1987 used to detect it. The dashed lines represent the crests of a wave with 160 km
wavelength. Each leg of the tomographic array functions as a linear array with maximum sensitivity to an incident plane wave
propagating perpendicular to the leg (i.e., with wave crests aligned parallel to the leg). The beam pattern (in dB) of the 750-km northern
leg for a model-1 incident wave with a wavelength of 160 km is indicated. The satellite altimeter data that subsequently confirmed the
tomographic observations are also shown. High-pass filtered M2 surface elevations (cm) are plotted along ten ascending TOPEX/
POSEIDON ground tracks. Adapted from Dushaw BD, Cornuelle BD, Worcester PF, Howe BM and Luther DS (1995) Barotropic and
baroclinic tides in the central North Pacific Ocean determined from long-range reciprocal acoustic transmissions. Journal of Physical
Oceanography 25: 631–647; Ray RD and Mitchum GT (1996) Surface manifestation of internal tides generated near Hawaii.
Geophysical Research Letters 23: 2101–2104.

of the baroclinic tide average in range and depth, transmissions from Europe to Africa (Figure 11).
suppressing internal-wave noise and providing en- Although it is normally difficult to obtain heat
hanced estimates of the deterministic part of the in- content measurements comparable to those pro-
ternal-tide signal compared to measurements made vided by the acoustic data, in this case one of the
at a point, such as by moored thermistors. transmission paths was intentionally aligned with
the route of a commercial ship, from which ex-
pendable bathythermograph (XBT) measurements
Heat Content
were made at two-week intervals. The acoustic
Acoustic methods have been used to measure the average of potential temperature between 0 and
heat content of the ocean and its variability on basin 2000 m depth over the 600-km path and the cor-
scales, taking advantage of the integrating nature of responding XBT average between 0 and 800 m
acoustic transmissions to rapidly and repeatedly depth agreed within the expected uncertainty of
make range- and depth-averaged temperature meas- 0.031C (Figure 11). Further, the evolution of the
urements at ranges out to about 5000 km. three-dimensional heat content of the western
In the Mediterranean Sea, for example, a net- Mediterranean estimated from the acoustic data
work of seven tomographic instruments was de- was found to be consistent with the integral of the
ployed for nine months during 1994 in the surface heat flux provided by European Centre for
THETIS-2 experiment, including cross-basin Medium-range Weather Forcasts (ECMWF), after
464 TOMOGRAPHY

USA

1000 2000

Cuba 4000
80°
W
4000

K1
75°

t ur
4 5 Bermuda

nin
gl
6 1

ati
70° 3

tud
P.R

Puerto 5500
2
.T

e
Rico
ren

5500
ch

65°
30° N
25°
5500
60°
(A) 20°

28°
K1 5 49, 89
o 1
N o
6
,2
69, 71 o

30
6o

44,

76
26° ,3
9,

,4
53, 70 o

86

26
20

85

o
o
o

51
,3
o 6 101, 2 14 o
282 62 o
2
76,
46, 53

24° o
4 94, 252 o
o

148
53,
110 o 3 6,

22° 2000 m 3
Caicos Is.

Silver Bk.
20°
4000 m
Hispaniola 2000 m
18° Puerto Rico
_1
(B) 72° W 70° 68° 66° 64° 1

Figure 10 (A) Schematic diagram showing the predicted displacement of the lowest internal mode for the resonant diurnal (K1)
internal tide north of Puerto Rico and the six-element tomographic array deployed during 1991–1992 used to observe it. The
tomographic array is about 670 km in diameter. (B) The predicted displacement of the diurnal (K1) internal tide and the measured
harmonic constants (amplitude and phase) for each acoustic path. Adapted from Dushaw BD and Worcester PF (1998) Resonant
diurnal internal tides in the North Atlantic. Geophysical Research Letters 25: 2189–2192.
 TOMOGRAPHY 465

44˚ correction for the heat flux through the Straits of

N
Gibraltar and Sicily (Figure 11). The acoustic data
42˚
were subsequently combined with satellite altimeter
data and an ocean general circulation model to
generate a consistent description of the basin-scale
40˚ temperature and flow fields in the western Medi-
terranean and their evolution over time. Acoustic
and altimetric data are complementary for this
38˚
purpose, with the acoustic data providing infor-
mation on the ocean interior with moderate vertical
36˚ resolution and the altimetric data providing detailed
(A) 2˚W 0 2˚E 4˚ 6˚ 8˚ 10˚ 12˚ horizontal coverage of the ocean surface.
Similar measurements have been made in the
Arctic Ocean in the Transarctic Acoustic Propa-
gation (TAP) experiment during 1994 and in the
0.15
Arctic Climate Observations using Underwater
0.10 Sound (ACOUS) project beginning in 1999. During
Potential temperature

the TAP experiment, ultralow-frequency (19.6 Hz)

anomaly (˚C)

0.05
acoustic transmissions propagated across the entire
0 Arctic basin from a source located north of Sval-
_ 0.05 bard to a receiving array located in the Beaufort Sea
at a range of about 2630 km. Modal travel time
_ 0.10 Inversion
CTD sections measurements yielded the surprising result that the
XBT sections
Atlantic Intermediate Water layer was about 0.41C
(B) 1 Feb 1 Apr 1 Jun 1 Aug 1 Oct warmer than expected from historical data. This
result was subsequently confirmed by direct meas-
urements made from icebreakers and submarines.
0.15 Acoustic data collected on a similar path during
April 1999 as part of the ACOUS project indicated
0.10 further warming of about 0.51C, which was again
Potential temperature

confirmed by direct measurements made from sub-

anomaly (˚C)

0.05
marines. Acoustic methods can provide the long-
0
term, continuous observations in ice-covered re-
_ 0.05 gions that are difficult to obtain using other
_ 0.10
3-D average tomography approaches.
ECMWF with straits corr.
Climatology Finally, measurements of basin-scale heat content
_ 0.15 in the Northeast Pacific were made intermittently
1 Feb 1 Apr 1 Jun 1 Aug 1 Oct 1 Dec
(C)
from 1983 through 1989 using transmissions from
1994
an acoustic source located near Kaneohe, Hawaii,
Figure 11 (A) Geometry of the THETIS-2 experiment in the and more recently from 1996 through 1999 during
western Mediterranean Sea, showing the instrument locations the Acoustic Thermometry of Ocean Climate
and acoustic transmission paths. The transmission path from (ATOC) project using sources located off central
source H to receiver W3 (heavy solid) coincided with an XBT
California and north of Kauai, Hawaii. Data from
section occupied every two weeks. (B) Range- and depth-
averaged potential temperature (relative to 13.1111C) over 0– the ATOC project have shown that ray travel times
2000 m depth and over the 600 km path from source H to receiver may be used for acoustic thermometry at least out to
W3 derived from the acoustic data, from CTD data, and from XBT ranges of about 5000 km. The estimated uncertainty
data. The shaded band indicates the uncertainty in the in range- and depth-averaged temperature estimates
temperature estimates derived from the acoustic data. (C)
made from the acoustic data at these ranges is only
Evolution of the three-dimensional average heat content for the
western Mediterranean during 1994 derived from the acoustic about 0.01 1C. Comparisons between sea-surface
data, from the ECMWF surface heat fluxes corrected for heat height measurements made with a satellite altimeter
transport through the Straits of Gibraltar and Sicily, and from and sea-surface height estimates derived using the
climatology. Adapted from Send U, Krahmann G, Mauuary D et range-averaged temperatures computed from the
al. (1997) Acoustic observations of heat content across the
acoustic data indicate that thermal expansion alone
Mediterranean Sea. Nature 385: 615–617.
466 TOMOGRAPHY

cm (rms)
l 34
k 32
n 30
28
26
o 24
22
v1
20
18
16
v2 14
12
10
8
6
4
2

(A)

5 l

_5
Sea surface height (cm)

k
5

_5

5 o

_5

(B) Jan 93 Jan 95 Jan 97

Figure 12 (A) The ATOC acoustic array is superimposed on a map of the root-mean-square (rms) sea level anomaly from altimetric
measurements. Transmission paths from sources off central California and north of Kauai to a variety of receivers are shown. (B) The
range-averaged sea-surface height anomaly along several of the acoustic sections from satellite altimeter data (solid black), inferred
from the acoustic data (solid red), computed from climatological temperature fluctuations (dashed), and derived from an ocean general
circulation model (solid blue). Uncertainties are indicated for the acoustic estimates. Adapted from ATOC Consortium (1998) Ocean
climate change: comparison of acoustic tomography, satellite altimetry, and modeling. Science 281: 1327–1332.

is inadequate to account for all of the observed model indicates that the differences result largely
changes in sea level (Figure 12). Analysis of the re- from a barotropic redistribution of mass, with vari-
sults obtained when the acoustic and altimetric data able salt anomalies a contributing, but smaller,
were used to constrain an ocean general circulation factor.
 TOMOGRAPHY 467

Appendix: Conversion from Sound 0.45

Speed to Temperature
Tomgraphic methods fundamentally provide infor-
0.40
mation on the oceanic sound-speed and water-vel-
ocity fields. For most oceanographic purposes,
however, temperature T and salinity Sa are of more
interest than sound speed. Although sound speed C is 0.35
a function of both T and Sa (as well as pressure),

∂T / ∂C (˚C / m s−1)
temperature perturbations are normally by far the
most important contributor to sound-speed 0.30
perturbations.
A simple nine-term equation for sound speed due
to Mackenzie is
0.25
2 2
CðT; Sa; DÞ ¼ 1448:96 þ 4:591T  5:304  10 T
þ 2:374  104 T 3 þ 1:340ðSa  35Þ
0.20
þ 1:630  102 D þ 1:675  107 D2
 1:025  102 T ðSa  35Þ
 7:139  1013 TD3 ½37 0.15
5 10 15 20
where C is in ms1, T is in degrees Celsius, Sa is in T (˚C)
parts per thousand (ppt), and D is the depth (positive Figure 13 The derivative @T =@C as a function of temperature.
down) in meters. Differentiating, Adapted from Dushaw BD, Worcester PF, Cornuelle BD and Howe
BM (1993) Variability of heat content in the central North-Pacific
@C in summer 1987 determined from long-range acoustic trans-
¼ 4:59  0:106T þ 7:12  104 T 2 missions. Journal of Physical Oceanography 23: 2650–2666.
@T ½38
1:03  102 ðSa  35Þ
sound speed is then
@C DC=C ¼ aDT ð1 þ mb=aÞ ½44
¼ 1:34  1:03  102 T ½39
@Sa
In midlatitudes a typical value for m might be
where a slight depth dependence has been dropped. 0.1(ppt)(1C)1, giving mb=aE0:03. Thus the sound-
The derivative @T=@C ¼ 1=ð@C=@TÞ varies signifi- speed perturbation DC depends to first order only on
cantly with temperature (Figure 13). the temperature perturbation DT.
To first order, the fractional change in sound speed The sound-speed perturbation profile DCðzÞ
is then derived from the acoustic data can be easily converted
to the corresponding temperature perturbation profile
DC=C ¼ aDT þ bDSa ½40
ZCðÞþDC
where @T
DT ðzÞ ¼ dC; ½45
@C
1 @C CðÞ
a¼ E2:4  103 ðo CÞ1 ½41
C @T where the integral allows for the dependence of
@T=@C on temperature.
1 @C
b¼ E0:8  103 ðpptÞ1 ½42
C @Sa
at 101C. For a locally linear temperature–salinity
See also
relation, Acoustics, Arctic. Acoustics in Marine Sediments.
Sa ¼ SaðT ðÞÞ þ mDT ½43
Further Reading
where TðÞ is the reference temperature profile
corresponding to the reference sound-speed profile Khil’ko AI, Caruthers JW, and Sidorovskaia NA (1998)
CðÞ and DT ¼ T  TðÞ. The fractional change in Ocean Acoustic Tomography: A Review with Emphasis
468 TOMOGRAPHY

on the Russian Approach. Nizhny Novgorod: Institute application of tomographic methods to studying the
of Applied Physics, Russian Academy of Sciences. ocean.)
Munk W, Worcester P, and Wunsch C (1995) Ocean Munk Wand Wunsch C (1979) Ocean acoustic tomography:
Acoustic Tomography, Cambridge: Cambridge a scheme for large scale monitoring. Deep-Sea Research
University Press. (The review given here draws heavily 26: 123--161.
from, and uses the same notation as, this monograph, Worcester PF (1977) Reciprocal acoustic transmission in a
which provides a comprehensive account of the mid-ocean environment. Journal of the Acoustical
elements of oceanography, acoustics, signal processing, Society of America 62: 895--905.
and inverse methods necessary to understand the
SATELLITE, AIRCRAFT, OR SHIP-BORNE
REMOTE SENSING
 AIRCRAFT REMOTE SENSING
L. W. Harding, Jr. and W. D. Miller, University of will be discussed in more detail in subsequent sec-
Maryland, College Park, MD, USA tions. Other important uses of aircraft remote sens-
R. N. Swift, and C. W. Wright, NASA Goddard ing are to test instruments slated for deployment on
Space Flight Center, Wallops Island, VA, USA satellites, to calibrate and validate space-based sen-
Copyright & 2001 Elsevier Ltd. sors using aircraft-borne counterparts, and to make
‘under-flights’ of satellite instruments and assess the
efficacy of atmospheric corrections applied to data
from space-based observations.
Aircraft have some advantages over satellites for
oceanography, including the ability to gather data
Introduction under cloud cover, high spatial resolution, flexibility
The use of aircraft for remote sensing has steadily of operations that enables rapid responses to ‘events’,
grown since the beginnings of aviation in the early and less influence of atmospheric effects that com-
twentieth century and today there are many appli- plicate the processing of satellite data. Aircraft re-
cations in the Earth sciences. A diverse set of remote mote sensing provides nearly synoptic data and
sensing uses in oceanography developed in parallel information on important oceanographic properties
with advances in aviation, following increased air- at higher spatial resolution than can be achieved by
craft capabilities and the development of instru- most satellite-borne instruments. Perhaps the great-
mentation for studying ocean properties. Aircraft est advantage of aircraft remote sensing is the ability
improvements include a greatly expanded range of to provide consistent, high-resolution coverage at
operational altitudes, development of the Global larger spatial scales and more frequent intervals than
Positioning System (GPS) enabling precision navi- are practical with ships, making it feasible to use
gation, increased availability of power for instru- aircraft for monitoring change.
ments, and longer range and duration of missions. Disadvantages of aircraft remote sensing include
Instrumentation developments include new sensor the relatively limited spatial coverage that can be
technologies made possible by microelectronics, obtained compared with the global coverage avail-
small, high-speed computers, improved optics, and able from satellite instruments, the repeated expense
increased accuracy of digital conversion of electronic of deploying multiple flights, weather restrictions on
signals. Advances in these areas have contributed operations, and lack of synopticity over large scales.
significantly to the maturation of aircraft remote Combination of the large-scale, synoptic data that are
sensing as an oceanographic tool. accessible from space with higher resolution aircraft
Many different types of aircraft are currently used surveys of specific locations is increasingly recognized
for remote sensing of the oceans, ranging from bal- as an important and useful marriage that takes ad-
loons to helicopters, and from light, single engine vantages of the strengths of both approaches.
piston-powered airplanes to jets. The data and in- This article begins with a discussion of sensors that
formation collected on these platforms are com- use lasers (also called active sensors), including air-
monly used to enhance sampling by traditional borne laser fluorosensors that have been used to
oceanographic methods, giving increased spatial and measure chlorophyll (chl-a) and other properties;
temporal resolution for a number of important continues with discussions of lidar sensors used for
properties. Contemporary applications of aircraft topographic and bathymetric mapping; describes
remote sensing to oceanography can be grouped into passive (sensors that do not transmit or illuminate,
several areas, among them ocean color, sea surface but view naturally occurring reflections and emis-
temperature (SST), sea surface salinity (SSS), wave sions) ocean color remote sensing directed at quan-
properties, near-shore topography, and bathymetry. tifying phytoplankton biomass and productivity;
Prominent examples include thermal mapping using moves to available or planned hyper-spectral aircraft
infrared (IR) sensors in both coastal and open ocean instruments; briefly describes synthetic aperture
studies, lidar and visible radiometers for ocean color radar applications for waves and wind, and closes
measurements of phytoplankton distributions and with a discussion of passive microwave measure-
‘algal blooms’, and passive microwave sensors to ments of salinity. Readers are directed to the Further
make observations of surface salinity structure of Reading section if they desire additional information
estuarine plumes in the coastal ocean. These topics on individual topics.

471
472 AIRCRAFT REMOTE SENSING

Active Systems The AOL has supported major oceanographic

studies throughout the 1980s and 1990s, extending
Airborne Laser Fluorosensing
the usefulness of shipboard measurements over wide
The concentrations of certain waterborne constitu- areas to permit improved interpretation of the ship-
ents, such as chl-a, can be measured from their derived results. Examples of data from the AOL
fluorescence, a relationship that is exploited in show horizontal structure of laser-induced fluor-
shipboard sensors such as standard fluorometers and escence converted to chl-a concentration (Figure 1).
flow cytometers that are discussed elsewhere in this Prominent oceanographic expeditions that have
encyclopedia. NASA first demonstrated the meas- benefited from aircraft coverage with the AOL in-
urement of laser-induced chl-a fluorescence from a clude the North Atlantic Bloom Experiment (NABE)
low-flying aircraft in the mid-1970s. Airborne laser of the Joint Global Ocean Flux Study (JGOFS), and
fluorosensors were developed shortly thereafter in the Iron Enrichment Experiment (IRONEX) of the
the USA, Canada, Germany, Italy, and Russia, and Equatorial Pacific near the Galapagos Islands. This
used for measuring laser-induced fluorescence of a system has been flown on NASA P-3B aircraft in
number of marine constituents in addition to chl-a. open-ocean missions that often exceeded 6 h in dur-
Oceanic constituents amenable to laser fluorosensing ation and collected hundreds of thousands of spectra.
include phycoerythrin (photosynthetic pigment in Each ‘experiment’ is able to generate both active and
some phytoplankton taxa), chromophoric dissolved passive data in ‘pairs’ that are used for determining
organic matter (CDOM), and oil films. Airborne ocean color and recovering chl-a and other con-
laser fluorosensors have also been used to follow stituents, and that are also useful in the development
dyes such as fluorescein and rhodamine that are of algorithms for measuring these constituents from
introduced into water masses to trace their oceanic radiance spectra.
movement.
The NASA Airborne Oceanographic Lidar (AOL)
Airborne Lidar Coastal Mapping
is the most advanced airborne laser fluorosensor. The
transmitter portion features a dichroic optical device The use of airborne lidar (light detection and ran-
to spatially separate the temporally concurrent 355 ging) sensors for meeting coastal mapping require-
and 532 nm pulsed-laser radiation, followed by in- ments is a relatively new and promising application
dividual steering mirrors to direct the separated of laser-ranging technology. These applications in-
beams to respective oceanic targets separated by clude high-density surveying of coastline and beach
B1 m when flown at the AOL’s nominal 150 m morphology and shallow water bathymetry. The
operational altitude. The receiver focal plane con- capability to measure distance accurately with lidar
taining both laser-illuminated targets is focused onto sensors has been available since the early 1970s, but
the input slits of the monochromator. The mono- their application to airborne surveying of terrestrial
chromator output focal planes are viewed by custom- features was seriously hampered by the lack of
made optical fibers that transport signal photons knowledge of the position of the aircraft from which
from the focal planes to the photo-cathode of each the measurement was made. The implementation of
photo-multiplier module (PMM) where the con- the Department of Defense GPS constellation of
version from photons to electrons takes place with a satellites in the late 1980s, coupled with the devel-
substantial gain. Time-resolved waveforms are col- opment of GPS receiver technology, has resulted in
lected in channels centered at 404 nm (water Raman) the capability to provide the position of a GPS an-
and 450 nm (CDOM) from 355 nm laser excitation, tenna located on an aircraft fuselage in flight to an
and at 560 nm and 590 nm (phycoerythrin), 650 accuracy approaching 5 cm using kinematic differ-
(water Raman), and 685 nm (chl-a) from the 532 nm ential methodology. These methods involve the use
laser excitation. The water Raman from the re- of a fixed receiver (generally located at the staging
spective lasers is the red-shifted emission from the airport) and a mobile receiver that is fixed to the
OH bonds of water molecules resulting from radi- aircraft fuselage. The distance between mobile and
ation with the laser pulse. The strength of the water fixed receivers, referred to as the baseline, is typically
Raman signal is directly proportional to the number on the order of tens of kilometers, and can be ex-
of OH molecules accessed by the laser pulse. Thus, tended to hundreds of kilometers by using dual fre-
the water Raman signal is used to normalize the quency survey grade GPS receivers aided by tracking
fluorescence signals to correct for variations in water the phase code of the carrier from each frequency.
attenuation properties in the surface layer of the Modern airborne lidars are capable of acquiring
ocean. The AOL is described in more detail on http:// 5000 or more discrete range measurements per sec-
lidar.wff.nasa.gov. ond. Aircraft attitude and heading information are
 AIRCRAFT REMOTE SENSING 473

Figure 1 Cross-section profiles flown across a large oceanic front west of the Galapagos Islands on 25 October and 3 November
1993. (A) A dramatic horizontal displacement of the front as measured with an infrared radiometer; (B) and (C) show corresponding
changes in laser-induced fluorescence of chl-a and phycoerythrin. This flight was made as part of the original IRONEX investigation in
late 1993. NFU, normalized fluorescence units.

used along with the GPS-determined platform pos- topographic surveys exist for most of the Atlantic
ition to locate the position of the laser pulse on the and Gulf coasts between central Maine and Texas
Earth’s surface to a vertical accuracy approaching 10 and for large sections of the Pacific coast. Affected
cm with some highly accurate systems and o30 cm sections of coastline are re-occupied following major
for most of these sensors. The horizontal accuracy is coastal storms, such as hurricanes and ‘Nor’easters’,
generally 50–100 cm. Depending on the pulse repe- to determine the extent of erosion and depositional
tition rate of the laser transmitter, the off-nadir patterns resulting from the storms. Additional details
pointing angle, and the speed of the aircraft plat- on the ATM and some results of investigations on
form, the density of survey points can exceed one coastal morphology can be found on websites (http://
sample per square meter. lidar.wff.nasa.gov and http://aol.wff.nasa.gov/aoltm/
NASA’s Airborne Topographic Mapper (ATM) is projects/beachmap/98results/).
an example of a topographic mapping lidar used for Other airborne lidar systems have been used to
coastal surveying applications. An example of data survey coastal morphology, including Optec lidar
from ATM shows shoreline features off the east coast systems by Florida State University and the
of the USA (Figure 2). ATM was originally developed University of Texas at Austin. Beyond these airborne
to measure changes in the elevation of Arctic ice lidar sensors, there are considerably more instru-
sheets in response to global warming. The sensor was ments with this capability that are currently in use
applied to measurements of changes in coastal in the commercial sector for surveying metropoli-
morphology beginning in 1995. Presently, baseline tan areas, flood plains, and for other terrestrial
474 AIRCRAFT REMOTE SENSING

Figure 2 (Left) Map of coastal topography around Pacifica, CA, USA, derived from lidar data obtained in April 1998, after a winter of
severe storms associated with El Niño. Map insert shows the Esplanade Drive area of Pacifica rendered from lidar data gridded at 2 m
resolution and colored according to elevation. (Right) Cross-sections derived from lidar data of October 1997 and April 1998 at
locations marked in the map inset. The profiles in (A) show a stable cliff and accreting beach, whereas about 200 m to the south the
profiles in (B) show erosion of the sea cliff and adjacent beach resulting in undermining of houses. Each profile shows individual laser
spot elevations that fall within a 2 m wide strip oriented approximately normal to shoreline. (Reproduced with permission from
Sallenger et al., 1999.)

applications. At the last count (early 2000) there technology has matured with the commercial avail-
were approximately 60 airborne lidars in operation ability of FRR instruments that can give vertical pro-
worldwide, with most engaged in a variety of survey files or operate in a continuous mode while underway.
applications generally outside the field of coastal There have been several attempts to develop airborne
mapping. lidar instruments to determine phytoplankton photo-
synthetic characteristics from aircraft in the past dec-
Pump and Probe Fluorometry ade. NASA scientists have deployed a pump and probe
fluorometer from aircraft, wherein the AOL laser
Several sections in this chapter describe recoveries of (described above) acts as the pump and a second laser
phytoplankton biomass as chl-a by active and passive with variable power options and rapid pulsing cap-
measurements. Another recent accomplishment is an abilities (10 ns) functions as the probe.
active, airborne laser measurement intended to aid in
remote detection of photosynthetic performance, an
important ingredient of primary productivity compu- Passive Systems
tations. Fluorometric techniques, such as fast repe-
Multichannel Ocean Color Sensor (MOCS)
tition rate (FRR) fluorometry (explained in another
article of this encyclopedia), have provided an alter- Passive ocean color measurements using visible
native approach to 14C assimilation and O2 evolution radiometers to measure reflected natural sunlight
in the measurement of primary productivity. This from the ocean have been made with a number of
 AIRCRAFT REMOTE SENSING 475

instruments in the past two decades. These instru- 50 m in processing, and interpolated to 1 km2 for
ments include the Multichannel Ocean Color Sensor visualization. Imagery derived from ODAS and SAS
(MOCS) that was flown in studies of Nantucket flights is available on a web site of the NOAA Che-
Shoals in the early 1980s, the passive sensors of the sapeake Bay Office for the main stem of the Bay
AOL suite that have been used in many locations (http://noaa.chesapeakebay.net), and for two con-
around the world, and more recently, simple radi- trasting tributaries, the Choptank and Patuxent Riv-
ometers that have been deployed on light aircraft in ers on a web site of the Coastal Intensive Sites
regional studies of Chesapeake Bay (see below). Network (CISNet) (http://www.cisnet-choptank.org).
MOCS was one of the earliest ocean color sensors Data from ODAS and SAS have provided detailed
used on aircraft. It provided mesoscale data on shelf information on the timing, position, and magnitude
and slope chl-a in conjunction with shipboard studies of blooms in Chesapeake Bay, particularly the spring
of physical structure, nutrient inputs, and phyto- diatom bloom that dominates the annual phyto-
plankton primary productivity. plankton cycle. This April–May peak of chl-a rep-
resents the largest accumulation of phytoplankton
biomass in the Bay and is a proximal indicator of
Ocean Data Acquisition System (ODAS) and
over-enrichment by nutrients. Data from SeaWiFS
SeaWiFS Aircraft Simulator (SAS)
for spring 2000 show the coast-wide chl-a distri-
Few aircraft studies have obtained long time-series bution for context, while SAS III data illustrate the
sufficient to quantify variability and detect secular high-resolution chl-a maps that are obtained re-
trends. An example is ocean color measurements gionally (Figure 3). A well-developed spring bloom
made from light aircraft in the Chesapeake Bay re- corresponding to a year of relatively high freshwater
gion for over a decade, providing data on chl-a and flow from the Susquehanna River, the main tributary
SST from 4250 flights. Aircraft over-flights of the feeding the estuary, is apparent in the main stem Bay
Bay using the Ocean Data Acquisition System chl-a distribution. Estimates of primary productivity
(ODAS) developed at NASA’s Goddard Space Flight are now being derived from shipboard observations
Center commenced in 1989. ODAS was a nadir- of key variables combined with high-resolution air-
viewing, line-of-flight, three-band radiometer with craft measurements of chl-a and SST for the Bay.
spectral coverage in the blue-green region of the
visible spectrum (460–520 nm), a narrow 1.51 field-
of-view, and a 10 Hz sampling rate. The ODAS in- Hyper-spectral Systems
strument package included an IR temperature sensor
Airborne Visible/Infrared Imaging Spectrometer
(PRT-5, Pyrometrics, Inc.) for measuring SST. The
(AVIRIS)
system was flown for B7 years over Chesapeake Bay
on a regular set of tracks to determine chl-a and SST. The Airborne Visible/Infrared Imaging Spectrometer
Over 150 flights were made with ODAS between (AVIRIS) was originally designed in the late 1980s by
1989 and 1996, coordinated with in situ obser- NASA at the Jet Propulsion Laboratory (JPL) to
vations from a multi-jurisdictional monitoring pro- collect data of high spectral and spatial resolution,
gram and other cruises of opportunity. anticipating a space-based high-resolution imaging
ODAS was flown together with the SeaWiFS Air- spectrometer (HIRIS) that was planned for launch in
craft Simulator (SAS II, III, Satlantic, Inc., Halifax, the mid-1990s.
Canada) beginning in 1995 and was retired soon Because the sensor was designed to provide data
thereafter and replaced with the SAS units. SAS III is similar to satellite data, flight specifications called for
a multi-spectral (13-band, 380–865 nm), line-of- both high altitude and high speed. AVIRIS flies al-
flight, nadir viewing, 10 Hz, passive radiometer with most exclusively on a NASA ER-2 research aircraft
a 3.51 field-of-view that has the same wavebands as at an altitude of 20 km and an airspeed of 732 km
the SeaWiFS satellite instrument, and several add- h1. At this altitude and a 301 field of view, the swath
itional bands in the visible, near IR, and UV. The SAS width is almost 11 km. The instantaneous field of
systems include an IR temperature sensor (Heimann view is 1 mrad, which creates individual pixels at a
Instruments, Inc.). Chl-a estimates are obtained using resolution of 20 m2. The sensor samples in a whisk-
a curvature algorithm applied to water-leaving radi- broom fashion, so a mirror scans back and forth,
ances at wavebands in the blue-green portion of the perpendicular to the line-of-flight, at a rate of 12
visible spectrum with validation from concurrent times per second to provide continuous spatial cov-
shipboard measurements. Flights are conducted at erage. Each pixel is then sent to four separate spec-
B50–60 m s1 (100–120 knots), giving an along- trometers by a fiber optic cable. The spectrometers
track profile with a resolution of 5–6 m averaged to are arranged so that they each cover a part of the
476 AIRCRAFT REMOTE SENSING

Chesapeake Bay chlorophyll

Mid-Atlantic chlorophyll 14 May 2000
8 May 2000
64.0 _
39˚30' Chl [mg m 3]
40

39 10.0 80

38 60
1.0 39˚00'
40
37

0.1 30
36

25
38˚30'
35
0.01
20
34 76 74 72 70
(A) 16

38˚00' 12

10

37˚30'
6

2
37˚00'

(B) 77˚00' 76˚30' 76˚00' 75˚30'

Patuxent chlorophyll
1 May 2000

38˚35'

Choptank chlorophyll
1 May 2000

38˚45'
38˚30'

38˚40'
38˚25'

38˚35'
38˚20'

(C) 76˚40' 76˚30' 76˚20' 76˚10' 76˚00'

(D)

Figure 3 Spring chl-a (mg m3) in: (A) the mid-Atlantic region from SeaWiFS; (B) Chesapeake Bay; (C) Patuxent R; (D) Choptank R
from SAS III.

spectrum from 0.40 to 2.4 mm, providing continuous flight parameters, this creates an image roughly
spectral coverage at 10 nm intervals over the entire 500 km long and 11 km wide. Data are encoded at
spectrum from visible to near IR. Data are recorded 12-bits for a high degree of discrimination. The
to tape cassettes for storage until rectification, at- physical dimensions of AVIRIS are quite large, 84 cm
mospheric correction, and processing at JPL. A wide  160 cm long  117 cm tall at a weight of 720
typical AVIRIS ‘scene’ is a 40 min flight line. At ER-2 pounds.
 AIRCRAFT REMOTE SENSING 477

Data collected with AVIRIS have been used for format detector arrays that have enabled increased
terrestrial, marine, and atmospheric applications. spectral resolution and higher signal-to-noise ratios
Accomplishments of AVIRIS include separation of for imaging spectrographs, extending the use of this
the chl-a signature from bottom reflectance for clear technology in low-albedo coastal waters. The ocean
lake waters of Lake Tahoe and turbid waters near PHILLS operates in a push-broom scanned mode
Tampa Bay, interpretation of spectral signals from whereby cross-track ground pixels are imaged with a
resuspended sediment and dissolved organic ma- camera lens onto the entrance slit of the spec-
terials in W. Florida, and of suspended sediment and trometer, and new lines of the along-track ground
kelp beds in S. California. Recent efforts have fo- pixels are attained by aircraft motion. The Navy’s
cused on improving atmospheric correction pro- interest in hyper-spectral imagers for coastal appli-
cedures for both AVIRIS and satellite data, providing cations centers on the development of methods for
inputs for bio-optical models which determine in- determining shallow water bathymetry, topography,
herent optical properties (IOPs) from reflectance, bottom type composition, underwater hazards, and
algorithm development, and sporadic attempts at visibility. PHILLS precedes a planned hyper-spectral
water quality monitoring (e.g., chl-a, suspended satellite instrument, the Coastal Ocean Imaging
sediment, diffuse attenuation coefficient, kd). AVIRIS Spectrometer (COIS) that is planned to launch on the
data have recently been used as an input variable to a Naval Earth Map Observer (NEMO) spacecraft.
neural network model developed to estimate water
depth. The model was able to separate the contri-
butions of different components to the total water- Radar Altimetry
leaving radiance and to provide relatively accurate
Ocean applications of airborne radar altimetry sys-
estimates of depth (rms error ¼ 0.48 m).
tems include several sensors that retrieve information
on wave properties. Two examples are the Radar
Compact Airborne Spectrographic Imager (CASI)
Ocean Wave Spectrometer (ROWS), and the Scan-
The Compact Airborne Spectrographic Imager ning Radar Altimeter (SRA), systems designed to
(CASI) is a relatively small, lightweight hyper-spec- measure long-wave directional spectra and near-sur-
tral sensor that has been used on a variety of light face wind speed. ROWS is a Ku-band system de-
aircraft. CASI was developed by Itres Research Ltd veloped at NASA’s Goddard Space Flight Center in
(Alberta, Canada) in 1988 and was designed for a support of present and future satellite radar missions.
variety of remote sensing applications in forestry, Data obtained from ROWS in a spectrometer mode
agriculture, land-use planning, and aquatic moni- are used to derive two-dimensional ocean spectral
toring. By allowing user-defined configurations, the wave estimates and directional radar backscatter.
12-bit, push-broom-type sensor (333 scan lines per Data from the pulse-limited altimeter mode radar
second) using a charge-coupled detector (CCD) can yield estimates of significant wave height and surface
be adapted to maximize either spatial (37.81 across wind speed.
track field of view, 0.0771 along-track, 512 pixels –
pixel size varies with altitude) or spectral resolution
(288 bands at 1.9 nm intervals between 400 and Synthetic Aperture Radar (SAR)
1000 nm). Experiments have been conducted using
Synthetic Aperture Radar (SAR) systems emit
CASI to determine bottom type, benthic cover, sub-
microwave radiation in several bands and collect the
merged aquatic vegetation, marsh type, and in-water
reflected radiation to gain information about sea
constituents such as suspended sediments, chl-a, and
surface conditions. Synthetic aperture is a technique
other algal pigments.
that is used to synthesize a long antenna by com-
bining signals, or echoes, received by the radar as it
Portable Hyper-spectral Imager for Low-Light
moves along a flight track. Aperture refers to the
Spectroscopy (PHILLS)
opening that is used to collect reflected energy and
The Portable Hyper-spectral Imager for Low-Light form an image. The analogous feature of a camera to
Spectroscopy (PHILLS) has been constructed by the the aperture would be the shutter opening. A syn-
US Navy (Naval Research Laboratory) for imaging thetic aperture is constructed by moving a real
the coastal ocean. PHILLS uses a backside-illumin- aperture or antenna through a series of positions
ated CCD for high sensitivity, and an all-reflective along a flight track.
spectrograph with a convex grating in an Offner NASA’s Jet Propulsion Laboratory and Ames Re-
configuration to produce a distortion-free image. The search Center have operated the Airborne SAR
instrument benefits from improvements in large- (AIRSAR) on a DC-8 since the late 1980s. The radar
478 AIRCRAFT REMOTE SENSING

of AIRSAR illuminates the ocean at three microwave Microwave Radiometer (SLFMR), and the Elec-
wavelengths: C-band (6 cm), L-band (24 cm), and P- tronically Thinned Array Radiometer (ESTAR).
band (68 cm). Brightness of the ocean (the amount of SLFMR was used recently in the estuarine plume
energy reflected back to the antenna) depends on the of Chesapeake Bay on the east coast of the USA to
roughness of the surface at the length scale of the follow the buoyant outflow that dominates the near-
microwave (Bragg scattering). The primary source of shore density structure and constitutes an important
roughness, and hence brightness, at the wavelengths tracer of water mass movement. SLFMR is able to
used is capillary waves associated with wind. recover SSS at an accuracy of about 1 PSU (Figure 4).
Oceanographic applications derive from the respon- This resolution is too coarse for the open ocean, but
siveness of capillary wave amplitude to factors that is quite suitable for coastal applications where sig-
affect surface tension, such as swell, atmospheric nificant gradients occur in regions influenced by
stability, and the presence of biological films. For freshwater inputs. SLFMR has a bandwidth of 25
example, the backscatter characteristics of the ocean MHz, a frequency of 1.413 GHz, and a single an-
are affected by surface oil and slicks can be observed tenna with a beam width of approximately 161 and
in SAR imagery as a decrease of radar backscatter; six across-track positions at 761, 7221 and 7391.
SAR imagery appears dark in an area affected by an Tests of SLFMR off the Chesapeake Bay demon-
oil spill, surface slick, or biofilm, as compared with strated its effectiveness as a ‘salinity mapper’ by
areas without these constituents. characterizing the trajectory of the Bay plume from
surveys using light aircraft in joint operations with
ships. Flights were conducted at an altitude of 2.6
km, giving a resolution of about 1 km. The accuracy
Microwave Salinometers of SSS in this example is B0.5 PSU.
Passive microwave radiometry (L-band) has been ESTAR is an aircraft instrument that is the
tested for the recovery of SSS from aircraft, and it may prototype of a proposed space instrument for meas-
be possible to make these measurements from space. uring SSS. This instrument relies on an interfero-
Salinity affects the natural emission of EM radiation metric technique termed ‘aperture synthesis’ in the
from the ocean, and the microwave signature can be across-track dimension that can reduce the size of the
used to quantify SSS. Two examples of aircraft in- antenna aperture needed to monitor SSS from space.
struments that have been used to measure SSS in the It has been described as a ‘hybrid of a real and a
coastal ocean are the Scanning Low-Frequency synthetic aperture radiometer.’ Aircraft surveys of

(PSU)

Figure 4 Sea surface salinity from an airborne microwave salinity instrument for (A) 14 September 1996; (B) 20 September 1996.
Images reveal strong onshore-offshore gradients in salinity from the mouth of Chesapeake Bay to the plume and shelf, and the effect
of high rainfall and freshwater input on the salinity distribution over a 1-week interval. (Adapted with permission from Miller et al.,
1998.)
 AIRCRAFT REMOTE SENSING 479

SSS using ESTAR in the coastal current off Maryland Campbell JW and Esaias WE (1985) Spatial patterns in
and Delaware showed good agreement with ther- temperature and chlorophyll on Nantucket Shoals from
mosalinograph measurements from ships in the airborne remote sensing data, May 7–9, 1981. Journal
range of 29–31 PSU. of Marine Research 43: 139--161.
Harding LW Jr, Itsweire EC, and Esaias WE (1994)
Estimates of phytoplankton biomass in the Chesapeake
Bay from aircraft remote sensing of chlorophyll
See also concentrations, 1989–92. Remote Sensing Environment
49: 41--56.
Bio-Optical Models. Fluorometry for Biological
Le Vine DM, Zaitzeff JB, D’Sa EJ, et al. (2000) Sea surface
Sensing. Inherent Optical Properties and
salinity: toward an operational remote-sensing system.
Irradiance. IR Radiometers. Optical Particle
In: Halpern D (ed.) Satellites, Oceanography and
Characterization. Satellite Oceanography, History
Society, pp. 321--335. Elsevier Science.
and Introductory Concepts. Satellite Remote
Miller JL, Goodberlet MA, and Zaitzeff JB (1998)
Sensing of Sea Surface Temperatures. Satellite
Airborne salinity mapper makes debut in coastal zone.
Remote Sensing SAR.
EOS Transactions of the American Geophysical Union
79: 173--177.
Sallenger AH Jr, Krabill W, Brock J, et al. (1999) EOS
Further Reading Transactions of the American Geophysical Union 80:
89--93.
Blume H-JC, Kendall BM, and Fedors JC (1978)
Sandidge JC and Holyer RJ (1998) Coastal bathymetry
Measurement of ocean temperature and salinity via
from hyper-spectral observations of water radiance.
microwave radiometry. Boundary Layer Meteorology
Remote Sensing Environment 65: 341--352.
13: 295--308.
 IR RADIOMETERS
C. J. Donlon, Space Applications Institute, principles described are applicable to satellite sensors
Ispra, Italy treated elsewhere in this volume.
Copyright & 2001 Elsevier Ltd.

Infrared Measurement Theory

Infrared (IR) radiation is heat energy that is emitted
from all objects that have a temperature above 0 K
Introduction (  273.161C). It includes all wavelengths of the
electromagnetic spectrum between 0.75 mm and
Measurements of sea surface temperature (SST) are
B100 mm (Figure 1) and has the same optical prop-
most important for the investigation of the processes
erties as visible light, being capable of reflection,
underlying heat and gas exchange across the air–sea
refraction, and forming interference patterns.
interface, the surface energy balance, and the general
The following total quantities, conventional sym-
circulation of both the atmosphere and the oceans.
bols and units provide the theoretical foundation for
Complementing traditional subsurface contact tem-
the measurement of IR radiation and are schemat-
perature measurements, there is a wide variety of
ically shown in Figure 2. Spectral quantities can be
infrared radiometers, spectroradiometers, and ther-
represented by restricting each to a specific
mal imaging systems that can be used to determine
waveband.
the SST by measuring thermal emissions from the sea
surface. However, the SST determined from thermal • Radiant energy Q, is the total energy radiated
emission can be significantly different from the sub- from a point source in all directions in units of
surface temperature (471 K) because the heat flux joules (J).
passing through the air–sea interface typically results • Radiant flux f ¼ dQ/dt is the flux of all energy
in a strong temperature gradient. Radiometer sys- radiated in all directions from a point source in
tems deployed on satellite platforms provide daily units of watts (W).
global maps of SSST (sea surface temperature) at • Emittance M ¼ df/dA is the radiant flux density
high spatial resolution (B1 km) whereas those de- from a surface area A in units of W m2. This is an
ployed from ships and aircraft provide data at small integrated flux (i.e., independent of direction) and
spatial scales of centimeters to meters. In particular, will therefore vary with orientation relative to a
the development of satellite radiometer systems nonuniform source.
providing a truly synoptic view of surface ocean • Radiant intensity I ¼ df/do is the radiant flux of a
thermal features has been pivotal in the description point source per solid angle o (steradian, sr) and is
and understanding of the global oceans. a directional flux in units of W sr1.
This article reviews the infrared properties of • Radiance L ¼ dI/d(A cos y) is the radiant intensity
water and some of the instruments developed to of an extended source per unit solid angle in a
measure thermal emission from the sea surface. It given direction y, per unit area of the source pro-
focuses on in situ radiometers although the general jected in the same y. It has units of W sr1 m2.

Figure 1 Schematic diagram of the electromagnetic spectrum showing the location and interval of the infrared waveband.

480
 IR RADIOMETERS 481

y (surface normal)

Radiant
flux 

Surface area, A

Figure 3 Black-body radiance as a function of wavelength

computed using eqns [1] and [2] for different target temperatures.
z
(A)
where h ¼ 6.626  1034 J s is Planck’s constant,
y c ¼ 2.998  108 m s1 is the speed of light and
Area = A k ¼ 1.381  1023 J K1 is Boltzmann’s constant. The
Solid angle Radiant sea surface is considered a Lambertian source (i.e.,
 = A /R
2 flux  uniform radiance in all directions) so that the spec-
tral radiance Ll is related to Ml by

Ml
Ll ¼ ½2
R
p
x
Figure 3 shows Ll computed for several tempera-
tures as a function of wavelength. Considering tem-
peratures of 273–310 K as representative of the
z
(B)
global ocean, maximum emission occurs at a wave-
length of 9.3–10.7 mm. Atmospheric attenuation is
y (surface normal) minimal at B3.5 mm, 9.0 mm and 11.0 mm that are
the spectral intervals often termed atmospheric
d
Radiant ‘windows’. Instruments operating within these
flux  intervals are optimal for sea surface measurements –
especially in the case of satellite deployment where
Surface area, A 
atmospheric attenuation can be significant. In the
Direction,  3–5 mm spectral region. Ll is a strong function of
x
temperature (Figure 3) highlighting the possibility to
increase in radiometer sensitivity by utilizing this
Projected source
area, dA cos  spectral interval.
z By measuring Ll using eqn [2] and inverting eqn
(C) [1], the spectral brightness temperature, B(T,l), rather
than the temperature is determined because these
Figure 2 Schematic definition of (A) Emittance, E; (B) radiant
intensity, I; (C) radiance, L.
equations assume that sea water is a perfect emitter
or black body. In practice, the sea surface does not
Planck’s law describes the emittance of a perfectly behave as a black body (it is slightly reflective in the
emitting surface (or black body) at a temperature T infrared) and therefore its spectral and geometric
in Kelvin. It is the radiant flux (f) per unit band- properties need to be considered. The emittance of a
width centered at wavelength l leaving a unit area of perfect emitter at the actual temperature T, wave-
surface in any direction in units of W m2 m1. length l, and view angle y is given by

2phc2 pLðT;l;yÞ
Ml;T ¼ ½1 MðT;l;yÞ ¼ ½3
l5 ðehc=lkT  1Þ eðl;yÞ
482 IR RADIOMETERS

where the emissivity, eðl;yÞ , can be calculated using B11 mm (rE0.0015) and following Kirchoff’s law

MðT;l;yÞ measured rl þ tl þ el ¼ 1 ½5

eðl;yÞ ¼ ½4
MðT;l;yÞ blackbody
where rl is the spectral reflectivity and tl is the
spectral transmissivity. The e-folding penetration
which has a strong dependence on wavelength and depth (i.e., the depth of 63% emission) or optical
viewing geometry. The effective emissivity, e, inte- depth at a typical wavelength of 11 mm is E 10 mm,
grates eðl;yÞ over all wavelengths of interest for radi- tl can be neglected and el can be calculated using
ometer view angle y. Figure 4A shows the calculated
normal reflectivity, r, of pure water as a function of el ¼ 1  rl ½6
wavelength for the spectral region 1–100 mm. Pure
water differs only slightly from sea water in this Although the actual optical depth is wavelength
context. Note that r is minimal at a wavelength of dependent, it is clear that IR radiometers determine

0.100
 (zenith angle = 40˚)

0.010

(h+v)
h
v

0.001
10 100
(A) Wavelength (μm)

1.00
0˚
40˚
Sea surface emissivity ( = 1 _ )

50˚
0.95 60˚
70˚

0.90

0.85

0.80
10 100
(B) Wavelength (μm)

Figure 4 (A) The normal reflectivity, r, of the pure water as a function of wavelength (full line). Also shown are the horizontal
polarized (rh) and the vertically polarized (rv) components of r. (B) The spectral emissivity of pure water as a function of viewing zenith
angle. h, height above sea surface.
 IR RADIOMETERS 483

Figure 5 Schematic diagram showing the radiance components measured by an IR radiometer viewing the sea surface. Fov, Field
of view.

the temperature of a very thin ‘skin’ layer of the surface because diffuse downwelling sky radiance
ocean. This temperature is termed the sea surface measured by a radiometer after reflection at the sea
skin temperature (SSST) and is distinct (although surface is polarized.
related) to the subsurface SST. Note that this is in
contrast to the situation for short-wave solar radi-
ation (having wavelengths of B0.4–0.7 mm) which, IR Radiometer Design
for clear water, penetrates to a depth of B100 m.
Figure 4B shows the el of pure water computed There are four fundamental components to all IR
from eqn [6] as a function of both viewing zenith radiometer instruments described below.
angle and spectral wavelength. Inspection of Figure 4
reveals that the best radiant temperature measure- Detector and Electronics System to Measure
ment will be made when viewing a calm sea surface Radiance and Control the Radiometer
at an angle of 0–401 from nadir.
A detector system provides an output proportional to
An IR radiometer is an optical instrument de-
the target radiance incident on the detector. There
signed to measure Ll entering an instrument aperture
are two main types of detector: thermal detectors
(Figure 5). The radiance measured by radiometer,
that respond to direct heating and quantum detectors
LðT;l;yÞ , having a spectral bandwidth l, viewing the
that respond to a photon flux. In general, thermal
sea surface at a zenith angle y and temperature T is
detectors have a response that is weakly dependent
given by:
on wavelength and can be operated at ambient
Z temperatures whereas rapid response quantum de-
a     tectors require cooling and are wavelength
LðT;l;y ¼ xl ½eðl;yÞ B Tsurf ; l þ 1  eðl;yÞ BðTatm ; lÞ
0 dependent.
þLpathðh;l;yÞ dl
½7 Fore-optics System to Filter, Direct and Focus
Radiance
where xl is the spectral response of the radiometer, All optical components have an impact on radi-
BðTsurf ; l and BðTatm ; l are the Planck function for ometer reliability and accuracy. Mirrors should be
surface temperature Tsurf, and atmospheric tem- free of aberration to minimize unwanted stray radi-
perature Tatm, and Lpathðh;l;yÞ is the radiance emitted ance reaching the detector. Several materials have
by the atmosphere between the radiometer at height good reflection characteristics in the IR including,
h above the sea surface reflected into the radiometer gold, polished aluminum, and cadmium. Care should
field-of-view (FoV) at the sea surface. Note that the be exercised when choosing an appropriate mirror
horizontal and vertical polarization components of substrate and reflection coating to avoid decay in the
reflectivity shown in Figure 4 are unequal. It is im- marine atmosphere. A glass substrate having a ‘hard’
portant to consider the polarization of surface re- scratch-resistant polished gold surface provides
flectance when making measurements of the sea 498% reflectance and good environmental wear.
484 IR RADIOMETERS

1.0
Detector transmission
Window transmission
Total transmission

0.8
Normalized transmission

0.6

0.4

0.2

Figure 7 Spectral transmission for common IR window

0.0 materials. (A) Germanium, (B) sodium chloride, (C) zinc selenide.
6 8 10 12 14 16
Waveband (μm)

Figure 6 Normalized spectral transmission of an IR window

and resistant to thermal shock but is soft and re-
and detector shown together with the combined total response. quires a protective ‘hard’ surface finish coating.
Certain window materials achieve better per-
formance when antireflection (AR) coatings are used
Spectral filter windows and lenses require spectral to minimize reflection from the window. For ex-
properties that, together with the detector charac- ample, when an AR coating is used on a ZnSe win-
teristics, define the overall spectral characteristics of dow the transmission increases from B70% to
a radiometer. Figure 6 shows the combined spectral B90%. Other coatings provide windows that po-
response for a broadband radiometer together with larize the incident radiance signal such as optically
the component spectral response of the window and thin interference coatings and wire grid diffraction
detector. polarizers.
Deposition of marine NaCl on all optical com-
Environmental System to Protect and Thermally ponents (especially calibration targets) presents an
Stabilize the Radiometer unavoidable problem. Although NaCl itself has good
infrared transmission properties (Figure 7), con-
For any optical instrument intended for use in the taminated surfaces may become decoupled from
harsh marine environment, adequate environmental temperature sensors and the noise introduced by the
protection is critical. Rain, sea-water spray, and high optical system will increase. Finally, adequate ther-
humidity can destroy a poorly protected instrument mal control using reflective paint together with sub-
rapidly and components such as electrical con- stantial instrument mass is required so that
nections and fore-optics should be resistant to these instruments are not sensitive to thermal shock. In
effects. In situ radiometer windows are particularly higher latitudes, it may be necessary to provide an
important in this context. They should not signifi- extensive antifreeze capability.
cantly reduce the incoming signal or render it noisy,
and be strong enough to resist mechanical, thermal
Calibration System to Quantify the Radiometer
and chemical degradation. Figure 7 describes several
Output
common materials. Germanium (Ge) windows have
good transmission characteristics but are very brittle. The role of a calibration system is to quantify the
Sodium chloride (NaCl) is a low-cost, low-ab- instrument output in terms of the measured radiance
sorption material but is of little use in the marine incident on the detector. Calibration techniques are
environment because it is water-soluble. Zinc selen- specific to the particular design of radiometer and
ide (ZnSe) has high transmission, is nonhygroscopic vary considerably from simple bias corrections to
 IR RADIOMETERS 485

systems providing automatic precision two-point radiance relative to a stable reference junction and a
blackbody calibrations. Proper calibration accounts mV signal is produced. The response of a thermopile
for the following primary sources of error: has little wavelength dependence and a hemispheric
dome having a filter (B3–50 mm) deposited on its
• the effect of fore-optics;
inner surface typically defines the spectral response.
• unavoidable drifts in detector gain and bias;
Direct compensation for thermal drift using a tem-
• long-term degradation of components.
perature sensor located at the thermopile reference
Finally, careful radiometer design and configuration junction is sometimes used but regular calibration
can avoid many measurement errors. Examples in- using an independent laboratory blackbody is man-
clude: poor focusing and optical alignment; filters datory. An accuracy of o10 W m2 is possible after
having transmission above or below the stated spec- significant correction for instrument temperature
tral bandwidth (termed ‘leaks’); inadequate pro- drift and stray radiance contribution using additional
tection against thermal shock, stability, and reliability on-board temperature sensors.
of the calibration system; poor electronics and the
general decay of optomechanical components. Narrow Beam Filter Radiometers
Narrow beam filter radiometers are often used to
Application of IR Radiometers determine the SSST for air–sea interaction studies
and for the validation of satellite derived SSST and
There are many different radiometer designs and
there are several low-cost instruments that provide
deployment strategies ranging from simple single-
suitable accuracy and spectral characteristics. Many
channel hand-held devices to complex spectro-
of these use a simple thermopile or thermistor de-
radiometers. Instrument accuracy, sensitivity and
tector together with a low-cost broadband-focusing
stability depends on both the deployment scenario
lens. Typically, they have simple self-calibration
and radiometer design. Accordingly, the following
techniques based on the temperature of the instru-
sections describe several different examples.
ment and/or detector. Consequently they have poor
resistance to thermal shock and have fore-optics that
Broadband Pyrgeometers
readily degrade in the marine atmosphere. However,
A pyrgeometer is an integrating hemispheric radi- handled with care, these devices are accurate to
ometer which, by definition, measures the band- 70.1 K, albeit with limited sensitivity.
limited spectral emittance, E, so that the angle de- Precision narrow beam filter radiometers often use
pendency in eqn [7] is redundant. They are used to pyroelectric detectors that produce a small electrical
determine the long-wave heat flux at the air–sea current in response to changes in detector temperature
interface by measuring the difference between at- forced by incident radiation. They have a fast response
mospheric and sea surface radiance either using two at ambient temperatures but require a modulated sig-
individual sensors (Figure 8A) or as a single com- nal to operate. Modulation is accomplished by using
bined sensor (Figure 8B). A thermopile detector is an optical chopper having high reflectivity ‘vanes’ to
often used which is a collection of thermocouple alternately view a reference radiance source by re-
detectors composed of two dissimilar metallic con- flection and a free path to the target radiance. The
ductors connected together at two ‘junctions.’ The most common chopper systems are rotary systems
measurement junction is warmed by incident driven by a small electric motor phase locked to the

Figure 8 (A) A typical design of a long-wave pyrgeometer. (B) A net-radiation pyrgeometer for determination of the net long-wave
flux at the sea surface.
486 IR RADIOMETERS

Figure 9 (A) Schematic layout of a rotary chopper; (B) schematic layout of a tuning fork chopper, (a) open, (b) closed.

detector output by an optoelectronic sensor Figure 9A. Calibration cycles should be made at regular inter-
An alternative design driven by a small oscillating vals so that gain changes can be accurately moni-
electromagnetic coil called a tuning fork chopper is tored and calibration sources need to be viewed
shown in Figure 9B. As the coil resonates, the re- using the same optical path as that used to view the
flective vanes of the chopper oscillate alternately sea surface.
opening and closing an aperture ‘gap.’ A basic ‘black-body’ calibration strategy uses an
Dynamic detector bias compensation is inherent external bath of sea water as a high e (40.95) refer-
when using an optical chopper. The detector alter- ence as shown in Figure 10. In this scheme, the radi-
nately measures radiance from the sea surface Lsrc ometer periodically views the water bath that is stirred
and a reference blackbody, Lbb (sometimes this is the vigorously to prevent the development of a thermal
detector itself) reflected by the chopper vanes re- skin temperature deviation. The view geometry for the
sulting in two signals water bath and the sea surface are assumed to be
identical and, by measuring the temperature of the
S1 ¼ Lbb þ d ½8
water bath the radiometer can be absolutely cali-
brated. An advantage of this technique is that eðl;yÞ is
S2 ¼ Lsrc þ d ½9
not required to determine the SSST. However, in
Assuming Lbb remains constant during a short practice, it is difficult to continuously operate a water
chopping cycle, the bias term d in eqns [8] and [9] is bath at sea and surface roughness differences between
eliminated the bath and sea surface are ignored.
DS ¼ S1  S2 ¼ Lbb  Lsrc ½10 On reflection at the sea surface, diffuse sky radi-
ance is polarized and, at Brewster’s angle (B501
It is important to recognize the advantages to this from nadir at a wavelength of 11 mm), the vertical v-
technique, which is widely used: polarization is negligible for a given wavelength
(Figure 11). Only the horizontal h-polarization
• There is minimal thermal drift of the detector; component remains so that if the radiometer filter
• The detector is dynamically compensated for response is v-polarized (i.e., only passes v-polarized
thermal shock; radiance), negligible reflected sky radiance is meas-
• A precise modulated signal is generated well sui- ured by the radiometer. In practice, because Brew-
ted to selective filtering providing excellent noise ster’s angle is very sensitive to the geometry of a
suppression and signal stability. particular deployment (approximately721) this
technique is only applicable to deployments from
However, in order to compensate for instrument fixed platforms and when the sea surface is relatively
gain changes, an additional blackbody sources(s) is calm. Further, the use of a polarizing filter will sig-
required. These are periodically viewed by the de- nificantly reduce the signal falling on the detector
tector to provide a mechanism for absolute cali- increasing the signal-to-noise ratio.
bration. Either the black body is moved into the The use of fabricated black-body cavities
detector FoV or an adjustable mirror reflects radi- (Figure 12A) provides an accurate, versatile
ance from the black body on to the detector. and, compact calibration system. Normally, two
 IR RADIOMETERS 487

Radiometer Radiometer beam

Inner water bath Water overflow Water bath moved out from
radiometer FoV which no
views the sea surface
Gimball mounting

Overflow drain

Sea-water input

(A) (B)

Figure 10 Schematic diagram showing the stirred water bath calibration scheme. (A) The radiometer in calibration mode, (B) the
radiometer viewing the sea surface after the water bath has been moved out of the field of view (FoV).

black-body cavities are used, one of which follows and Figure 12C provides a schematic diagram of a
the ambient temperature of the instrument and a typical output signal.
second is heated to a nominal temperature above Note that for all calibration schemes, larger errors
this. High e (40.99) is attained by a combination of are expected beyond the calibrated temperature
specialized surface finish and black-body geometry. range which can be a problem for sky radiance
The cavity radiance is determined as a function of the measurements where clear sky temperatures of
black body temperature that is easily measured. o200 K are common.
Figure 12B shows a schematic outline of a typical
black-body radiometer design using a rotary chopper
Multichannel Radiometers
The terms in eqn [7] are directly influenced by the
 at  = 11μm for angles 0 _ 90˚ height, h, of the radiometer above the sea surface and
1.0000
are different in magnitude for in situ and spacecraft
deployments. In the case of a sea surface in situ
radiometer deployment, Lpathðh;l;yÞ is typically neg-
lected because h is normally o10 m unless the at-
0.1000
mosphere has a heavy water vapor loading (e.g.,
490%) or an aircraft deployment is considered.
However, for a spacecraft deployment, this is a sig-
nificant term requiring explicit correction. Con-
 at  = 11 μm

versely, the BðTatm ; lÞ term is critical to the accuracy

0.0100 of an in situ radiometer deployment but of little
impact (except perhaps at the edge of clouds) for a
satellite instrument deployment because Lpathðh;l;yÞ
dominates the signal. A multispectral capability can
be used to explicitly account for Lpathðh;l;yÞ in eqn [7]
0.0010 because of unequal atmospheric attenuation for dif-
ferent spectral wavebands. Multichannel radiometers
are exclusively used on satellite platforms for this
reason. Many in situ multichannel radiometers are
h
v designed primarily for the radiant calibration or
0.0001 validation of specific satellite radiometers and the
0 20 40 60 80 development of satellite radiometer atmospheric
Zenith angle (˚)
correction algorithms. They often have several se-
Figure 11 Polarization of sea surface reflection at 11 mm as a lectable filters matched to those of the satellite
function of view angle. Total polarization is shown as a solid line. sensor.
488 IR RADIOMETERS

Figure 12 (A) A section through a black-body calibration cavity of the re-entrant cone design. (B) A typical black-body calibration
radiometer using a rotary optical chopper. (C) A schematic diagram of a typical detector output signal, showing sea, reference, and
calibration signals.

It is worth noting that multiangle view radi-

ometers are also capable of providing an explicit
correction for atmospheric attenuation. Often oper-
ated from satellite and aircraft, these instruments or satellite
provide a direct measure of atmospheric attenuation
by making two views of the same sea surface area at
different angles using a geometry that doubles the
atmospheric pathlength (Figure 13). The assumption
is made that atmospheric and oceanic conditions are
stationary in the time between each measurement.

Spectroradiometers
1+dt e=t1
A recent development is the use of Fourier transform
infrared spectrometers (FTIR) that are capable of Figure 13 Schematic diagram of dual view, double
accurate (B0.05 K). High spectral resolution atmospheric path radiometer deployment geometry.
 IR RADIOMETERS 489

(B0.5 cm1) measurements over a broad spectral spectrum to minimize the band-limited variance of
range (typically B3–18 mm) as shown in Figure 14. the BðTsea ; lÞ spectrum (Figure 14B).
The FTIR provides a unique tool for the development Finally, direct measurement of the thermal gradi-
of new IR measurement techniques and investigation ent at the air–sea interface to obtain the net heat flux
of processes at the air–sea interface. For example, the has been demonstrated using an FTIR in the la-
Marine-Atmospheric Emitted Radiance Interfer- boratory. The FTIR uses the 3.3–4.1 mm spectral
ometer (M-AERI) has pioneered a SSST algorithm interval that has an effective optical depth (EOD)
that uses a narrow spectral region centered at 7.7 mm depending on the wavelength (EOD ¼ 0 mm at 3.3 mm
that is less susceptible to the influence of cloud cover whereas at 3.8 mm EOD ¼ 65 mm) demonstrating the
and sky emissions that at 10–12 mm. The FTIR can versatility of the FTIR. However, measurement in-
also be used to measure air temperatures by viewing tegration times are long and further progress is re-
the atmosphere at B15 mm (a spectral region opaque quired before this technique is applicable for normal
due to CO2 emission) that are accurate to o0.1 K. field operations.
Of considerable interest is the ability of an FTIR
provide an indirect estimate of rðl;yÞ so that by using
Thermal Imagers
eqn [5], eðl;yÞ can be computed. The sky radiance
spectrum has particular structures associated with Another recent development is the application of IR
atmospheric emission–absorbance lines (Figure 14A) imagers and thermal cameras for high-resolution
that are physically uncorrelated with the smooth process studies such as fine-scale variability of SSST,
spectrum of rðl;yÞ (Figure 4). The spectrum of rðl;yÞ wave breaking (Figure 15), and understanding air–
can be derived by subtracting a scaled BðTarm ; lÞ sea gas and heat transfer. They are also used during

Figure 14 Spectra of emitted sky and sea view radiation measured by the M-AERI FTIR in the tropical Western Pacific Ocean on
March 24, 1996. (A) Spectrum of sky radiance and (B) spectrum of corresponding sea radiance Sky measurements were made at 451
and zenith (red) above the horizon and ocean measurements were made at 451 below the horizon. The cold temperatures in the sky
spectra show where the atmosphere is relatively transparent. The ‘noise’ in the 5.5–7 mm range is caused by the atmosphere being so
opaque that the radiometer does not ‘see’ clearly the instrument internal black-body targets and calibration is void. The spectrum of
upwelling radiation (B) consists of emission from the sea surface, reflected sky emission and emission from the atmospheric
pathlength between the sea surface and the radiometer.
490 IR RADIOMETERS

Figure 15 A thermal image of a breaking wave. Each pixel is B25  25 cm and the wavelength of the camera is 8–12 mm. (Courtesy
of D. Woolf.)

air–sea rescue operations providing a nighttime eventually provide extremely versatile instru-
capability for detecting warm objects such as a life mentation for the investigation of fine-scale sea sur-
raft or survivors. face emission.
These instruments use a focal plane array (FPA)
detector (a matrix of individual detectors, e.g.,
256  256) located at the focal point of incoming
radiance together with a charge couple device that is
Future Direction and Conclusions
used to ‘read’ the FPA. This type of ‘staring array’ In the last 10 years, considerable progress has been
system generates a two-dimensional image either by made in the development and application of IR
using a mechanical scanning system (in the case of a sensors to study the air–sea interface. The continued
small FPA array) or as an instantaneous image. development and use of FTIR sensors will provide
Larger FPA arrays are much more power efficient, the capability to accurately investigate the spectral
lighter and smaller than more elaborate mechanical characteristics of the sea surface in order to optimize
scanning systems. Rapid image acquisition (415 the spectral intervals used by space sensors to de-
frames s1) is typical of these instruments that are termine SSST. It can be expected that in the near
available in a wide variety of spectral configurations future, new algorithms will emerge for the direct
and a typical accuracy of B70.1 K. However, con- measurement of the air–sea heat flux using multi-
siderable problems are encountered when obtaining spectral sounding techniques and the accurate in situ
sky radiance data due to the difficulty of geometric- determination of sea surface emissivity. Although
ally matching sea and sky radiance data. still in their infancy, the development and use of
The major problem with FPA detector technology thermal cameras will provide valuable insight into
is nonuniformity between FPA detector elements and the fine-resolution two-dimensional spatial and
drifts in detector gain and bias. Many innovative temporal variability of the ocean surface. These data
self-calibration methods which range in quality are will be useful in developing and understanding the
used to correct for these problems. For example, a sampling limitations of large footprint satellite sen-
small heated calibration plate assumed to be at an sors and in the refinement of validation protocols.
isothermal temperature is periodically viewed by the Finally, as satellite radiometers are now providing
detector to provide an absolute calibration. How- consistent and accurate observations of the SSST
ever, further development of this technology will (e.g., ATSR), there is a need for autonomous
 IR RADIOMETERS 491

operational in situ radiometer systems for ongoing Jessup AT, Zappa CJ, and Yeh H (1997) Defining and
validation of their data. Such intelligent systems that quantifying micro-scale wave breaking with infrared
are extremely robust against the harsh realities of the imagery. Journal of Geophysical Research 102:
marine environment are currently being developed. 23145--23153.
McKeown W and Asher W (1997) A radiometric method
to measure the concentration boundary layer thickness
at an air–water interface. Journal of Atmospheric and
See also Oceanic Technology 14: 1494--1501.
Shaw JA (1999) Degree of polarisation in spectral
Radiative Transfer in the Ocean. Satellite Remote
radiances from water viewing infrared radiometers.
Sensing of Sea Surface Temperatures.
Applied Optics 15: 3157--3165.
Smith WL, Knuteson RO, Rivercombe HH, et al. (1996)
Observations of the infrared radiative properties of the
Further Reading ocean – implications for the measurement of sea surface
temperature via satellite remote sensing. Bulletin of the
Bertie JE and Lan ZD (1996) Infrared intensities of liquids:
American Meteorological Society 77: 41--51.
the intensity of the OH stretching band revisited, and
Suarez MJ, Emery WJ, and Wick GA (1997) The multi-
the best current values of the optical constants H2O (1)
channel infrared sea truth radiometric calibrator
at 251C between 15,000 and 1 cm1. Applied
(MISTRC). Journal of Atmospheric and Oceanic
Spectroscopy 50: 1047--1057.
Technology 14: 243--253.
Donlon CJ, Keogh SJ, Baldwin DJ, et al. (1998) Solid state
Thomas JP, Knight RJ, Roscoe HK, Turner J, and Symon C
measurements of sea surface skin temperature. Journal
(1995) An evaluation of a self-calibrating infrared
of Atmospherical Oceanic Technology 15: 775--787.
radiometer for measuring sea surface temperature.
Donlon CJ and Nightingale TJ (2000) The effect of
Journal of Atmospheric and Oceanic Technology 12:
atmospheric radiance errors in radiometric sea surface
301--316.
skin temperature measurements. Applied Optics 39:
2392--2397.
 REMOTE SENSING OF COASTAL WATERS
N. Hoepffner and G. Zibordi, Institute for Coastal Water Optics
Environment and Sustainability, Ispra, Italy
The radiance leaving the water at a given wavelength,
& 2009 Elsevier Ltd. All rights reserved.
l, Lw(l), in the direction of the remote sensor results
from three major processes of unequal magnitude: (1)
the reflected sunlight at the surface of the water; (2) the
light that has entered the sea surface and retransmitted
back to the atmosphere through scattering; and (3) the
Introductory Comments fluorescence of some specific material in suspension in
the upper water column. In addition, in shallow and
Coastal waters occupy at the most 8–10% of the
relatively clear waters, remote sensing techniques are
ocean surface and only 0.5% of its volume, but
often affected by the ocean bottom.
represent an important fraction in terms of eco-
As the photons enter the water column, the spec-
nomic, social, and ecological value. With growing
tral signature of the sunlight is altered depending on
concerns about the rapid and negative changes of the
the optical properties of the medium itself and its
coastal areas and marine resources, there has been
constituents. Optically complex waters are com-
over the last decade a pressing request for the de-
monly labeled as ‘case 2’ waters in a bipartite clas-
velopment of quantitative and cost-effective meth-
sification scheme (see Bio-Optical Models) where
odologies to detect and characterize both long-term
‘case 1’ refers to marine waters in which phyto-
changes and short-term events in the coastal en-
plankton and covarying materials are the principal
vironment. The challenge, however, is that the
components responsible for the changes in the op-
properties of coastal waters are controlled by com-
plex interactions and fluxes of material between tical properties. Optical properties of case 2 waters
land, ocean, and atmosphere. As a result, coastal are influenced by substances that vary independently
zones are among the most changeable environment from phytoplankton. Such substances are inorganic
on Earth, typically involving phenomena which vary particles in suspension (e.g., sediment), and some
along time and space scales shorter than those in the colored or chromophoric dissolved organic materials
open ocean. (CDOMs), also called yellow substances.
In this view, remote sensing techniques have a key The water-type classification is based on the rela-
role to play, providing consistent products for a wide tive contribution of these three types of substances
range of coastal applications over scales otherwise affecting the light field, irrespectively of possible
inaccessible from standard ship surveys. In spite of changes in the concentration of each substance. For
this advantage, remote sensing techniques have also example, case 1 waters can range from clear, oligo-
their limitations and should not be seen as a substi- trophic open ocean waters with chlorophyll con-
tute to field measurements but rather complementary centration less than 0.1 mg m 3, to some eutrophic
to them in terms of both sampling scales and vari- conditions where chlorophyll can reach a concen-
ables to be measured. tration of 10 mg m 3.
In this article, particular emphasis is given to the Accordingly, all coastal water samples can be op-
Space-borne sensors measuring radiances in the tically charted in a triangular diagram (Figure 1) in
visible and near-infrared, enabling the observation of which the axes represent the fractional contributions
biogeochemical processes in the marine environment. due to each of the three types of optically active
Coastal waters are optically more complex than the components (AOCs). As the data point moves
open oceanic waters because of the large influence by toward one of the three apices, the sample will be
the land system and catchment areas delivering sig- classified as either case 1 water, case 2 ‘yellow-
nificant amount of dissolved and particulate ma- substance’ dominated water, or case 2 ‘suspended
terial. Thus, relative to other Earth-observing material’ dominated water. This classification may
techniques, optical remote sensing of coastal waters appear rather restricted considering the huge variety
requires more specificity in the treatment of the sig- of living and nonliving compounds that contributes
nal, and in the choice of the algorithms, to to the optical variability of marine waters, but it is
support any quantitative assessment of relevant important in remote sensing of coastal waters to se-
biogeochemical variables and their associated lect the type of algorithms which are best suited to
processes. retrieve biogeochemical variables.

492
 REMOTE SENSING OF COASTAL WATERS 493

The Color of Coastal Waters property (AOP) determined by the ratio between the
As for any water body, the color of coastal waters is upwelling and downwelling irradiances at a given
defined by the spectral variations in reflectance, R(l), wavelength (see Satellite Remote Sensing: Ocean
at the sea surface. This is an apparent optical Colour). This quantity is strictly related to the nor-
malized water-leaving radiance, LWN(l), which can
be determined from the total radiance measured by a
P space-borne sensor. The reflectance varies depending
on the illumination and viewing geometry, as well as
on the inherent optical properties (IOPs) of the water
constituents. In the context of remote sensing, the
IOPs of relevance are the absorption coefficient, a(l),
and the backscattering coefficient, bb(l), which rep-
s

Case 1
ce

resents the integral of the volume scattering function,

n

Ph
sta

yto b(l,j), in the backward direction from 901 to 1801.

b
su

p
Accordingly, the reflectance at the sea surface is
lan
w
llo

kto
formulated as
Ye

RðlÞ ¼ f ½bb ðlÞ=ðaðlÞ þ bb ðlÞÞ ½1

Case 2
where f is a function of the solar zenith angle and
Y S
depends also on the shape of the volume scattering
Suspended material function.
In coastal waters, the optically active constituents
Figure 1 Ternary or triangular diagram illustrating the relative
contribution of phytoplankton, dissolved organic matter (yellow may vary in space and time, independently of each
substances) and suspended material (nonalgal particles), and the other, and consequently, this variation affects the
division between case 1 and case 2 waters. Copyright: IOCCG. color (i.e., reflectance) of the water (Figure 2).

0.020

0.018

Coastal waters with high

0.016 sediment and CDOM
concentrations
(sr −1)

0.014
Clear waters
Remote sensing reflectance

0.012 Waters with moderate

chlorophyll and sediment
0.010 concentrations

0.008

0.006

0.004

0.002

0.000
400 450 500 550 600 650 700 750 800
Wavelength (nm)

Figure 2 Examples of remote sensing reflectance spectra for different water types. Maximum reflectance shifts to higher wavelength
in coastal waters with high concentration of nonalgal particles and CDOMs. Adapted from IOCCG.
494 REMOTE SENSING OF COASTAL WATERS

Optical remote sensing in coastal waters thus ne- calculation. In addition, the presence of fine mineral
cessitates a thorough assessment of the water bulk particles with high refractive index represents there a
optical properties, quantifying the role of each indi- crucial factor controlling the reflectance variability
vidual component. To this end, the total spectral when compared with open ocean waters. Specific
absorption and backscattering coefficients are div- blooms of phytoplankton (e.g., coccolithophores)
ided into as many additive components: and bubbles entrained by breaking waves can also
play a significant role in shaping the backscattering
aðlÞ ¼ aw ðlÞ þ aphy ðlÞ þ aNAP ðlÞ þ aCDOM ðlÞ ½2 coefficient of coastal waters.

Bottom Reflectance
bb ðlÞ ¼ bb w ðlÞ þ bb phy ðlÞ þ bb NAP ðlÞ ½3 In shallow and clear coastal waters, the ocean color
signal is affected by the light reflected off the bottom.
where the subscript ‘w’ refers to water molecules
The influence of the bottom depends on the depth of
(pure water), ‘phy’ to phytoplankton, ‘NAP’ to
the water body, the bottom type, and, of course, the
nonalgal particle (e.g., sediment), and ‘CDOM’ to
optical state of the water column above. In clear
colored dissolved organic matter (i.e., yellow sub-
waters, a 30-m-deep bottom feature can affect the
stances, assumed to have negligible scattering
remote sensing signal over a narrow spectral band in
properties).
the blue part of the spectrum. As the bottom be-
The basic theory developed for open ocean (see
comes shallower, the surface water reflectance is af-
Bio-Optical Models) to parametrize the wavelength
fected over a wider spectral region (assuming the
dependence of each term in eqns [2] and [3] is
optical state of the water column above remains
directly applicable in coastal waters. The imple-
unchanged).
mentation of these models require, however, differ-
To account for the bottom, the water-leaving ra-
ent strategies to account for unrelated distributions
diance in shallow waters is usually modeled using a
between the various optically active components as
two-flow system where the upwelled irradiance is
well as differences in the values of some model
made of a water column and a bottom component of
parameters. A notable difference, for example, is in
known reflectance. The seabed, however, may be
the contribution of nonalgal particles which can be
made of sand, rocks, mud, or covered with benthic
significantly higher in coastal waters due to a supply
organisms such as algae, seagrasses, and coral reefs.
of sediment either from land or resuspension at the
All of these reflect light in different ways, which add
bottom.
to the complexity in using remote sensing technique
The spectral absorption by nonalgal particles and
in nearshore waters, albeit open to new challenges to
CDOMs obeys similar exponential shape defined by
apply satellite data for mapping benthic features.
their respective absorption coefficients at a reference
wavelength and the slope of the absorption curve in
Influence of Inelastic Processes
the lower part of the spectrum (400–440 nm). In
coastal areas influenced by river runoff, CDOMs In marine waters, the water reflectance can also be
mainly consist of terrestrial humic acids with lower affected by some transpectral (or inelastic) processes
slope values than those observed in open ocean and such as those associated with solar-stimulated fluor-
in coastal waters not directly connected with terres- escence by organic compounds, and Raman (mo-
trial sources. In the case of nonalgal particles, the lecular) scattering. Fluorescence is mostly generated
variability in the exponential slope in marine waters by phytoplankton chlorophyll pigments, and by
is restricted to a narrow range of values (B0.01 phycoerythrin-containing cells. At high concen-
to o0.02 nm  1). A potential distinction between tration (typically, Chl41 mg m 3), chlorophyll can
coastal and open ocean waters remains therefore have a significant effect on the reflectance signal
difficult, even though assemblage of nonalgal par- within a narrow band in the red part of the spectrum
ticles may be dominated by mineral sediments or centered at 685 nm, whereas phycoerythrin fluor-
organic debris. escence occurs over a larger band centered at
The analysis of the backscattering coefficient is 585 nm.
still advancing because of the difficulty in measuring In coastal ‘CDOM-dominated’ waters, fluor-
it directly. Our knowledge largely benefits from escence by the dissolved organic matter can also in-
theoretical studies, which require a good estimate of fluence the blue part of the spectrum, c. 430–450 nm.
the particle-size distribution and their refractive In all cases, the integration of fluorescence into
index. In coastal waters, the large variety of particle bio-optical models requires a good knowledge of the
type and size increases the complexity of the absorption coefficients of each component at the
 REMOTE SENSING OF COASTAL WATERS 495

excitation wavelengths, as well as their fluorescence been used for their development. In a highly
quantum yield. Note that in coastal waters, the effect changeable environment such as coastal waters,
of Raman scattering on the surface reflectance is re- these algorithms are clearly restricted to specific re-
duced as the water turbidity increases. gions, and even specific periods. Some improvements
can be made to increase the efficiency of these al-
gorithms in turbid waters, by using more than one
Algorithms for Coastal Waters band ratio, within spectral regions that are more
Established algorithms are currently on a routine sensitive to the optical conditions observed locally.
basis by the space agencies and others to generate For example, additional wavelengths in the blue part
multi-sensors archives of satellite biogeochemical of the spectrum could be used to separate the dis-
products over the global ocean. These algorithms solved material from the particulate matter, whereas
were developed to perform as efficiently as possible other bands in the green and orange part may pro-
in the open ocean. The complexity of coastal waters, vide information on specific phytoplankton blooms
as well as the atmosphere above, requires more often observed in coastal waters (see Table 1). In
complex algorithms capable of handling nonlinear reality, the spectral characteristics of combined sea-
multivariate bio-optical systems. In addition, the water constituents are not unique, suggesting that no
large variability in time and space of the bulk prop- single wavelength can really be representing infor-
erties of coastal waters hampers any implementation mation on just one constituent. Under these con-
of a single and global algorithm that would work ditions, more complex modeling approaches, and
with the same efficiency in all regions. optimization techniques are envisaged to solve the
problem.
Atmospheric Correction Analytical (or semi-) algorithms are based on the
inversion of a forward radiance model describing
The atmospheric correction process is applied to
both the relationship between water constituents and
remove the effects of the atmosphere that contribute
the water reflectance at the surface (typically as in
to the signal measured by a satellite sensor. The ob-
eqn [1]), and the light propagation through the water
jective of this process is the discrimination, from
and atmosphere. This approach makes use of known
top-of-atmosphere radiance, of the signal emerging
optical properties for the specific components, even
from the sea carrying information on the materials
though the spectral characteristics of these com-
suspended and dissolved in seawater. The atmos-
ponents are usually determined from empirical for-
pheric correction of coastal data is challenged by the
mulation, thus justifying the term ‘semi-analytical’ to
presence of continental aerosols, bottom reflectance,
describe these algorithms. In coastal waters, the
and adjacency of land. Minimizing these perturbing
number of unknowns in each set of equations may
effects, which generally are site-specific, requires
increase rapidly with the number of independent
knowledge of the regional aerosol and bottom op-
variables (i.e., water constituents), causing dif-
tical properties. Specifically, effects of continental
ficulties in solving the system without applying some
aerosols are minimized by properly accounting for
their scattering phase function and single scattering
albedo; the increase in radiance due to bottom re- Table 1 Example of spectral range for coastal waters
flectance can be removed through iterative processes applications
knowing the water depth and spectral reflectance of
the seabed; and the adjacency effects are minimized Product name Critical spectral range
by determining the reduction in image contrast as a (nm)
function of the aerosol and of the land reflectance.
Chlorophyll, and other pigment 443–445 (max. abs.),
concentration 550–560 (min. abs.)
Coastal Water Algorithms 490, 548, 640 (for other
pigments)
In optical remote sensing of marine waters, the al- Red tides 510
gorithms used to retrieve the concentrations of water Sediment concentration 530–650, NIR
constituents are either analytical (or semi-) or em- CDOM 410–420
pirical. Empirical algorithms (so-called blue–green Seabed reflectance 490–580
ratio algorithms; see Satellite Remote Sensing: Ocean Chlorophyll fluorescence 680–685
Aerosol optical properties 700–NIR
Colour) have been implemented in the processing of Atmospheric corrections
satellite data over open ocean with various degrees of Sea surface temperature (SST) TIR
success. But the performance of these algorithms
remains limited to the range of input values that have NIR, near-infrared; TIR, thermal infrared.
496 REMOTE SENSING OF COASTAL WATERS

approximation that eventually leads to a lower ac- Data Validation and Measurement
curacy of the algorithms. Protocols
Nonlinear optimization techniques and other ma-
chine learning methods such as artificial neural net- The normalized water-leaving radiance, LWN(l), is
works provide an interesting alternative to examine the primary ocean color remote sensing product
complex coastal waters and to handle multivariate obtained from top-of-atmosphere data corrected for
systems. Models based on these methods determine the atmospheric perturbations. Higher-level marine
the output values (e.g., the concentrations of optic- products, like the concentration of phytoplankton
ally active components) from input data (e.g., pigments or seawater inherent optical properties
water reflectance at various wavelengths) through (i.e., absorption, scattering, and backscattering co-
nonlinear multidimensional parametric functions efficients), are determined from LWN(l). The accur-
(Figure 3). The determination of the model par- acy of coastal remote sensing products may be
ameters, as well as the assessment of the model affected by the presence of turbid waters or ab-
performance rely on a reference data set. sorbing aerosols. Because of this, the validation of
Remote sensing of coastal waters and algorithm remote sensing LWN(l) and aerosol optical thickness
development are still topics of ongoing research in data, resulting from the atmospheric correction
spite of substantial effort and promising results that process, is a first fundamental step in assessing the
have taken place within the last few years. Im- accuracy of ocean color products. Additional steps
provements remain to be achieved on several aspects include the validation of derived products to assess
of both the atmospheric correction and in-water bio- the consistency of the whole process leading to the
optical modeling to ensure an operational retrieval of determination of higher-level products.
the products at a reasonable accuracy. These in- Validation measurements can result from oceano-
vestigations rely on field experiments and the col- graphic campaigns, or from continuous data col-
lection of high-quality optical data for a wide range lection by autonomous systems deployed on fixed
of water bodies. platforms or ships of opportunity. Oceanographic
With the rapid changes occurring in coastal campaigns ensure a wide spatial data collection with
waters, a unique algorithm that will serve all optical a comprehensive optical and biogeochemical char-
conditions with the same accuracy is doubtful. On acterization of each sampling site, but with a
the other hand, the implementation of more than one poor time resolution. Autonomous systems enable
regional algorithm into a unique processing scheme continuous a highly repeated collection of a re-
remains a challenge, as that approach could result in stricted set of parameters at a specific site or along a
significant discontinuities at the region boundaries. given route. Obviously, measurement campaigns and

Input units Hidden units

Output units
Bias x0 Bias z0

y1 Chl-a

R(490) x1 z1

w (1)
ij w (2)
jk y2 CDOM

R(555) x2 zj

y3 NAP

R(665) x3 z10
First layer Second layer
of weights of weights

Figure 3 Representation of a multilayer perceptron (MLP) neural network adopted for the retrieval of the optically active seawater
components. In this particular case, the network architecture consists of three inputs (reflectances at three satellite wavelengths), 10
hidden units, and three outputs (chlorophyll a, colored dissolved organic matter, and nonalgal particulate matter). An additional input is
implemented for the bias parameter. Note that the MLP architecture presented here has only an illustrative purpose. Adapted from
D’Alimonte D and Zibordi G (2003) Phytoplankton determination in an optically complex coastal region using a multilayer perceptron
neural network. IEEE Transactions on Geoscience and Remote Sensing 41(12): 2861–2868.
 REMOTE SENSING OF COASTAL WATERS 497

autonomous systems are complementary and both Time, spatial, and spectral coincident pairs of
contribute to the objective of validating remote satellite and in situ data, so-called matchups, are the
sensing products. basis for any validation analysis. Due to the intrinsic
Within the framework of in situ optical technolo- higher variability exhibited by coastal waters, when
gies commonly applied for the determination of the compared to oceanic waters, particular care must be
fundamental quantity LWN(l), both in- and above- given to the spatial variability of satellite data and to
water radiometry can produce individual measure- the temporal variability of in situ observations
ments or time series of optical data. In this context, (Figure 5).
various in-water autonomous systems have been
widely exploited in open ocean regions where the
biofouling perturbation is low. Differently, the use of
autonomous above-water radiometry is more suit-
Remote Sensing Coastal Applications
able to produce long-term measurements for the Remote sensing of coastal waters satisfies on its own
validation of ocean color data in optically complex a large number of applications, owing to substantial
coastal waters. progress in sensor technology, as well as algorithm
Offshore fixed platforms like lighthouses, performance, made over the last few years. Import-
oceanographic towers, and oil rigs, generally located ant contributions have been accomplished in various
at several nautical miles from the main land in disciplines of marine sciences, like marine physics
coastal areas (i.e., regions permanently or occasion- and biogeochemistry, process modeling, as well as
ally affected by bottom resuspension, coastal ero- fisheries and coastal management. More importantly,
sion, river inputs, or by relevant anthropogenic satellites provide continuous long time series of bio-
impact), are suitable for the deployment of autono- optical and geophysical variables, thus representing a
mous above-water radiometers provided that (1) cost-effective tool to monitor changes in coastal
their distance from the coast makes negligible the waters that might occur as a direct or indirect result
adjacency effects in the remote sensing data; and (2) of external pressures (e.g., climate change and
the bottom structure and type have no effect human activities). In parallel, the scientific and
(Figure 4). technical community is progressing in the search for

Figure 4 Lighthouse tower in the Baltic Sea used for the deployment of above-water radiometer systems supporting the validation of
ocean color products in coastal areas.
498 REMOTE SENSING OF COASTAL WATERS

3.00 4.00 0.70

443 551 667

LMOD 2.25 3.00 0.50

LMOD

LMOD
WN

WN

WN
1.50 2.00 0.30

0.75 1.00 0.10

N = 291 N = 291 N = 291
0.00 0.00 −0.10
0.00 0.75 1.50 2.25 3.00 0.00 1.00 2.00 3.00 4.00 −0.10 0.10 0.30 0.50 0.70
LPRS
WN
LPRS
WN LPRS
WN

2
Figure 5 Scatter plots of satellite-derived LMOD MOD
WN vs. in situ LWN normalized water-leaving radiances in units of mW cm mm  1s  1
at the 443, 551, and 667 nm MODIS center-wavelengths.

0.01 0.03 0.1 0.3 1

m−1

Figure 6 Light attenuation coefficient Kd in units of m  1 for the Black Sea derived from SeaWiFS monthly composite (Jan. 2003).
The coefficient is calculated using a semi-analytical algorithm to retrieve the IOPs of the water constituents. Source: Joint Research
Centre of the European Commission & Joint Research Centre, 2006.

the optimal model to be applied in these complex total suspended matter, and dissolved organic sub-
waters, taking benefit from advanced instruments stances, with accuracies as good as in open ocean
with higher spectral resolution and better cali- (c. 30–40% in biomass) when using regional dedi-
bration. Recent development of hyperspectral sensor cated algorithms. These components, in turn, influ-
prototypes, providing the full spectral coverage over ence the light attenuation coefficient in the water
the visible wavelengths, offers new opportunities to which is also directly retrieved from satellite
discriminate optical disparities, and thus to classify (Figure 6), providing information on water trans-
turbid waters and quantify individual seawater parency. The latter represents an important quality
components down to phytoplankton species. parameter to assess the ecological status of lakes and
Optical remote sensing in coastal waters typically coastal waters. Remotely sensed chlorophyll con-
provides quantitative estimates of phytoplankton centration and light attenuation are also required to
biomass (indexed by chlorophyll concentrations), estimate water-column primary production, which
 REMOTE SENSING OF COASTAL WATERS 499

represents a critical indicator to assess coastal eu- coastal fisheries to locate target areas with high fish
trophication, and to analyze the carbon budget dir- densities. Many pelagic species aggregate along
ectly associated with the continental shelf pump. frontal structures and upwelling areas that bear dis-
The concentrations of suspended sediment and dis- tinctive signals on the sea surface temperature and
solved organic matter can be used to track river primary production (or phytoplankton biomass)
plumes and other coastal dynamics, and to calibrate maps. Moreover, many aspects of the life cycle of
and validate sediment transport models. Recently, the fishes (e.g., reproduction and early-stage feeding)
availability of IOPs, that is, absorption and back- depends on physical and biological processes, and
scattering coefficients, as direct products of satellite discerning these from remote sensing is important for
ocean color, has given an alternative approach to ad- marine conservation and sustainable management of
dress changes in the water constituents, as well as the marine coastal resources.
water transparency, further extending satellite appli- Satellite instruments, other than optical sensors,
cations to particle size and composition characteristics. can provide additional information of importance to
Several algal species can be present in coastal coastal water management. Microwave sensors col-
waters as intense blooms (e.g., coccolithophorid lect data either passively or actively in that part of
species, red tides, Trichodesmium), which can be the electromagnetic spectrum ranging from 3 mm to
readily distinguished on the basis of their differences 1 m wavelength. These instruments provide signifi-
in pigment composition and optical properties cant advantages in viewing the Earth under all-
(Figure 7). In that case, remote sensing acts as an weather conditions, even through clouds; but their
efficient technique to monitor these anomalous spatial resolution, which depends on the physical
blooms; some are associated with harmful con- dimension of the antenna, is often not adequate for
sequences on the surrounding ecosystem, causing coastal applications. Sophisticated synthetic aperture
mass mortalities of marine organisms. radars (SARs) are an exception to that rule and can
In shallow coastal waters, remote sensing is being retrieve surface features at extremely fine resolution
used to monitor benthic structure and classify bot- of the order of meters. This remote sensing technique
tom types. At present, satellite data most commonly is used for fisheries applications to monitor fishing
exploited for that purpose are equipped with radio- vessels in coastal waters, contributing thus to the
meters, such as the SPOT-HRV and Landsat TM management of the stocks through evaluation of the
series with typically 10–30 m spatial resolution, or fishing effort. In addition to ship survey, SAR data
even IKONOS with spatial resolution better than are exploited in coastal waters to monitor sea surface
10 m. Even though these sensors have a limited set of roughness and to detect pollution at the surface. Oil
wavelengths (three to four broad spectral bands in spill detection is of particular importance to coastal
the visible), the spectral contrast between various management. The presence of oil slicks on the sea
bottom structures (e.g., corals and noncoral objects) surface increases the surface tension of seawater, re-
can be used to construct classification algorithms ducing its natural roughness which appears as a
based on differential reflectance (Figure 8). black signature on a SAR image. Multiple techniques
Surface features simultaneously observed from using, for example, ocean color and SAR can be
ocean color and thermal imagery have assisted helpful to discriminate between dense algal blooms

20
18
16
14
12
10
8
6
4
2
SMHI
0

Figure 7 Cyanobacteria bloom in the Baltic Sea. Left: MODIS-Aqua 30 Jul. 2003. True color image of the Baltic Sea showing a
bloom of Nodularia spumigena. Source: http://oceancolor.gsfc.nasa.gov. Right: Occurrence of cyanobacterial accumulations in the
Baltic during 2003 expressed as number of days (data based on satellite images). Source: http://www.helcom.fi
500 REMOTE SENSING OF COASTAL WATERS

Glovers
(10 km × 10 km)

Glovers (5 km × 10 km)
Dense seagrass
Deep medium-dense brown algae
Seagrass and brown algae (Lobophora sp.)
Deep lagoon floor, sparse seagrass
Shallow lagoon floor, sparse seagrass
Sparse brown algae
Shallow medium-dense brown algae
Dense brown algae
Sand
Shallow forereef
Deep forereef

Figure 8 Classification of tropical coral reef environment (Belize, Caribbean Sea). Top: RGB color composite based on the red,
green, and blue bands of IKONOS sensor. Bottom: Eleven-class scheme classification of the benthic features. Reproduced from
Andréfouët S, Kramer P, Torres-Pulliza D, et al. (2003) Multi-site evaluation of IKONOS data for classification of tropical coral reef
environments. Remote Sensing of Environment 88: 128–143, with permission from Elsevier.

and surface slicks. The low revisit cycle (e.g., 35 days Further Reading
for the European Remote Sensing satellites ERS-1
Andréfouët S, Kramer P, Torres-Pulliza D, et al. (2003)
and ERS-2 instruments) limits, however, the use of
Multi-site evaluation of IKONOS data for classification
SAR to monitor rapidly changing events.
of tropical coral reef environments. Remote Sensing of
In conclusion, there is a large and ever-increasing Environment 88: 128--143.
interest in applying remote sensing technique in Babin M, Stramski D, Ferrari GM, et al. (2003) Variations
coastal waters. It offers a unique solution to monitor in the light absorption coefficients of phytoplankton,
this environment at scales that are compatible with nonalgal particles, and dissolved organic matter in coastal
many biological and physical processes. At the same waters around Europe. Journal of Geophysical Research
time, the number of satellite products is growing 108(C7): 3211 (doi:10.1029/2001JC000882).
rapidly with the advent of new sensors, better algo- Berthon J-F, Mélin F, and Zibordi G (2007) Ocean colour
rithms and validation experiments, providing a range remote sensing of the optically complex European seas.
of key indicators to support scientific investigations, In: Barale V and Gade M (eds.) Remote Sensing of
European Seas, pp. 35--52. Dordrecht: Springer.
as well as operations for coastal management.
Bukata RP, Jerome JH, Kondratyev KY, and Pozdnyakov
DV (eds.) (1995) Optical Properties and Remote
Sensing of Inland and Coastal Waters. Boca Raton, FL:
See also CRC Press.
Bio-Optical Models. Inherent Optical Properties D’Alimonte D and Zibordi G (2003) Phytoplankton
and Irradiance. Satellite Remote Sensing: Ocean determination in an optically complex coastal region
Colour. using a multilayer perceptron neural network. IEEE
 REMOTE SENSING OF COASTAL WATERS 501

Transactions on Geoscience and Remote Sensing International Ocean-Colour Coordinating Group, No.
41(12): 2861--2868. 5. Dartmouth, NS: IOCCG.
Gordon HR (2002) Inverse methods in hydrologic optics. Miller RL, Del Castillo CE, and McKee BA (eds.) (2005)
Oceanologia 44: 9--58. Remote Sensing of Coastal Aquatic Environments:
Hansson M (2004) Cyanobacterial blooms in the Baltic Sea. Technologies, Techniques and Applications. Dordrecht:
HELCOM Indicator Fact Sheets 2004. Online (2008- Springer.
05-06) http://www.helcom.fi/environment2/ifs/archive/ Oceanography Magazine (2004) Special Issue: Coastal
ifs2004/enGB/cyanobacteria (accessed July 2008). Ocean Optics and Dynamics. The Official Magazine of
IOCCG (2000) Remote sensing of ocean colour in coastal, the Oceanography Society 17(2).
and other optically-complex waters. In: Sathyendranath Zibordi G, Holben B, Hooker SB, et al. (2006) A network
S (ed.) Report of the International Ocean-Colour for standardized ocean color validation measurements.
Coordinating Group, No. 3. Dartmouth, NS: IOCCG. EOS Transactions, American Geophysical Union
IOCCG (2006) Remote sensing of inherent optical 87(30): 293--297.
properties: Fundamentals, tests of algorithms and
applications. In: Lee ZP (ed.) Report of the
 SATELLITE ALTIMETRY
R. E. Cheney, Laboratory for Satellite Altimetry, relative to a fixed Earth coordinate system. Com-
NOAA, Silver Spring, Maryland, USA bining these two measurements yields profiles of sea
Copyright & 2001 Elsevier Ltd. surface topography, or sea level, with respect to the
reference ellipsoid (a smooth geometric surface
which approximates the shape of the Earth).
In practice, the various measurement systems are
highly sophisticated and require expertise at the
Introduction cutting edge of instrument and modeling capabilities.
This is because accuracies of a few centimeters must
Students of oceanography are usually surprised to be achieved to properly observe and describe the
learn that sea level is not very level at all and that the various oceanographic and geophysical phenomena
dominant force affecting ocean surface topography is of interest. Figure 1 shows a schematic of the Topex/
not currents, wind, or tides; rather it is regional Poseidon (T/P) satellite altimeter system. Launched
variations in the Earth’s gravity. Beginning in the in 1992 as a joint mission of the American and
1970s with the advent of satellite radar altimeters, French Space agencies (and still operating as of
the large-scale shape of the global ocean surface 2001), T/P is the most accurate altimeter flown to
could be observed directly for the first time. What the date. Its microwave radars measure the distance to
data revealed came as a shock to most of the the sea surface with a precision of 2 cm. Two dif-
oceanographic community, which was more accus- ferent frequencies are used to solve for the path delay
tomed to observing the sea from ships. Profiles tele- due to the ionosphere, and a downward-looking
metered back from NASA’s pioneering altimeter, microwave radiometer provides measurements of the
Geos-3, showed that on horizontal scales of hun- integrated water vapor content which must also be
dreds to thousands of kilometers, the sea surface is known. Meteorological models must be used to es-
extremely complex and bumpy, full of undulating timate the attenuation of the radar pulse by the at-
hills and valleys with vertical amplitudes of tens to mosphere, and other models correct for biases
hundreds of meters. None of this came as a surprise created by ocean waves. Three different tracking
to geodesists and geophysicists who knew that the systems (a laser reflector, a Global Positioning Sys-
oceans must conform to these shapes owing to spa- tem receiver, and a ‘DORIS’ Doppler receiver) de-
tial variations in marine gravity. But for the termine the satellite orbit to within 2 cm in the radial
oceanographic community, the concept of sea level direction. The result of all these measurements is a
was forever changed. During the following two set of global sea level observations with an absolute
decades, satellite altimetry would provide exciting accuracy of 3–4 cm at intervals of 1 s, or about 6 km,
and revolutionary new insights into a wide range of along the satellite track. The altimeter footprint is
earth science topics including marine gravity, ba- exceedingly small – only 2–3 km – so regional maps
thymetry, ocean tides, eddies, and El Niño, not to or ‘images’ can only be derived by averaging data
mention the marine wind and wave fields which can collected over a week or more.
also be derived from the altimeter echo. This chapter
briefly addresses the technique of satellite altimetry
and provides examples of applications. Gravitational Sea Surface Topography
Sea surface topography associated with spatial vari-
ations in the Earth’s gravity field has vertical ampli-
Measurement Method tudes 100 times larger than sea level changes
In concept, radar altimetry is among the simplest of generated by tides and ocean currents. To first order,
remote sensing techniques. Two basic geometric therefore, satellite altimeter data reveal information
measurements are involved. In the first, the distance about marine gravity. Within 1–2% the ocean
between the satellite and the sea surface is deter- topography follows a surface of constant gravi-
mined from the round-trip travel time of microwave tational potential energy known as the geoid or the
pulses emitted downward by the satellite’s radar and equipotential surface, shown schematically in Fig-
reflected back from the ocean. For the second ure 1. Gravity can be considered to be constant in
measurement, independent tracking systems are used time for most purposes, even though slight changes
to compute the satellite’s three-dimensional position do occur as the result of crustal motions,

502
 SATELLITE ALTIMETRY 503

Figure 1 Schematic diagram of satellite radar altimeter system. Range to the sea surface together with independent determination
of the satellite orbit yields profiles of sea surface topography (sea level) relative to the Earth’s reference ellipsoid. Departures of sea
level from the geoid drive surface geostrophic currents.

redistribution of terrestrial ice and water, and other dimensional images of the ocean floor – as if all the
slowly varying phenomena. An illustration of the water were drained away (Figure 3). For many
gravitational component of sea surface topography is oceanic regions, especially in the Southern Hemi-
provided in Figure 2, which shows a T/P altimeter sphere, these data have provided the first reliable
profile collected in December 1999 across the maps of bottom topography. This new data set has
Marianas Trench in the western Pacific. The trench many scientific and commercial applications, from
represents a deficit of mass, and therefore a negative numerical ocean modeling, which requires realistic
gravity anomaly, so that the water is pulled away bottom topography, to fisheries, which have been
from the trench axis by positive anomalies on either able to take advantage of new fishing grounds over
side. Similarly, seamounts represent positive gravity previously uncharted seamounts.
anomalies and appear at the ocean surface as
mounds of water. The sea level signal created by
ocean bottom topography ranges from B1 m for
seamounts to B10 m for pronounced features like
Dynamic Sea Surface Topography
the Marianas Trench, and the peak-to-peak ampli- Because of variations in the density of sea water over
tude for the large-scale gravity field is nearly 200 m. the globe, the geoid and the mean sea surface are not
Using altimeter data collected by several different exactly coincident. Departures of the sea surface
satellites over a period of years, it is possible to create with respect to the geoid have amplitudes of about 1
global maps of sea surface topography with extra- m and constitute what is known as ‘dynamic topo-
ordinary accuracy and resolution. When these maps graphy’. These sea surface slopes drive the geo-
are combined with surface gravity measurements, strophic circulation: a balance between the surface
models of the Earth’s crust, and bathymetric data slope (or surface pressure gradient) and the Coriolis
collected by ships, it is possible to construct three- force (created by the Earth’s rotation). The
504 SATELLITE ALTIMETRY

65

Altimeter profile:
60 Marianas Trench

55

50
0 100 200 300 400 500
Distance (km)

Figure 2 Sea surface topography across the Marianas Trench in the western Pacific measured by Topex/Poseidon. Heights are
relative to the Earth’s reference ellipsoid.

60˚N

30˚N

0˚

30˚S

60˚S

0˚ 30˚E 60˚E 90˚E 120˚E 150˚E 180˚ 150˚W 120˚W 90˚W 60˚W 30˚W

Figure 3 Topography of the ocean bottom determined from a combination of satellite altimetry, gravity anomalies, and bathymetric
data collected by ships. (Courtesy of Walter H. F. Smith, NOAA, Silver Spring, MD, USA.)

illustration in Figure 4 shows an estimate of the Variations are between  110 cm (deep blue) and
global geostrophic circulation derived by combining 110 cm (white). The surface flow is along lines of
a mean altimeter-derived topography with a geoid equal dynamic topography (red arrows). In the
computed from independent gravity measurements. Northern Hemisphere, the flow is clockwise around
 SATELLITE ALTIMETRY 505

cm

_ 110 _ 90 _ 70 _ 50 _ 30 _ 10 10 30 50 70 90 100

Figure 4 Surface geostrophic circulation determined from a combination of satellite altimetry and a model of the marine gravity field.
(Courtesy of Space Oceanography Division, CLS, Toulouse, France.)

the topography highs, while in the Southern Hemi- measurement of the change in gravity as a function
sphere, the flow is counter-clockwise around the of time will provide new information about the
highs. The map shows all the features of the general global hydrologic cycle and perhaps shed light on the
circulation such as the ocean gyres and associated factors contributing to global sea level rise. For ex-
western boundary currents (e.g. Gulf Stream, Kur- ample, how much of the rise is due simply to heating
oshio, Brazil/Malvinas Confluence) and the Antarctic and how much to melting of glaciers? Together with
Circumpolar Current. complementary geophysical data, satellite gravity
At the time of writing, global geoid models are not data represent a new frontier in studies of the Earth
sufficiently accurate to reveal significant new infor- and its fluid envelope.
mation about the surface circulation of the ocean.
However, extraordinary gravity fields will soon be
available from dedicated satellite missions such as
Sea Level Variability
the Challenging Minisatellite Payload (CHAMP: At any given location in the ocean, sea level rises and
2000 launch), the Gravity Recovery and Climate falls over time owing to tides, variable geostrophic
Experiment (GRACE: 2002 launch), and the Gravity flow, wind stress, and changes in temperature and
Field and Steady-state Ocean Circulation Explorer salinity. Of these, the tides have the largest signal
(GOCE: 2005 launch). These satellite missions, amplitude, on the order of 1 m in mid-ocean. Satel-
sponsored by various agencies in the USA and Eur- lite altimetry has enabled global tide models to be
ope, will employ accelerometers, gravity gradi- dramatically improved such that mid-ocean tides can
ometers, and the Global Positioning System to now be predicted with an accuracy of a few cm (see
virtually eliminate error in marine geoid models at )In studying ocean dynamics, the contribution of the
spatial scales larger than 300 km and will thereby tides is usually removed using these models so that
have a dramatic impact on physical oceanography. other dynamic ocean phenomena can be isolated.
Not only will it be possible to accurately compute The map in Figure 5 shows the variability of global
global maps of dynamic topography and geostro- sea level for the period 1992–98. It is derived from
phic surface circulation, but the new gravity models three satellite altimeter data sets: ERS-1, T/P, and
will also allow recomputation of orbits for past ERS-2 (ERS is the European Space Agency Remote
altimetric satellites back to 1978, permitting studies Sensing Satellite), from which the tidal signal has
of long, global sea level time-series. Furthermore, been removed. The map is dominated by mesoscale
506 SATELLITE ALTIMETRY

0 2 4 6 8 10 12 14 16 18 20

Figure 5 Variability of sea surface topography over the period 1992–98 from three satellite altimeters: Topex/Poseidon, ERS-1, and
ERS-2. Highest values (cm) correspond to western boundary currents which meander and generate eddies. (Courtesy of Space
Oceanography Division, CLS, Toulouse, France.)

35

30
Tide gauge
25
vs Topex/Poseidon
20 altimeter

15

10
cm
5

0
_5

_ 10

_ 15
rms = 2.3 cm
_ 20
93 94 95 96 97 98 99

Figure 6 Monthly mean sea level deviation near the Galapagos Islands derived from tide gauge data and altimeter data. The B2 cm
agreement demonstrates the accuracy of altimetry for observing sea level variability. The effect of the 1997–98 El Niño is apparent.

(100–300 km) variability associated with the western signals such as El Niño are the dominant contributor.
boundary currents, where the rms variability can be The smallest variability is found in the eastern por-
as high as 30 cm. This is due to a combination of tions of the major ocean basins where values are o5
current meandering, eddies, and seasonal heating cm rms.
and cooling. Other bands of relative maxima (10–15 To examine a sample of the sea level signal more
cm) can be seen in the tropics where interannual closely, Figure 6 shows the record from the region of
 SATELLITE ALTIMETRY 507

NOAA / Laboratory for Satellite Altimetry

_ 32 _ 28 _ 24 _ 20 _ 16 _ 12 _8 _4 0 4 8 12 16 20 24 28 32
H (cm)

Figure 7 Global sea level anomaly observed by the Topex/Poseidon altimeter at the height of the 1997–98 El Niño. High (low) sea
level corresponds to areas of positive (negative) heat anomaly in the ocean’s upper layers.

the Galapagos Islands in the eastern equatorial Pacific. 1.8 mm y1. Unfortunately, because these data are
The plot includes two time-series: one from the T/P relatively sparse and contain large interdecadal
altimeter and the other from an island tide gauge, fluctuations, the observations must be averaged over
both averaged over monthly time periods. These in- 50–75 years in order to obtain a stable mean value. It
dependent records agree at the level of 2 cm, an in- is therefore not possible to say whether sea level rise
dication of the remarkable reliability of satellite is accelerating in response to recent global warming.
altimetry. The plot also illustrates changes associated Satellite altimeter data have the advantage of dense,
with the El Niño event which took place during global coverage and may offer new insights on the
1997–98. During El Niño, relaxation of the Pacific problem in a relatively short period of time. Based on
trade winds cause a dramatic redistribution of heat in T/P data collected since 1992, it is thought that 15
the tropical oceans. In the eastern Pacific, sea level years of continuous altimeter measurements may be
during this event rose to 30 cm above normal by sufficient to obtain a reliable estimate of the current
December 1977 and fell by a corresponding amount rate of sea level rise. This will require careful cali-
in the far western Pacific. The global picture of sea bration of the end-to-end altimetric system, not to
level deviations observed by the T/P altimeter at this mention cross-calibration of a series of two or three
time is shown in Figure 7. Because sea level changes missions (which typically last only 5 years). Fur-
can be interpreted as changes in heat (and to a lesser thermore, in order to interpret and fully understand
extent, salinity) in the upper layers, altimetry provides the sea level observations, the various components of
important information for operational ocean models the global hydrologic system must be taken into ac-
which are used for long-range El Niño forecasts. count, for example, polar and glacial ice, ground
water, fresh water stored in man-made reservoirs,
and the total atmospheric water content. It is a
Global Sea Level Rise complicated issue, but one which may yield to the
Tide gauge data collected over the last century increasingly sophisticated observational systems that
indicate that global sea level is rising at about are being brought to bear on the problem.
508 SATELLITE ALTIMETRY

Wave Height and Wind Speed assimilated in near-real-time to help forecast El

Niño, monitor coastal circulation, and predict hur-
In addition to sea surface topography, altimetry
ricane intensity. Although past missions have been
provides indirect measurements of ocean wave
flown largely for research purposes, altimetry is
height and wind speed (but not wind direction). This
rapidly moving into the operational domain and will
is made possible by analysis of the shape and inten-
become a routine component of international satel-
sity of the reflected radar signal: a calm sea sends the
lite systems during the twenty-first century.
signal back almost perfectly, like a flat mirror,
whereas a rough sea scatters and deforms it. Wave
height measurements are accurate to about 0.5 m or
10% of the significant wave height, whichever is See also
larger. Wind speed can be measured with an accuracy Satellite Remote Sensing Microwave Scattero-
of about 2 m s1. For additional information, see meters. Satellite Remote Sensing SAR.
articles 69 and 129.

Conclusions Further Reading

Satellite altimetry is somewhat unique among ocean Cheney RE (ed.) (1995) TOPEX/POSEIDON: Scientific
remote sensing techniques because it provides much Results. Journal of Geophysical Research 100: 24
more than surface observations. By measuring sea 893–25 382
Douglas BC, Kearney MS, and Leatherman SP (eds.)
surface topography and its change in time, altimeters
(2001) Sea Level Rise: History and Consequences.
provide information on the Earth’s gravity field, the London: Academic Press.
shape and structure of the ocean bottom, the inte- Fu LL and Cheney RE (1995) Application of satellite
grated heat and salt content of the ocean, and geo- altimetry to ocean circulation studies, 1987–1994.
strophic ocean currents. Much progress has been Reviews of Geophysics Suppl: 213--223.
made in the development of operational ocean ap- Fu LL and Cazenave A (eds.) (2001) Satellite Altimetry and
plications, and altimeter data are now routinely Earth Sciences. London: Academic Press.
 SATELLITE OCEANOGRAPHY, HISTORY AND
INTRODUCTORY CONCEPTS
W. S. Wilson, NOAA/NESDIS, Silver Spring, MD, USA the long term, they can provide oceanographers with
E. J. Lindstrom, NASA Science Mission Directorate, an observational capability conceptually similar to
Washington, DC, USA that which meteorologists use on a daily basis to
J. R. Apelw, Global Ocean Associates, Silver Spring, forecast atmospheric weather.
MD, USA
Our ability to understand and forecast oceanic
Published by Elsevier Ltd. variability, how the oceans and atmosphere interact,
critically depends on an ability to observe the three-
dimensional global oceans on a long-term basis. In-
deed, the increasing recognition of the role of the
Oceanography from a satellite – the words themselves ocean in weather and climate variability compels us
sound incongruous and, to a generation of scientists ac- to implement an integrated, operational satellite and
customed to Nansen bottles and reversing thermometers, in situ observing system for the ocean now – so that
the idea may seem absurd. it may complement the system which already exists
Gifford C. Ewing (1965) for the atmosphere.

Introduction: A Story of Two The Early Era

Communities The origins of satellite oceanography can be traced
The history of oceanography from space is a story of back to World War II – radar, photogrammetry, and
the coming together of two communities – satellite the V-2 rocket. By the early 1960s, a few scientists
remote sensing and traditional oceanography. had recognized the possibility of deriving useful
For over a century oceanographers have gone to
sea in ships, learning how to sample beneath the
surface, making detailed observations of the vertical
distribution of properties. Gifford Ewing noted that
oceanographers had been forced to consider ‘‘the
class of problems that derive from the vertical dis-
tribution of properties at stations widely separated in
space and time.’’
With the introduction of satellite remote sensing in
the 1970s, traditional oceanographers were provided
with a new tool to collect synoptic observations
of conditions at or near the surface of the global
ocean. Since that time, there has been dramatic
progress; satellites are revolutionizing oceanography.
(Appendix 1 provides a brief overview of the prin-
ciples of satellite remote sensing.)
Yet much remains to be done. Traditional sub-
surface observations and satellite-derived observations
of the sea surface – collected as an integrated set of
observations and combined with state-of-the-art
models – have the potential to yield estimates of the
three-dimensional, time-varying distribution of prop- Figure 1 Thermal infrared image of the US southeast coast
erties for the global ocean. Neither a satellite nor an showing warmer waters of the Gulf Stream and cooler slope
in situ observing system can do this on its own. Fur- waters closer to shore taken in the early 1960s. While the
resolution and accuracy of the TV on Tiros were not ideal, they
thermore, if such observations can be collected over
were sufficient to convince oceanographers of the potential
usefulness of infrared imagery. The advanced very high resolu-
tion radiometer (AVHRR) scanner (see text) has improved images
w
Deceased considerably. Courtesy of NASA.

509
510 SATELLITE OCEANOGRAPHY, HISTORY AND INTRODUCTORY CONCEPTS

oceanic information from the existing aerial sensors. this report. Dr. Ewing has since come to be justifiably
These included (1) the polar-orbiting meteorological regarded as the father of oceanography from space.
satellites, especially in the 10–12-mm thermal infra- A second important step occurred in 1969 when
red band; and (2) color photography taken by the Williamstown Conference was held at Williams
astronauts in the Mercury, Gemini, and Apollo College in Massachusetts. The ensuing Kaula report
manned spaceflight programs. Examples of the kinds set forth the possibilities for a space-based geodesy
of data obtained from the National Aeronautics and mission to determine the equipotential figure of the
Space Administration (NASA) flights collected in the Earth using a combination of (1) accurate tracking of
1960s are shown in Figures 1 and 2. satellites and (2) the precision measurement of sat-
Such early imagery held the promise of deriving ellite elevation above the sea surface using radar
interesting and useful oceanic information from altimeters. Dr. William Von Arx of WHOI realized
space, and led to three important conferences on the possibilities for determining large-scale oceanic
space oceanography during the same time period. currents with precision altimeters in space. The re-
In 1964, NASA sponsored a conference at the quirements for measurement precision of 10-cm
Woods Hole Oceanographic Institution (WHOI) to height error in the elevation of the sea surface with
examine the possibilities of conducting scientific re- respect to the geoid were articulated. NASA scien-
search from space. The report from the conference, tists and engineers felt that such accuracy could be
entitled Oceanography from Space, summarized achieved in the long run, and the agency initiated the
findings to that time; it clearly helped to stimulate a Earth and Ocean Physics Applications Program, the
number of NASA projects in ocean observations and first formal oceans-oriented program to be estab-
sensor development. Moreover, with the exception of lished within the organization. The required accur-
the synthetic aperture radar (SAR), all instruments acy was not to be realized until 1992 with TOPEX/
flown through the 1980s used techniques described in Poseidon, which was reached only over a 25-year

Figure 2 Color photograph of the North Carolina barrier islands taken during the Apollo-Soyuz Mission (AS9-20-3128). Capes
Hatteras and Lookout, shoals, sediment- and chlorophyll-bearing flows emanating from the coastal inlets are visible, and to the right,
the blue waters of the Gulf Stream. Cloud streets developing offshore the warm current suggest that a recent passage of a cold polar
front has occurred, with elevated air–sea evaporative fluxes. Later instruments, such as the coastal zone color scanner (CZCS) on
Nimbus-7 and the SeaWiFS imager have advanced the state of the art considerably. Courtesy of NASA.
 SATELLITE OCEANOGRAPHY, HISTORY AND INTRODUCTORY CONCEPTS 511

period of incremental progress that saw the flights of approximately 5000 km down to 1 cm. The confer-
five US altimetric satellites of steadily increasing ence identified several data requirements in ocean-
capabilities: Skylab, Geos-3, Seasat, Geosat, and ography that could be addressed with space-based
TOPEX/Poseidon (see Figure 3 for representative radar and radiometers. These included determination
satellites). of surface currents, Earth and ocean tides, the shape
A third conference, focused on sea surface topo- of the marine geoid, wind velocity, wave refraction
graphy from space, was convened by the National patterns and spectra, and wave height. The confer-
Oceanic and Atmospheric Administration (NOAA), ence established a broad scientific justification for
NASA, and the US Navy in Miami in 1972, with space-based radar and microwave radiometers, and
‘sea surface topography’ being defined as undu- it helped to shape subsequent national programs in
lations of the ocean surface with scales ranging from space oceanography.

NOAA /TIROS

NROSS

Seasat Nimbus-7 Geosat

Figure 3 Some representative satellites: (1) Seasat, the first dedicated oceanographic satellite, was the first of three major launches
in 1978; (2) the Tiros series of operational meteorological satellites carried the advanced very high resolution radiometer (AVHRR)
surface temperature sensor; Tiros-N, the first of this series, was the second major launch in 1978; (3) Nimbus-7, carrying the CZCS
color scanner, was the third major launch in 1978; (4) NROSS, an oceanographic satellite approved as an operational demonstration
in 1985, was later cancelled; (5) Geosat, an operational altimetric satellite, was launched in 1985; and (6) this early version of TOPEX
was reconfigured to include the French Poseidon; the joint mission TOPEX/Poseidon was launched in 1992. Courtesy of NASA.
512 SATELLITE OCEANOGRAPHY, HISTORY AND INTRODUCTORY CONCEPTS

The First Generation convinced most workers of the validity of the data.
Today, higher-resolution, temperature-calibrated in-
Two first-generation ocean-viewing satellites, Skylab frared imagery constitutes a valuable data source
in 1973 and Geos-3 in 1975, had partially responded used frequently by ocean scientists around the world.
to concepts resulting from the first two of these During the same period, spacecraft and air-
conferences. Skylab carried not only several astro- craft programs taking ocean color imagery were
nauts, but a series of sensors that included the delineating the possibilities and difficulties of deter-
S-193, a radar-altimeter/wind-scatterometer, a long- mining sediment and chlorophyll concentrations re-
wavelength microwave radiometer, a visible/infrared motely. Figure 2 is a color photograph of the North
scanner, and cameras. S-193, the so-called Rad/Scatt, Carolina barrier islands taken with a hand-held
was advanced by Drs. Richard Moore and Willard camera, with Cape Hatteras in the center. Shoals and
Pierson. These scientists held that the scatterometer sediment- and chlorophyll-bearing flows emanating
could return wind velocity measurements whose ac- from the coastal inlets are visible, and to the right,
curacy, density, and frequency would revolutionize the blue waters of the Gulf Stream. Cloud streets
marine meteorology. Later aircraft data gathered by developing offshore the warm stream suggest a re-
NASA showed that there was merit to their asser- cent passage of a cold polar front and attendant in-
tions. Skylab’s scatterometer was damaged during creases in air–sea evaporative fluxes.
the opening of the solar cell panels, and as a con-
sequence returned indeterminate results (except for
passage over a hurricane), but the altimeter made
observations of the geoid anomaly due to the Puerto
The Second Generation
Rico Trench. The combination of the early data and advances in
Geos-3 was a small satellite carrying a dual-pulse scientific understanding that permitted the exploit-
radar altimeter whose mission was to improve the ation of those data resulted in spacecraft sensors
knowledge of the Earth’s marine geoid, and coinci- explicitly designed to look at the sea surface. Infor-
dentally to determine the height of ocean waves via mation returned from altimeters and microwave
the broadening of the short transmitted radar pulse radiometers gave credence and impetus to dedicated
upon reflection from the rough sea surface. Before microwave spacecraft. Color measurements of the
the end of its 4-year lifetime, Geos-3 was returning sea made from aircraft had indicated the efficacy of
routine wave height measurements to the National optical sensors for measurement of near-surface
Weather Service for inclusion in its Marine Waves chlorophyll concentrations. Infrared radiometers re-
Forecast. Altimetry from space had become a clear turned useful sea surface temperature measurements.
possibility, with practical uses of the sensor imme- These diverse capabilities came together when, dur-
diately forthcoming. The successes of Skylab and ing a 4-month interval in 1978, the USA launched a
Geos-3 reinforced the case for a second generation of triad of spacecraft that would profoundly change the
radar-bearing satellites to follow. way ocean scientists would observe the sea in the
The meteorological satellite program also pro- future. On 26 June, the first dedicated oceanographic
vided measurements of sea surface temperature using satellite, Seasat, was launched; on 13 October,
far-infrared sensors, such as the visible and infrared TIROS-N was launched immediately after the cata-
scanning radiometer (VISR), which operated at strophic failure of Seasat on 10 October; and on
wavelengths near 10 mm, the portion of the terrestrial 24 October, Nimbus-7 was lofted. Collectively they
spectrum wherein thermal radiation at terrestrial carried sensor suites whose capabilities covered vir-
temperatures is at its peak, and where coincidentally tually all known ways of observing the oceans re-
the atmosphere has a broad passband. The coarse, motely from space. This second generation of
5-km resolution of the VISR gave blurred tempera- satellites would prove to be extraordinarily success-
ture images of the sea, but the promise was clearly ful. They returned data that vindicated their pro-
there. Figure 1 is an early 1960s TV image of the ponents’ positions on the measurement capabilities
southeastern USA taken by the NASA TIROS pro- and utility, and they set the direction for almost all
gram, showing the Gulf Stream as a dark signal. subsequent efforts in satellite oceanography.
While doubts were initially held by some ocean- In spite of its very short life of 99 days, Seasat
ographers as to whether such data actually repre- demonstrated the great utility of altimetry by meas-
sented the Gulf Stream, nevertheless the repeatability uring the marine geoid to within a very few meters,
of the phenomenon, the verisimilitude of the pos- by inferring the variability of large-scale ocean sur-
itions and temperatures with respect to conventional face currents, and by determining wave heights. The
wisdom, and their own objective judgment finally wind scatterometer could yield oceanic surface wind
 SATELLITE OCEANOGRAPHY, HISTORY AND INTRODUCTORY CONCEPTS 513

velocities equivalent to 20 000 ship observations per part of NASA, the US Navy, and NOAA – the Na-
day. The scanning multifrequency radiometer also tional Oceanographic Satellite System (NOSS). This
provided wind speed and atmospheric water content was to be an operational system, with a primary and
data; and the SAR penetrated clouds to show fea- a backup satellite, along with a fully redundant
tures on the surface of the sea, including surface and ground data system. NOSS was proposed shortly
internal waves, current boundaries, upwellings, and after the failure of Seasat, with a first launch ex-
rainfall patterns. All of these measurements could pected in 1986. NASA formed a ‘science working
be extended to basin-wide scales, allowing ocean- group’ (SWG) in 1980 under Francis Bretherton to
ographers a view of the sea never dreamed of before. define the potential that NOSS offered the oceano-
Seasat stimulated several subsequent generations of graphic community, as well as to recommend sensors
ocean-viewing satellites, which outline the chron- to constitute the 25% of its payload allocated for
ologies and heritage for the world’s ocean-viewing research. However, with oceanographers essentially
spacecraft. Similarly, the early temperature and color brought in as junior partners, the job of securing a
observations have led to successor programs that new start for NOSS fell to the operational com-
provide large quantities of quantitative data to munity – which it proved unable to do. NOSS was
oceanographers around the world. canceled in early 1981. The prevailing and realistic
view was that the greater community was not ready
to implement such an operational system.
The Third Generation
The second generation of spacecraft would demon- Science Working Groups
strate that variables of importance to oceanography
could be observed from space with scientifically useful During this period, SWGs were formed to look at
accuracy. As such, they would be characterized as each promising satellite sensing technique, assess its
successful concept demonstrations. And while both potential contribution to oceanographic research,
first- and second-generation spacecraft had been ex- and define the requirements for its future flight. The
clusively US, international participation in demon- notable early groups were the TOPEX SWG formed
strating the utility of their data would lead to the in 1980 under Carl Wunsch for altimetry, Satellite
entry of Canada, the European Space Agency (ESA), Surface Stress SWG in 1981 under James O’Brien for
France, and Japan into the satellite program during scatterometry, and Satellite Ocean Color SWG in
this period. This article focuses on the US effort. 1981 under John Walsh for color scanners. These
Additional background on US third-generation mis- SWGs were true partnerships between the remote
sions covering the period 1980–87 can be found in the sensing and oceanographic communities, developing
series of Annual Reports for the Oceans Program consensus for what would become the third gener-
(NASA Technical Memoranda 80233, 84467, 85632, ation of satellites.
86248, 87565, 88987, and 4025).

Partnership with Oceanography Partnership with Field Centers

Up to 1978, the remote sensing community had been Up to this time, NASA’s Oceans Program had been a
the prime driver of oceanography from space and collection of relatively autonomous, in-house activ-
there were overly optimistic expectations. Indeed, ities run by NASA field centers. In 1981, an overrun
the case had not yet been made that these obser- in the space shuttle program forced a significant
vational techniques were ready to be exploited for budget cut at NASA headquarters, including the
ocean science. Consequently, in early 1979, the Oceans Program. This in turn forced a reprioritiza-
central task was establishing a partnership with the tion and refocusing of NASA programs. This was a
traditional oceanographic community. This meant blessing in disguise, as it provided an opportunity to
involving them in the process of evaluating the per- initiate a comprehensive, centrally led program –
formance of Seasat and Nimbus-7, as well as build- which would ultimately result in significant funding
ing an ocean science program at NASA headquarters for the oceanographic as well as remote-sensing
to complement the ongoing remote sensing effort. communities. Outstanding relationships with indi-
viduals like Mous Chahine in senior management at
the Jet Propulsion Laboratory (JPL) enabled the
National Oceanographic Satellite System
partnership between NASA headquarters and the
This partnership with the oceanographic community two prime ocean-related field centers (JPL and
was lacking in a notable and early false start on the the Goddard Space Flight Center) to flourish.
514 SATELLITE OCEANOGRAPHY, HISTORY AND INTRODUCTORY CONCEPTS

Partnerships in Implementation program. If requisite funding was to be secured to

A milestone policy-level meeting occurred on 13 July pay for proposed missions, it was critical that
1982 when James Beggs, then Administrator of government policymakers, the Congress, the greater
NASA, hosted a meeting of the Ocean Principals oceanographic community, and the public had a
Group – an informal group of leaders of the ocean- good understanding of oceanography from space
related agencies. A NASA presentation on oppor- and its potential benefits. In response to this need, a
tunities and prospects for oceanography from space set of posters, brochures, folders, and slide sets
was received with much enthusiasm. However, when was designed by Payson Stevens of Internetwork
asked how NASA intended to proceed, Beggs told Incorporated and distributed to a mailing list which
the group that – while NASA was the sole funding grew to exceed 3000. These award-winning ma-
agency for space science and its missions – numerous terials – sharing a common recognizable identity –
agencies were involved in and support oceanography. were both scientifically accurate and esthetically
Beggs said that NASA was willing to work with other pleasing.
agencies to implement an ocean satellite program, At the same time, dedicated issues of magazines
but that it would not do so on its own. Beggs’ and journals were prepared by the community of in-
statement defined the approach to be pursued in volved researchers. The first example was the issue
implementing oceanography from space, namely, a of Oceanus (1981), which presented results from the
joint approach based on partnerships. second-generation missions and represented a first
step toward educating the greater oceanographic
community in a scientifically useful and balanced way
Research Strategy for the Decade about realistic prospects for satellite oceanography.
As a further step in strengthening its partnership with
the oceanographic community, NASA collaborated
with the Joint Oceanographic Institutions Incorpor- Implementation Studies
ated (JOI), a consortium of the oceanographic insti- Given the SWG reports taken in the context of
tutions with a deep-sea-going capability. At the the Research Strategy, the NASA effort focused on
time, JOI was the only organization in a position to the following sensor systems (listed with each are the
represent and speak for the major academic oceano- various flight opportunities which were studied):
graphic institutions. A JOI satellite planning com-
mittee (1984) under Jim Baker examined SWG • altimetry – the flight of a dedicated altimeter
reports, as well as the potential synergy between the mission, first TOPEX as a NASA mission, and
variety of oceanic variables which could be measured then TOPEX/Poseidon jointly with the French
from space; this led to the idea of understanding the Centre Nationale d’Etudes Spatiales (CNES);
ocean as a system. (From this, it was a small leap to • scatterometry – the flight of a NASA scatte-
understanding the Earth as a system, the goal of rometer (NSCAT), first on NOSS, then on the
NASA’s Earth Observing System (EOS).) Navy Remote Ocean Observing Satellite
The report of this Committee, Oceanography (NROSS), and finally on the Advanced Earth
from Space: A Research Strategy for the Decade, Observing Satellite (ADEOS) of the Japanese
1985–1995, linked altimetry, scatterometry, and National Space Development Agency (NASDA);
ocean color with the major global ocean research • visible radiometry – the flight of a NASA color
programs being planned at that time – the World scanner on a succession of missions (NOSS,
Ocean Circulation Experiment (WOCE), Tropical NOAA-H/-I, SPOT-3 (Système Pour l’Obser-
Ocean Global Atmosphere (TOGA) program, and vation de la Terre), and Landsat-6) and finally the
Joint Global Ocean Flux Study (JGOFS). This strat- purchase of ocean color data from the Sea-viewing
egy, still being followed today, served as a catalyst to Wide Field-of-view Sensor (SeaWiFS) to be flown
engage the greater community, to identify the most by the Orbital Sciences Corporation (OSC);
important missions, and to develop an approach for • microwave radiometry – a system to utilize data
their prioritization. Altimetry, scatterometry, and from the series of special sensor microwave ima-
ocean color emerged from this process as national ger (SSMI) radiometers to fly on the Defense
priorities. Meteorological Satellite Program satellites;
• SAR – a NASA ground station, the Alaska SAR
Facility, to enable direct reception of SAR data
Promotion and Advocacy
from the ERS-1/-2, JERS-1, and Radarsat satel-
The Research Strategy also provided a basis for lites of the ESA, NASDA, and the Canadian Space
promoting and building an advocacy for the NASA Agency, respectively.
 SATELLITE OCEANOGRAPHY, HISTORY AND INTRODUCTORY CONCEPTS 515

New Starts after Seasat; and SeaWiFS was launched in 1997 – 19

years after Nimbus-7. (In addition to the missions
Using the results of the studies listed above, the
mentioned in Table 1, the Japanese ADEOS-1 in-
Oceans Program entered the new start process at
cluded the US NSCAT in its sensor complement, and
NASA headquarters numerous times attempting to
the US Aqua included the Japanese advanced micro-
secure funds for implementation of elements of the
wave scanning radiometer (AMSR); the United States
third generation. TOPEX was first proposed as a
provided a launch for the Canadian RADARSAT-1.)
NASA mission in 1980. However, considering lim-
In fact, these missions came so late that they had
ited prospects for success, partnerships were sought
limited overlap with the field phases of the major
and the most promising one was with the French.
ocean research programs (WOCE, TOGA, and
CNES initially proposed a mission using a SPOT bus
JGOFS) they were to complement. Why did it take
with a US launch. However, NASA rejected this be-
so long?
cause SPOT, constrained to be Sun-synchronous,
would alias solar tidal components. NASA proposed
Understanding and Consensus
instead a mission using a US bus capable of flying in
a non-Sun-synchronous orbit with CNES providing First, it took time to develop a physically un-
an Ariane launch. The NASA proposal was accepted ambiguous understanding of how well the satellite
for study in fiscal year (FY) 1983, and a new start sensors actually performed, and this involved learn-
was finally secured for the combined TOPEX/ ing to cope with the data – satellite data rates being
Poseidon in FY 1987. orders of magnitude larger than those encountered in
In 1982 when the US Navy first proposed NROSS, traditional oceanography. For example, it was not
NASA offered to be a partner and provide a scatte- until 3 years after the launch of Nimbus-7 that CZCS
rometer. The US Navy and NASA obtained new data could be processed as fast as collected by the
starts for both NROSS and NSCAT in FY 1985. satellite. And even with only a 3-month data set from
However, NROSS suffered from a lack of strong Seasat, it took 4 years to produce the first global
support within the navy, experienced a number of maps of variables such as those shown in Figure 4.
delays, and was finally terminated in 1987. Even In evaluating the performance of both Seasat and
with this termination, NASA was able to keep Nimbus-7, it was necessary to have access to the
NSCAT alive until establishing the partnership with data. Seasat had a free and open data policy; and
NASDA for its flight on their ADEOS mission. after a very slow start, the experiment team concept
Securing a means to obtain ocean color obser- (where team members had a lengthy period of ex-
vations as a follow-on to the coastal zone color clusive access to the data) for the Nimbus-7 CZCS
scanner (CZCS) was a long and arduous process, fi- was replaced with that same policy. Given access to
nally coming to fruition in 1991 when a contract was the data, delays were due to a combination of sorting
signed with the OSC to purchase data from the flight out the algorithms for converting the satellite ob-
of their SeaWiFS sensor. By that time, a new start had servations into variables of interest, as well as being
already been secured for NASA’s EOS, and ample constrained by having limited access to raw com-
funds were available in that program for the Sea- puting power.
WiFS data purchase. In addition, the rationale for the third-generation
Finally, securing support for the Alaska SAR Fa- missions represented a major paradigm shift. While
cility (now the Alaska Satellite Facility to reflect its earlier missions had been justified largely as dem-
broader mission) was straightforward; being small in onstrations of remote sensing concepts, the third-
comparison with the cost of flying space hardware, generation missions would be justified on the basis of
its funding had simply been included in the new start their potential contribution to oceanography. Hence,
that NSCAT obtained in FY 1985. Also, funding for the long time it took to understand sensor perform-
utilization of SSMI data was small enough to be ance translated into a delay in being able to convince
covered by the Oceans Program itself. traditional oceanographers that satellites were an
important observational tool ready to be exploited
for ocean science. As this case was made, it was
Implementing the Third Generation possible to build consensus across the remote sensing
and oceanographic communities.
With the exception of the US Navy’s Geosat, these
third-generation missions would take a very long time
Space Policy
to come into being. As seen in Table 1, TOPEX/
Poseidon was launched in 1992 – 14 years after Seasat; Having such consensus reflected at the highest levels of
NSCAT was launched on ADEOS in 1996 – 18 years government was another matter. The White House
516 SATELLITE OCEANOGRAPHY, HISTORY AND INTRODUCTORY CONCEPTS

Table 1 Some major ocean-related missions

Year USA Russia Japan Europe Canada Other

1968 Kosmos 243

Nimbus-3
1970 Nimbus-4 Kosmos 384
1972 Nimbus-5
1974 Skylab
Nimbus-6, Geos-3
1976
1978 Nimbus-7, Seasat
Kosmos 1076
1980 Kosmos 1151
1982
Kosmos 1500
1984 Kosmos 1602
Geosat
1986 Kosmos 1776
Kosmos 1870 MOS 1A
1988 OKEAN 1
1990 OKEAN 2 MOS 1B
Almaz-1, ERS-1
OKEAN 3
1992 Topex/Poseidona JERS-1
1994 OKEAN 7
OKEAN 8 ERS-2 RADARSAT-1
1996 ADEOS-1
SeaWiFS TRMM b
1998 GFO
Terra, QuikSCAT OKEAN-O #1 OCEANSAT-1 i
c
2000 CHAMP
Meteor-3 #1 Jason-1 e
2002 Aqua; GRACE d ENVISAT HY-1A j
WINDSAT ADEOS-2
2004 ICESat SICH-1M f
CRYOSat
2006 ALOS GOCE, MetOp-1 HY-1B j
Meteor-3M #2 SMOS RADARSAT-2 OCEANSAT-2 i
2008 OSTM/Jason-2 g
NPP; Aquariush CryoSat-2 HY-1C, HY-2A j
2010 GCOM-W Sentinel-3, MetOp-2 OCEANSAT-3 i

a
US/France TOPEX/Poseidon.
b
Japan/US TRMM.
c
German CHAMP.
d
US/German GRACE.
e
France/US Jason-1.
f
Russia/Ukraine Sich-1M.
g
France/US Jason-2/OSTM.
h
US/Argentina Aquarius.
i
India OCEANSAT series.
j
China HY series.
Updated version of similar data in Wilson WS, Fellous JF, Kawamura H, and Mitnik L (2006) A history of oceanography from space. In:
Gower JFR (ed.) Manual of Remote Sensing, Vol. 6: Remote Sensing of the Environment, pp. 1–31. Bethesda, MD: American Society
for Photogrammetry and Remote Sensing.

Fact Sheet on US Civilian Space Policy of 11 October applications.’’ Landsat was commercialized in 1979 as
1978 states, ‘‘y emphasizing space applications y part of this space policy. As Robert Stewart explains,
will bring important benefits to our understanding ‘‘Clearly the mood at the presidential level was that
of earth resources, climate, weather, pollution y earth remote sensing, including the oceans, was a
and provide for the private sector to take an practical space application more at home outside the
increasing responsibility in remote sensing and other scientific community. It took almost a decade to get an
 SATELLITE OCEANOGRAPHY, HISTORY AND INTRODUCTORY CONCEPTS 517

Figure 4 Global sea surface topography c. 1983. This figure shows results computed from the 70 days of Seasat altimeter data in
1978. Clearly visible in the mean sea surface topography, the marine geoid (upper panel), are the Mid-Atlantic Ridge (1) and
associated fracture zones (2), trenches in the western Pacific (3), the Hawaiian Island chain (4), and the Emperor seamount chain (5).
Superimposed on the mean surface is the time-varying sea surface topography, the mesoscale variability (lower panel), associated
with the variability of the ocean currents. The largest deviations (10–25 cm), yellow and orange, are associated with the western
boundary currents: Gulf Stream (6), Kuroshio (7), Agulhas (8), and Brazil/Falkland Confluence (9); large variations also occur in the
West Wind Drift (10). Courtesy of NASA.

understanding at the policy level that scientific needs space science program. However, implementing a
were also important, and that we did not have the satellite mission across different organizations,
scientific understanding necessary to launch an oper- countries, and cultures is both challenging and time-
ational system for climate.’’ The failures of NOSS, and consuming. An enormous amount of time and energy
later NROSS, were examples of an effort to link re- was invested in studies of various flight options,
mote sensing directly with operational applications many of which fizzled out, but some were imple-
without the scientific underpinning. mented. With the exception of the former Soviet
The view in Europe was not dissimilar; govern- Union, NASA’s third-generation missions would be
ments felt that cost recovery was a viable financial joint with each nation having a space program at
scheme for ocean satellite missions, that is, the data that time, as well as with a private company.
have commercial value and the user would be willing
to pay to help defray the cost of the missions. The Geosat Exception
Geosat was the notable US exception, having been
Joint Satellite Missions
implemented so quickly after the second generation.
It is relatively straightforward to plan and implement It was approved in 1981 and launched in 1985 in
missions within a single agency, as with NASA’s order to address priority operational needs on the
518 SATELLITE OCEANOGRAPHY, HISTORY AND INTRODUCTORY CONCEPTS

part of the US Navy. During the second half of its (e.g., real-time SeaWiFS for nonresearch purposes,
mission, data would become available within 1–2 RADARSAT), along with a variety of intermediate
days. As will be discussed below, Geosat shared a cases (e.g., ERS, ALOS, and ENVISAT). From a
number of attributes with the meteorological satel- scientific perspective, full and open access is the
lites: it had a specific focus; it met priority oper- preferred route, in order to obtain the best under-
ational needs for its user; experience was available standing of how systems perform, to achieve the full
for understanding and using the observations; and its potential of the missions for research, and to lay the
implementation was done in the context of a single most solid foundation for an operational system. Full
organization. and open access is also a means to facilitate the de-
velopment of a healthy and competitive private sec-
tor to provide value-added services. Further, if the
Challenges Ahead international community is to have an effective ob-
Scientific Justification serving system for climate, a full and open data
policy will be needed, at least for that purpose.
As noted earlier, during the decade of the 1980s,
there was a dearth of ocean-related missions in the
United States, it being difficult to justify a mission In Situ Observations
based on its contribution to ocean science. Then later Satellites have made an enormous contribution
in that decade, NASA conceived of the EOS and was enabling the collection of in situ observations from
able to make the case that Earth science was suf- in situ platforms distributed over global oceans. The
ficient justification for a mission. Also noted earlier, Argos (plural spelling; not to be confused with Argo
ESA initially had no appropriate framework for profiling floats) data collection and positioning sys-
Earth science missions, and a project like ERS-1 was tem has flown on the NOAA series of polar-orbiting
pursued under the assumption that it would help operational environmental satellites continuously
develop commercial and/or operational applications since 1978. It provides one-way communication
of remote sensing of direct societal benefit. Its suc- from data collection platforms, as well as positioning
cessor, ERS-2, was justified on the basis of needing of those platforms. While an improved Argos
continuity of SAR coverage for the land surface, ra- capability (including two-way communications) is
ther than the need to monitor ocean currents. And coming with the launch of MetOp-1 in 2006,
the ENVISAT was initially decided by ESA member oceanographers are looking at alternatives – Iridium
states as part of the Columbus program of the being one example – which offer significant higher
International Space Station initiative. The advent of data rates, as well as two-way communications.
an Earth Explorer program in 1999 represented a In addition, it is important to note that the Inter-
change in this situation. As a consequence, new governmental Oceanographic Commission and the
Earth science missions – GOCE, CryoSat, and World Meteorological Organization have established
SMOS – all represent significant steps forward. the Joint Technical Commission for Oceanography
This ESA program, together with similar efforts at and Marine Meteorology (JCOMM) to bring a focus
NASA, are leading to three sets of ground-breaking to the collection, formatting, exchange, and archival
scientific missions which have the potential to sig- of data collected at sea, whether they be oceanic
nificantly impact oceanography. GOCE and GRACE or atmospheric. JCOMM has established a center,
will contribute to an improved knowledge of the JCOMMOPS, to serve as the specific institutional
Earth’s gravity field, as well as the mass of water on focus to harmonize the national contributions of
the surface of the Earth. CryoSat-2 and ICESat will Argo floats, surface-drifting buoys, coastal tide gau-
contribute to knowledge of the volume of water ges, and fixed and moored buoys. JCOMMOPS will
locked up in polar and terrestrial ice sheets. Finally, play an important role helping contribute to ‘inte-
SMOS and Aquarius will contribute to knowledge of grated observations’ described below.
the surface salinity field of the global oceans. Toge-
ther, these will be key ingredients in addressing the
global water cycle. Integrated Observations
To meet the demands of both the research and the
Data Policy
broader user community, it will be necessary to
The variety of missions described above show a mix focus, not just on satellites, but ‘integrated’ ob-
of data policies, from full and open access without serving systems. Such systems involve combinations
any period of exclusive use (e.g., TOPEX/Poseidon, of satellite and in situ systems feeding observations
Jason, and QuikSCAT) to commercial distribution into data-processing systems capable of delivering a
 SATELLITE OCEANOGRAPHY, HISTORY AND INTRODUCTORY CONCEPTS 519

comprehensive view of one or more geophysical observations. From the operational perspective, it is
variables (sea level, surface temperature, winds, etc.). necessary to implement proven, scientifically sound,
Three examples help illustrate the nature of inte- cost-effective observing systems – where the un-
grated observing systems. First, consider global sea interrupted supply of real-time data is critical. This is
level rise. The combination of the Jason-1 altimeter, a big challenge to be met by the space systems because
its precision orbit determination system, and the of the demand for higher reliability and redundancy,
suite of precision tide gauges around the globe allow at the same time calling for stringent calibration and
scientists to monitor changes in volume of the accuracy requirements. Meeting these sometimes
oceans. The growing global array of Argo profiling competing, but quite complementary demands will be
floats allows scientists to assess the extent to which the challenge and legacy of the next generation of
those changes in sea level are caused by changes in ocean remote-sensing satellites.
the temperature and salinity structure of the upper
ocean. ICESat and CryoSat will provide estimates of Meteorological Institutional Experience
changes in the volume of ice sheets, helping assess the
With the launch in 1960 of the world’s first me-
extent to which their melting contributes to global
teorological satellite, the polar-orbiting Tiros-1
sea level rise. And GRACE and GOCE will provide
carrying two TV cameras, the value of the resulting
estimates of the changes in the mass of water on the
imagery to the operational weather services was
Earth’s surface. Together, systems such as these will
recognized immediately. The very next year a Na-
enable an improved understanding of global sea level
tional Operational Meteorological System was im-
rise and, ultimately, a reduction in the wide range of
plemented, with NASA to build and launch the
uncertainty in future projections.
satellites and the Weather Bureau to be the operator.
Second, global estimates of vector winds at the sea
The feasibility of using satellite imagery to locate and
surface are produced from the scatterometer on
track tropical storms was soon demonstrated, and by
QuikSCAT, a global array of in situ surface buoys,
1969 this capability had become a regular part of
and the Seawinds data processing system. Delivery of
operational weather forecasting. In 1985, Richard
this product in real time has significant potential
Hallgren, former Director of the National Weather
to improve marine weather prediction. The third ex-
Service, stated, ‘‘the use of satellite information
ample concerns the Jason-1 altimeter together with
simply permeates every aspect of the [forecast and
Argo. When combined in a sophisticated data as-
warning] process and all this in a mere 25 years.’’
similation system – using a state-of-the-art ocean
In response to these operational needs, there has been
model – these data enable the estimation of the
a continuing series of more than 50 operational,
physical state of the ocean as it changes through time.
polar-orbiting satellites in the United States alone!
This information – the rudimentary ‘weather map’
The first meteorological satellites had a specific
depicting the circulation of the oceans – is a critical
focus on synoptic meteorology and weather fore-
component of climate models and provides the fun-
casting. Initial image interpretation was straight-
damental context for addressing a broad range of
forward (i.e., physically unambiguous), and there was
issues in chemical and biological oceanography.
a demonstrated value of resulting observations in
meeting societal needs. Indeed, since 1960 satellites
Transition from Research to Operations have ensured that no hurricane has gone undetected.
In addition, the coupling between meteorology and
The maturing of the discipline of oceanography in-
remote sensing started very near the beginning. An
cludes the development of a suite of global oceano-
‘institutional mechanism’ for transition from research
graphic services being conducted in a manner similar
to operations was established almost immediately.
to what exists for weather services. The delivery of
Finally, recognition of this endeavor extended to the
these services and their associated informational
highest levels of government, resulting in the financial
products will emerge as the result of the successes in
commitment needed to ensure success.
ocean science (‘research push’), as well as an in-
creasing demand for ocean analyses and forecasts
Oceanographic Institutional Issue
from a variety of sectors (‘user pull’).
From the research perspective, it is necessary to Unlike meteorology where there is a National Wea-
‘transition’ successfully demonstrated (‘experimental’) ther Service in each country to provide an insti-
observing techniques into regular, long-term, system- tutional focus, ocean-observing systems have
atic (‘operational’) observing systems to meet a broad multiple performer and user institutions whose
range of user requirements, while maintaining the interests must be reconciled. For oceanography,
capability to collect long-term, ‘research-quality’ this is a significant challenge working across
520 SATELLITE OCEANOGRAPHY, HISTORY AND INTRODUCTORY CONCEPTS

‘institutions’, where the in situ research is in one or operational basis since 2001 (Mercator). This part-
more agencies, the space research and development is nership could serve as a model in the effort to de-
in another, and operational activities in yet another – velop operational oceanography in other countries.
with possibly separate civil and military systems. Drawing from this experience working together
In the United States, the dozen agencies with ocean- within France, IFREMER is leading the European
related responsibilities are using the National integrated project, MERSEA, aimed at establishing a
Oceanographic Partnership Program and its Ocea- basis for a European center for ocean monitoring and
n.US Office to provide a focus for reconciling forecasting.
such interests. In the United Kingdom, there is the
Interagency Committee on Marine Science and
Ocean Climate
Technology.
In France, there is the Comité des Directeurs If we are to adequately address the issue of global
d’Organismes sur l’Océanographie, which gathers climate change, it is essential that we are able to
the heads of seven institutions interested in the de- justify the satellite systems required to collect the
velopment of operational oceanography, including global observations ‘over the long term’. Whether it
CNES, meteorological service, ocean research insti- be global sea level rise or changes in Arctic sea ice
tution, French Research Institute for Exploitation of cover or sea surface temperature, we must be able to
the Sea (IFREMER), and the navy. This group of sustain support for the systems needed to produce
agencies has worked effectively over the past 20 climate-quality data records, as well as ensure the
years to establish a satellite data processing and continuing involvement of the scientific community.
distribution system (AVISO), the institutional sup- Koblinsky and Smith have outlined the inter-
port for a continuing altimetric satellite series national consensus for ocean climate research needs
(TOPEX/Poseidon, Jason), the framework for the and identified the associated observational require-
French contribution to the Argo profiling float pro- ments. In addition to their value for research, we are
gram (CORIOLIS), and to create a public corpor- compelled by competing interests to demonstrate the
ation devoted to ocean modeling and forecasting, value of such observations in meeting a broad range
using satellite and in situ data assimilation in an of societal needs. Climate observations pose

1 GHz 1 MHz

1018 1015 1012 109 106

Frequency (Hz)

Microwave radars

Spectral windows Microwave emissions

used for remote
sensing
Thermal infrared
Near infrared
Visible
100%

Transmittance
through the
atmosphere

0%
1 nm 1 μm 1 mm 1m 1 km

10−9 10−6 10 −3 1 103

Wavelength (m)

Figure 5 The electromagnetic spectrum showing atmospheric transmitance as a function of frequency and wavelength, along with
the spectral windows used for remote sensing. Microwave bands are typically defined by frequency and the visible/infrared by
wavelength. Adapted from Robinson IS and Guymer T (1996) Observing oceans from space. In: Summerhayes CP and Thorpe SA
(eds.) Oceanography: An Illustrated Guide, pp. 69–87. Chichester, UK: Wiley.
 SATELLITE OCEANOGRAPHY, HISTORY AND INTRODUCTORY CONCEPTS 521

challenges, since they require operational discipline DC, following a recommendation adopted at the G-8
to be collected in a long-term systematic manner, yet meeting held in Evian the previous month; this sum-
also require the continuing involvement of the re- mit proposed to ‘plan and implement’ a Global Earth
search community to ensure their scientific integrity, Observation System of Systems (GEOSS). Four add-
and have impacts that may not be known for decades itional summits have been held, with participation
(unlike observations that support weather forecast- having grown to include 60 nations and 40 inter-
ing whose impact can be assessed within a matter national organizations; the GEOSS process provides
of hours or days). Together, the institutional and the political visibility – not only to implement the
observational challenges for ocean climate have been plans developed within the IGOS Partnership – but to
difficult to surmount. do so in the context of an overall Earth observation
framework. This represents a remarkable opportunity
International Integration to develop an improved understanding of the oceans
and their influence on the Earth system, and to con-
The paper by the Ocean Theme Team prepared tribute to the delivery of improved oceanographic
under the auspices of the Integrated Global Ob- products and services to serve society.
serving Strategy (IGOS) Partnership represents how
the space-faring nations are planning for the col-
lection of global ocean observations. IGOS partners Appendix 1: A Brief Overview of
include the major global research program sponsors,
global observing systems, space agencies, and inter-
Satellite Remote Sensing
national organizations. Unlike the severe attenuation in the sea, the atmos-
On 31 July 2003, the First Earth Observations phere has ‘windows’ in which certain electro-
Summit – a high-level meeting involving ministers magnetic (EM) signals are able to propagate. These
from over 20 countries – took place in Washington, windows, depicted in Figure 5, are defined in terms

a b

Visible
Infrared Microwave
radiometer
radiometer radiometer

Sunlight scattered
from below the surface

Infrared and microwave emission from the surface

c d

Satellite orbit Radar scatterometer

Radar
altimeter

Microwaves scattered
Microwaves from short waves
reflected from on the surface
the surface

Figure 6 Four techniques for making oceanic observations from satellites: (a) visible radiometry, (b) infrared and microwave
radiometry, (c) altimetry, and (d) scatterometry. Adapted from Robinson IS and Guymer T (1996) Observing oceans from space. In:
Summerhayes CP and Thorpe SA (eds.) Oceanography: An Illustrated Guide, pp. 69–87. Chichester, UK: Wiley.
522 SATELLITE OCEANOGRAPHY, HISTORY AND INTRODUCTORY CONCEPTS

Passive sensors (radiometers) Active sensors (microwave radars)

Sensor Infrared Microwave Altimetry Scatterometry SAR

Visible
type

Measured Solar radiation Infrared emission Microwave Travel time, shape, Strength of return Strength and phase
backscattered from the sea emission from the and strength of pulse when of return pulse
physical from beneath the surface sea surface reflected pulse illuminated from
variable sea surface different directions

Ocean color; Surface Ice cover, Surface topography Surface vector Surface roughness at
chlorophyll; temperature; age and motion; for geostrophic winds; fine spatial scales;
Applications primary production; ice cover sea surface currents and tides; ice cover surface and internal
water clarity; temperature; bathymetry; wave patterns;
shallow-water wind speed oceanic geoid; bathymetric patterns;
bathymetry wind and wave ice cover and motion
conditions

Figure 7 Measured physical variables and applications for both passive and active sensors, expressed as a function of sensor type.

of atmospheric transmittance – the percentage of an surface winds. The companion articles on this topic
EM signal which is able to propagate through the address various aspects of Figure 7 in more detail, so
atmosphere – expressed as a function of wavelength only this general overview is given here.
or frequency.
Given a sensor on board a satellite observing the
ocean, it is necessary to understand and remove the
Acknowledgments
effects of the atmosphere (such as scattering and at- The authors would like to acknowledge contri-
tenuation) as the EM signal propagates through it. butions to this article from Mary Cleave, Murel
For passive sensors (Figures 6(a) and 6(b)), it is then Cole, William Emery, Michael Freilich, Lee Fu, Rich
possible to relate the EM signals collected by the Gasparovic, Trevor Guymer, Tony Hollingsworthw,
sensor to the associated signals at the bottom of the Hiroshi Kawamura, Michele Lefebvre, Leonid Mit-
atmosphere, that is, the natural radiation emitted or nik, Jean-Franc¸ois Minster, Richard Moore, William
reflected from the sea surface. Note that passive Patzert, Willard Piersonw, Jim Purdom, Keith Raney,
sensors in the visible band are dependent on the Sun Payson Stevens, Robert Stewart, Ted Strub, Tasuku
for natural illumination. Tanaka, William Townsend, Mike Van Woert, and
Active sensors, microwave radar (Figures 6(c) and Frank Wentz.
6(d)), provide their own source of illumination and
have the capability to penetrate clouds and, to a
certain extent, rain. Atmospheric correction must be See also
done to remove effects for a round trip from the
satellite to the sea surface. Aircraft Remote Sensing. IR Radiometers. Satellite
With atmospheric corrections made, measure- Altimetry. Satellite Remote Sensing: Ocean
ments of physical variables are available: emitted Colour. Satellite Remote Sensing of Sea Surface
radiation for passive sensors, and the strength, phase, Temperatures. Satellite Remote Sensing SAR.
and/or travel time for active sensors. Figure 7 shows
typical measured physical variables for both types of Further Reading
sensors in their respective spectral bands, as well as Apel JR (ed.) (1972) Sea surface topography from space.
applications or derived variables of interest – ocean NOAA Technical Reports: ERL No. 228, AOML No. 7.
color, surface temperature, ice cover, sea level, and Boulder, CO: NOAA.
 SATELLITE OCEANOGRAPHY, HISTORY AND INTRODUCTORY CONCEPTS 523

Cherny IV and Raizer VY (1998) Passive Microwave www.igospartners.org/docs/theme_reports/IGOS-

Remote Sensing of Oceans. Chichester, UK: Praxis. Oceans-Final-0101.pdf (accessed Mar. 2008).
Committee on Earth Sciences (1995) Earth Observations Purdom JF and Menzel WP (1996) Evolution of satellite
from Space: History, Promise, and Reality. Washington, observations in the United States and their use in
DC: Space Studies Board, National Research Council. meteorology. In: Fleming JR (ed.) Historical Essays on
Ewing GC (1965) Oceanography from space. Proceedings Meteorology: 1919–1995, pp. 99--156. Boston, MA:
of a Conference held at Woods Hole, 24–28 August American Meteorological Society.
1964. Woods Hole Oceanographic Institution Ref. No. Robinson IS and Guymer T (1996) Observing oceans from
65-10. Woods Hole, MA: WHOI. space. In: Summerhayes CP and Thorpe SA (eds.)
Fu L-L, Liu WT, and Abbott MR (1990) Satellite remote Oceanography: An Illustrated Guide, pp. 69--87.
sensing of the ocean. In: Le Méhauté B (ed.) The Sea, Chichester, UK: Wiley.
Vol. 9: Ocean Engineering Science, pp. 1193--1236. Victorov SV (1996) Regional Satellite Oceanography.
Cambridge, MA: Harvard University Press. London: Taylor and Francis.
Guymer TH, Challenor PG, and Srokosz MA (2001) Wilson WS (ed.) (1981) Special Issue: Oceanography from
Oceanography from space: Past success, future Space. Oceanus 24: 1--76.
challenge. In: Deacon M (ed.) Understanding the Wilson WS, Fellous JF, Kawamura H, and Mitnik L (2006)
Oceans: A Century of Ocean Exploration, pp. 193--211. A history of oceanography from space. In: Gower JFR
London: UCL Press. (ed.) Manual of Remote Sensing, Vol. 6: Remote
JOI Satellite Planning Committee (1984) Oceanography Sensing of the Environment, pp. 1--31. Bethesda, MD:
from Space: A Research Strategy for the Decade, 1985– American Society for Photogrammetry and Remote
1995, parts 1 and 2. Washington, DC: Joint Oceano- Sensing.
graphic Institutions. Wilson WS and Withee GW (2003) A question-based
Kaula WM (ed.) (1969) The terrestrial environment: solid- approach to the implementation of sustained, systematic
earth and ocean physics. Proceedings of a Conference observations for the global ocean and climate, using sea
held at William College, 11–21 August 1969, NASA level as an example. MTS Journal 37: 124--133.
CR-1579. Washington, DC: NASA.
Kawamura H (2000) Era of ocean observations using Relevant Websites
satellites. Sokko-Jiho 67: S1--S9 (in Japanese).
http://www.aviso.oceanobs.com
Koblinsky CJ and Smith NR (eds.) (2001) Observing the
– AVISO.
Oceans in the 21st Century. Melbourne: Global Ocean
http://www.coriolis.eu.org
Data Assimilation Experiment and the Bureau of
– CORIOLIS.
Meteorology.
http://www.eohandbook.com
Masson RA (1991) Satellite Remote Sensing of Polar
– Earth Observation Handbook, CEOS.
Regions. London: Belhaven Press.
http://www.igospartners.org
Minster JF and Lefebvre M (1997) TOPEX/Poseidon
– IGOS.
satellite altimetry and the circulation of the oceans.
http://wo.jcommops.org
In: Minster JF (ed.) La Machine Océan, pp. 111--135
– JCOMMOPS.
(in French). Paris: Flammarion.
http://www.mercator-ocean.fr
Ocean Theme Team (2001) An Ocean Theme for the
– Mercator Ocean.
IGOS Partnership. Washington, DC: NASA. http://
 SATELLITE PASSIVE-MICROWAVE
MEASUREMENTS OF SEA ICE
C. L. Parkinson, NASA Goddard Space Flight Center, covering an area approximately the size of North
Greenbelt, MD, USA America, and it is highly dynamic, experiencing a
Published by Elsevier Ltd. prominent annual cycle in both polar regions and
many short-term fluctuations as it is moved by winds
and waves, melted by solar radiation, and aug-
mented by additional freezing. It is a major player in
and indicator of the polar climate state and has
Introduction multiple impacts on all levels of the polar marine
Satellite passive-microwave measurements of sea ice ecosystems. Consequently it is highly desirable to
have provided global or near-global sea ice data for monitor the sea ice cover on a routine basis. In view
most of the period since the launch of the Nimbus 5 of the vast areal coverage of the ice and the harsh
satellite in December 1972, and have done so with polar conditions, the only feasible means
horizontal resolutions on the order of 25–50 km and of obtaining routine monitoring is through satellite
a frequency of every few days. These data have been observations. Visible, infrared, active-microwave,
used to calculate sea ice concentrations (percent areal and passive-microwave satellite instruments are all
coverages), sea ice extents, the length of the sea ice proving useful for examining the sea ice cover, with
season, sea ice temperatures, and sea ice velocities, the passive-microwave instruments providing the
and to determine the timing of the seasonal onset of longest record of near-complete sea ice monitoring
melt as well as aspects of the ice-type composition of on a daily or near-daily basis.
the sea ice cover. In each case, the calculations
are based on the microwave emission characteristics Theory
of sea ice and the important contrasts between the
microwave emissions of sea ice and those of the The tremendous value of satellite passive-microwave
surrounding liquid-water medium. measurements for sea ice studies results from the
The passive-microwave record is most complete combination of the following four factors:
since the launch of the scanning multichannel 1. Microwave emissions of sea ice differ noticeably
microwave radiometer (SMMR) on the Nimbus 7 from those of seawater, making sea ice generally
satellite in October 1978; and the SMMR data and readily distinguishable from liquid water on the
follow-on data from the special sensor microwave basis of the amount of microwave radiation re-
imagers (SSMIs) on satellites of the United States ceived by the satellite instrument. For example,
Defense Meteorological Satellite Program (DMSP) Figure 1 presents color-coded images of the data
have been used to determine trends in the ice covers from one channel on a satellite passive-microwave
of both polar regions since the late 1970s. The data instrument, presented in units (termed ‘brightness
have revealed statistically significant decreases in temperatures’) indicative of the intensity of emitted
Arctic sea ice coverage and much smaller magnitude microwave radiation at that channel’s frequency,
increases in Antarctic sea ice coverage. 19.4 GHz. The ice edge, highlighted by the bro-
ken white curve, is readily identifiable from the
Background on Satellite Passive- brightness temperatures, with open-ocean values
Microwave Sensing of Sea Ice of 172–198 K outside the ice edge and sea ice
values considerably higher, predominantly greater
Rationale
than 230 K, within the ice edge.
Sea ice is a vital component of the climate of the 2. The microwave radiation received by Earth-
polar regions, insulating the oceans from the at- orbiting satellites derives almost exclusively from
mosphere, reflecting most of the solar radiation in- the Earth system. Hence, microwave sensing does
cident on it, transporting cold, relatively fresh water not require sunlight, and the data can be collected
toward equator, and at times assisting overturning in irrespective of the level of daylight or darkness.
the ocean and even bottom water formation through This is a major advantage in the polar latitudes,
its rejection of salt to the underlying water. Further- where darkness lasts for months at a time, cen-
more, sea ice spreads over vast distances, globally tered on the winter solstice.

524
 SATELLITE PASSIVE-MICROWAVE MEASUREMENTS OF SEA ICE 525

(a) (b)
>282.5 K

270 K

250 K

230 K

210 K

190 K

170 K

150 K

<132.5 K
Figure 1 Late-winter brightness temperature images of 19.4-GHz vertically polarized (19.4 V) data from the DMSP SSMI for (a) the
north polar region on 15 Mar. 1998 and (b) the south polar region on 15 Sep. 1998, showing near-maximum sea ice coverage in each
hemisphere. The broken white curve has been added to indicate the location of the sea ice edge. Black signifies areas of no data; the
absence of data poleward of 87.61 latitude results from the satellite’s near-polar orbit and is consistent throughout the SSMI data set.

3. Many of the microwave data are largely un- corresponding frequency of 19.35 GHz. It collected
affected by atmospheric conditions, including the good-quality data for much of the 4-year period from
presence or absence of clouds. Storm systems January 1973 through December 1976, although
can produce atmospheric interference, but, at with some major data gaps, including one that lasted
selected wavelengths, the microwave signal from for 3 months, from June through August 1975. Being
the ice–ocean surface can pass through most a single-channel instrument, it did not allow some of
nonprecipitating clouds essentially unhindered. the more advanced studies that have been done with
Hence, microwave sensing of the surface does not subsequent instruments, but its flight was a highly
require cloud-free conditions. successful proof-of-concept mission, establishing the
4. Satellite passive-microwave instruments can ob- value of satellite passive-microwave technology for
tain a global picture of the sea ice cover at least observing the global sea ice cover and other vari-
every few days with a resolution of 50 km or ables. The ESMR data were used extensively in the
better, providing reasonable spatial resolution and determination and analysis of sea ice conditions in
extremely good temporal resolution for most both the Arctic and the Antarctic over the 4 years
large-scale or climate-related studies. 1973–76. Emphasis centered on the determination of
ice concentrations (percent areal coverages of ice)
and, based on the ice concentration results, the cal-
Satellite Passive-Microwave culation of ice extents (integrated areas of all grid
elements with ice concentration Z15%). This 4-year
Instruments data set established key aspects of the annual cycles
The first major satellite passive-microwave imager of the polar sea ice covers, including the non-
was the electrically scanning microwave radiometer uniformity of the growth and decay seasons and the
(ESMR) launched on the Nimbus 5 satellite of the marked interannual differences even within a 4-year
United States National Aeronautics and Space Ad- data set.
ministration (NASA) in December 1972, preceded by The Nimbus 5 ESMR was followed by a less
a nonscanning passive-microwave radiometer laun- successful ESMR on the Nimbus 6 satellite and then
ched on the Russian Cosmos satellite in September by the more advanced 10-channel SMMR on board
1968. The ESMR was a single-channel instrument NASA’s Nimbus 7 satellite and a sequence of seven-
recording radiation at a wavelength of 1.55 cm and channel SSMIs on board satellites of the DMSP.
526 SATELLITE PASSIVE-MICROWAVE MEASUREMENTS OF SEA ICE

Nimbus 7 was launched in late October 1978, and With a single channel of microwave data, taken at
the SMMR on board was operational through mid- a radiative frequency and polarization combination
August 1987. The first of the DMSP SSMIs was that provides a clear distinction between ice and
operational as of early July 1987, providing a wel- water, approximate sea ice concentrations can be
come data overlap with the Nimbus 7 SMMR and calculated by assuming a uniform radiative brightness
thereby allowing intercalibration of the SMMR and temperature TBw for water and a uniform radia-
SSMI data sets. SSMIs continue to operate into the tive brightness temperature TBI for ice, with both
twenty-first century. There was also an SMMR on brightness temperatures being appropriate for the
board the short-lived Seasat satellite in 1978; and values received at the altitude of the satellite, that is,
there was a two-channel microwave scanning radio- incorporating an atmospheric contribution. Assuming
meter (MSR) on board the Japanese Marine Obser- no other surface types within the field of view, the
vation Satellites starting in February 1987. Each of observed brightness temperature TB is given by
these successor satellite passive-microwave instru-
ments, after the ESMR, has been multichannel, TB ¼ Cw TBw þ CI TBI ½1
allowing both an improved accuracy in the ice con-
centration derivations and the calculation of add- Cw is the percent areal coverage of water and CI is the
itional sea ice variables, including ice temperature ice concentration. With only the two surface types,
and the concentrations of separate ice types. Cw þ CI ¼ 1, and eqn [1] can be expressed as
The Japanese developed a 12-channel advanced
microwave scanning radiometer (AMSR) for the TB ¼ ð1  CI ÞTBw þ CI TBI ½2
Earth Observing System’s (EOS) Aqua satellite (for-
merly named the PM-1 satellite), launched by NASA This is readily solved for the ice concentration:
in May 2002, and for the Advanced Earth Observing
Satellite II (ADEOS-II), launched by the Japanese TB  TBw
National Space Development Agency in December CI ¼ ½3
TBI  TBw
2002. ADEOS-II prematurely ceased operations in
October 2003, but the Aqua AMSR, labeled AMSR-E Equation [3] is the standard equation used for the
in recognition of its place in the EOS and to dis- calculation of ice concentrations from a single channel
tinguish it from the ADEOS-II AMSR, has collected a of microwave data, such as the data from the ESMR
multiyear data record. The AMSR-E has a major instrument. A major limitation of the formulation is
advantage over the SMMR and SSMI instruments in that the polar ice cover is not uniform in its microwave
allowing sea ice measurements at a higher spatial emission, so that the assumption of a uniform TBI
resolution (12–25 km vs. 25–50 km for the major for all sea ice is only a rough approximation, far less
derived sea ice products). It furthermore has an add- justified than the assumption of a uniform TBw for
itional advantage over the SMMR in having channels seawater, although that also is an approximation.
at 89 GHz in addition to its lower-frequency chan- Multichannel instruments allow more sophisti-
nels, at 6.9, 10.7, 18.7, 23.8, and 36.5 GHz. cated, and generally more accurate, calculation of
the ice concentrations. They additionally allow many
options as to how these calculations can be done. To
Sea Ice Determinations from Satellite illustrate the options, two algorithms will be de-
Passive-Microwave Data scribed, both of which assume two distinct ice types,
thereby advancing over the assumption of a single ice
Sea Ice Concentrations
type made in eqns [1]–[3] and being particularly
Ice concentration is among the most fundamental appropriate for the many instances in which two ice
and important parameters for describing the sea ice types dominate the sea ice cover. For this approxi-
cover. Defined as the percent areal coverage of ice, it mation, the assumption is that the field of view
is directly critical to how effectively the ice cover contains only water and two ice types, type 1 ice and
restricts exchanges between the ocean and the at- type 2 ice (see the section ‘Sea ice types’ for more
mosphere and to how much incoming solar radiation information on ice types), and that the three surface
the ice cover reflects. Ice concentration is calculated types have identifiable brightness temperatures, TBw,
at each ocean grid element, for whichever grid is TBI1, and TBI2, respectively. Labeling the concen-
being used to map or otherwise display the derived trations of the two ice types as CI1 and CI2, re-
satellite products. A map of ice concentrations pre- spectively, the percent coverage of water is
sents the areal distribution of the ice cover, to the 1 – CI1 – CI2, and the integrated observed brightness
resolution of the grid. temperature is
 SATELLITE PASSIVE-MICROWAVE MEASUREMENTS OF SEA ICE 527

TB ¼ ð1  CI1  CI2 ÞTBw þ CI1 TBI1 þ CI2 TBI2 ½4 concentration calculations. This algorithm is com-
plemented by a weather filter that sets to 0 all ice
With two channels of information, as long as ap- concentrations at locations and times with a GR
propriate values for TBw, TBI1, and TBI2 are known value exceeding 0.07. The weather filter eliminates
for each of the two channels, eqn [4] can be used many of the erroneous calculations of sea ice pres-
individually for each channel, yielding two linear ence arising from the influence of storm systems on
equations in the two unknowns CI1 and CI2. These the microwave data.
equations are immediately solvable for CI1 and CI2, For the SSMI data, the same basic NASA Team al-
and the total ice concentration CI is then given by gorithm is used, although 18 V and 18 H in eqns [6]
and [7] are replaced by 19.4 V and 19.4 H, reflecting
CI ¼ CI1 þ CI2 ½5 the placement on the SSMI of channels at a frequency
of 19.4 GHz rather than the 18-GHz channels on the
Although the scheme described in the preceding SMMR. Also, because the data from the 19.4-GHz
paragraph is a marked advance over the use of a channels tend to be more contaminated by water
single-channel calculation (eqn [3]), most algorithms vapor absorption/emission and other weather effects
for sea ice concentrations from multichannel data than the 18-GHz data, the weather filter for the SSMI
make use of additional channels and concepts to calculations incorporates a threshold level for the GR
further improve the ice concentration accuracies. A calculated from the 22.2-GHz vertically polarized
frequently used algorithm (termed the NASA Team data and 19.4-V data as well as a threshold for the GR
algorithm) for the SMMR data employs three of the calculated from the 37 V and 19.4-V data. To illustrate
10 SMMR channels, those recording horizontally the results of this ice concentration algorithm,
polarized radiation at a frequency of 18 GHz and Figure 2 presents the derived sea ice concentrations for
vertically polarized radiation at frequencies of 18 15 March 1998 in the Northern Hemisphere and
and 37 GHz. The algorithm is based on both the for 15 September 1998 in the Southern Hemisphere,
polarization ratio (PR) between the 18-GHz ver- the same dates as used in Figure 1.
tically polarized data (abbreviated 18 V) and the As mentioned, there are several alternative ice
18-GHz horizontally polarized data (18 H) and the concentration algorithms in use. Contrasts from the
spectral gradient ratio (GR) between the 37-GHz NASA Team algorithm just described include: use
vertically polarized data (37 V) and the 18-V data. of different microwave channels; use of regional
PR and GR are defined as: tie points rather than hemispherically applicable tie
points; use of cluster analysis on brightness tem-
TBð18 VÞ  TBð18 HÞ perature data, without PR and GR formulations; use
PR ¼ ½6
TBð18 VÞ þ TBð18 HÞ of iterative techniques whereby an initial ice con-
centration calculation leads to refined atmospheric
TBð37 VÞ  TBð18 VÞ temperatures and opacities, which in turn lead to
GR ¼ ½7
TBð37 VÞ þ TBð18 VÞ refined ice concentrations; use of iterative techni-
ques involving surface temperature, atmospheric
Substituting into eqns [6] and [7] expanded forms of water vapor, cloud liquid water content, and wind
TB(18 V), TB(18 H), and TB(37 V) obtained from speed; incorporation of higher-frequency data to
eqn [4], the result yields equations for PR and GR in reduce the effects of snow cover on the compu-
the two unknowns CI1 and CI2. Solving for CI1 and tations; and use of a Kalman filtering technique in
CI2 yields two algebraically messy but compu- conjunction with an ice model. The various techni-
tationally straightforward equations for CI1 and CI2 ques tend to obtain very close overall distributions
based on PR, GR, and numerical coefficients deter- of where the sea ice is located, although sometimes
mined exclusively from the brightness temperature with noticeable differences (up to 20%, and on
values assigned to water, type 1 ice, and type 2 ice for occasion even higher) in the individual, local ice
each of the three channels (these assigned values are concentrations. The differences can often be mark-
termed ‘tie points’ and are determined empirically). edly reduced by adjustment of tunable parameters,
These are the equations that are then used for the such as the algorithm tie points, in one or both of
calculation of the concentrations of type 1 and type 2 the algorithms being compared. However, the lack
ice once the observations are made and are used to of adequate ground data often makes it difficult to
calculate PR and GR from eqns [6] and [7]. The total know which tuning is most appropriate or which
ice concentration CI is then obtained from eqn [5]. algorithm is yielding the better results. To help re-
The use of PR and GR in this formulation reduces solve the uncertainties, in situ and aircraft meas-
the impact of ice temperature variations on the ice urements are being made in both the Arctic and the
528 SATELLITE PASSIVE-MICROWAVE MEASUREMENTS OF SEA ICE

100%
15 Mar. 1998

15 Sep. 1998
80%

60%

40%

20%

<12%

Figure 2 North and south polar sea ice concentration images for 15 Mar. 1998 and 15 Sep. 1998, respectively. The ice
concentrations are derived from the data of the DMSP SSMI, including the 19.4-V data depicted in Figure 1.

Antarctic to help validate and improve the AMSR-E percentage). Sea ice extents are now routinely cal-
sea ice products. culated for the north polar region as a whole, for the
Whichever ice concentration algorithm is used, the south polar region as a whole, and for each of several
result provides estimates of ice coverage, not ice subregions within the two polar domains, using
thickness. In cases where ice thickness data are also ice concentration maps determined from satellite
available, the combination of ice concentration and passive-microwave data.
ice thickness allows the calculation of ice volume. Ice A major early result from the use of satellite passive-
thickness data, however, are quite limited both spa- microwave data was the detailed determination of
tially and over time. Furthermore, they are generally the seasonal cycle of ice extents in each hemisphere.
not derived from the passive-microwave obser- Incorporating the interannual variability observed
vations and so are not highlighted in this article. The from the 1970s through the early twenty-first
limited ice thickness data traditionally have come century, the Southern Ocean ice extents vary from
from in situ and submarine measurements, although about 2–4  106 km2 in February to about
some recent data are also available from satellite 17–20  106 km2 in September, and the north polar
radar altimetry and, since the January 2003 launch ice extents vary from about 5–8  106 km2 in Sep-
of the Ice, Cloud and land Elevation Satellite (ICE- tember to about 14–16  106 km2 in March. The
Sat), from satellite laser altimetry. The laser techni- exact timing of minimum and maximum ice coverage
que appears promising, although the laser on board and the smoothness of the growth from minimum to
ICESat operates only for short periods, and hence the maximum and the decay from maximum to mini-
full value of the technique will not be realized until a mum vary noticeably among the different years.
new laser is launched with a longer operational As the data sets lengthened, a major goal became
lifetime. Average sea ice thickness in the Antarctic is the determination of trends in the ice extents and the
estimated to be in the range 0.4–1.5 m, and average placement of these trends in the context of other
sea ice thickness in the Arctic is estimated to be in the climate variables and climate change. Because of the
range 1.5–3.5 m. lack of a period of data overlap between the ESMR
and the SMMR, matching of the ice extents derived
Sea Ice Extents and Trends from the ESMR data to those derived from the
SMMR and SSMI data has been difficult and un-
Sea ice extent is defined as the total ocean area of all certain. Consequently, most published results re-
grid cells with sea ice concentration of at least 15% garding trends found from the SMMR and SSMI
(or, occasionally, an alternative prescribed minimum data sets do not include the ESMR data.
 SATELLITE PASSIVE-MICROWAVE MEASUREMENTS OF SEA ICE 529

The SMMR/SSMI record from late 1978 until the of an oscillatory pattern and would soon reverse.
early twenty-first century indicates an overall de- However, although the picture remained complicated
crease in Arctic sea ice extents of about 3% per by interannual variability and regional differences,
decade and an overall increase in Antarctic sea ice the Arctic decreases overall continued (with fluctu-
extents of about 1% per decade. The Arctic ice de- ations) through the 1990s and into the twenty-first
creases have received particular attention because century, and by the middle of the first decade of the
they coincide with a marked warming in the Arctic twenty-first century no large-scale regions of the
and likely are tied closely to that warming. Although Arctic showed overall increases since late 1978.
the satellite data reveal significant interannual vari- Moreover, by the early twenty-first century the Arctic
ability in the ice cover, even as early as the late 1980s sea ice decreases were apparent in all seasons of the
it had become clear from the satellite record that the year, and the decreases in ice extent found from
Arctic as a whole had lost ice since the late 1970s. satellite data were complemented by decreases in ice
The picture was mixed regionally, with some regions thickness found from submarine and in situ data.
having lost ice and others having gained ice; and with The satellite-derived Arctic sea ice decreases are
such a short data record, there was a strong possi- illustrated in Figure 3 with 26-year March and Sep-
bility that the decreases through the 1980s were part tember time series for the Northern Hemisphere sea

20 10

Northern Hemisphere Arctic Ocean

March ice extents
8
•

•
•

•
•

March ice extents

•
•
•

•
•

15
•
•

•
•

•
•
•

•
•

•
•

•
•

•
Sea ice extent (106 km2)

Sea ice extent (106 km2)

•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•

•
6
•

•
•
•
•

•
•
•

•
•
•
•

•
•
•

•
•
10

•
•

•
•

•
•
September ice extents
4
•

•
•

•
•

•
•

•
•
•
•

•
•

•
•
•
•

•
•

•
•

September ice extents

•
•

5
2

0 0
1980 1983 1986 1989 1992 1995 1998 2001 2004 1980 1983 1986 1989 1992 1995 1998 2001 2004

3 3
Seas of Okhotsk and Japan Kara and Barents Seas
Sea ice extent (106 km2)

Sea ice extent (106 km2)

March ice extents

•

2 2
•
•

•
•
•

•
•

•
•
•

March ice extents

•

•
•
•

•
•

•
•
•
•

•
•
•
•

•
•

•
•
•

•
•

•
•
•

•
•
•

•
•

1 1
•
•

•
•

•
•

•
•
•

September ice extents

•
•
•

•
•

•
•
•

•
•
•
•

•
•

September ice extents

•
•

•
•
•

•
•
•
•

0
•

0
1980 1983 1986 1989 1992 1995 1998 2001 2004 1980 1983 1986 1989 1992 1995 1998 2001 2004

Figure 3 Time series of monthly average 1979–2004 March and September sea ice extents for the Northern Hemisphere and the
following three regions within the Northern Hemisphere: the Arctic Ocean, the Seas of Okhotsk and Japan, and the Kara and Barents
Seas. All ice extents are derived from the Nimbus 7 SMMR and DMSP SSMI satellite passive-microwave data. The trend lines are
linear least squares fits through the data points, and the slopes of the trend lines for the Northern Hemisphere total are
 29 50075400 km2 yr  1 (  1.970.35% per decade) for the March values and  51 300710 800 km2 yr  1 ( 6.971.5% per decade)
for the September values. For the Seas of Okhotsk and Japan, all the September ice extents are 0 km2, as the ice cover fully
disappeared from these seas by the end of summer in each of the 26 years.
530 SATELLITE PASSIVE-MICROWAVE MEASUREMENTS OF SEA ICE

ice cover as a whole and for three regions within the regional conditions. In Figure 4, the 26-year Febru-
Northern Hemisphere ice cover, those being the ary (generally the month of Antarctic sea ice min-
Arctic Ocean, the Seas of Okhotsk and Japan, and imum) and September (generally the month of
the Kara and Barents Seas. March and September are Antarctic sea ice maximum) ice extent time series are
typically the months of maximum and minimum plotted for the Southern Hemisphere total, the
Northern Hemisphere sea ice coverage. Among the Weddell Sea, the Bellingshausen and Amundsen Seas,
regional and seasonal differences visible in the plots, and the Ross Sea. The region of the Bellingshausen
the Arctic Ocean shows little or no variation in and Amundsen Seas shows statistically significant
March ice extents because of consistently being fully (99% confidence level) February sea ice decreases
covered with ice in March, to at least 15% ice cov- but shows very slight (not statistically significant)
erage in each grid square, but shows noticeable September sea ice increases. The Weddell Sea instead
fluctuations and overall decreases in September ice shows statistically significant (95% confidence level)
coverage. In contrast, the region of the Seas of increases in February ice coverage and slight (not
Okhotsk and Japan has no variability in September, statistically significant) decreases in September ice
because of having no ice in any of the Septembers, coverage, while the Ross Sea shows ice increases in
but for March shows marked fluctuations and both months, with the February increases being
slight (not statistically significant) overall decreases, statistically significant at the 99% confidence level
interrupted by prominent increases from 1994 to (Figure 4). In line with the mixed pattern of ice ex-
2001. The Kara and Barents Seas exhibit significant tent increases and decreases, the Antarctic has also
variability in both the March and September values, experienced a mixed pattern of temperature in-
although with slight (not statistically significant) creases and decreases, with the Antarctic Peninsula
overall decreases for both months (Figure 3). The (separating the Bellingshausen and Weddell seas)
March and September ice extent decreases for the being the one region of the Antarctic with a prom-
Northern Hemisphere as a whole are both statistic- inent warming signal.
ally significant at the 99% level, as are the September
decreases for the Arctic Ocean region.
Sea Ice Types
The lack of uniformity in the Arctic sea ice losses
(e.g., Figure 3) complicates making projections into The sea ice covers in both polar regions are mix-
the future. Nonetheless, in the early twenty-first cen- tures of various types of ice, ranging from individual
tury, several scientists have offered projections in light ice crystals to coherent ice floes several kilometers
of the expectation of continued warming of across. Common ice types include frazil ice (fine
the climate system. These projections – some based spicules of ice suspended in water); grease ice
on extrapolation from the data, others on computer (a soupy layer of ice that looks like grease from a
modeling – suggest continued Arctic sea ice decreases, distance); slush ice (a viscous mass formed from a
with the effects of warming dominating over oscil- mixture of water and snow); nilas (a thin elastic sheet
latory behavior and other fluctuations. Some studies of ice 0.01–0.1-m thick); pancake ice (small, roughly
project a totally ice-free late summer Arctic Ocean by circular pieces of ice, 0.3–3 m across and up to 0.1-m
the end of the century, the middle of the century, or thick); first-year ice (ice at least 0.3-m thick that has
even, in one projection, by as early as the year 2013. not yet undergone a summer melt period); and
An ice-free Arctic would greatly ease shipping but multiyear ice (ice that has survived a summer melt).
would also have multiple effects on the Arctic climate Because different ice types have different microwave
(by seasonally eliminating the highly reflective and emission characteristics, once these differences are
insulating ice cover) and on Arctic ecosystems (by understood, appropriate satellite passive-microwave
seasonally removing both the habitat for the organ- data can be used to distinguish ice types. The ice types
isms that live within the ice and the platform on which most frequently distinguished with such data are first-
a variety of polar animals depend). year ice and multiyear ice in the Arctic Ocean.
In contrast to the Arctic sea ice, the Antarctic sea In fact, the NASA Team algorithm described earlier
ice cover as a whole does not show a warming signal was initially developed specifically for first-year and
over the period from the start of the multichannel multiyear ice, with the resulting calculations yielding
satellite passive-microwave record, in late 1978, the concentrations, CI1 and CI2, of those two ice types.
through the end of 2004. Over this period, some First-year and multiyear ice are distinguishable in their
regions in the Antarctic experienced overall ice cover microwave signals because the summer melt process
increases and other regions experienced overall ice drains some of the salt content downward through the
cover decreases, with the hemispheric 1% per decade ice, reducing the salinity of the upper layers of the ice
ice extent increases incorporating contrasting and thereby changing the microwave emissions; these
 SATELLITE PASSIVE-MICROWAVE MEASUREMENTS OF SEA ICE 531

25 10

Southern Hemisphere Weddell Sea

20 8

•
•
•
•
•

•
•
•
•

•
•

•
•

•
•
•

•
•
•

•
•
•
•

•
Sea ice extent (106 km2)

Sea ice extent (106 km2)

•
•

•
•

•
•
•
•

•
•

•
•

•
•

•
•
•
•
15

•
6

•
September ice extents

10 4

5 2
February ice extents

•
February ice extents

•
•
•
•
•

•
•

•
•
•

•
•

•
•
•

•
•
•

•
•
•

•
•
•
•

•
•
•
•
•

•
•

•
•

•
•
•

•
•
•
•
•

•
•

•
0 0
1980 1983 1986 1989 1992 1995 1998 2001 2004 1980 1983 1986 1989 1992 1995 1998 2001 2004

4 6
Bellingshausen and Amundsen Seas Ross Sea
Sea ice extent (106 km2)
Sea ice extent (106 km2)

•
•
•
September ice extents

•
•
•
•

•
4

•
•
•

•
•

•
•

•
•

•
•

•
•

•
•
•
•
•

•
•

•
•
•

•
•

September ice extents

•

•
•
•

•
•

•
•

2
•
•

•
•

2
1 February ice extents
•
•

•
•

•
•

•
•

•
•

•
•

•
•
•

•
•
•

•
•
•

•
•
•

•
•

•
•
•

•
•

February ice extents

•
•

•
•

•
•
•

•
•

•
•

•
•

0
•
•

0
1980 1983 1986 1989 1992 1995 1998 2001 2004 1980 1983 1986 1989 1992 1995 1998 2001 2004

Figure 4 Time series of monthly average 1979–2004 February and September sea ice extents for the Southern Hemisphere and the
following three regions within the Southern Hemisphere: the Weddell Sea, the Bellingshausen and Amundsen Seas, and the Ross
Sea. All ice extents are derived from the Nimbus 7 SMMR and DMSP SSMI satellite passive-microwave data. The trend lines are
linear least squares fits through the data points, and the slopes of the trend lines for the Southern Hemisphere total are
14 10078000 km2 yr  1 (5.072.8% per decade) for the February values and 700078200 km2 yr  1 (0.470.5% per decade) for the
September values.

changes are dependent on the frequency and polar- water than for either ice type. Furthermore, the GR
ization of the radiation. To illustrate the differences, (eqn [7], revised for 19.4-GHz data) is positive for
Figure 5 presents a plot of the tie points employed in water, slightly negative for first-year ice, and con-
the NASA Team algorithm for the Arctic ice. Tie siderably more negative for multiyear ice (Figure 5).
points were determined empirically and are included The differences allow the sorting out, either through
for each of the three SSMI channels used in the the calculation of CI1 and CI2 as described earlier
calculation of ice concentrations prior to the appli- or through alternative algorithms, of the first-year
cation of the weather filter. The plot shows that ice and multiyear ice percentages in the satellite field
while the transition from first-year to multiyear ice of view.
lowers the brightness temperatures for each of the
three channels, the reduction is greatest for the 37-V
Other Sea Ice Variables: Season Length,
data and least for the 19.4-V data. The plot further
Temperature, Melt, Velocity
reveals that the polarization PR (eqn [6], revised
for 19.4 GHz rather than 18-GHz data) is larger for Although ice concentrations, ice extents, and, to a
multiyear ice than for first-year ice, and larger for lesser degree, ice types have been the sea ice variables
532 SATELLITE PASSIVE-MICROWAVE MEASUREMENTS OF SEA ICE

300 atmosphere and the presence of multiple ice types.

The derived temperature is not the surface tem-
250 • FY perature but the temperature of the radiating por-
• FY
Brightness temperature (K)

• FY
• MY tion of the ice, for whichever radiative frequency
200 • MY •W being used. The ice temperature fields derived from
•W • MY passive-microwave data complement those derived
150 from satellite infrared data, which have the advan-
tages of generally having finer spatial resolution and
•W
100 of more nearly approaching surface temperatures
but have the disadvantage of greater contamination
by clouds. The passive-microwave and infrared data
50
are occasionally used together, iteratively, for an
0
enhanced ice temperature product.
19.4 H 19.4 V 37 V The seasonal melting of portions of the sea ice and
Data type its overlying snow cover generally produces marked
changes in microwave emissions, first increasing the
Figure 5 Typical brightness temperatures of first-year ice (FY),
multiyear ice (MY), and liquid water (W) at three channels of
emissions as liquid water emerges in the snow, then
SSMI data from the DMSP F13 satellite, specifically those for decreasing the emissions once the snow has melted
19.4-GHz horizontally polarized data (19.4 H), 19.4-GHz ver- and meltwater ponds cover the ice. Because of the
tically polarized data (19.4 V), and 37-GHz vertically polarized emission changes, these events on the ice surface
data (37 V). These are the values used as tie points for the Arctic frequently become detectable through time series of
calculations in the NASA Team algorithm described in the text.
Data from Cavalieri DJ, Parkinson CL, Gloersen P, Comiso JC,
the satellite passive-microwave data. The onset of
and Zwally HJ (1999) Deriving long-term time series of sea ice melt in particular can often be identified, and hence
cover from satellite passive-microwave multisensor data sets. yearly maps can be created of the dates of melt onset.
Journal of Geophysical Research 104(C7): 15803–15814. Melt ponds, however, present greater difficulties, as
they can have similar microwave emissions to
those of the water in open spaces between ice floes,
so that a field of heavily melt-ponded ice can easily
most widely calculated and used from satellite be confused in the microwave data with a field of
passive-microwave data, several additional variables low-concentration ice. The ambiguities can be re-
have also been obtained from these data, including duced through analysis of the passive-microwave
the length of the sea ice season, sea ice temperature, time series and comparisons with active-microwave,
sea ice melt, and sea ice velocity. The length of the visible, and infrared data. Still, because of these
sea ice season for any particular year is calculated complications under melt conditions, the passive-
directly from that year’s daily maps of sea ice con- microwave-derived ice concentrations tend to have
centrations, by counting, at each grid element, the larger uncertainties for summertime ice than for
number of days with ice coverage of at least some wintertime ice.
predetermined (generally 15% or 30%) ice concen- The calculation of sea ice velocities from satellite
tration. Trends in the length of the sea ice season data has in general relied upon data with fine enough
from the late 1970s to the early twenty-first century resolution to distinguish individual medium-sized
show coherent spatial patterns in both hemispheres, ice floes, such as visible data and active-microwave
with a predominance of negative values (shortening data rather than the much coarser resolution passive-
of the sea ice season) in the Northern Hemisphere microwave data. However, in the 1990s, several
and in the vicinity of the Antarctic Peninsula and a groups devised methods of determining ice velocity
much lesser predominance of positive values in the fields from passive-microwave data, some using
rest of the Southern Hemisphere’s sea ice region, techniques based on cross-correlation of brightness
consistent with the respective hemispheric trends in temperature fields and others using wavelet analysis.
sea ice extents. These techniques have yielded ice velocity maps
The passive-microwave-based ice temperature on individual dates for the entire north and south
calculations generally depend on the calculated sea polar sea ice covers. Comparisons with buoy and
ice concentrations, empirically determined ice emis- other data have been quite encouraging regarding
sivities, a weighting of the water and ice tempera- the potential of using the passive-microwave satellite
tures within the field of view, and varying levels of data for long-term records and monitoring of ice
sophistication in incorporating effects of the polar motions.
 SATELLITE PASSIVE-MICROWAVE MEASUREMENTS OF SEA ICE 533

Looking toward the Future Further Reading

Monitoring of the polar sea ice covers through sat- Barry RG, Maslanik J, Steffen K, et al. (1993) Advances in
ellite passive-microwave technology is ongoing with sea-ice research based on remotely sensed passive
the operational SSMI instruments on the DMSP microwave data. Oceanography 6(1): 4--12.
satellites and the Japanese AMSR-E instrument on Carsey FD (ed.) (1992) Microwave Remote Sensing
of Sea Ice. Washington, DC: American Geophysical Union.
NASA’s Aqua satellite. Both Japan and the United
Cavalieri DJ, Parkinson CL, Gloersen P, Comiso JC, and
States anticipate launching additional passive-
Zwally HJ (1999) Deriving long-term time series of sea
microwave instruments to maintain an uninterrupted ice cover from satellite passive-microwave multisensor
satellite passive-microwave data record. The re- data sets. Journal of Geophysical Research 104(C7):
sulting lengthening sea ice records should conti- 15803--15814.
nue to provide an improved basis with which Comiso JC, Yang J, Honjo S, and Krishfield RA (2003)
scientists can examine trends in the sea ice cover and Detection of change in the Arctic using satellite and in
interactions between the sea ice and other ele- situ data. Journal of Geophysical Research 108(C12):
ments of the climate system. For instance, the 3384 (doi:10.1029/2002JC001347).
lengthened records will be essential to answering Gloersen P, Campbell WJ, Cavalieri DJ, et al. (1992)
many of the questions raised concerning whether Arctic and Antarctic Sea Ice, 1978–1987: Satellite Passive-
Microwave Observations and Analysis. Washington, DC:
the negative overall trends found in Arctic sea ice
National Aeronautics and Space Administration.
extents for the first quarter century of the SMMR/
Jeffries MO (ed.) (1998) Antarctic Sea Ice: Physical
SSMI record will continue and how these trends re- Processes, Interactions and Variability. Washington,
late to temperature trends, in particular to climate DC: American Geophysical Union.
warming, and to oscillations within the climate Johannessen OM, Bengtsson L, Miles MW, et al. (2004)
system, in particular the North Atlantic Oscillation, Arctic climate change: Observed and modelled
the Arctic Oscillation, and the Southern Oscillation. temperature and sea-ice variability. Tellus 56A(4):
In addition to covering a longer period, other 328--341.
expected improvements in the satellite passive- Kramer HJ (2002) Observation of the Earth and Its
microwave record of sea ice include further algo- Environment, 4th edn. Berlin: Springer.
rithm developments, following additional analyses Lubin D and Massom R (2006) Polar Remote Sensing, Vol.
1: Atmosphere and Oceans. Berlin: Springer-Praxis.
of the microwave properties of sea ice and liquid
Parkinson CL (1997) Earth from Above: Using Color-
water. Such analyses are likely to lead both to
Coded Satellite Images to Examine the Global
improved algorithms for the variables already Environment. Sausalito, CA: University Science Books.
examined and to the development of techniques for Parkinson CL (2004) Southern Ocean sea ice and its wider
calculating additional sea ice variables from the linkages: Insights revealed from models and
satellite data. observations. Antarctic Science 16(4): 387--400.
Smith WO, Jr. and Grebmeier JM (eds.) (1995) Arctic
Oceanography: Marginal Ice Zones and Continental
Glossary Shelves. Washington, DC: American Geophysical
brightness temperature Unit used to express the Union.
intensity of emitted microwave radiation received Thomas DN and Dieckmann GS (eds.) (2003) Sea Ice: An
Introduction to Its Physics, Chemistry, Biology and
by the satellite, presented in temperature units (K)
Geology. Oxford, UK: Blackwell Science.
following the Rayleigh–Jeans approximation to
Ulaby FT, Moore RK, and Fung AK (eds.) (1986) Monitoring
Planck’s law, whereby the radiation emitted from a sea ice. In: Microwave Remote Sensing: Active and Passive,
perfect emitter at microwave wavelengths is Vol. III: From Theory to Applications, pp. 1478–1521.
proportional to the emitter’s physical temperature. Dedham, MA: Artech House.
sea ice concentration Percent areal coverage of sea Walsh JE, Anisimov O, Hagen JOM, et al. (2005) Cryosphere
ice. and hydrology. In: Symon C, Arris L, and Heal B (eds.)
sea ice extent Integrated area of all grid elements Arctic Climate Impact Assessment, pp. 183--242.
with sea ice concentration of at least 15%. Cambridge, UK: Cambridge University Press.

Relevant Websites
See also http://www.awi-bremerhaven.de
– Alfred Wegener Institute for Polar and Marine
Acoustics, Arctic. Satellite Oceanography, History
Research.
and Introductory Concepts. Satellite Remote
Sensing SAR.
534 SATELLITE PASSIVE-MICROWAVE MEASUREMENTS OF SEA ICE

http://www.antarctica.ac.uk http://www.spri.cam.ac.uk
– Antarctica, British Antarctic Survey. – Scott Polar Research Institute.
http://www.arctic.noaa.gov http://www.nasa.gov
– Arctic Change, NOAA Arctic Theme Page. – US National Aeronautics and Space Administration.
http://www.aad.gov.au http://www.natice.noaa.gov
– Australian Antarctic Division. – US National Ice Center.
http://www.dcrs.dtu.dk http://nsidc.org
– Danish Center for Remote Sensing. – US National Snow and Ice Data Center.
http://www.jaxa.jp
– Japan Aerospace Exploration Agency.
 SATELLITE REMOTE SENSING MICROWAVE
SCATTEROMETERS
W. J. Plant, Applied Physics Laboratory, University of beams across the patch, a process that requires as
Washington, Seattle, WA, USA much as 4 min. Thus an additional assumption of
Copyright & 2001 Elsevier Ltd. scatterometry is that average winds over kilometer-
scale patches of ocean surface are stationary for
several minutes.
With these assumptions and an adequate definition
of the wind being measured (discussed below), sat-
Introduction ellite scatterometers have proven to be able to
measure winds over the ocean with accuracies as
Microwave scatterometers are instruments that
good as or better than in situ measurement techni-
transmit low-power pulses of radiation toward the
ques. Because oceans cover most of the Earth, this
Earth’s surface at intermediate incidence angles and
means that microwave scatterometers carried on
measure the intensity of the signals scattered back at
satellites can monitor the wind field over most of the
the same angles from surface areas a few kilometers
globe every few days. Table 1 gives typical specifi-
on a side. Satellite scatterometers operate continu-
cations that a satellite scatterometer can be expected
ously and therefore scatter from land and ice as well
to meet.
as the ocean. Useful information is available in the
The spatial coverage offered by satellite scattero-
signals from land and ice, but will not be discussed
metry is far better than can be achieved by in situ
here. This article will concentrate on the primary goal
measurements. It allows scatterometers to provide
of satellite scatterometers: the measurement of near-
data to study global weather patterns, monitor storm
surface wind speed and direction over the ocean.
intensities, improve meteorological forecasts, impact
Scatterometers achieve this goal by measuring the
global ocean circulation models, facilitate climate
intensity, or cross-section, of microwave signals
prediction, and much more. In addition to intro-
backscattered from the ocean surface. Common fre-
ducing the basics of scatterometry, this article will
quencies of the transmitted signals for satellite scat-
provide examples of these benefits of satellite scat-
terometers are near 5.3 GHz (C-band) on European
terometry and indicate how continued improvements
instruments and 14 GHz (Ku-band) on US ones. At
in the technique may be expected to provide even
these frequencies, microwaves penetrate only a few
better results in the future.
millimeters into sea water, so all backscatter origin-
ates at the surface and is caused by the roughness of
the surface; a perfectly calm sea surface produces no
detectable scattering in the direction of the incident Satellite Scatterometers
radiation. Changes in the average roughness of the Other instruments such as radar altimeters that look
ocean surface over scales of several kilometers are straight down and real and synthetic aperture radars
caused primarily, but not exclusively, by changes in (RARs and SARs) that image surface scenes at
the wind speed or direction at the ocean surface. resolutions of meters to kilometers have been oper-
Standard assumptions of scatterometry are that the ated in space and are capable of measuring wind
backscatter cross-section, usually called s0 , over such
scales depends only on parameters of the scatte-
rometer and on the mean wind, increases with wind Table 1 Expected specifications of the NSCAT satellite
scatterometer
speed, is a maximum when the antenna looks up-
wind, and is a minimum when the antenna looks Parameter Value Accuracy comment
nearly perpendicular to the wind, or crosswind.
These assumptions allow the wind speed and dir- Wind speed 3–30 m s1 2 m s1 or 10%
ection to be determined from cross-sections meas- Wind direction 3–30 m s1 201
ured for the same patch of the ocean, but with the Spatial resolution 50 km Wind cells
Location accuracy 25/10 km Absolute/relative
antenna directed at several different azimuth angles. Coverage 90% of ice-free ocean Every 2 d
For satellite scatterometry a given patch of ocean can Mission duration 3y Includes check out
be viewed from several different directions only by
allowing the scatterometer to sweep its antenna (Data from Naderi et al. 1991.)

535
536 SATELLITE REMOTE SENSING MICROWAVE SCATTEROMETERS

speed or direction, but not both simultaneously and Sensing Satellite (ERS-1) in 1991, a continuous series
routinely. Microwave radiometers, passive instru- of global wind vector measurements was begun. As
ments that measure the naturally occurring radiation Table 2 shows, this series has continued to the pre-
from the ocean surface, are presently being de- sent through the launch of ERS-2 and the subsequent
veloped as spaceborne anemometers capable of decommissioning of ERS-1. If present plans are car-
measuring wind speed and direction simultaneously. ried out, at least two microwave scatterometers (and
However, only microwave scatterometers, active in- one or more microwave radiometers) will continue
struments that both transmit and receive radiation, to produce global wind vector information into the
have a history of wind vector measurement from foreseeable future.
space. Because they are microwave, scatterometers While most scatterometers in space have consisted
can make their measurements both day and night of multiple, fixed waveguide (stick) antennas, more
and in most kinds of weather. Only very heavy recent US scatterometers have used rotating para-
rainfall, as discussed below, can hinder a scatter- bolic antennas. All satellite scatterometers to date
ometer’s view of the surface. have been in orbits with inclinations near 98.51,
For these reasons, microwave scatterometers have except for Seasat which was 1081. Spatial resolutions
been the instruments of choice for measuring near- have usually been 50  50 km, although 25  25 km
surface winds from space. Table 2 lists the scatte- is becoming more common. Table 3 addresses the
rometers that have been in space to date and those primary characteristics of the scatterometers listed in
that are planned for the future. As the table shows, Table 2. As Table 3 shows, ERS scatterometers have
the first scatterometer in space specifically designed had about half the spatial coverage of their NASA
to measure winds was the one on Seasat in 1978. An counterparts because their antennas view only one
earlier microwave radar on Skylab viewed the ocean side of the subsatellite path. As discussed below,
surface at intermediate incidence angles from space however, uniformly reliable wind vectors are not
in 1973, but did not produce multiple looks at a obtained over the whole swath of the SeaWinds in-
single ocean patch from which wind vector infor- struments. Also a gap is present between the two
mation could be obtained. With the launch of the swaths on either side of the subsatellite path of the
C-band scatterometer on the first European Remote NASA stick-type scatterometers because the response

Table 2 Scatterometers in space to date and planned for the future

Satellite Country/agency Scatterometer Launch date Status

Seasat USA/NASA SASS June 1978 Failed, October 1978

ERS-1 Europe/ESA CSCAT(AMI) July 1991 Standby, June 1996
ERS-2 Europe/ESA CSCAT(AMI) April 1995 Operational
ADEOS-I USA/NASA NSCAT August 1996 ADEOS failed
Japan/NASDA June 1997
QuikSCAT USA/NASA SeaWinds-1 June 1999 Operational
ADEOS-II USA/NASA Japan/NASDA SeaWinds-2 November 2001 Approved
ASCAT Europe/ESA Adv.CSCAT 2003 Proposed

(Adapted from Patzert and Van Woert, private communication.)

Table 3 Principal characteristics of satellite scatterometers

Satellite Type Frequency (GHz) Polarization Incidence angle (1) Swath (km) Altitude

Seasat 4 stick 14.6 4 VV, 4 HH 25–55 475, 475 800

0–4 140
ERS-1 3 stick 5.3 3 VV 18–57 500 785
ERS-2 3 stick 5.3 3 VV 18–57 500 785
ADEOS-I 6 stick 14.0 6 VV, 2 HH 15–63 600, 600 797
QuikSCAT 1 rotating 13.4 1 VV, 1 HH 46(H), 54(V) 1800 803
ADEOS-II 1 rotating 13.4 1 VV, 1 HH 46(H), 54(V) 1800 803
METOP 3 stick – – – – 835
 SATELLITE REMOTE SENSING MICROWAVE SCATTEROMETERS 537

of the cross-section to the wind vector is weak at low The Normalized Radar Cross-section
incidence angles. The Seasat scatterometer did have a of the Sea
mode that observed the surface at very low incidence
angles, but could not get wind direction in this swath The basis of a scatterometer’s ability to measure the
without extrapolation from the wider swaths. near-surface wind vector is the dependence of the
Figure 1 shows the fixed-stick ERS-1 scatterometer normalized radar cross-section of the sea, so , on this
(Figure 1A) and the rotating-antenna QuickSCAT vector. This cross-section is defined through the radar
scatterometer (Figure 1B). equation as follows:

Active microwave instrument (AMI)

Microwave sounder (MS)

Along track scanning
Radar altimeter (RA) radiometer (ATSR)
Global ozone monitoring
Laser retro-reflector experiment (GOME)
(LRR)

Precise range and range

rate experiment (PRARE)

Instrument data handling and

transmission (IDHT)

Solar array

(A)

Figure 1 (A) ERS-1 satellite with two scatterometer waveguide antennas directed upwards. (B) QuikSCAT satellite with parabolic
antenna with dual beams pointed downwards.
538 SATELLITE REMOTE SENSING MICROWAVE SCATTEROMETERS

Figure 1 continued

Pt G2 l2 Aso parameters and depends on l only because the

Pr ¼ þ Pn ½1 mechanism of backscatter from the sea surface is
ð4pÞ3 R4
weakly dependent on l. Two other system par-
where Pr is the received power, Pt is the transmitted ameters upon which so depends due to the surface
power, G is antenna gain, l is microwave length, A is scattering mechanism are the incidence angle and the
the total illuminated area on the sea surface, and R is polarization. The strength of backscatter from the
the range to the surface. Pn is the noise signal due to ocean surface increases with decreasing incidence
thermal noise in the components of the system and angle and depends on the direction of the electric
natural radiation of the earth to the receiving an- field of the incident radiation, its polarization.
tenna. In a real scatterometer, the received power is The radar equation must be solved in order to
reduced below the above level by system losses that obtain so . This requires a knowledge of Pn , which is
must be taken into account. Also, for greater accur- usually obtained by sampling only noise signals on a
acy, the variation of G and R over the scatterometer small fraction of the transmitted pulses, typically
footprint are taken into account by integration over about 15%. The noise level obtained in this manner
the footprint. By relating so rather than Pr to the is subtracted from Pr and the difference is multiplied
wind vector, the importance of system-specific par- or divided by the other quantities which are known.
ameters, Pt , G, R, and l, are greatly reduced. In fact, This yields so values that are accurate down to very
so is independent of the first three of these low signal-to-noise ratios, but that can become
 SATELLITE REMOTE SENSING MICROWAVE SCATTEROMETERS 539

Upwind V-POL
Upwind H-POL
10 10

0 0

_ 10 _ 10

, dB

, dB
0

0
_ 20 _ 20
Symbol  (˚) Symbol  (˚)
0 0
10 10
_ 30 20 _ 30 20
30 30
40 40
50 50

_ 40 _ 40
1 10 100 1 10 100
_ _1
(A) Wind speed (m s 1) (B) Wind speed (m s )

_6 _6
_1
m sec
_1
_ 10 15.0 _ 10 m sec

15.0

_ 14 _ 14

VV, dB

HH, dB

6.5
0

_ 18
0

_ 18 6.5

3.0
_ 22 _ 22 3.0
Upwind Crosswind Downwind Upwind Crosswind Downwind

_ 26 _ 26
0 60 120 180 240 300 360 0 60 120 180 240 300 360
Azimuth relative to wind vector (˚) Azimuth relative to wind vector (˚)
(C) (D)

Figure 2 (A) Normalized radar cross section, s0 versus wind speed for VV polarization and an upwind look direction (y is incidence
angle); (B) same as (A) for HH polarization; (C) azimuthal dependence of s0 for VV polarization and a 301 incidence angle; (D) same
as (C) for HH polarization. (Reproduced with permission from Jones WL et al. Aircraft measurements of the microwave scattering
signature of the ocean. IEEE Journal of Oceanic Engineering ^ 1977 IEEE.)

negative due to sampling variability. This is an im- pointing and the direction from which the wind
portant consideration in the measurement of very comes, has generally been found to fit a three-term
low wind speeds. Fourier cosine series in w very well. The relationship
Figure 2 shows the measured dependance of so on between so and incidence angle, yi , polarization, p,
polarization and incidence angle as well as on the wind speed, U, and wind direction, w, is called the
wind vector. Two polarizations are indicated in the geophysical model function. Its general form is taken
figure: the electric field vertical on both transmission to be the following:
and reception (VV) and the electric field horizontal
on both (HH). These are the two polarizations of so ¼ Ao ðU; yi ; pÞ þ A1 ðU; yi ; pÞcosw
importance in scatterometry. The characteristics of ½2
þA2 ðU; yi ; pÞcos2w
so that make it useful for wind vector measurement
are obvious from this figure: for any given incidence Usually the A coefficients are specified in tabular
angle and polarization, it depends on both wind form.
speed and direction. In general the slope of so versus Some characteristics of the dependence of so on w
wind speed is smaller at lower incidence angles. The can be easily discerned in Figure 2. Cross-sections
dependence of so on azimuth angle, w, defined as the measured with the antenna looking nearly per-
angle between the horizontal direction the antenna is pendicular to the wind direction (w ¼ 901 or 2701)
540 SATELLITE REMOTE SENSING MICROWAVE SCATTEROMETERS

1500 1500

1000 1000

500 500
Along track distance (km)

Along track distance (km)

0 0

_ 500 _ 500

_ 1000 _ 1000

_ 1500 _ 1500
_ 1500 _ 1000 _ 500 0 500 1000 1500 _ 1500 _ 1000 _ 500 0 500 1000 1500
Cross track distance (km) (B) Cross track distance (km)
(A)

Figure 3 Lines on which scatterometer surface footprints lie showing crossing where multiple look directions occur. Along track
distance expanded by a factor of 5. (A) ERS-1/2, (B) SeaWinds.

are always lower than those measured with the an- Unfortunately this simple technique will not work
tenna directed upwind (w ¼ 01) or downwind from satellites because of their high speeds. By the
(w ¼ 1801). Furthermore, so is larger when the an- time the antenna rotates one revolution, the satellite
tenna looks upwind than when it looks downwind, has travelled several kilometers, so the footprint on
except perhaps for VV polarization at small inci- the surface samples many different, widely separated
dence angles. areas. The solution to this problem is to use so values
measured in different directions at different times so
that nearly the same surface area is illuminated in
Spaceborne Scatterometer Wind each direction. Figure 3 indicates how this might be
accomplished in the case of three, fixed stick an-
Measurements tennas (Figure 3A) and in the case of a rotating an-
Given the behavior of so shown in Figure 2, a scat- tenna with two beams (Figure 3B). The first case
terometer fixed on Earth can easily yield the wind corresponds to the ERS scatterometers while the
vector. If the antenna is rotated and the direction of latter corresponds to SeaWinds whose single rotating
maximum so is determined, then the level of so in antenna transmits beams at two different incidence
this direction yields the wind speed. In most cases, angles. Figure 3 shows the lines on which the surface
the direction of maximum so would be the wind resolution cells are located. For clarity, the vertical
direction. The exception to this might be at VV po- distance traveled by the satellite has been expanded
larization for some incidence angles where the by a factor of 5 in Figure 3. Thus crossings of the
downwind look direction could yield the maximum lines, where cross-sections can be measured from
so . Thus a 1801 ambiguity might exist in the wind different directions, are more frequent than indicated
direction. in the figure. The two parts of Figure 3 have the same
 SATELLITE REMOTE SENSING MICROWAVE SCATTEROMETERS 541

16

260°

180° 345°
14
80°

12
1000*
o (linear)

10

0 50 100 150 200 250 300 350

Scatterometer look direction (°N)

Figure 4 Angular dependence of s0 for various wind directions (indicated in the figure). Wind speeds vary little. Circles show
possible data points; curves crossing at data points produce ambiguities in wind direction.

horizontal scale to emphasize the much wider swath like the Seasat scatterometer, then the other two
of the SeaWinds scatterometer compared with ERS. wind directions, 801 and 2601 would also fit the
For regions of the swath near the center, however, the data, provided that the wind speeds were slightly
azimuth angles at which the SeaWinds beam views higher for the 2601 direction and slightly lower for
the surface are nearly opposite or parallel. Further- the 801 direction. Ambiguous wind vectors always
more, near the edges, only two nearly parallel looks exist in the output of scatterometers. Attempts to
at a given surface area can be obtained. Estimation of resolve these ambiguities have ranged from using
wind speed and direction is consequently more dif- human analysts to insure consistency of the wind
ficult near the center and edge of the swath due to the fields to using median filters in which a given wind
ambiguous nature of the model function. vector is forced to be in the direction of the median
The ambiguities inherent in the form of the model of its surrounding vectors. While automatic techni-
function are illustrated in Figure 4. The figure shows ques such as median filtering are now producing
the angular dependence of so for winds coming from skills upwards of 90% in selecting the correct am-
801, 1801, 2601, and 3451 with respect to north. biguity, incorrect assignment of vector directions in
Each curve maximizes when the antenna is directed some locations is still a problem for scatterometry.
into the wind direction. The circles in the figure Retrieval of the ambiguous wind vectors is ac-
represent measurements of the cross-section made by complished in practice by noting that any single
an ERS-type scatterometer where the three beams are measurement of so is an average of measurements
451 apart. As the figure shows, winds from 1801 and from several scatterometer pulses and therefore has a
3451 fit the measurements equally well. If the scat- probability distribution near Gaussian. The variance,
terometer had only two beams that were 901 apart, d2 , of this distribution may be determined from the
542 SATELLITE REMOTE SENSING MICROWAVE SCATTEROMETERS

characteristics of the scatterometer, especially its the friction velocity, the square root of the stress
noise level, along with so . Thus the probability of divided by the density of air. However, to date this
measuring a given so value, call it ŝo , is given by procedure has not been adopted by any agency op-
0 1 erating a scatterometer. Rather, official winds pro-

duced by scatterometers are the winds measured at
1 ŝo s2o =2d2
P@ŝo A ¼ pffiffiffiffiffiffi 2 e ½3 10 m above the ocean surface that would yield the
2pd same surface stress under neutrally buoyant atmos-
pheric stratification. The only exception was the
where so is the true normalized cross-section. Since Seasat scatterometer, whose output winds were spe-
this is unknown in practice, the value of so given by cified at a height of 19.5 m above the surface; these
the geophysical model function is used instead. are approximately 6% higher than winds at 10 m.
If all the measured values of ŝo from every view of Thus wind fields produced by scatterometers are not
the surface that is within the specified resolution of necessarily the wind fields that a fixed array of in situ
the system (for the NASA scatterometer (NSCAT) anemometers would measure, even if they were all at
with a 50  50 km resolution this could be up to 24) a height of 10 m. Not only are scatterometer wind
then the joint probability of measuring these values is fields the neutrally buoyant equivalent winds, but
the product of many Gaussian distributions such as evidence is growing that these wind fields are the
eqn [3]. The wind speed and direction, and therefore ones that would be measured under neutral con-
so , are then varied to maximize this joint probability. ditions by an anemometer drifting along with the
This is the same as minimizing the following ob- ambient ocean current.
jective function: Although physical models of backscatter from the
2 wind-roughened ocean have been developed, they
!2 3
X
N ŝoi  soi ðU; wÞ have not proven to be sufficiently accurate to be used
JðU; wÞ ¼ 4ln d2 þ 5 ½4 as geophysical model functions. Therefore wind re-
i
di
i¼1 trieval from scatterometers flown to date has been
achieved using empirical model functions developed
Good knowledge of both the geophysical model by various means. Experiments have been mounted
function and the variances associated with the dif- to relate cross-sections measured by airborne scat-
ferent looks is essential for a good retrieval. terometers to in situ measurements of winds con-
verted to neutral conditions. The Europeans noted
that if so depends on no other environmental vari-
Calibration ables than U and w, then so values measured by three
In principle, so values obtained by scatterometers antennas looking in three different directions must
need not be calibrated in an absolute sense; they fall on a well-defined surface when so values from
could simply be related to the surface wind that the three separate antennas are plotted on three
produced them. In order to cross-check the operation orthogonal axes. This observation has allowed them
of different scatterometers and to further the devel- to determine the properties of the geophysical model
opment of rough surface scattering theories, how- function to within a constant calibration factor,
ever, all scatterometers flown in space have produced which was obtained from comparisons with buoy
normalized radar cross-sections that have been cali- measurements. US geophysical model functions have
brated against a standard target. Usually signals from also been developed by comparison of cross-sections
a geophysical target that has been calibrated by air- with buoy measurements and with winds measured
borne measurements and found to be isotropic are by other remote sensing instruments, such as
used for calibration of so from spaceborne scatte- microwave radiometers. Most recently, however, US
rometers. The Amazon Rain Forest is the most geophysical model functions have been developed by
common geophysical target chosen, although areas binning satellite scatterometer so values according to
of Antarctica have also been used. U and w values produced by numerical weather pre-
Calibration and verification of scatterometer diction models. The idea is that errors in the nu-
winds require an exact definition of the wind that a merical model will cancel out in the mean so that the
scatterometer measures. Since a scatterometer really correct model function can be obtained even in the
responds to surface roughness, the suggestion has presence of these errors.
often been made that a scatterometer measures the After a model function has been developed by
stress of the wind on the ocean surface rather than some means, a period of validation of the scatte-
the wind vector itself. In fact, a geophysical model rometer wind fields always follows. Generally scat-
function has been developed to relate so directly to terometer wind fields are compared with those
 SATELLITE REMOTE SENSING MICROWAVE SCATTEROMETERS 543

measured by buoys moored in fixed locations whose directions produced by NSCAT with winds measured
anemometer readings have been corrected to 10 m by anemometers on buoys operated by the US Na-
and neutral conditions. As an indication of the tional Data Buoy Center (NDBC). To produce the
accuracy that can be obtained by scatterometers, wind speed comparison in Figure 5A, buoy wind
Figure 5 shows a comparison of wind speeds and speeds have been binned into 0.5 m s1 bins and the

25
10 m neutral buoy winds

20
NSCAT speed (m s )
_1

15

10

0
0 5 10 15 20 25
_1
(A) NDBC speed (m s )

0.4 _1
3.5 m s
_1
7.5 m s
_1
10.5 m s
0.3 _1
Fractional occurrence

15.5 m s

0.2

0.1

0
_ 100 0 100
(B) Direction difference (NSCAT buoy, deg)

Figure 5 (A) Sample mean NSCAT wind speeds in 0.5 m s1 buoy wind speed bins. All NSCAT measurements were located within
50 km and 30 min of buoy measurements. Triangles denote 4100 NSCAT measurements in the mean; asterisks denote 5–99
samples. (B) Distributions of directional differences (NSCAT buoy) for 1 m s1 buoy wind speed bins centered on the indicated wind
speeds. (Reproduced with permission from Freilich and Dunbar, 1999.)
544 SATELLITE REMOTE SENSING MICROWAVE SCATTEROMETERS

corresponding NSCAT winds measured near the observations better than those using other wind
buoy have been averaged for each bin. The nonzero fields. Comparisons of sea levels predicted by the
values of NSCAT wind speed at zero buoy wind modular ocean model (MOM) with those of the
speed are known to be a result of comparing mag- TOPEX/Poseidon radar altimeter have indicated that
nitudes of vectors whose components are Gaussian more accurate predictions are achieved using ERS-1
distributed so the figure shows good agreement. wind fields than using NCEP fields. Similarly, pre-
Figure 5B shows the distribution of differences be- diction of the 1997–98 El Niño by the Lamont-
tween NSCAT wind directions and buoy wind dir- Doherty Earth Observatory model has been shown
ections for several different wind speeds and again to be improved by using NSCAT wind fields rather
agreement is good. than the surface observations collected by Florida
State University (FSU). This is illustrated in Figure 7.
Here sea surface temperature anomolies (with re-
spect to FSU climatology) in the eastern tropical
Applications of Scatterometry
Pacific predicted by the model with FSU and NSCAT
Further indications of the accuracy and usefulness of wind fields are compared with those actually ob-
data on winds from satellite scatterometers come served in 1997 and 1998. NSCAT wind fields have
from studies that compare these winds with, or use improved the prediction.
them in, atmospheric, oceanic, or climate models to
assess the improvements possible through the use of
scatterometer winds. Comparisons between wind
fields from satellite scatterometers and numerical
Limitations and Improvements
models have shown that significant differences in The examples above indicate the usefulness of sat-
these fields often exist. Cyclones predicted by nu- ellite scatterometry in oceanic and atmospheric
merical models have been found to disagree in lo- modeling. However, limitations still exist in the sat-
cation and intensity with those observed in satellite ellite wind fields which make them less useful than
scatterometry, sometimes by 300 km and 10 mbar or they could be. Ambiguities have been mentioned
more. The location of the Intertropical Convergence earlier. The temporal and spatial sampling patterns
Zone (ITCZ) has been shown to be 11–21 farther inherent in the data collection of individual satellites
south in NSCAT wind fields than in wind fields can obscure geophysical effects that occur on spatial
predicted by the numerical model of the European scales less than about 200 km and temporal scales
Centre for Medium-Range Weather Forecasts less than a day or two. This effectively limits the
(ECMWF). Furthermore, the ITCZ observed by application of satellite scatterometry to low-fre-
NSCAT was stronger and narrower than that pre- quency, large-scale events. Interpolation of the raw
dicted by ECMWF. An example of the differences data can reduce this problem somewhat but may
that can occur between scatterometer wind vectors introduce false signals. The obvious way to alleviate
and those of numerical models is given in Figure 6A. the problem is to put fleets of satellite scatterometers
Here NSCAT measurements in the South China Sea into orbit, an idea that may not be completely im-
are compared with predictions of the numerical possible in this day of smaller, cheaper satellites.
model of the National Centers for Environmental Another limitation on satellite scatterometry that
Prediction (NCEP). Obviously the fields can be quite has been observed many times is the mismeasure-
different. ment of surface wind fields in the presence of rain.
In the South China Sea, the Princeton Ocean Without an accompanying microwave radiometer, it
Model (POM) was run using wind fields from is difficult to ascertain from scatterometer measure-
NSCAT and from NCEP, some of which are shown in ments alone when rain is occurring on the surface.
Figure 6A. The results indicate significant differences Since heavy rainfall can attenuate a Ku-band scatte-
in the output of the POM depending on the wind rometer signal by more than a factor of 10 every
field used, although these differences were generally kilometer, serious mismeasurement of surface wind
smaller than those of the wind fields themselves. This fields can occur when rain is present. Furthermore,
is illustrated in Figure 6B, which shows surface cur- microwave return from the sea surface is also chan-
rents predicted by POM for the two different wind ged by rainfall, and this change is not well under-
fields and their differences. While this study did not stood. Measurements indicate that so becomes more
assess which POM prediction was the most accurate, isotropic in the presence of rain, and this effect is
other studies have indicated that the predictions of being utilized in an attempt to develop a rain flag
ocean models and coupled atmosphere–ocean mod- from scatterometer data alone. Such a flag would
els using scatterometer wind fields agreed with allow users to determine when rain effects might be
 SATELLITE REMOTE SENSING MICROWAVE SCATTEROMETERS 545

_1 _1
NCEP wind (m s ) Nov. 12 NCEP current (m s ) Nov. 12

20°N 20°N

0.200°E + 02 0.100°E + 01
Maximum vector Maximum vector

10°N 10°N

0°EQ 0°EQ

100°E 110°E 120°E 100°E 110°E 120°E

_ _
NSCAT wind (m s 1) Nov. 12 NSCAT current (m s 1) Nov. 12

20°N 20°N

0.200°E + 02 0.100°E + 01
Maximum vector Maximum vector

10°N 10°N

0°EQ 0°EQ

100°E 110°E 120°E 100°E 110°E 120°E

NSCAT_NCEP wind (m s ) Nov. 12

_1
NSCAT_NCEP current (m s ) Nov. 12
_1

20°N 20°N

0.200°E + 02 0.100°E + 01
Maximum vector Maximum vector

10°N 10°N

0°EQ 0°EQ

100°E 110°E 120°E 100°E 110°E 120°E

(A) (B)

Figure 6 (A) NCEP and NSCAT wind vectors in the South China Sea and their differences. (B) Surface currents in the South China
Sea from NCEP model under NCEP and NSCAT wind forcing. (Reproduced with permission from Chu et al., 1999.)

present in scatterometer data, but not what the geophysical model functions are smoothly decreasing
proper wind vectors are. functions of wind speed that are nonzero even at zero
Winds below 3 m s1 are difficult to measure ac- wind speeds. The Europeans have chosen not to
curately using satellite scatterometry. Present provide wind speed estimates below 3 m s1 using
546 SATELLITE REMOTE SENSING MICROWAVE SCATTEROMETERS

Initialized with FSU winds Initialized with NSCAT winds

observed Nino3 observed Nino3
4

3
SSTA (°C)

_1
ON D J FMAM J J A SON D J FMAM J ON D J FMAM J J A SON D J FMAM J
1997 1998 1997 1998
Time
(A) (B)

Figure 7 Forecasts of sea surface temperature anomalies in the eastern tropical Pacific for the 1997–98 El Niño by the Lamont-
Doherty Earth Observatory model using (A) Florida State University winds and (B) NSCAT winds. Heavy curves are observations;
other lines are predictions made at various times (Reproduced with permission from Chen et al., 1999.)

their ERS scatterometers. Recent studies have shown The result is that wind speeds above this value tend
that, in addition to the directional variability in- to be underestimated in scatterometry wind re-
herent in low winds, microwave backscatter is vir- trievals. Proposals have been offered for improved
tually zero below some threshold wind speed in the high wind speed model functions based on simul-
neighborhood of 2 m s1 to 4.5 m s1, depending on taneous scatterometry/radiometry measurements
incidence angle and water temperature. Variability of from aircraft flying through hurricanes. Implemen-
the wind over the surface footprint of the scatte- tation of such model functions promises to yield
rometer, however, obscures this threshold in most better wind retrievals at high wind speeds.
satellite scatterometry measurements. The geo-
physical model function at these low wind speeds
therefore appears to depend on both the mean wind
vector over the footprint and its variability. Sugges-
Conclusion
tions for improving low wind speed measurements Satellite-based microwave scatterometry is a mature
have included specifying a variability-dependent technology that has proven itself capable of yielding
model function that depends on geographic region global oceanic wind speeds of unprecedented accur-
and season, as well as using details of the probability acy and spatial coverage. Satellites currently in orbit
distribution on so . Implementation of such sugges- and planned for future missions promise a continu-
tions in future retrieval schemes offers hope of im- ous long-term series of global wind measurements
proving low wind speed measurements. that can aid in climate studies. Based on the time-
Finally, high wind speeds have also proven to be a series presently available, satellite scatterometer
problem for scatterometry. Recent studies indicate measurements have proven themself capable of
that so increases less rapidly with wind speed above improving present oceanographic and atmospheric
about 25 m s1 than current model functions predict. models. Future improvements in scatterometry
 SATELLITE REMOTE SENSING MICROWAVE SCATTEROMETERS 547

promise to make this technology even more valuable Freilich MH (1997) Validation of vector magnitude
in studying the dynamics of the atmosphere and datasets: effects of random component errors. Journal
oceans. of Atmospheric and Oceanic Technology 14: 695--703.
Freilich MH and Dunbar RS (1999) The accuracy of the
NSCAT 1 vector winds: Comparisons with National
Data Buoy Center buoys. Journal of Geophysical
See also Research 104: 11231--11246.
Aircraft Remote Sensing. Satellite Altimetry. Fu Lee-Lueng and Yi Chao (1997) The sensitivity of a
Satellite Remote Sensing SAR. Satellite Remote global ocean model to wind forcing. A test using sea
Sensing of Sea Surface Temperatures. Sensors for level and wind observations from satellites and
Mean Meteorology. Sensors for Micrometeorological operational wind analysis. Geophysical Research
and Flux Measurements. Letters 24: 1783--1786.
Graf J, Sasaki C, Winn C, et al. (1998) NASA
scatterometer experiment. Acta Astronautica 43:
397--407.
Further Reading Jones WL, Schroeder LC, and Mitchell JL (1977) Aircraft
measurements of the microwave scattering signature of
Brown RA and Zeng L (1994) Estimating central pressures the ocean. IEEE Journal of Oceanic Engineering OE-2:
of oceanic midlatitude cyclones. Journal of Applied 52--61.
Meteorology 33: 1088--1095. Kelly KA, Dickinson S, and Yu Z (1999) NSCAT tropical
Chen D, Cane MA, and Zebiak SE (1999) The impact of wind stress maps: implications for improving ocean
NSCAT winds on predicting the 1997/1998 El Niño: A modeling. Journal of Geophysical Research 104:
case study with the Lamont-Doherty Earth Observatory 11291--11310.
model. Journal of Geophysical Research 104: Moore RK and Pierson WJ (1996) Measuring Sea State and
11322--11327. Estimating Surface Winds from a Polar Orbiting
Chu PC, Lu S, and Liu WT (1999) Uncertainty of South Satellite. In: Proceedings of the International
China Sea prediction using NSCAT and National Symposium on Electromagnetic Sensing of Earth from
Centers for Environmental Prediction winds during Satellites, Miami Beach, FL, 1966.
tropical storm Ernie, 1996. Journal of Geophysical Moore RK and Fung AK (1979) Radar determination of
Research 104: 11273--11289. winds at sea. Proceedings of the IEEE 67: 1504--1521.
Donelan MA and Pierson WJ (1987) Radar scattering and Naderi FM, Freilich MH, and Long DG (1991) Spaceborne
equilibrium ranges in wind-generated waves with radar measurement of wind velocity over the ocean –
application to scatterometry. Journal of Geophysical An overview of the NSCAT Scatterometer system.
Research 92: 4971--5029. Proceedings of the IEEE 79: 850--866.
Fisher RE (1972) Standard deviation of scatterometer Pierson WJ Jr (1983) The measurement of the synoptic
measurements from space. IEEE Transactions Geo- scale wind over the ocean. Journal of Geophysical
science Electronics GE-10: 106--113. Research 88: 1682--1780.
 SATELLITE REMOTE SENSING OF SEA SURFACE
TEMPERATURES
P. J. Minnett, University of Miami, Miami, FL, USA flux) can be calculated by the Planck equation:
Copyright & 2001 Elsevier Ltd.
 1
Bl ðT Þ ¼ 2hc2 l5 ehc=ðlkT Þ  1 ½1

where h is Planck’s constant, c is the speed of light in

a vacuum, k is Boltzmann’s constant, l is the wave-
Introduction length and T is the temperature. The spectral inter-
The ocean surface is the interface between the two vals (wavelengths) are chosen where three conditions
dominant, fluid components of the Earth’s climate are met: (1) the sea emits a measurable amount of
system: the oceans and atmosphere. The heat moved radiant energy, (2) the atmosphere is sufficiently
around the planet by the oceans and atmosphere transparent to allow the energy to propagate to the
helps make much of the Earth’s surface habitable, spacecraft, and (3) current technology exists to build
and the interactions between the two, that take place radiometers that can measure the energy to the re-
through the interface, are important in shaping the quired level of accuracy within the bounds of size,
climate system. The exchange between the ocean and weight, and power consumption imposed by the
atmosphere of heat, moisture, and gases (such as spacecraft. In reality these constrain the instruments
CO2) are determined, at least in part, by the sea to two relatively narrow regions of the infrared part
surface temperature (SST). Unlike many other crit- of the spectrum and to low-frequency microwaves.
ical variables of the climate system, such as cloud The infrared regions, the so-called atmospheric
cover, temperature is a well-defined physical variable windows, are situated between wavelengths of
that can be measured with relative ease. It can also 3:524:1mm and 8212mm (Figure 1); the microwave
be measured to useful accuracy by instruments on measurements are made at frequencies of 6–12 GHz.
observation satellites. As the electromagnetic radiation propagates
The major advantage of satellite remote sensing of through the atmosphere, some of it is absorbed and
SST is the high-resolution global coverage provided scattered out of the field of view of the radiometer,
by a single sensor, or suite of sensors on similar thereby attenuating the original signal. If the at-
satellites, that produces a consistent data set. By the tenuation is sufficiently strong none of the radiation
use of onboard calibration, the accuracy of the time- from the sea reaches the height of the satellite, and
series of measurements can be maintained over years, such is the case when clouds are present in the field of
even decades, to provide data sets of relevance to view of infrared radiometers. Even in clear-sky con-
research into the global climate system. The rapid ditions a significant fraction of the sea surface
processing of satellite data permits the use of the emission is absorbed in the infrared windows. This
global-scale SST fields in applications where the energy is re-emitted, but at a temperature charac-
immediacy of the data is of prime importance, such teristic of that height in the atmosphere. Con-
as weather forecasting – particularly the prediction sequently the brightness temperatures measured
of the intensification of tropical storms and through the clear atmosphere by a spacecraft radi-
hurricanes. ometer are cooler than would be measured by a
similar device just above the surface. This atmos-
pheric effect, frequently referred to as the tempera-
Measurement Principle ture deficit, must be corrected accurately if the
derived sea surface temperatures are to be used
The determination of the SST from space is based on quantitatively.
measuring the thermal emission of electromagnetic
radiation from the sea surface. The instruments,
Infrared Atmospheric Correction Algorithms
called radiometers, determine the radiant energy
flux, Bl , within distinct intervals of the electro- The peak of the Planck function for temperatures
magnetic spectrum. From these the brightness tem- typical of the sea surface is close to the longer
perature (the temperature of a perfectly emitting wavelength infrared window, which is therefore well
‘black-body’ source that would emit the same radiant suited to SST measurement (Figure 1). However, the

548
 SATELLITE REMOTE SENSING OF SEA SURFACE TEMPERATURES 549

1.00

transmission 0.75
Atmosphere

Polar
0.50 Mid-latitude
Tropical
0.25

30
Spectral radiance
(W m 2 sr 1 μm 1)

NOAA-14
_

AVHRR T = 30°C
20
_ _

10
Ch 3 T = 0°C Ch 4 Ch 5

0
0 2 4 6 8 10 12 14
Wavelength (μm)

Figure 1 Spectra of atmospheric transmission in the infrared (wavelengths 1–14 mm) calculated for three typical atmospheres from
diverse parts of the ocean; polar, mid-latitude and tropical with integrated water vapor content of 7 kg m2 (polar), 29 kg m2
(mid-latitude) and 54 kg m2 (tropical). Regions where the transmission is high are well suited to satellite remote sensing of SST. The
lower panel shows the electromagnetic radiative flux for four sea surface temperatures (0, 10, 20, and 301C) with the relative spectral
response functions for channels 3, 4, and 5 of the AVHRR on the NOAA-14 satellite. The so-called ‘split-window’ channels, 4 and 5,
are situated where the sea surface emission is high, and where the atmosphere is comparatively clear but exhibits a strong
dependence on atmospheric water vapor content.

main atmospheric constituent in this spectral interval Further, by assuming that the atmospheric at-
that contributes to the temperature deficit is water tenuation is small in these channels, so that the ra-
vapor, which is very variable both in space and time. diative transfer can be linearized, and that the
Other molecular species that contribute to the tem- channels are spectrally close so that Planck’s function
perature deficit are quite well mixed throughout the can be linearized, the algorithm can be expressed in
atmosphere, and therefore inflict a relatively con- the very simple form:
stant temperature deficit that is simple to correct.
The variability of water vapor requires an atmos- SSTij ¼ ao þ ai Ti þ aj Tj ½3
pheric correction algorithm based on the information
contained in the measurements themselves. This is
where are ao, ai, and aj are coefficients. These are
achieved by making measurements at distinct spec-
determined by regression analysis of either coinci-
tral intervals in the windows when the water vapor
dent satellite and in situ measurements, mainly from
attenuation is different. These spectral intervals are
buoys, or of simulated satellite measurements de-
defined by the characteristics of the radiometer and
rived by radiative transfer modeling of the propa-
are usually referred to as bands or channels (Fig-
gation of the infrared radiation from the sea surface
ure 1). By invoking the hypothesis that the difference
through a representative set of atmospheric profiles.
in the brightness temperatures measured in two
The simple algorithm has been applied for many
channels, i and j, is related to the temperature deficit
years in the operational derivation of the sea surface
in one of them, the atmospheric correction algorithm
from measurements of the Advanced Very High
can be formulated thus:
Resolution Radiometer (AVHRR, see below), the
product of which is called the multi-channel SST
  (MCSST), where i refers to channel 4 and j to
SSTij  Ti ¼ f Ti  Tj ½2
channel 5.
More complex forms of the algorithms have been
where SSTij is the derived SST and Ti, Tj are the developed to compensate for some of the short-
brightness temperatures in channels i, j. comings of the linearization. One such widely
550 SATELLITE REMOTE SENSING OF SEA SURFACE TEMPERATURES

applied algorithm takes the form: correct for extraneous effects, and black-body cali-
bration targets to ensure the accuracy of the meas-
 
SSTij ¼ bo þ b1 Ti þ b2 Ti  Tj SSTr urements. The suite of channels is selected to include
  ½4 sensitivity to the parameters interfering with the SST
þb3 Ti  Tj ðsecy  1Þ
measurements, such as cloud droplets and surface
where SSTr is a reference SST (or first-guess tem- wind speed, which occurs with microwaves at higher
perature), and y is the zenith angle to the satellite frequencies. A simple combination of the brightness
radiometer measured at the sea surface. When ap- temperature, such as eqn [2], can retrieve the SST.
plied to AVHRR data, with i and j referring to The relative merits of infrared and microwave
channels 4 and 5 derived SST is called the nonlinear radiometers for measuring SST are summarized in
SST (NLSST). A refinement is called the Pathfinder Table 1.
SST (PFSST) in the research program designed to
post-process AVHRR data over a decade or so to Characteristics of Satellite-derived SST
provide a consistent data set for climate research. In
Because of the very limited penetration depth of in-
the PFSST, the coefficients are derived on a monthly
frared and microwave electromagnetic radiation in
basis for two different atmospheric regimes, dis-
sea water the temperature measurements are limited
tinguished by the value of the T4–T5 differences
to the sea surface. Indeed, the penetration depth is
being above or below 0.7 K, by comparison with
typically less than 1 mm in the infrared, so that
measurements from buoys.
temperature derived from infrared measurements is
The atmospheric correction algorithms work ef-
characteristic of the so-called skin of the ocean. The
fectively only in the clear atmosphere. The presence
skin temperature is generally several tenths of a de-
of clouds in the field of view of the infrared radi-
gree cooler than the temperature measured just
ometer contaminates the measurement so that such
below, as a result of heat loss from the ocean to
pixels must be identified and removed from the SST
atmosphere. On days of high insolation and low
retrieval process. It is not necessary for the entire
wind speed, the absorption of sunlight causes an
pixel to be obscured, even a portion as small as 3–
increase in near surface temperature so that the
5%, dependent on cloud type and height, can pro-
water just below the skin layer is up to a few degrees
duce unacceptable errors in the SST measurement.
warmer than that measured a few meters deeper,
Thin, semi-transparent cloud, such as cirrus, can
beyond the influence of the diurnal heating. For those
have similar effects to subpixel geometric obscur-
people interested in a temperature characteristic of a
ation by optically thick clouds. Consequently, great
depth of a few meters or more, the decoupling of the
attention must be paid in the SST derivation to the
skin and deeper, bulk temperatures is perceived as a
identification of measurements contaminated by even
disadvantage of using satellite SST. However, algo-
a small amount of clouds. This is the principle dis-
rithms generated by comparisons between satellite
advantage to SST measurement by spaceborne in-
frared radiometry. Since there are large areas of
cloud cover over the open ocean, it may be necessary Table 1 Relative merits of infrared and microwave radiometers
to composite the cloud-free parts of many images to for sea surface temperature measurement
obtain a complete picture of the SST over an ocean
Infrared Microwave
basin.
Similarly, aerosols in the atmosphere can introduce Good spatial resolution Poor spatial resolution
significant errors in SST measurement. Volcanic (B1 km) (B50 km)
aerosols injected into the cold stratosphere by violent Surface obscured by clouds Clouds largely transparent, but
eruptions produce unmistakable signals that can bias measurement perturbed by
heavy rain
the SST too cold by several degrees. A more insidious
No side-lobe contamination Side-lobe contamination
problem is caused by less readily identified aerosols prevents measurements close
at lower, warmer levels of the atmosphere that can to coasts or ice
introduce systematic errors of a much smaller Aperture is reasonably small; Antenna is large to achieve
amplitude. instrument can be compact spatial resolution from polar
for spacecraft use orbit heights (B800 km above
the sea surface)
Microwave Measurements 4 km resolution possible from Distance to geosynchronous
geosynchronous orbit; can orbit too large to permit useful
Microwave radiometers use a similar measurement provide rapid sampling data spatial resolution with current
principle to infrared radiometers, having several antenna sizes
spectral channels to provide the information to
 SATELLITE REMOTE SENSING OF SEA SURFACE TEMPERATURES 551

Table 2 Spectral characteristics of current and planned satellite-borne infrared radiometers

AVHRR ATSR MODIS OCTS GLI

l (mm) NEDT (K) l (mm) NEDT (K) l (mm) NEDT (K) l (mm) NEDT (K) l (mm) NEDT (K)

3.75 0.12 3.7 0.019 3.75 0.05 3.7 0.15 3.715 o0.15
3.96 0.05
4.05 0.05
8.55 0.05 8.52 0.15 8.3 o0.1
10.5 0.12 10.8 0.028 11.03 0.04 10.8 0.15 10.8 o0.1
11.5 0.12 12.0 0.025 12.02 0.04 11.9 0.15 12 o0.1

and in situ measurements from buoys include a mean equivalent temperature difference (NEDT) to the
skin effect masquerading as part of the atmospheric levels shown in Table 2.
effect, and so the application of these results in an
estimate of bulk temperatures.
The Advanced Very High Resolution Radiometer
The greatest advantage offered by satellite remote
(AVHRR)
sensing is, of course, coverage. A single, broad-
swath, imaging radiometer on a polar-orbiting sat- The satellite instrument that has contributed the
ellite can provide global coverage twice per day. An most to the study of the temperature of the ocean
imaging radiometer on a geosynchronous satellite surface is the AVHRR that first flew on TIROS-N
can sample much more frequently, once per half- launched in late 1978. AVHRRs have flown on suc-
hour for the Earth’s disk, or smaller segments every cessive satellites of the NOAA series from NOAA-6
few minutes, but the spatial extent of the data is to NOAA-14, with generally two operational at any
limited to that part of the globe visible from the given time. The NOAA satellites are in a near-polar,
satellite. sun-synchronous orbit at a height of about 780 km
The satellite measurements of SST are also rea- above the Earth’s surface and with an orbital period
sonably accurate. Current estimates for routine of about 100 min. The overpass times of the two
measurements show absolute accuracies of 70.3 to NOAA satellites are about 2.30 a.m. and p.m. and
70.5 K when compared to similar measurements about 7.30 a.m. and p.m. local time. The AVHRR
from ships, aircraft, and buoys. has five channels: 1 and 2 at B0.65 and B0.85 mm
are responsive to reflected sunlight and are used to
detect clouds and identify coastlines in the images
from the daytime part of each orbit. Channels 4 and
Spacecraft Instruments
5 (Table 2 and Figure 1) are in the atmospheric
All successful instruments have several attributes in window close to the peak of the thermal emission
common: a mechanism for scanning the Earth’s sur- from the sea surface and are used primarily for the
face to generate imagery, good detectors, and a measurement of sea surface temperature. Channel 3,
mechanism for real-time, in-flight calibration. Cali- positioned at the shorter wavelength atmospheric
bration involves the use of one or more black-body window, is responsive to both surface emission and
calibration targets, the temperatures of which are reflected sunlight. During the nighttime part of each
accurately measured and telemetered along with the orbit, measurements of channel 3 brightness tem-
imagery. If only one black-body is available a peratures can be used with those from channels 4
measurement of cold dark space provides the and 5 in variants of the atmospheric correction al-
equivalent of a very cold calibration target. Two gorithm to determine SST. The presence of reflected
calibration points are needed to provide in-flight sunlight during the daytime part of the orbit prevents
calibration; nonlinear behavior of the detectors is much of these data from being used for SST meas-
accounted for by means of pre-launch instrument urement. Because of the tilting of the sea surface by
characterization measurements. waves, the area contaminated by reflected sunlight
The detectors themselves inject noise into the data (sun glitter) can be quite extensive, and is dependent
stream, at a level that is strongly dependent on their on the local surface wind speed. It is limited to the
temperature. Therefore, infrared radiometers require point of specular reflection only in very calm seas.
cooled detectors, typically operating from 78 K The images in each channel are constructed by
(  1951C) to 105 K (  1681C) to reduce the noise scanning the field of view of the AVHRR across the
552 SATELLITE REMOTE SENSING OF SEA SURFACE TEMPERATURES

AVHRR

Flight

Sub-satellite
point

Sub-satellite
track

Scan
line
km
= 2700
swath
AVHR R

(A)

Figure 2 Scan geometries of AVHRR (A) and ATSR (B). The continuous wide swath of the AVHRR is constructed by linear scan
lines aligned across the direction of motion of the subsatellite point. The swaths of the ATSR are generated by an inclined conical
scan, which covers the same swath through two different atmospheric path lengths. The swath is limited to 512 km by geometrical
constraints. Both radiometers are on sun-synchronous, polar-orbiting satellites.

Earth’s surface by a mirror inclined at 451 to the made in quick succession, so that the SST and at-
direction of flight (Figure 2A). The rate of rotation, mospheric conditions do not change in the time
6.67 Hz, is such that successive scan lines are interval. This approach is that used by the ATSR,
contiguous at the surface directly below the satellite. two of which have flown on the European satellites
The width of the swath (B2700 km) means that the ERS-1 and ERS-2.
swaths from successive orbits overlap so that The ATSR has infrared channels in the atmos-
the whole Earth’s surface is covered without gaps pheric windows comparable to those of AVHRR, but
each day. the rotating scan mirror sweeps out a cone inclined
from the vertical by its half-angle (Figure 2B). The
field of view of the ATSR sweeps out a curved path
The Along-Track Scanning Radiometer (ATSR)
on the sea surface, beginning at the point directly
An alternative approach to correcting the effects of below the satellite, moving out sideways and for-
the intervening atmosphere is to make a brightness wards. Half a mirror revolution later, the field of
temperature measurement of the same area of sea view is about 900 km ahead of the sub-satellite track
surface through two different atmospheric path in the center of the ‘forward view’. The path of the
lengths. The pairs of such measurements must be field of view returns to the sub-satellite point, which,
 SATELLITE REMOTE SENSING OF SEA SURFACE TEMPERATURES 553

ATSR

Flight

Sub-satellite
point

ATSR swath Sub-satellite

= 512 km track

NADIR
scan

Forward
scan
(B)

Figure 2 continued

during the period of the mirror rotation, has moved 1 three atmospheric window channels and the set of c
km ahead of the starting point. Thus the pixels are coefficients. The coefficients, derived by radiative
forming the successive swaths through the nadir transfer simulations, have an explicit latitudinal
point are contiguous. The orbital motion of the sat- dependence.
ellite means that the nadir point overlays the center Accurate calibration of the brightness tempera-
of the forward view after about 2 min. The atmos- tures is achieved by using two onboard black-body
pheric path length of the measurement at nadir is cavities, situated between the apertures for the nadir
simply the thickness of the atmosphere, whereas the and forward views such that they are scanned each
slant path to the center of the forward view is almost rotation of the mirror. One calibration target is at the
double that, resulting in colder brightness tempera- spacecraft ambient temperature while the other is
tures. The differences in the brightness temperatures heated, so that the measured brightness temperatures
between the forward and nadir swaths are a direct of the sea surface are straddled by the calibration
measurement of the effect of the atmosphere and temperatures.
permit a more accurate determination of the sea The limitation of the simple scanning geometry
surface temperature. The atmospheric correction al- of the ATSR is a relatively narrow swath width
gorithm takes the form: of 512 km. The ERS satellites have at various
times in their missions been placed in orbits with
X X
SST ¼ co þ cn:i Tn;i þ cf ;i Tf ;i ½5 repeat patterns of 3, 35, and 168 days, and given
i i the narrow ATSR swath, complete coverage of
the globe has been possible only for the 35 and 186
where the subscripts n and f refer to measurements day cycles. This disadvantage is offset by the in-
from the nadir and forward views, i indicates two or tended improvement in absolute accuracy of the
554 SATELLITE REMOTE SENSING OF SEA SURFACE TEMPERATURES

atmospheric correction, and of its better insensitiv- polarizations. The 10.65 GHz channels confer a
ity to aerosol effects. sensitivity to SST, at least in the higher SST range
found in the tropics, that has been absent in micro-
The Moderate Resolution Imaging wave radiometers since the SMMR (Scanning Mul-
Spectroradiometer (MODIS) tifrequency Microwave Radiometer) that flew on the
short-lived Seasat in 1978 and on Nimbus-7 from
The MODIS is a 36-band imaging radiometer on the 1978 to 1987. Although SSTs were derived from
NASA Earth Observing System (EOS) satellites Terra, SMMR measurements, these lacked the spatial reso-
launched in December 1999, and Aqua, planned for lution and absolute accuracy to compete with those
launch by late 2001. MODIS is much more complex of the AVHHR. The TMI complements AVHRR data
than other radiometers used for SST measurement, by providing SSTs in the tropics where persistent
but uses the same atmospheric windows. In addition clouds can be a problem for infrared retrievals. In-
to the usual two bands in the 10–12 mm interval, stead of a rotating mirror, TMI, like other microwave
MODIS has three narrow bands in the 3.7–4.1 mm imagers, uses an oscillating parabolic antenna to
windows, which, although limited by sun-glitter direct the radiation through a feed-horn into the
effects during the day, hold the potential for much radiometer.
more accurate measurement of SST during the night. The swath width of TMI is 759 km and the orbit
Several of the other 31 bands of MODIS contribute to of TRMM restricts SST measurements to within
the SST measurement by better identification of 38.51 of the equator. The beam width of the 10.65
residual cloud and aerosol contamination. GHz channels produces a footprint of 37  63 km,
The swath width of MODIS, at 2330 km, is but the data are over-sampled to produce 104 pixels
somewhat narrower than that of AVHRR, with the across the swath.
result that a single day’s coverage is not entire, but
the gaps from one day are filled in on the next. The
spatial resolution of the infrared window bands is 1 Applications
km at nadir.
With absolute accuracies of satellite-derived SST
fields of B0.5 K or better, and even smaller relative
The GOES Imager
uncertainties, many oceanographic features are re-
SST measurements from geosynchronous orbit are solved. These can be studied in a way that was
made using the infrared window channels of the hitherto impossible. They range from basin-scale
GOES Imager. This is a five-channel instrument that perturbations to frontal instabilities on the scales of
remains above a given point on the Equator. The tens of kilometers. SST images have revealed the
image of the Earth’s disk is constructed by scanning great complexity of ocean surface currents; this
the field of view along horizontal lines by an oscil- complexity was suspected from shipboard and air-
lating mirror. The latitudinal increments of the scan craft measurements, and by acoustically tacking
line are done by tilting the axis of the scan mirror. The neutrally buoyant floats. However, before the advent
spatial resolution of the infrared channels is 2.3 km of infrared imagery the synoptic view of oceanic
(east–west) by 4 km (north–south) at the subsatellite variability was elusive, if not impossible.
point. There are two imagers in orbit at the same time
on the two GOES satellites, covering the western El Niño
Atlantic Ocean (GOES-East) and the eastern Pacific
Ocean (GOES-West). The other parts of the global The El Niño Southern Oscillation (ENSO) phe-
oceans visible from geosynchronous orbit are covered nomenon has become a well-known feature of the
by three other satellites operated by Japan, India, and coupled ocean–atmosphere system in terms of per-
the European Meteorological Satellite organization turbations that have a direct influence on people’s
(Eumetsat). Each carries an infrared imager, but with lives, mainly by altering the normal rainfall patterns
lesser capabilities than the GOES Imager. causing draughts or deluges – both of which imperil
lives, livestock, and property.
The normal SST distribution in the topical Pacific
TRMM Microwave Imager (TMI)
Ocean is a region of very warm surface waters in the
The TMI is a nine-channel microwave radiometer on west, with a zonal gradient to cooler water in the
the Tropical Rainfall Measuring Mission satellite, east; superimposed on this is a tongue of cool surface
launched in 1997. The nine channels are centered at water extending westward along the Equator. This
five frequencies: 10.65, 19.35, 21.3, 37.0, and 85.5 situation is associated with heavy rainfall over the
GHz, with four of them being measured at two western tropical Pacific, which is in turn associated
 SATELLITE REMOTE SENSING OF SEA SURFACE TEMPERATURES 555

with lower level atmospheric convergence and deep the shape, area, position, and movement of the
atmospheric convection. The atmospheric con- Warm Pool can be provided from satellite-derived
vergence and convection are part of the large-scale SST to help study the phenomenon and forecast its
global circulation. The warm area of surface water, consequences.
enclosed by the 281C isotherm, is commonly referred Figure 3 shows part of the global SST fields de-
to as the ‘Warm Pool’ and in the normal situation is rived from the Pathfinder SST algorithm applied to
confined to the western part of the tropical Pacific. AVHRR measurements. The tropical Pacific SST field
During an El Niño event the warm surface water, and in the normal situation (December 1993) is shown in
associated convection and rainfall, migrate eastward the upper panel, while the lower panel shows the
perturbing the global atmospheric circulation. El anomalous field during the El Niño event of 1997–
Niño events occur up to a few times per decade and 98. This was one of the strongest El Niños on record,
are of very variable intensity. Detailed knowledge of but also the best documented and forecast. Seasonal

180˚W 120˚W 60˚W 0˚ 60˚E 120˚E 180˚E

December 1993

60˚N 60˚N

0˚ 0˚

60˚S 60˚S

December 1997

60˚N 60˚N

0˚ 0˚

60˚S 60˚S

180˚W 120˚W 60˚W 0˚ 60˚E 120˚E 180˚E

RSMAS
_2 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34
Sea surface temperature ˚C

Figure 3 Global maps of SST derived from the AVHRR Pathfinder data sets. These are monthly composites of cloud-free pixels and
show the normal situation in the tropical Pacific Ocean (above) and the perturbed state during an El Niño event (below).
556 SATELLITE REMOTE SENSING OF SEA SURFACE TEMPERATURES

predictions of disturbed patterns of winds and rain- ‘pinched off’ from the flow and evolve as independ-
fall had an unprecedented level of accuracy and ent features, migrating away from the currents.
provided improved useful forecasts for agriculture in When these occur on the equator side of the current
many affected areas. Milder than usual hurricane these are called ‘Warm Core Rings’ and can exist for
and tropical cyclone seasons were successfully fore- many months; in the case of the Gulf Stream these
cast, as were much wetter winters and severe coastal can propagate into the Sargasso Sea.
erosion on the Pacific coasts of the Americas. Figure 5 shows a series of instabilities along the
boundaries of the Equatorial current system in the
Hurricane Intensification Pacific Ocean. The extent and structure of these
features were first described by analysis of satellite
The Atlantic hurricane season in 1999 was one of the
SST images.
most damaging on record in terms of land-falling
storms in the eastern USA, Caribbean, and Central
Coral Bleaching
America. Much of the damage was not a result of
high winds, but of torrential rainfall. Accurate fore- Elevated SSTs in the tropics have adverse influences
casting of the path and intensity of these land-falling on living coral reefs. When the temperatures exceed
storms is very important, and a vital component of the local average summertime maximum for several
this forecasting is detailed knowledge of SST patterns days, the symbiotic relationship between the coral
in the path of the hurricanes. The SST is indicative of polyps and their algae breaks down and the reef-
the heat stored in the upper ocean that is available to building animals die. The result is extensive areas
intensify the storms, and SSTs of 4261C appear to where the coral reef is reduced to the skeletal struc-
be necessary to trigger the intensification of the ture without the living and growing tissue, giving the
hurricanes. Satellite-derived SST maps are used in the reef a white appearance. Time-series of AVHRR-de-
prediction of the development of storm propagation rived SST have been shown to be valuable predictors
across the Atlantic Ocean from the area off Cape of reef areas around the globe that are threatened by
Verde where atmospheric disturbances spawn the warmer than usual water temperatures. Although it
nascent storms. Closer to the USA and Caribbean, is not possible to alter the outcome, SST maps have
the SST field is important in determining the sudden been useful in determining the scale of the problem
intensification that can occur just before landfall. and identifying threatened, or vulnerable reefs.
After the hurricane has passed, they sometimes leave
a wake of cooler water in the surface that is readily The ‘Global Thermometer’
identifiable in the satellite-derived SST fields.
Some of the most pressing problems facing environ-
mental scientists are associated with the issue of
Frontal Positions
global climate change: whether such changes are
One of the earliest features identified in infrared natural or anthropogenic, whether they can be
images of SST were the positions of ocean fronts, forecast accurately on regional scales over decades,
which delineate the boundaries between dissimilar and whether undesirable consequences can be avoi-
surface water masses. Obvious examples are western ded. The past decade has seen many air temperature
boundary currents, such as the Gulf Stream in the records being surpassed and indeed the planet ap-
Atlantic Ocean (Figure 4) and the Kuroshio in the pears to be warming on a global scale. However, the
Pacific Ocean, both of which transport warm surface air temperature record is rather patchy in its distri-
water poleward and away from the western coast- bution, with most weather stations clustered on
lines. In the Atlantic, the path of the warm surface Northern Hemisphere continents.
water of the Gulf Stream can be followed in SST Global SST maps derived from satellites provide
images across the ocean, into the Norwegian Sea, an alternative approach to measuring the Earth’s
and into the Arctic Ocean. The surface water loses temperature in a more consistent fashion. However,
heat to the atmosphere, and to adjacent cooler because of the very large thermal inertia of the ocean
waters on this path from the Gulf of Mexico to the (it takes as much heat to raise the temperature of
Arctic, producing a marked zonal difference in the only the top meter of the ocean through one degree
climates of the opposite coasts of the Atlantic and as it does for the whole atmosphere), the SST
Greenland-Norwegian Seas. Instabilities in the fronts changes indicative of global warming are small.
at the sides of the currents have been revealed in Climate change forecast models indicate a rate of
great detail in the SST images. Some of the large- temperature increase of only a few tenths of a degree
scale instabilities can lead to loops on scales of a few per decade, and this is far from certain because of
tens to hundreds of kilometers that can become our incomplete understanding of how the climate
 SATELLITE REMOTE SENSING OF SEA SURFACE TEMPERATURES 557

40

38

36

34

˚N
77 75 73 ˚W

5 25
12 μm brightness temperature / ˚C

Figure 4 Brightness temperature image derived from the measurements of the ATSR on a nearly cloud-free day over the eastern
coast of the USA. The warm core of the Gulf Stream is very apparent; it departs from the coast at Cape Hatteras. The cool, shelf water
from the north entrains the warmer outflows from the Chesapeake and Delaware Bays. The north wall of the Gulf Stream reveals very
complex structure associated with frontal instabilities that lead to exchanges between the Gulf Stream and inshore waters. The small-
scale multicolored patterns over the warm Gulf Stream waters to the south indicate the presence of cloud. This image was taken at
15.18 UTC on 21 May 1992, and is derived from nadir view data from the 12 mm channel. (Generated from data ^ NERC/ESA/RAL/
BNSC, 1992)
558 SATELLITE REMOTE SENSING OF SEA SURFACE TEMPERATURES

30 60 90 120 150 180 210 240 270 300 330 0 30

40 40

20 20

0 0

_ 20 _ 20

_ 40 _ 40
30 60 90 120 150 180 210 240 270 300 330 0 30

16 20 24 28 32
Sea surface temperature ˚C

Figure 5 Tropical SSTs produced by microwave radiometer measurements from the TRMM (Tropical Rainfall Measuring Mission)
Microwave Imager (TMI). This is a composite generated from data taken during the week ending December 22, 1999. The latitudinal
extent of the data is limited by the orbital geometry of the TRMM satellite. The measurement is much less influenced by clouds than
those in the infrared, but the black pixels in parts of the oceans where there are no islands indicate areas of heavy rainfall. The image
reveals the cold tongue of surface water along the Equator in the Pacific Ocean and cold water off the Pacific coast of South America,
indicating a non-El Niño situation. Note that the color scale is different from that used in Figure 3. The image was produced by Remote
Sensing Systems, sponsored in part by NASA’s Earth Science Information Partnerships (ESIP) (a federation of information sites for
Earth science); and by the NOAA/NASA Pathfinder Program for early EOS products; principal investigator: Frank Wentz.

system functions, especially in terms of various SST fields, particularly in the presence of atmos-
feedback factors such as those involving changes in pheric aerosols. This will involve the incorporation
cloud and aerosol properties. Such a rate of tem- of information from additional spectral channels,
perature increase will require SST records of several such as those on MODIS or other EOS era satellite
decades length before the signal, if present, can be instruments. Improvements in SST coverage, at least
unequivocally identified above the uncertainties in in the tropics, can be expected in areas of heavy,
the accuracy of the satellite-derived SSTs. Further- persistent cloud cover by melding SST retrievals from
more, the inherent natural variability of the global high-resolution infrared sensors with those from
SST fields tends to mask any small, slow changes. microwave radiometers, such as the TMI.
Difficult though this task may be, global satellite- Continuing improvements in methods of valid-
derived SSTs are an important component in climate ating the SST retrieval algorithms will improve our
change research. understanding of the error characteristics of the SST
fields, guiding both the appropriate applications of
Air-sea Exchanges the data and also improvements to the algorithms.
On the hardware front, a new generation of in-
The SST fields play further indirect roles in the cli- frared radiometers designed for SST measurements
mate system in terms of modulating the exchanges of will be introduced on the new operational satellite
heat and greenhouse gases between the ocean and series, the National Polar-Orbiting Environmental
atmosphere. Although SST is only one of several Satellite System (NPOESS) that will replace both the
variables that control these exchanges, the SST dis- civilian (NOAA-n) and military (DMSP, Defense
tributions, and their evolution on seasonal timescales Meteorological Satellite Program) meteorological
can help provide insight into the global patterns of satellites. The new radiometer, called VIIRS (the
the air–sea exchanges. An example of this is the Visible and Infrared Imaging Radiometer Suite), will
study of tropical cloud formation over the ocean, a replace the AVHRR and MODIS. The prototype
consequence of air–sea heat and moisture exchange, VIIRS will fly on the NPP (NPOESS Preparatory
in terms of SST distributions. Program) satellite scheduled for launch in late 2005.
At present, the design details of the VIIRS are not
finalized, but the physics of the measurement con-
Future Developments strains the instrument to use the same atmospheric
Over the next several years continuing improvement window channels as previous and current instru-
of the atmospheric correction algorithms can be an- ments, and have comparable, or better, measurement
ticipated to achieve better accuracies in the derived accuracies.
 SATELLITE REMOTE SENSING OF SEA SURFACE TEMPERATURES 559

The ATSR series will continue with at least one Ikeda M and Dobson FW (1995) Oceanographic
more model, called the Advanced ATSR (AATSR) to Applications of Remote Sensing. London: CRC Press.
fly on Envisat to be launched in 2001. The SST Kearns EJ, Hanafin JA, Evans RH, Minnett PJ, and Brown
capability of this will be comparable to that of its OB (2000) An independent assessment of Pathfinder
predecessors. AVHRR sea surface temperature accuracy using the
Marine-Atmosphere Emitted Radiance Interferometer
Thus, the time-series of global SSTs that now ex-
(M-AERI). Bulletin of the American Meteorological
tends for two decades will continue into the future to Society. 81: 1525--1536.
provide invaluable information for climate and Kidder SQ and Vonder Haar TH (1995) Satellite
oceanographic research. Meteorology: An Introduction. London: Academic
Press.
Legeckis R and Zhu T (1997) Sea surface temperature from
See also the GEOS-8 geostationary satellite. Bulletin of the
American Meteorological Society 78: 1971--1983.
IR Radiometers. Radiative Transfer in the Ocean. May DA, Parmeter MM, Olszewski DS, and Mckenzie BD
Satellite Altimetry. Satellite Remote Sensing SAR. (1998) Operational processing of satellite sea surface
temperature retrievals at the Naval Oceanographic
Office. Bulletin of the American Meteorological Society,
Further Reading 79: 397--407.
Robinson IS (1985) Satellite Oceanography: An
Barton IJ (1995) Satellite-derived sea surface temperatures: Introduction for Oceanographers and Remote-sensing
Current status. Journal of Geophysical Research 100: Scientists. Chichester: Ellis Horwood.
8777--8790. Stewart RH (1985) Methods of Satellite Oceanography.
Gurney RJ, Foster JL, and Parkinson CL (eds.) (1993) Berkeley, CA: University of California Press.
Atlas of Satellite Observations Related to Global Victorov S (1996) Regional Satellite Oceanography.
Change. Cambridge: Cambridge University Press. London: Taylor and Francis.
 SATELLITE REMOTE SENSING SAR
A. K. Liu and S. Y. Wu, NASA Goddard images using the radar cross-section. The ability of a
Space Flight Center, Greenbelt, MD, USA SAR to provide valuable information on the type,
Copyright & 2001 Elsevier Ltd. condition, and motion of the sea ice and surface
signatures of swells, wind fronts, and eddies near the
ice edge has also been amply demonstrated.
With all-weather, day/night imaging capability,
SAR penetrates clouds, smoke, haze, and darkness to
acquire high quality images of the Earth’s surface.
Introduction This makes SAR the frequent sensor of choice for
Synthetic aperture radar (SAR) is a side-looking cloudy coastal regions. Space agencies from the USA,
imaging radar usually operating on either an aircraft Canada, and Europe use SAR imagery on an oper-
or a spacecraft. The radar transmits a series of short, ational basis for sea ice monitoring, and for the de-
coherent pulses to the ground producing a footprint tection of icebergs, ships, and oil spills. However,
whose size is inversely proportional to the antenna there can be considerable ambiguity in the inter-
size, its aperture. Because the antenna size is gener- pretation of physical processes responsible for the
ally small, the footprint is large and any particular observed ocean features. Therefore, the SAR imaging
target is illuminated by several hundred radar pulses. mechanisms of ocean features are briefly described
Intensive signal processing involving the detection of here to illustrate how SAR imaging is used oper-
small Doppler shifts in the reflected signals from ationally in applications such as environmental
targets to the moving radar produces a high reso- monitoring, fishery support, and marine surveillance.
lution image that is equivalent to one that would
have been collected by a radar with a much larger
History
aperture. The resulting larger aperture is the ‘syn-
thetic aperture’ and is equal to the distance traveled The first spaceborne SAR was flown on the US sat-
by the spacecraft while the radar antenna is col- ellite Seasat in 1978. Although Seasat only lasted 3
lecting information about the target. SAR techniques months, analysis of its data confirmed the sensitivity
depend on precise determination of the relative of SAR to the geometry of surface features. On
position and velocity of the radar with respect to the March 31, 1991 the Soviet Union became the next
target, and on how well the return signal is country to operate an earth-orbiting SAR with the
processed. launch of Almaz-1. Almaz-1 returned to earth in
SAR instruments transmit radar signals, thus 1992 after operating for about 18 months. The
providing their own illumination, and then measure European Space Agency (ESA) launched its first re-
the strength and phase of the signals scattered back mote sensing satellite, ERS-1, with a C-band SAR on
to the instrument. Radar waves have much longer July 17, 1991. Shortly thereafter, the JERS-1 satellite,
wavelengths compared with light, allowing them to developed by the National Space Development
penetrate clouds with little distortion. In effect, Agency of Japan (NASDA), was launched on Feb-
radar’s longer wavelengths average the properties of ruary 11, 1992 with an L-band SAR. This was fol-
air with the properties and shapes of many individual lowed a few years later by Radarsat-1, the first
water droplets, and are only affected while entering Canadian remote sensing satellite, launched in No-
and exiting the cloud. Therefore, microwave radar vember 1995. Radarsat-1 has a ScanSAR mode with
can ‘see’ through clouds. a 500 km swath and a 100 m resolution, an in-
SAR images of the ocean surface are used to detect novative variation of the conventional SAR (with a
a variety of ocean features, such as refracting surface swath of 100 km and a resolution of 25 m). ERS-2
gravity waves, oceanic internal waves, wind fields, was launched in April 1995 by ESA, and Envisat-1
oceanic fronts, coastal eddies, and intense low pres- with an Advanced SAR is underway with a scheduled
sure systems (i.e. hurricanes and polar lows), since launch date in July 2001. The Canadian Space
they all influence the short wind waves responsible Agency (CSA) has Radarsat-2 planned for 2002, and
for radar backscatter. In addition, SAR is the only NASDA has Advanced Land Observing Satellite
sensor that provides measurements of the directional (ALOS) approved for 2003. Table 1 shows all major
wave spectrum from space. Reliable coastal wind ocean-oriented spaceborne SAR missions worldwide
vectors may be estimated from calibrated SAR from 1978 to 2003.

560
 SATELLITE REMOTE SENSING SAR 561

Table 1 Major ocean-oriented spaceborne SAR missions system performance and obtain sample data of the
land using various incidence angles. The SIR-B ex-
Platform Nation Launch Band a Status periment provided a unique opportunity for studying
ocean wave spectra due to the relatively lower orbit
Seasat USA 1978 L Ended
Almaz-1 USSR 1991 S Ended
of the Shuttle as compared with satellites. The low
ERS-1 Europe 1991 C Standby orbital altitude increases the frequency range of
JERS-1 Japan 1992 L Ended ocean waves that could be reliably imaged, because
ERS-2 Europe 1995 C Operational blurring of the detected waves caused by the motion
Radarsat-1 Canada 1995 C Operational
of ocean surface during the imaging process is re-
Envisat-1 Europe 2001 C Launch scheduled
Radarsat-2 Canada 2002 C Approved duced. The final SIR mission, SIR-C in April and
ALOS Japan 2003 L Approved October 1994, simultaneously recorded SAR data at
three wavelengths (L-, C-, and X-bands). These
a
Some frequently used radar wavelengths are: 3.1 cm for X-band, multiple-frequency data from SIR-C improved our
5.66 cm for C-band, 10.0 cm for S-band, and 23.5 cm for L-band. understanding of the radar scattering properties of
the ocean surface.
Aside from these free-flying missions, a number of
early spaceborne SAR experiments in the USA were Imaging Mechanism of Ocean
conducted using shuttle imaging radar (SIR) systems
flown on NASA’s Space Shuttle. The SIR-A and SIR-
Features
B experiments, in November 1981 and October For a radar with an incidence angle of 201–501, such
1984, respectively, were designed to study radar as all spaceborne SARs, backscatter from the ocean

Sea ice

Sea ice

Oil spills

Point
Barro
w

0 25
km

Figure 1 Radarsat ScanSAR image of oil spills off Point Barrow, Alaska, collected on November 2, 1997. (^ CSA 1997.)
562 SATELLITE REMOTE SENSING SAR

surface is produced primarily by the Bragg resonant Bragg-scattering scales. Thus, SARs have proven to
scattering mechanism. That is, surface waves tra- be an excellent means of mapping ocean features.
veling in the radar range (across-track) direction For ocean current features, the essential element of
with a wavelength of l=ð2 sin yÞ, called the Bragg the surface manifestation is the interaction between
resonant waves, account for most of the back- the current field and the wind-driven ocean surface
scattering. In this formula, l is the radar wavelength, waves. The effect of the surface current is to alter the
and y is the incidence angle. In general, the Bragg short-wave spectrum from its equilibrium value,
resonant waves are short gravity waves with wave- while the natural processes of wave energy input
lengths in the range of 3–30 cm, depending on the from the wind restores the ambient equilibrium
radar wavelength or band, as shown in Table 1. spectrum. A linear SAR system is one for which the
Because SAR is most sensitive to waves of this variation of the SAR image intensity is proportional
wavelength, or roughness of this scale, any ocean to the gradient of the surface velocity. The pro-
phenomenon or process that produces modulation in portionality depends on radar wavelength, radar in-
these particular wavelengths is theoretically detect- cidence angle, angle between the radar look direction
able by SAR. The radar cross-section of the ocean and the current direction, azimuth angle, and the
surface is affected by any geophysical variable, wind velocity. Under high wind condition, large wind
such as wind stress, current shear, or surface slicks, waves may overwhelm the weaker current feature.
that can modulate the ocean surface roughness at When current flows in the cross-wind direction, the

0 10
N km

Sand waves
Valley

Continental Shelf

Figure 2 ERS-1 SAR image of shallow water bathymetry at Taiwan Tan acquired on July 27, 1994. (^ ESA 1994.)
 SATELLITE REMOTE SENSING SAR 563

wave–current interaction is relatively weak, causing and are thus similar to the appearance of areas of low
a weak radar backscattering signal. winds. The distinctive shape and sharp boundary of
For ocean frontal features, the change in surface localized surface films allows them to be distinguished
brightness across a front in a SAR image is caused by from the relatively large regions of low wind.
the change in wind stress exerted onto the ocean
surface. The wind stress in turn depends on wind
speed and direction, air–sea temperature difference, Examples of Ocean Features from
and surface contamination. The effects of wind stress
SAR Applications
upon surface ripples and therefore upon radar cross-
section, have been modeled and demonstrated as A number of important SAR applications have
shown in the example below. In the high wind stress emerged recently, particularly since ERS-1/2, and
area, the ocean surface is rougher and appears as a Radarsat-1 data became available and the ability to
brighter area in a SAR image. On the other side of the process SAR data has improved. In the USA, the
front where the wind is lower, the surface is smoother National Oceanic and Atmospheric Administration
and appears as a darker area in a SAR image. (NOAA) and the National Ice Center use SAR im-
The reason why surface films are detectable on agery on an operational basis for sea ice monitoring,
radar images is that oil films have a dampening effect iceberg detection, fishing enforcement, oil spill de-
on short surface waves. Radar is remarkably sensitive tection, wind and storm information. In Canada, sea
to small changes in the roughness of sea surface. Oil ice surveillance is now a proven near-real-time op-
slicks also have a dark appearance in radar images eration, and new marine and coastal applications for

Wind boundary

Polar low

0 50
km

Figure 3 Radarsat ScanSAR image of a polar low in the Bering Sea collected on February 5, 1998. (^ CSA 1998.)
564 SATELLITE REMOTE SENSING SAR

SAR imagery are still emerging. In Europe, research detection, monitoring, containment, and clean up of
on SAR imaging of ocean waves has received great oil spills are crucial to the protection of the
attention in the past 10 years, and has contributed to environment.
better global ocean wave forecasts. However, the role Under favorable wind conditions with strong tidal
of SAR in the coastal observing system still remains current, the surface signature of bottom topography
at the research and development stage. For reference, in shallow water has often been observed in SAR
examples of some typical ocean features from SAR images. Figure 2, showing an ERS-1 SAR image of
applications are provided below. the shallow water bathymetry of Taiwan Tan col-
For marine environmental monitoring, features lected on July 27, 1994, is such an example. Taiwan
such as oil spills, bathymetry, and polar lows are Tan is located south west of Taiwan in the Taiwan
important for tracking and can often be identified Strait. Typical water depth there is around 30 m with
easily with SAR. In early November 1997, Radarsat’s a valley in the middle and the continental shelf break
SAR sensor captured an oil spill off Point Barrow, to the south. An extensive sand wave field (Figure 2)
Alaska. The oil slicks showed up clearly on the is developed at Taiwan Tan regularly by wind, tidal
ScanSAR imagery on November 2, 3 and 9. The oil current, and surface waves. Monitoring the changes
spill is suspected to be associated with the Alaskan of bathymetry is critical to ship navigation, especially
Oil Pipeline. Figure 1 shows a scene containing the in the areas where water is shallow and ship traffic is
oil slicks cropped out from the original ScanSAR heavy.
image for a closer look. Tracking oil spills using SAR Intense low pressure systems in polar regions,
is useful for planning clean-up activities. Early often referred to as polar lows, may develop above

la
insu
pen
ka N
Alas

Eddy

0 10
km

Figure 4 ERS-1 SAR image of lower Shelikof Strait, acquired on October 23, 1991 showing a spiral eddy. (^ ESA 1991.)
 SATELLITE REMOTE SENSING SAR 565

regions between colder ice/land and warmer ocean continental shelf and ultimately recruit to nearshore
during cold air outbreaks. These intense polar lows nursery areas. Features such as fronts and eddies can
are formed off major jet streams in cold air masses. retain larval patches within the shelf zone. Figure 4
Since they usually occur near polar regions where shows an ERS-1 SAR image acquired on October 23,
data are sparse, SAR images have been a useful tool 1991 (centered at 56.691N, 156.071W) in lower
for studying these phenomena. Figure 3 shows a Shelikof Strait, the Gulf of Alaska. In this image, an
Radarsat ScanSAR image of a polar low in the Bering eddy with a diameter of approximately 20 km is
Sea (centered at 58.01N, 174.91E) collected on Feb- visible due to low wind conditions. The eddy is
ruary 5, 1998. It has a wind boundary to the north characterized by spiraling curvilinear lines which are
spiraling all the way to the center of the storm that most likely associated with current shears, surface
separates the high wind (bright) area from the low films, and to a lesser extent temperature contrasts.
wind (dark) area. The rippled character along the SAR has the potential to locate these eddies over
wind boundary indicates the presence of an in- extensive areas in coastal oceans.
stability disturbance induced by the shear flow, A Radarsat ScanSAR image over the Gulf of
which in turn is caused by the substantial difference Mexico taken on November 23, 1997 (Figure 5)
in wind speed across the boundary. shows a distinct, nearly straight front stretching at
Ocean features such as eddies, fronts, and ice least 300 km in length. The center of the front in this
edges can result in changes in water temperature, scene is in the Gulf of Mexico some 400 km south
turbulence, or transport and may be the primary west of New Orleans. The frontal orientation is
determinant of recruitment to fisheries. The survival about 761 east of north. Closer inspection reveals
of larvae is enhanced if they remain on the that there are many surface film-like filaments on the

Front

0 25
km

Figure 5 Radarsat ScanSAR image collected over the Gulf of Mexico on November 23, 1997 showing a frontal boundary. (^ CSA
1997.)
566 SATELLITE REMOTE SENSING SAR

south side of the front. Concurrent wind data suggest clearly seen as the bright area because sea ice surface
that surface currents converge along the front. is rougher than ocean surface (dark area). In the
Therefore, the formation of this front is probably same image, a front is also visible and may or may
caused by the accumulation of natural surface films not be associated with the ice edge to the north. The
brought about by the convergence around the front. cold water near the ice edge dampens wave action
This example highlights SAR’s sensitivity to the and appears as a darker area compared with the
changes of wind speed and the presence of surface other side of the front, where it shows up as brighter
films. area due to higher wind and higher sea states.
The edge of the sea ice has been found to be highly Information on surface and internal waves, as well
productive for plankton spring bloom and fishery as ship wakes, are very important and valuable for
feeding. In the Bering Sea, fish abundance is highly marine surveillance and ship navigation. The prin-
correlated with yearly ice extent because for their cipal use of SAR for oceanographic studies has been
survival many species of fish prefer the cold pools left for the detection of ocean waves. The wave direction
behind after ice retreat. SAR images are very useful and height derived from SAR data can be incorpor-
for tracking the movement of the ice edge. Figure 6 ated into models of wind–wave forecast and other
shows a Radarsat ScanSAR image near the ice edge applications such as wave–current interaction. Fig-
in the Bering Sea collected on February 29, 2000 ure 7 shows an ERS-1 SAR image of long surface
(centered at 59.61N and 177.31W). The sea ice pack waves (or swells) in the lower Shelikof Strait col-
with ice bands extending from the ice edge can be lected on October 17, 1991 (centered at 56.111N,

Sea ice

Front

0 50
km

Figure 6 Radarsat ScanSAR image collected over the Bering Sea on February 29, 2000 showing a front near the ice edge. (^ CSA
2000.)
 SATELLITE REMOTE SENSING SAR 567

N 0 5
km

Surface waves

Figure 7 ERS-1 SAR image of lower Shelikof Strait, obtained on October 17, 1991 showing long surface gravity waves refracted by
current. (^ ESA 1991.)

156.361W). The location is close to that of the spiral in the transfer of energy from tides to ocean mixing.
eddy shown in Figure 4. Because of the higher winds These nonlinear internal waves are apparently gen-
and higher sea states at the time the image was ac- erated by internal mixing as tidal currents flow over
quired, the eddy is less conspicuous in this SAR bottom features and propagate in the open ocean. In
image taken 6 days earlier. Although the direct sur- the South China Sea near DongSha Island, enormous
face signature of the eddy cannot be discerned clearly westward propagating internal waves from the open
in this image, the wave refraction in the eddy area ocean are often confronted by coral reefs on the
can still be observed. The rays of the wave field can continental shelf. As a result, the waves are diffracted
be traced out directly from the SAR image. The ray upon passing the reefs. Figure 8 shows a Radarsat
pattern provides information on the wave refraction ScanSAR image collected over the northern South
pattern and on the relative variation of wave energy China Sea on April 26, 1998, showing at least three
along a ray through wave–current interaction. packets of internal waves. Each packet consists of a
Tidal currents flowing over submarine topo- series of internal waves, and the pattern of each wave
graphic features such as a sill or continental shelf in a is characterized by a bright band followed immedi-
stratified ocean can generate nonlinear internal ately by a dark band. The bright/dark bands indicate
waves of tidal frequency. This phenomenon has been the contrast in ocean rough/smooth surfaces caused
studied by many investigators. Direct observations by convergence/divergence areas induced by the in-
have lent valuable insight into the internal wave ternal waves. At times, the wave ‘crest’ as observed
generation process and explained the role they play by SAR from the length of bands can be over 200 km
568 SATELLITE REMOTE SENSING SAR

N
Internal waves

DongSha Island

Coral reefs

0 25
km

Figure 8 Radarsat ScanSAR image collected over the South China Sea on April 26, 1998 showing three internal wave packets.
(^ CSA 1998.)

long. After passing the DongSha coral reefs, the wake is clearly visible. The turbulent wake dampens
waves regroup themselves into two separate packets any short waves, resulting in an area with low back-
of internal waves. Later, they interact with each other scattering as indicated by the arrow A in Figure 9.
and emerge as a single wave packet again. SAR can Near the ship, the dark wake is accompanied by a
be a very useful tool for studying these shelf pro- bright line which may be caused by the vortex shed by
cesses and the effect of the internal waves on oil the ship into its wake. The ship track follows the busy
drilling platforms, nutrient mixing, and sediment shipping lane between Hong Kong, Taiwan, and
transport. Japan. The ambient dark slicks are natural surface
Ships and their wakes are commonly observable in films induced by upwelling on the continental shelf. In
high-resolution satellite SAR imagery. Detection of the lower part of the image, another ship turbulent
ships and ship wakes by means of remote sensing can wake (long and dark linear feature oriented east–west)
be useful in the areas of national defense intelligence, can be identified near the location B in Figure 9. A
shipping traffic, and fishing enforcement. Figure 9 is faint bright line connects to the end of this turbulent
an ERS-1 SAR image collected on May 31, 1995 near wake, forming a V-shaped wake in the box B. The
the northern coast of Taiwan. The image is centered at faintness of this second ship may be caused by very
25.621N and 121.151E, approximately 30 km off- low backscattering of the ship configuration or the
shore in the East China Sea. A surface ship heading wake could have been formed by a submarine. In the
north east, represented by a bright spot, can be easily latter case, it must have been operating very close
identified. Behind this ship, a long dark turbulence to the ocean surface, since the surface wake is
 SATELLITE REMOTE SENSING SAR 569

0 5
km

Figure 9 ERS-1 SAR image of East China Sea, obtained on May 31, 1995 showing a surface ship and its wake (arrow A) and a
V-shaped wake in box B. (^ ESA 1995.)

observable. The ship wake is pointing to the east, areas at high latitudes, the uniformly cold sea surface
indicating that the faint ship was moving from temperature and persistent cloud cover preclude
mainland China toward the open ocean. optical and infrared measurement of surface tem-
perature features, and obscure ocean color obser-
vations. The mapping of ocean features by SAR in
these challenging coastal regions is, therefore, a po-
Discussion tentially major application for satellite-based SAR,
As mentioned earlier, SAR has the unique capability particularly for the wider swath ScanSAR mode.
of operating during the day or night and under all Furthermore, SAR data provide unique information
weather conditions. With repeated coverage, space- for studying the health of the Earth system, as well
borne SAR instruments provide the most efficient as critical data for natural hazards and resource
means to monitor and study the changes in important assessments.
elements of the marine environment. As demonst- The prospect of SAR data collection extending
rated by the above examples, the use of SAR-derived well into the twenty-first century gives impetus to
observations to track eddies, fronts, ice edges, and oil current research in SAR applications in ocean science
slicks can supply valuable information and can aid in and opens the doors to change detection studies on
the management of the fishing industry and the decadal timescales. The next step is to move into the
protection of the environment. In overcast coastal operational use of SAR data to complement ground
570 SATELLITE REMOTE SENSING SAR

measurements. The challenge is to increase cooper- Beal RC and Pichel WG (eds) (2000) Coastal and Marine
ation in the scheduling, processing, dissemination, Applications of Wide Swath SAR. Johns Hopkins APL
and pricing of SAR data from all SAR satellites be- Technical Digest, 21.
tween international space agencies. Such cooperation European Space Agency (1995) Scientific Achievements of
might permit near-real-time high-resolution coastal ERS-1. ESA SP-1176/I.
Fu L and Holt B (1982) Seasat Views Oceans and Sea Ice
SAR measurements of sufficient temporal and spatial
with Synthetic Aperture Radar, JPL Publication,
coverage to impact weather forecasting for selected pp. 81--120. Pasadena, CA: NASA, JPL/CIT.
heavily populated coastal regions. It is necessary to Hsu MK, Liu AK, and Liu C (2000) An internal wave
bear in mind that each satellite image is a snapshot study in the China Seas and Yellow Sea by SAR.
and can be complemented with buoy and ship Continental Shelf Research 20: 389--410.
measurements. Ultimately, these data sets should be Liu AK, Peng CY, and Schumacher JD (1994) Wave–
integrated by numerical models. Such validated and current interaction study in the Gulf of Alaska for
calibrated models will prove extremely useful in detection of eddies by SAR. Journal of Geophysical
understanding a wide variety of oceanic processes. Research 99: 10075--10085.
Liu AK, Peng CY, and Weingartner TJ (1994) Ocean–ice
interaction in the marginal ice zone using SAR. Journal
See also of Geophysical Research 99: 22391--22400.
Liu AK, Peng CY, and Chang YS (1996) Mystery ship
Aircraft Remote Sensing. Satellite Altimetry. detected in SAR image. EOS, Transactions, American
Satellite Oceanography, History and Introductory Geophysical Union 77: 17--18.
Concepts. Satellite Passive-Microwave Liu AK, Peng CY, and Chang YS (1997) Wavelet analysis
Measurements of Sea Ice. Satellite Remote of satellite images for coastal watch. IEEE Journal of
Sensing Microwave Scatterometers. Satellite Oceanic Engineering 22: 9--17.
Remote Sensing of Sea Surface Temperatures. Tsatsoulis C and Kwok R (1998) Analysis of SAR Data of
the Polar Oceans. Berlin: Springer-Verlag.

Further Reading
Alaska SAR Facility User Working Group (1999) The
Critical Role of SAR in Earth System Science. (http://
www.asf.alaska.edu/)
 SATELLITE REMOTE SENSING: OCEAN COLOUR
C. R. McClain, NASA Goddard Space Flight Center, example, chlorophyll a concentration and the diffuse
Greenbelt, MD, USA attenuation coefficient. The motivation for space-
& 2009 Elsevier Ltd. All rights reserved. borne observations of this kind lies in the need for
frequent high-resolution spatial measurements of
these geophysical parameters on regional and global
scales for addressing both research and operational
requirements associated with marine primary pro-
Introduction duction, ecosystem dynamics, fisheries management,
The term ‘ocean color’ refers to the spectral com- ocean dynamics, and coastal sedimentation and
position of the visible light field that emanates from pollution, to name a few. The first proof-of-concept
the ocean. The color of the ocean depends on the satellite ocean color mission was the Coastal Zone
solar irradiance spectra, atmospheric conditions, Color Scanner (CZCS) on the Nimbus-7 spacecraft
solar and viewing geometries, and the absorption which was launched in the summer of 1978. The
and scattering properties of water and the substances CZCS was intended to be a 1-year demonstration
that are dissolved and suspended in the water col- with very limited data collection, ground processing,
umn, for example, phytoplankton and suspended and data validation requirements. However, because
sediments. Water masses whose reflectance is deter- of the extraordinary quality and unexpected utility of
mined primarily by absorption due to water and the data for both coastal and open-ocean research,
phytoplankton are generally referred to as ‘case 1’ data collection continued until June 1986 when the
waters. In other situations, where scattering is the sensor ceased operating. The entire CZCS data set
dominant process or where absorption is dominated was processed, archived, and released to the research
by substances other than phytoplankton or their community by 1990. As a result of the CZCS ex-
derivatives, the term ‘case 2’ is applied. perience, a number of other ocean color missions
The primary optical variable of interest for remote have been launched, for example, the Ocean Color
sensing purposes is the water-leaving radiance, Lw, and Temperature Sensor (OCTS; Japan; 1996–97),
that is, the subsurface upwelled radiance (light the Sea-viewing Wide Field-of-view Sensor (Sea-
moving upward in the water column) propagating WiFS, 1997 and continuing; US), and two Moderate
through the air–sea interface, but not including the Resolution Imaging Spectroradiometer (MODIS on
downwelling irradiance (light moving downward the Terra spacecraft, 2000 and continuing; MODIS
through the atmosphere) reflected at the interface. To on the Aqua spacecraft, 2002 and continuing; US),
simplify the interpretation of ocean color, measure- with the expectation that continuous global obser-
ments of the water-leaving radiances are normalized vations will be maintained in the future.
by the surface downwelling irradiances to produce
‘remote sensing’ reflectances, which provide an un-
Ocean Color Theoretical and
ambiguous measure of the ocean’s subsurface optical
signature. Clear open-ocean reflectances have a Observational Basis
spectral peak at blue wavelengths because water Reflectance can be defined in a number of ways and
absorbs strongly in the near-infrared (NIR) and the relationship between the various quantities can be
scatters blue light more effectively than at longer confusing. The most common definition is irradiance
wavelengths. As the concentrations of microscopic reflectance, R, just below the surface, as given in eqn
green plants (phytoplankton) and suspended ma- [1], where Eu and Ed are the upwelling and down-
terials increase, absorption and scattering reduce the welling irradiances, respectively, and the superscript
reflectance at blue wavelengths and increase the re- minus sign implies the value just beneath the surface:
flectance at green wavelengths, that is, the color
shifts from blue to green and brown. This spectral Eu ðl; 0 Þ
shift in reflectance can be quantified and used to es- RðlÞ ¼ ½1
Ed ðl; 0 Þ
timate concentrations of optically active components
such as chlorophyll a. In general, irradiance and radiance are functions of
The goal of satellite ocean color analysis is to ac- depth (or altitude in the atmosphere) and viewing
curately estimate the water-leaving radiance spectra geometry with respect to the sun. R has been theo-
in order to derive other geophysical quantities, for retically related to the absorption and scattering

571
572 SATELLITE REMOTE SENSING: OCEAN COLOUR

properties of the ocean as in eqn [2], where Q is the Lw ðlÞ

Rrs ðlÞ ¼ ½4
ratio of Eu(l,0) divided by the upwelling radiance, Ed ðl; 0þ Þ
Lu(l,0), c1 ¼ 0.094 9, c2 ¼ 0.079 4, bb(l) is the
backscattering coefficient, and a(l) is the absorption and
coefficient.
Lw ðlÞ
Lwn ðlÞ ¼ Fo ðlÞ ½5
 i Ed ðl; 0þ Þ
X
2
bb ðlÞ
RðlÞ ¼ QðlÞ ci ½2 The term (1r)/n2 is approximately equal to 0.54.
i¼1
aðlÞ þ bb ðlÞ
Furthermore, it can be shown that Lwn(l) is related
Both bb(l) and a(l) represent the sum of the contri- to R(l) by eqn [6] where the /rS and /rS denote
butions of the various optical components (water, the angular mean Fresnel reflectances for down-
inorganic particulates, dissolved substances, phyto- welling irradiance above the surface and irradiance
plankton, etc.) which are often specified explicitly. If below the surface, respectively, and depend on the
the angular distribution of Eu were directionally angular distributions of those irradiances:
uniform, that is, Lambertian, Q would equal p.   
However, the irradiance distribution is not uniform 1  r 1  hri RðlÞ
Lwn ðlÞ ¼ Fo ðlÞ ½6
and is dependent on a number of variables. Some n2 1  Rhri QðlÞ
experimental results indicate that Q is roughly 4.5 Combining eqns [2] and [6] for case 1 waters yields
and, to a first approximation, independent of wave-
length. In case 1 water, the approximation,   
f ðlÞ bb ðlÞ
R(l)Bf[bb(l)/a(l)] is often used and f is assigned a Lwn ðlÞEFo ðlÞRðlÞ ½7
QðlÞ aðlÞ
constant value of 0.33. In reality, f is wavelength-
dependent. f and Q are also functions of the solar R is the product of the two bracketed terms in eqn
zenith and viewing angles. [6] and is largely dependent on surface roughness
When concentrations of substances, for example, (wind speed). f/Q is usually given in tables derived
chlorophyll a, are measured, the coefficients can be from Monte Carlo simulations.
specified in terms of the concentrations, for example, The actual surface reflectance includes not only
aj ðlÞ ¼ aj* ðlÞ½chl a where aj(l) and aj* ðlÞ are the Lw, but also the Fresnel reflection of photons not
phytoplankton absorption coefficient and specific scattered by the atmosphere (the direct component),
absorption coefficient, respectively, and [chl a] is the photons that are scattered by the atmosphere (sky-
chlorophyll a concentration. The absorption co- light or the indirect component), and light reflected
efficients are designated for phytoplankton rather off whitecaps. The angular distribution of total re-
than chlorophyll a because the actual absorption by flection broadens as wind speed increases and the sea
living cells can vary substantially for a fixed amount surface roughens, and the total surface whitecap
of chlorophyll a. Thus, R(l) can be expressed in coverage also increases with wind speed. For clear
terms of the specific absorption and scattering co- case 1 waters, that is, areas where the chlorophyll a is
efficients and pigment and particle concentrations. less than about 0.25 mg chl a/m3, the Lwn spectrum is
When R(l) is observed at a sufficient number of fairly constant, for example, Lwn(550)B0.28 mW/
wavelengths across the spectrum, the R(l) values can cm2  mm  sr, and the values are referred to as ‘clear
be inverted to provide estimates of the scattering and water normalized water-leaving radiances’.
absorption coefficients and pigment concentrations. Another optical parameter of interest is the diffuse
For satellite applications, the reflectances and ra- attenuation coefficient for upwelled radiance, KLu ,
diances just above the surface are more appropriate (eqn [8]):
to use than R(l). Therefore, water-leaving radiance,


remote sensing reflectance, Rrs(l), and normalized 1 dLu ðl; zÞ
KLu ðl; zÞ ¼  ½8
water-leaving radiance, Lwn(l), are commonly used. Lu ðl; zÞ dz
These are defined in eqns [3]–[5], respectively, where
r is the Fresnel reflectance of the air–sea interface, n Near the surface, where optical constituents are
is the index of refraction of seawater, Fo(l) is the relatively uniform and KLu is constant,
extraterrestial solar irradiance, and the plus sign
denotes the value just above the surface: Lu ðl; zÞ ¼ Lu ðl; 0 ÞeKLu ðlÞz ½9

  Similar K relationships hold for irradiance. For

1r convenience, K(l) will be used to denote KLu ðlÞ.
Lw ðlÞ ¼ Lu ðl; 0 Þ ½3
n2 The various upwelling and downwelling diffuse
 SATELLITE REMOTE SENSING: OCEAN COLOUR 573

2.5 0.4

Solar irradiance Solar irradiance (mw/cm2/nm)

2 0.3 Water absorption (I/m)

Absorption
Phytoplankton absorption (I/m)
Gelbstoff absorption (I/m)
1.5 0.2

1 0.1
Phytoplankton absorption
at 10 mg/m3 chlorophyll
0.5 0 ag(440) = 0.04, S = 0.14
350 400 450 500 550 600 650 700
Wavelength (nm)

Figure 1 Solar irradiance, chlorophyll specific absorption, yellow substance, and water absorption spectra.

attenuation coefficients for radiance and irradiance obtained, atmospheric correction and bio-optical
are commonly interchanged, but, strictly speaking, algorithms will improve and replace previous ver-
they are different quantities and are not equal. sions, and new data products will be defined by the
Key to the ocean color measurement technique are research community. Therefore, flight projects, such
the relationships between the solar irradiance, water as SeaWiFS, are prepared to periodically reprocess
absorption, and chlorophyll a absorption spectra, their entire data set. In fact, the SeaWiFS data was
chlorophyll a being the primary chemical associated reprocessed 5 times in the first 8 years. Figure 2
with photosynthesis. The solar spectrum peaks at provides a graphical depiction of all components of
blue wavelengths which correspond to the maximum the satellite measurement scenario, each aspect of
transparency of water and the peak in chlorophyll a which is discussed below.
absorption. Thus, phytoplankton photosynthesis
is tuned to the spectral range of maximum light.
Sensor Design and Performance
Figure 1 provides spectra for Fo(l), aj* ðlÞ, ocean
water absorption (aw(l)), and colored dissolved or- Sensor design and performance characteristics en-
ganic matter (ag(l); CDOM, also known as yellow compass many considerations which cannot be elab-
substance, gelbstoff, and gilvin). The spectrum for orated on here, but are essential to meeting the overall
CDOM has the form ag(l) ¼ ag(440) exp[S(l–440)], measurement accuracy requirements. Radiometric
where S ranges from 0.01 to 0.02 nm1 . There are factors include wavelength selection, bandwidth, sat-
other optical constituents besides phytoplankton and uration radiances, signal-to-noise ratios (SNRs), po-
CDOM, such as sediments, which are particularly larization sensitivity, temperature sensitivity, scan
important in coastal regions. angle dependencies (response vs. scan, RVS), stray-
light rejection, out-of-band contamination, field of
view (spatial resolution), band co-registration, and a
number of others, all of which must be accurately
Satellite Ocean Color Methodology
quantified (characterized) prior to launch and in-
In order to obtain accurate estimates of geophysical corporated into the data-processing algorithms. Other
quantities such as chlorophyll a and K(490) from design features may include sensor tilting for sunglint
satellite measurements, a number of radiometric avoidance and capacities for tracking the sensor sta-
issues must be addressed including (1) sensor design bility on-orbit (e.g., internal lamps, solar diffusers,
and performance; (2) postlaunch sensor calibration and lunar views), as instruments generally lose sensi-
stability; (3) atmospheric correction, that is, the re- tivity over time due to contamination of optical
moval of light due to atmospheric scattering, at- components, filter and detector degradation, etc.
mospheric absorption, and surface reflection; and (4) Also, there are a variety of spacecraft design criteria
bio-optical algorithms, that is, the transformation of including attitude control for accurate navigation,
Rrs(l) or Lwn(l) values into geophysical parameter power (solar panel and battery capacities), onboard
values. Items 24 represent developments that pro- data storage capacity, telemetry bandwidth (com-
gress over time during a mission. For example, over mand uplink and downlink data volumes, transmis-
the first 9 years of operation, the radiometric sensi- sion frequencies, and ground station compatibility
tivity of SeaWiFS degraded by as much as 18% in and contact constraints), and real-time data broadcast
one of the NIR bands. Also, as radiative transfer (high-resolution picture transmission (HRPT) station
theory develops and additional optical data are compatibility). Depending on the specifications,
574 SATELLITE REMOTE SENSING: OCEAN COLOUR

SeaWiFS Moon

Sun
F0()

Ozone

Aerosols

Key: Air molecule

O2 molecule
O3 molecule
Particulate
Water molecule
Phytoplankton

Light pathway
Calibration only
Details of multi-
ple interactions

Figure 2 Depiction of the various sensor, atmospheric, and oceanic optical processes which must be accurately accounted for in-
satellite ocean color data processing.

which must be accurately described and linked to the high-altitude geostationary platforms are being con-
final data product accuracy requirements, instruments sidered. Sun-synchronous orbits provide global cov-
can be built in a variety of ways to optimize per- erage as the Earth rotates so the data is collected at
formance. For example, the CZCS used a grating about the same local time, for example, local noon.
for spectral separation, while SeaWiFS and MODIS Typically, the satellite circles the Earth about every
use filters. Along with the specifications, the testing 90–100 min (14–15 orbits/day). Multiple views for a
procedures and test equipment must be carefully de- given day are possible only at high latitudes where
signed to ensure that the characterization data ad- the orbit tracks converge. Figure 3 shows the daily
equately reflect the true sensor performance. Sensor global area coverage (GAC) of SeaWiFS. The gaps
characterization can be a technically difficult, time- between the swaths are filled on the following day as
consuming, and expensive process. Table 1 provides a the ground track pattern progressively shifts. The
summary of past, present, and planned ocean color SeaWiFS scan is 7571, but the GAC is truncated to
sensors and certain observational attributes. It is im- 7451 and subsampled (every fourth pixel and line)
portant to note that only the two MODIS instruments on the spacecraft. The full-resolution local area
have a common design, although their behaviors on coverage (LAC, 1.1 km at nadir) is broadcast as
orbit were significantly different. HRPT data and includes the entire swath at full
To date, all ocean color missions, both previous resolution. Note that as the scan angle increases,
and currently approved, have been designed for low- pixels (the area on the surface being viewed) become
altitude Sun-synchronous orbits, although sensors on much larger and the atmospheric corrections are less
 SATELLITE REMOTE SENSING: OCEAN COLOUR 575

Table 1 Summary of satellite ocean color sensor and orbit characteristics

Mission instrument Visible bands Resolution Swath Tilt (deg) Onboard Orbit node
(launch yr) (km) (km) calibration (inclination)

Nimbus-7 CZCS 443, 520, 550, 670, 0.8 1600 0, 720 Internal lamps 12:00 noon
(1978) 750 11.5 mm (ascending)
ADEOS-I OCTS 412, 443, 490, 520, 0.7 1400 0, 720 Solar 10:30 am
(1996) 565, 670, 765, (descending)
865 nm
Internal lamps
ADEOS-I (1996) 443, 490, 565, 670, 6.2 2471 N/A None 10:30 am
ADEOS-II (2002) 763, 765, 865, (descending)
POLDER 910 nm
OrbView-2 SeaWiFS 412, 443, 490, 510, 1.1 (LAC) 4.5 2800 0, 720 Solar diffuser 12:00 noon
(1997) 555, 670, 765, (GAC: (LAC) (descending)
865 nm subsampled) 1500 Lunar imaging
(GAC)
Terra (2000) Aqua 412, 443, 488, 531, 1.0 1500 None Solar diffuser 10:30
(2002) MODIS 551, 667, 678, 748, (with stability (ascending)
869 nm monitor)
Spectral
radiometric
calibration
assembly
Envisat-1 MERIS 412, 443, 490, 510, 0.3 (LAC) 1.2 1150 None Solar diffusers 10:30
(2002) 560, 620, 665, 681, (GAC) (3) (descending)
709, 754, 760, 779,
870, 890, 900 nm
ADEOS-II GLI (2002) 400, 412, 443, 460, 1.0 1600 0, 720 Solar diffuser 10:30
490, 520, 545, 565, Internal lamps (descending)
625, 666, 680, 710,
749, 865 nm

Some sensors have more channels than are indicated which are used for other applications. There are a number of other ocean color
missions not listed because the missions are not designed to routinely generate global data sets. Instruments not mentioned in the text
are the Polarization and Directionality of the Earth’s Reflectances (POLDER) and the Global Imager (GLI).

from 201 to þ 201 to avoid viewing into the sun-

glint. The tilt operation is staggered on successive
days in order to ensure every-other-day coverage
of the gap. The MODIS and Medium Resolution
Imaging Spectrometer (MERIS) do not tilt.
Geostationary orbits, that is, orbits having a fixed
subsatellite (nadir) point on the equator, only allow
hemispheric coverage with decreased spatial reso-
lution away from nadir, but can provide multiple
Figure 3 An example of the daily global area coverage (GAC) views each day. Multiple views per day allow for the
from SeaWiFS. The SeaWiFS scan extends to 7571 and evaluation of tidal and other diurnal time-dependent
samples at about a 1-km resolution. This data is the LAC and biases in sampling to be evaluated, and also provide
is continuously broadcast for HRPT station reception. The GAC more complete sampling of a given location as cloud
data are subsampled at every fourth pixel and line over only the
patterns change.
center 7451 of the scan. GAC data are stored on the spacecraft
and downlinked to the NASA and Orbimage Corporation (the
company that owns SeaWiFS) ground stations twice per day.
Postlaunch Sensor Calibration Stability
reliable. MODIS has a swath similar to SeaWiFS The CZCS sensitivity at 443 nm changed by about
LAC, all of which is recorded and broadcast real- 40% during its 7.7 years of operation. Even the
time at full resolution. The data gaps in each swath relatively small changes noted above for SeaWiFS
about the subsolar point are where SeaWiFS is tilted undermine the mission objectives for global change
576 SATELLITE REMOTE SENSING: OCEAN COLOUR

research because they introduce spurious trends in Once sensor degradation is removed, field
the derived products. Quantifying changes in the measurements can be used to adjust the calibration
sensor can be very difficult, especially if the changes gain factors so the satellite retrievals of Lwn match
are gradual. In the case of the CZCS, there was independent radiometric field observations, the
no ongoing comprehensive validation program so-called ‘vicarious’ calibration (meaning that the
after its first year of operation, because the mission vicarious calibration replaces the original laboratory-
was a proof-of-concept. Subsequent missions have based prelaunch calibration), and involves the
some level of continuous validation. In the case of atmospheric correction (discussed below). This ad-
SeaWiFS, a combination of solar, lunar, and field justment is necessary because the prelaunch cali-
observations (oceanic and atmospheric) are used. bration is only accurate to within about 3%, the
The solar measurements are made daily using a solar sensor can change during launch and orbit raising,
diffuser to detect sudden changes in the sensor. The and any biases in the atmospheric correction can be
solar measurements cannot be used as an absolute removed in this way. The Marine Optical Buoy
calibration because the diffuser reflectance gradually (MOBY) was developed and located off Lanai,
changes over time. MODIS has a solar diffuser sta- Hawaii, to provide the SeaWiFS and MODIS vi-
bility monitor which accounts for diffuser degra- carious calibration data. Given cloud cover, sun glint
dation. SeaWiFS lunar measurements are made once contamination, and satellite sampling frequency at
a month at a fixed lunar phase angle (71) using a space- Hawaii, it can take up to 3 years to collect enough
craft pitch maneuver that allows the sensor to image high-quality comparisons (25–40 depending on
the Moon through the Earth-viewing optics. This the sensor and the wavelength), to derive statistically
process provides an accurate estimate of the sensor stable gain correction factors which should be in-
stability relative to the first lunar measurement dependent of the sensor-viewing geometry and the
(Figure 4). The data do require a number of cor- solar zenith angle. It is important to note that in-
rections, for example, Sun–Moon distance, satellite– water measurements cannot help in accessing biases
Moon distance, and lunar libration variations. The in the NIR band calibrations. Atmospheric mea-
lunar measurements are not used for an absolute surements of optical depth and other parameters
calibration because the moon’s surface reflectance is are needed to evaluate the calibration of these
not known to a sufficient accuracy. wavelengths.

SeaWiFS lunar calibrations

1.00

0.95
Normalized radiance

0.90

Band 1 Band 2
Band 3 Band 4
Band 5 Band 6
0.85
Band 7 Band 8

0 500 1000 1500 2000 2500 3000

Days since first image

Figure 4 SeaWiFS degradation as observed in the monthly lunar calibration data.

 SATELLITE REMOTE SENSING: OCEAN COLOUR 577

Atmospheric Correction Note that radiance lost to each absorbing gas is esti-
Solar irradiance propagates through the atmosphere mated separately and added to the satellite radiance
where it is attenuated by molecular (Rayleigh) and to yield Lt. The Rayleigh radiances can be estimated
aerosol scattering and absorption. Rayleigh scattering theoretically and the aerosol radiances are usually
can be calculated theoretically with a high degree of inferred from NIR wavelengths using aerosol models
accuracy. Aerosol scattering and absorption are much and the assumption that Lw is negligible. Determining
more difficult to estimate because their horizontal and values for all terms on the right side of eqn [10], with
vertical distributions are highly variable, as are their the exception of Lw, constitutes the ‘atmospheric
absorption and scattering properties. The estimation correction’ which allows eqn [10] to be solved for Lw.
of the aerosol effects on the upwelling radiance at the As mentioned earlier, if Lw is known, then Lt can be
top of the atmosphere is one of the most difficult as- adjusted to balance eqn [10] to derive a ‘vicarious’
pects of satellite remote sensing. Ozone (O3) is the calibration.
primary absorbing gas that must be considered. For-
Bio-Optical Algorithms
tunately, ozone is concentrated in a thin band near the
top of the atmosphere and its global distribution is Bio-optical algorithms are used to define relation-
mapped daily by sensors such as the Total Ozone ships between the water-leaving radiances or re-
Mapping Spectrometer (TOMS), Total Ozone Ver- flectances and constituents in the water. The basis for
tical Sounder (TOVS), and the Ozone Monitoring the chlorophyll a algorithm is the change in reflec-
Instrument (OMI) which have been deployed sepa- tance spectral slope with increasing concentration,
rately on various spacecraft. Continuous global sat- that is, as chlorophyll a increases, the blue end of
ellite ozone measurements have been made since 1978 the spectrum is depressed by pigment absorption and
when the first TOMS was flown with the CZCS on the red end is elevated by increased particle scat-
Nimbus-7. The other absorbing gases of interest are tering. In case 1 waters, R(510) shows little variation
O2 which has a strong absorption band (the A-band) with chlorophyll a and is alluded to as the ‘hinge
between 750 and 770 nm and NO2 which absorbs point’. Algorithms can be strictly empirical (sta-
ultraviolet (UV) and blue light. O2 is well mixed in the tistical regressions) or semi-empirical relationships,
atmosphere and only requires atmospheric pressure that is, based on theoretical expressions using
data for correction. NO2 is highly variable and re- measured values of certain optical variables. For
quires global concentration data from satellite sensors example, the empirical chlorophyll a (chl a) rela-
such as OMI. tionship, OC4v4, being used by the SeaWiFS mission
Light that reaches the surface is either reflected or is depicted in Figure 5 and is expressed in eqn [11],
penetrates through the air–sea interface. Simple where R ¼ log10[max(Rrs(443), Rrs(490), Rrs(510)/
Fresnel reflection off a flat interface is easily computed Rrs(555))]:
theoretically. However, once the surface is wind-
roughened and includes foam (whitecaps), the esti- chl a ¼ 100:3663:067Rþ1:930R
2
þ0:649R3 1:532R4
½11
mation of the reflected light is more complex and
empirical relationships must be invoked. Only a small This relationship was based on observed Rrs and
percentage of the light that enters the water column is chlorophyll a data from many locations and, there-
reflected upward back through the air–sea interface in fore, is not optimized to a particular biological or
the general direction of the satellite sensor. Of that optical regime, that is, a bio-optical province.
light, only a fraction makes its way back through the
atmosphere into the sensor. Lw accounts for no more Product Validation
than about 15% of the top-of-the-atmosphere radi-
ance, Lt. Therefore, each process must be accurately Validation of the derived products can be ap-
accounted for in estimating Lw. The radiances asso- proached in several ways. The most straightforward
ciated with each process are additive, to the first approach is to compare simultaneous field and sat-
order, and can be expressed as eqn [10] where the ellite data. Another approach is to make statistical
subscripts denote ‘total’ (t), ‘Rayleigh’ (r), ‘aerosol’ comparisons, for example, frequency distributions,
(a), ‘Rayleigh–aerosol interaction’ (ra), ‘sun glint’ (g), of large in situ and satellite data sets. Simultaneous
and ‘foam’ (f), T(l) is the direct transmittance, and comparisons can provide accurate error estimates,
t(l) is the diffuse transmittance: but typically only about 15% of the observations
result in valid matchups mainly because of cloud
cover, sun glint, spatial inhomogeneities, and time
Lt ðlÞ ¼ Lr ðlÞ þ La ðlÞ þ Lra ðlÞ þ TðlÞLg ðlÞ
differences. Figure 6 represents all comparisons for
þ tðlÞ½Lf ðlÞ þ Lw ðlÞ ½10 the first 9 years of SeaWiFS. Statistical comparisons
578 SATELLITE REMOTE SENSING: OCEAN COLOUR

100

OC4v4

10

Chl a (mg m−3)

0.1

Relative frequency n = 2804

0 5 10 15
0.01
0.1 1 10
443 490 510
max(Rrs 555, Rrs 555, Rrs 555)
Clear
water

Figure 5 An empirical chlorophyll a algorithm, OC4v4, based on a regression of in situ chlorophyll a vs. Rrs ratio observations. In this
algorithm, the exponent numerator is the Rrs that has the greatest value of the three shown.

of cumulative data sets allow utilization of much common is the use of filters, rather than prisms or
more data, but can be subject to sampling biases. gratings, to define the spectral bands. In some situ-
Differences in field measurements and satellite esti- ations, the design differences can be used to identify
mates can be due to a number of sources including problems in the data products. This point is illus-
erroneous satellite estimations of Lw, and inaccurate trated by two examples, polarization sensitivity and
in situ values. In the early 1990s, to minimize in situ noise characteristics.
measurement errors, the SeaWiFS project initiated a SeaWiFS has a polarization scrambler, which re-
calibration comparison program, the development duces the sensor polarization sensitivity to B0.25%.
and documentation of in situ measurement proto- SeaWiFS also incorporates a rotating telescope rather
cols, and a number of technology development than a mirror which reduces the amount of polar-
activities. These were continued under the Sensor ization introduced by the optical surfaces. MODIS
Intercomparison and Merger for Biological and has a rotating mirror and a polarization sensitivity of
Interdisciplinary Oceanic Studies (SIMBIOS) pro- several percent. Because the Rayleigh radiance is
gram until 2003. Much of the documentation from highly polarized, errors in the polarization charac-
SeaWiFS and SIMBIOS is available online at http:// terization can introduce substantial errors in the Lwn
oceancolor.gsfc.nasa.gov/DOCS/. values. In fact, an error in the MODIS polarization
An important aspect of satellite data validation is tables did initially cause large regional differences
the comparison of data from different missions to (450%) between the SeaWiFS and MODIS Lwn
assess the level of agreement in the derived products, values. This difference prompted a review of the
especially the Lwn and chlorophyll a values. Because MODIS calibration inputs resulting in the error
every sensor has a different design and, therefore, being corrected, greatly reducing the regional and
different sensitivities to the host of attributes listed seasonal differences in the SeaWiFS and MODIS
earlier, identifying the causes for any differences in products.
the Lwn values can be difficult. Good examples are While the MODIS and SeaWiFS mean Lwn and
SeaWiFS and MODIS, which have very different chlorophyll values are quite comparable, the
designs. The only design feature these sensors have in MODIS products have much lower variances (less
 SATELLITE REMOTE SENSING: OCEAN COLOUR 579

(a)
4.5 4.5
Lwn412 Lwn443

3.0 3.0

1.5 1.5

0.0 0.0

0.0 1.5 3.0 4.5 0.0 1.5 3.0 4.5

4.5 4.5
Lwn490 Lwn510

3.0 3.0

1.5 1.5

0.0 0.0

0.0 1.5 3.0 4.5 0.0 1.5 3.0 4.5

4.5 Lwn555 2.25 Lwn670

3.0 1.50

1.5 0.75

0.0 0.00

0.0 1.5 3.0 4.5 0.00 0.75 1.50 2.25

(b)
102
Chlorophyll a

101

100

10−1

10−2 −2
10 10−1 100 101 102

Figure 6 A comparison of simultaneous SeaWiFS-derived and in situ (a) Lwn and (b) chlorophyll a data. Only about 15% of the field
data collected are used in the comparisons, because many of the data are excluded due to cloud cover, sun glint, time difference, and
other rejection criteria. The dashed line is the least squares fit to the data.
580 SATELLITE REMOTE SENSING: OCEAN COLOUR

noise) providing a more reasonable estimate of variability. First, MODIS has higher SNRs. Also,
sampling variability in the global binned fields, that the MODIS global data set has about 20 times
is, data merged over on a fixed grid for various more pixels (samples) due to the SeaWiFS GAC
lengths of time (daily, weekly, monthly, etc.). There subsampling. The SeaWiFS subsampling allows small
are several reasons for SeaWiFS having higher clouds to escape detection in the GAC processing in

Autumn 1997−2005 Winter 1997−2006

Spring 1998−2006 Summer 1998−2005

Chlorophyll a concentration (mg m−3)

0.01 0.03 0.1 0.3 1 3 10 30 60

Figure 7 Seasonal average chlorophyll a concentrations from the 9 years of SeaWiFS operation. The composites combine all
chlorophyll a estimates within 9 km square ‘bins’ obtained during each 3-month period. A variety of quality control exclusion criteria are
applied before a sample (pixel) value is included in the average.

Jan. 1998

Jul. 1998

>0.01 0.02 0.03 0.05 0.1 0.2 0.3 0.5 1 2 3 5 10 15 20 30 50

Ocean: chlorophyll a concentration (mg m−3)

Figure 8 A comparison of the monthly average chlorophyll a concentrations at the peaks of the 1997 El Niño and the 1998 La Niña in
the equatorial Pacific Ocean.
 SATELLITE REMOTE SENSING: OCEAN COLOUR 581

which case stray light is uncorrected (stray light is aerosol correction. Undetected clouds in the GAC
scattered light within the instrument that contami- data and digitization truncation are thought to be the
nates measurements in adjacent pixels), thereby ele- primary reasons for ‘speckling’ in the SeaWiFS-
vating the Lt values. Finally, the SeaWiFS data is derived products. One strength of the SeaWiFS de-
truncated from 12 to 10 bits on the data recorder tector array or focal plane design is the bilinear gain
resulting in coarser digitization, especially in the NIR which prevents bright pixels from saturating any
bands where the SNRs are relatively low. Noise can band. The prelaunch characterization data provided
cause ‘jitter’ in the aerosol model selection which enough information for a stray light correction al-
amplifies the variability in visible Lwn values via the gorithm to be derived. This correction works well in

100° W 98° W 96° W 94° W 92° W 90° W 88° W 86° W

2° N 2° N

0 0

2° S 2° S

4° S 4° S

100° W 98° W 96° W 94° W 92° W 90° W 88° W 86° W

9 May 1998

100° W 98° W 96° W 94° W 92° W 90° W 88° W 86° W

2° N 2° N

0 0

2° S 2° S

4° S 4° S

100° W 98° W 96° W 94° W 92° W 90° W 88° W 86° W

24 May 1998

>0.01 0.02 0.05 0.1 0.2 0.5 1 2 5 10 20 50

SeaWiFS chlorophyll a concentration (mg m−3)

Figure 9 Mesoscale temporal and spatial chlorophyll a variability around the Galapagos Islands in the eastern equatorial Pacific
Ocean before and after the sudden onset of the 1998 La Niña. The sea surface temperature near the islands dropped nearly 10 1C
between 9 and 24 May.
582 SATELLITE REMOTE SENSING: OCEAN COLOUR

the LAC data processing and for correcting the ef- The 1998 transition from El Niño to La Niña
fects of large bright targets in the GAC. On the other occurred very rapidly. Figure 9 compares the
hand, the MODIS NIR bands saturate over bright chlorophyll a concentrations around the Galapagos
targets including the Moon. Islands in early and late May. During the time be-
tween these two high-resolution scenes, ocean tem-
Satellite Ocean Color Data Sets and peratures around the islands dropped by about
10 1C. The chlorophyll a concentrations jumped
Applications
dramatically as nutrient-rich waters returned and
In this section, examples of satellite ocean color data phytoplankton populations could grow unabated in
products are presented to illustrate some research the absence of zooplankton grazers.
applications. These include (1) an annual seasonal The final example illustrates that some phyto-
cycle of global chlorophyll a, (2) interannual vari- plankton have special optical properties which
ability due to the El Niño–Southern Oscillation allows them to be uniquely identified. Figure 10, a
(ENSO) cycle in the equatorial Pacific, (3) mesoscale composite of three Rayleigh-corrected SeaWiFS
variability (scenes from the Galápagos Islands), bands, shows an extensive bloom of coccolitho-
and (4) blooms of special phytoplankton groups phores in the Bering Sea. In their mature stage of
(coccolithophores). development, coccolithophores shed calcite platelets,
Plant growth in the ocean is regulated by the supply which turn the water a milky white. Under these
of macronutrients (e.g., nitrate, phosphate, and sili- conditions, algorithms for reflectance (eqn [2]) and
cate), micronutrients (iron, in particular), light, and chlorophyll a concentration (eqn [10]) are not valid.
temperature. Light is modulated by cloud cover and However, the anomalously high reflectance allows
time of year (solar zenith angle). Nutrient supply for their detection and removal from spatial and
and temperature are determined by ocean circulation temporal averages of the satellite-derived products.
and mixing, especially the vertical fluxes, and heat Since coccolithophores are of interest for a number
exchange with the atmosphere. Figure 7 provides of ecological and biogeochemistry pursuits, satellite
seasonal average chlorophyll a concentrations derived ocean color data can be used to map the temporal
from SeaWiFS. Areas such as the North Atlantic show and spatial distribution of these blooms. In the case
a clear seasonal cycle. The seasonality in the North of the Bering Sea, the occurrence of coccolithophores
Atlantic is the result of deep mixing in the winter, had been rare prior to 1997 when the bloom
which renews the surface nutrient supply because the
deeper waters are a reservoir for nitrate and other
macronutrients. Once illumination begins to increase
in the spring, the mixed layer shallows providing a
well-lit, nutrient-rich surface layer ideal for phyto-
plankton growth. A bloom results and persists into
the summer until zooplankton grazing and nutrient
depletion curtail the bloom.
Figure 8 depicts the effects of El Niño and La Niña
on the ecosystems of the equatorial Pacific during
1997–98. Under normal conditions, the eastern
equatorial Pacific is one of the most biologically
productive regions in the world ocean as westward
winds force a divergent surface flow resulting in
upwelling of nutrient-rich subsurface water into the
euphotic zone, that is, the shallow illuminated layer
where plant photosynthesis occurs. During El Niño,
warm, nutrient-poor water migrates eastward from
the western Pacific and replaces the nutrient-rich
water, resulting in a collapse of the ecosystem.
Eventually, the ocean–atmosphere system swings
back to cooler conditions, usually to colder than
normal ocean temperatures, causing La Niña. The
Figure 10 A true color depiction of a coccolithophore, Emiliania
result is an extensive bloom which eventually de- huxleyi, bloom in the Bering Sea. The true color composite is
clines to more typical concentrations as the atmo- formed by summing the Rayleigh radiance-corrected 412-, 555-,
sphere–ocean system returns to a more normal state. and 670-nm SeaWiFS images.
 SATELLITE REMOTE SENSING: OCEAN COLOUR 583

persisted for c. 6 months. The ecological impact of the Q Eu/Lu ratio (sr)
blooms in 1997 and 1998, which encompassed the R irradiance reflectance
entire western Alaska continental shelf, was dramatic (dimensionless)
and caused fish to avoid the bloom region resulting in Rrs remote sensing reflectance
extensive starvation of certain marine mammals and (sr1)
seabirds (those with limited foraging range), and S gelvin absorption spectra par-
formed a barrier for salmon preventing them from ameter (nm1)
spawning in the rivers along that coast. Research is t diffuse transmittance
presently being conducted on methods of identifying (dimensionless)
other types of blooms such as trichodesmium, im- T direct transmittance
portant for understanding nitrification in the ocean, (dimensionless)
and red tides (toxic algal blooms).

Conclusions See also

Bio-Optical Models. Inherent Optical Properties
Satellite ocean color remote sensing combines a and Irradiance. Remote Sensing of Coastal
broad spectrum of science and technology. In many Waters. Satellite Oceanography, History and
ways, the CZCS demonstrated that the technique Introductory Concepts.
could work. However, to advance the ocean biology
and biogeochemistry on a global scale and conduct
research on the effects of global warming, many
improvements in satellite sensor technology, atmo- Further Reading
spheric and oceanic radiative transfer modeling, field
Hooker SB and Firestone ER (eds.) (1996) SeaWiFS
observation methodologies, calibration metrology,
Technical Report Series, NASA Technical Memorandum
and other areas have been necessary and are con- 104566, vols. 1–43. Greenbelt, MD: NASA Goddard
tinuing to evolve. As these develop, new products Space Flight Center.
and applications will become feasible and will re- Hooker SB and Firestone ER (eds.) (2000) SeaWiFS
quire periodic reprocessing of the satellite data to Postlaunch Technical Report Series, NASA Technical
incorporate these advances. Ultimately, the goal of Memorandum Year-206892, vols. 1–29. Greenbelt,
the international ocean science community, working MD: NASA Goddard Space Flight Center.
with the various space agencies, is to develop a Jerlov NG (1976) Marine Optics, 231pp. New York:
continuous long-term global time series of highly Elsevier.
accurate and well-documented satellite ocean color Kirk JTO (1994) Light and Photosynthesis in Aquatic
Ecosystems, 509pp. Cambridge, UK: Cambridge
observations.
University Press.
Martin S (2004) An Introduction to Ocean Remote Sensing,
Nomenclature 426pp. New York: Cambridge University Press.
McClain CR, Feldman GC, and Hooker SB (2004) An
a, aj absorption coefficents (m1)
overview of the SeaWiFS project and strategies for
a*j specific absorption coefficient producing a climate research quality global ocean bio-
(m2/mg Chl a) optical time series. Deep-Sea Research II 51(1–3): 5--42.
bb backscattering coefficent (m1) Mobley CD (1994) Light and Water, Radiative Trans-
chl a chlorophyll a concentration fer in Natural Waters, 592 pp. New York: Academic
(mg (Chl a)/m3) Press.
Ed, Eu, Fo irradiances (mW/cm2 mm) Robinson IS (1985) Satellite Oceanography, 455pp.
K diffuse attenuation coefficient Chichester, UK: Wiley.
(m1) Shifrin KS (1988) Physical Optics of Ocean Water, 285pp.
La, Lf, Lg, Lr, radiances (mW/cm2 mm sr) New York: American Institute of Physics.
Stewart RH (1985) Methods of Satellite Oceanography,
Lra, Lr, Lu, Lw
360pp. Los Angeles: University of California Press.
Lwn normalized water-leaving radi-
ance (mW/cm2 mm sr)
n index of refraction
Relevant Website
(dimensionless) http://oceancolor.gsfc.nasa.gov
r, r Fresnel reflectances – OceanColor Documentation, OceanColor Home
(dimensionless) Page.
 SATELLITE REMOTE SENSING: SALINITY
MEASUREMENTS
G. S. E. Lagerloef, Earth and Space Research, sensors. Salinity is derived from conductivity, tem-
Seattle, WA, USA perature, and pressure with an international stand-
& 2009 Elsevier Ltd. All rights reserved. ard set of empirical equations known as the Practical
Salinity Scale, established in 1978 (PSS-78), which is
much easier to standardize and more precise than
previous chemical methods and which numerically
represents grams per kilogram. Accordingly, the
Introduction modern literature often quotes salinity measurements
in practical salinity units (psu), or refers to PSS-78.
Surface salinity is an ocean state variable which
Salinity ranges from near-zero adjacent to the
controls, along with temperature, the density of
mouths of major rivers to more than 40 in the Red
seawater and influences surface circulation and for-
Sea. Aside from such extremes, open ocean surface
mation of dense surface waters in the higher latitudes
values away from coastlines generally fall between
which sink into the deep ocean and drive the thermo-
32 and 37 (Figure 1).
haline convection. Although no satellite measure-
This global mean surface salinity field has been
ments are made at present, emerging new technology
compiled from all available oceanographic obser-
and a growing scientific need for global measure-
vations. A significant fraction of the 11 latitude and
ments have stimulated programs now underway to
longitude cells have no observations, requiring such
launch salinity-observing satellite sensors within the
maps to be interpolated and smoothed over scales of
present decade. Salinity remote sensing is possible
several hundred kilometers. Seasonal to interannual
because the dielectric properties of seawater which
salinity variations can only be resolved in very lim-
depend on salinity also affect the surface emission at
ited geographical regions where the sampling density
certain microwave frequencies. Experimental heri-
is suitable. Data are most sparse over large regions of
tage extends more than 35 years in the past, in-
the Southern Hemisphere. Remote sensing from
cluding laboratory studies, airborne sensors, and one
satellite will be able to fill this void and monitor
instrument flown briefly in space on Skylab. Re-
multiyear variations globally.
quirements for very low noise microwave radi-
ometers and large antenna structures have limited
Remote sensing theory Salinity remote sensing with
the advance of satellite systems, and are now being
microwave radiometry is likewise possible through
addressed. Science needs, primarily for climate
the electrically conductive properties of seawater. A
studies, dictate a resolution requirement of ap-
radiometric measurement of an emitting surface
proximately 100 km spatial grid, observed monthly,
is given in terms of a ‘brightness temperature’ (TB),
with approximately 0.1 error on the Practical
measured in kelvin (K). TB is related to the true
Salinity Scale (or 1 part in 10 000), which demand
absolute surface temperature (T) through the emis-
very precise radiometers and that several ancillary
sivity coefficient (e):
errors be accurately corrected. Measurements will
be made in the 1.413 GHz astronomical hydrogen
TB ¼ eT
absorption band to avoid radio interference.
For seawater, e depends on the complex dielectric
constant (e), the viewing angle (Fresnel laws), and
Definition and Theory surface roughness (due to wind waves). The complex
How Salinity Is Defined and Measured dielectric constant is governed by the Debye equation

Salinity represents the concentration of dissolved es ðS; TÞ  eN iCðS; TÞ

inorganic salts in seawater (grams salt per kilogram e ¼ eN þ 
1 þ i2pf tðS; TÞ 2pf e0
seawater, or parts per thousand, and historically
given by the symbol %). Oceanographers have de- and includes electrical conductivity (C), the static
veloped modern methods based on the electrical dielectric constant es, and the relaxation time t,
conductivity of seawater which permit accurate which are all sensitive to salinity and temperature
measurement by use of automated electronic in situ (S, T). The equation also includes radio frequency (f)

584
 SATELLITE REMOTE SENSING: SALINITY MEASUREMENTS 585

Global mean surface salinity

90° N

60° N

30° N 35 37
37
Latitude 35
0
36 37
30° S 35 36
35 35
34 34
60° S 34 34

90° S
0 60° E 120° E 180° 120° W 60° W
Longitude

Figure 1 Contour map of the mean global surface salinity field (contour interval 0.5) based on the World Ocean Atlas, 1998
(WOA98). Arctic Ocean salinities o32 are not contoured. Elevated mid-ocean salinities, especially the subtropical Atlantic ones, are
caused by excess evaporation. Data were obtained from the US Department of Commerce, NOAA, National Oceanographic Data
Center.

and terms for permittivity at infinite frequency (eN) c. 3 GHz. Considerations for selecting a measure-
which may vary weakly with T, and permittivity of ment radio frequency include salinity sensitivity,
free space (e0, a constant). The relation of electrical requisite antenna size (see below), and radio inter-
conductivity to salinity and temperature is derived ference from other (mostly man-made) sources. A
from the Practical Salinity Scale. The static dielectric compromise of these factors, dominated by the
and time constants have been modeled by making interference issue, dictates a choice of about a 27-
laboratory measurements of e at various frequencies, MHz-wide frequency band centered at 1.413 GHz,
temperatures, and salinities, and fitting es and t to which is the hydrogen absorption band protected by
polynomial expressions of (S, T) to match the e data. international treaty for radio astronomy research.
Different models in the literature show similar varia- This falls within a frequency range known as L-band.
tions with respect to (f, S, T). Atmospheric clouds have a negligible effect, allowing
Emissivity for the horizontal (H) and vertical (V) observations in all weather except possibly heavy
polarization state is related to e by Fresnel reflection: rain. Accompanying illustrations are based on ap-
plying f ¼ 1.413 GHz in the Debye equation and
" #2 using a model that included laboratory dielectric
cos y  ðe  sin2 yÞ1=2
eH ¼ 1  constant measurements at the nearby frequency
cos y þ ðe  sin2 yÞ1=2 1.43 GHz. Features of this model function and their
" #2 influence on measurement accuracy are discussed in
e cos y  ðe  sin2 yÞ1=2
eV ¼ 1  the section on resolution and error sources.
e cos y þ ðe  sin2 yÞ1=2

where y is the vertical incidence angle from which Antennas Unusually large radiometer antennas
the radiometer views the surface, and eH ¼ eV when will be required to be deployed on satellites for
y ¼ 0. The above set of equations provides a physi- measuring salinity. Radiometer antenna beam width
cally based model function relating TB to surface S, varies inversely with both antenna aperture and
T, y, and H or V polarization state for smooth water radio frequency. 1.413 GHz is a significantly lower
(no wind roughness). This can be inverted to retrieve frequency than found on conventional satel-
salinity from radiometric TB measurements provided lite microwave radiometers, and large antenna
the remaining parameters are known. The micro- structures are necessary to avoid excessive beam
wave optical depth is such that the measured emis- width and accordingly large footprint size. For
sion originates in the top 1 cm of the ocean, example, a 50 km footprint requires about a 6-m-
approximately. aperture antenna whereas conventional radiometer
The rate at which TB varies with salinity is sensi- antennas are around 1–2 m. To decrease the
tive to microwave frequency, achieving levels prac- footprint by a factor of 2 requires doubling the
tical for salinity remote sensing at frequencies below antenna size. Various filled and thinned array
586 SATELLITE REMOTE SENSING: SALINITY MEASUREMENTS

technologies for large antennas have now reached a waters to o32 near shore was measured, along with
development stage where application to salinity several frontal features visible in the satellite surface
remote sensing is feasible. temperature image from the same day. This Gulf
Stream transect demonstrated that small salinity
variations typical of the open ocean can be detected
as well as the strong salinity gradients in the coastal
History of Salinity Remote Sensing
and estuary settings demonstrated previously.
The only experiment to date to measure surface By the mid-1990s, a new airborne salinity mapper
salinity from space took place on the NASA Skylab scanning low frequency microwave radiometer
mission during the fall and winter 1973–74, when a (SLFMR) was developed for light aircraft and has
1.413 GHz microwave radiometer with a 1 m an- been extensively used by NOAA and the US Navy to
tenna collected intermittent data. A weak correlation survey coastal and estuary waters on the US East
was found between the sensor data and surface sal- Coast and Florida. A version of this sensor is now
inity, after correcting for other influences. There was being used in Australia, and a second-generation
no ‘ground truth’ other than standard surface charts, model is presently used for research by the US Navy.
and many of the ambient corrections were not as In 1999 a satellite project was approved by the
well modeled then as they could be today. Research European Space Agency for the measurement of Soil
leading up to the Skylab experiment began with Moisture and Ocean Salinity (SMOS), now with
several efforts during the late 1940s and early 1950s projected launch in 2009. The SMOS mission design
to measure the complex dielectric constant of saline emphasizes soil moisture measurement requirements,
solutions for various salinities, temperatures, and which is done at the same microwave frequency
microwave frequencies. These relationships provide for many of the same reasons as salinity. The TB
the physical basis for microwave remote sensing of dynamic range is about 70–80 K for varying soil
the ocean as described above. moisture conditions and the precision requirement is
The first airborne salinity measurements were therefore much less rigid than for salinity. SMOS will
demonstrated in the Mississippi River outflow and employ a large two-dimensional phased array an-
published in 1970. This led to renewed efforts during tenna system that will yield 40–90 km resolution
the 1970s to refine the dielectric constants and gov- across the measurement swath. The Aquarius/SAC-D
erning equations. Meanwhile, a series of airborne ex- mission is being jointly developed by the US (NASA)
periments in the 1970s mapped coastal salinity and Argentina (CONAE) and is due to be launched
patterns in the Chesapeake and Savannah river plumes in 2010. The Aquarius/SAC-D mission design puts
and freshwater sources along the Puerto Rico shore- primary emphasis on ocean salinity rather than soil
line. In the early 1980s, a satellite concept was sug- moisture, with the focus on optimizing salinity ac-
gested that might achieve an ideal precision of about curacy with a very precisely calibrated microwave
0.25 and spatial resolution of about 100 km. At that radiometer and key ancillary measurements for ad-
time, space agencies were establishing the oceanic dressing the most significant error sources.
processes remote sensing program around missions
and sensors for measuring surface dynamic topo-
graphy, wind stress, ocean color, surface temperature, Requirements for Observing Salinity
and sea ice. For various reasons, salinity remote sens- from Satellite
ing was then considered only marginally feasible from Scientific Issues
satellite and lacked a strongly defined scientific need.
Interest in salinity remote sensing revived in the Three broad scientific themes have been identified for
late 1980s with the development of a 1.4 GHz air- a satellite salinity remote sensing program. These
borne electrically scanning thinned array radiometer themes relate directly to the international climate
(ESTAR) designed primarily for soil moisture meas- research and global environmental observing pro-
urements. ESTAR imaging is done electronically with gram goals.
no moving antenna parts, thus making large antenna
structures more feasible. The airborne version was Improving seasonal to interannual climate pre-
developed as an engineering prototype and to pro- dictions This focuses primarily on El Niño fore-
vide the proof of concept that aperture synthesis can casting and involves the effective use of surface
be extrapolated to a satellite design. The initial ex- salinity data (1) to initialize and improve the
periment to collect ocean data with this sensor con- coupled climate forecast models, and (2) to study
sisted of a flight across the Gulf Stream in 1991 near and model the role of freshwater flux in the
Cape Hatteras. The change from 36 in the offshore formation and maintenance of barrier layers and
 SATELLITE REMOTE SENSING: SALINITY MEASUREMENTS 587

mixed layer heat budgets in the Tropics. Climate thermohaline circulation: 0.1, 100 km, 30 days; and
prediction models in which satellite altimeter sea (4) surface freshwater flux balance: 0.1, 300 km, 30
level data are assimilated must be adjusted for steric days. Thermohaline circulation and convection in the
height (sea level change due to ocean density) subpolar seas has the most demanding requirement,
caused by the variations in upper layer salinity. If and is the most technically challenging because of
not, the adjustment for model heat content is the reduced TB/salinity ratio at low seawater tempe-
incorrect and the prediction skill is degraded. ratures (see below). This can serve as a prime satellite
Barrier layer formation occurs when excessive mission requirement, allowing for the others to be
rainfall creates a shallow, freshwater-stratified, met by reduced mission requirements as appropriate.
surface layer which effectively isolates the deeper Aquarius/SAC-D is a pathfinder mission capable of
thermocline from exchanging heat with the meeting the majority of these requirements with a grid
atmosphere with consequences on the air–sea scale of 150 km and salinity error less than 0.2
coupling processes that govern El Niño dynamics. observed monthly.

Improving ocean rainfall estimates and global Resolution and Error Sources
hydrologic budgets Precipitation over the ocean is
still poorly known and relates to both the Model function Figure 2 shows that the dynamic
hydrologic budget and to latent heating of the range of TB is about 4 K over the range of typical open
overlying atmosphere. Using the ocean as a rain ocean surface salinity and temperature conditions. TB
gauge is feasible with precise surface salinity gradients are greater with respect to salinity than to
observations coupled with ocean surface current temperature. At a given temperature, TB decreases as
velocity data and mixed layer modeling. Such salinity increases, whereas the tendency with respect
calculations will reduce uncertainties in the surface to temperature changes sign. The differential of TB
freshwater flux on climate timescales and will with respect to salinity ranges from  0.2 to  0.7 K
complement satellite precipitation and evaporation per salinity unit. Corrected TB will need to be
observations to improve estimates of the global measured to 0.02–0.07 K precision to achieve 0.1
water and energy cycles. salinity resolution. The sensitivity is strongly affected
by temperature, being largest at the highest tempe-
Monitoring large-scale salinity events and thermo- ratures and yielding better measurement precision in
haline convection Studying interannual surface warm versus cold ocean conditions. Random error
salinity variations in the subpolar regions, particularly can be reduced by temporal and spatial averaging. The
the North Atlantic and Southern Oceans, is essential degraded measurement precision in higher latitudes
to long-time-scale climate prediction and model- will be somewhat compensated by the greater sam-
ing. These variations influence the rate of oceanic pling frequency from a polar orbiting satellite.
convection and poleward heat transport (thermo- The TB variation with respect to temperature
haline circulation) which are known to have been falls generally between 70.15 K 1C  1 and near zero
coupled to extreme global climate changes in the over a broad S and T range. Knowledge of the sur-
geologic record. Outside of the polar regions, salinity face temperature to within a few tenths of degrees
signals are stronger in the coastal ocean and marginal Celsius will be adequate to correct TB for tempera-
seas than in the open ocean in general, but large ture effects and can be obtained using data from
footprint size will limit near-shore applications of the other satellite systems. The optical depth for this
data. Many of the larger marginal seas which have microwave frequency in seawater is about 1–2 cm,
strong salinity signals might be adequately resolved and the remotely sensed measurement depends on
nonetheless, such as the East China Sea, Bay of the T and S in that surface layer thickness. TB for the
Bengal, Gulf of Mexico, Coral Sea/Gulf of Papua, and H and V polarizations have large variations with
the Mediterranean. incidence angle and spacecraft attitude will need to
be monitored very precisely.
Science requirements From the above science
themes, preliminary accuracy and spatial and Other errors Several other error sources will bias TB
temporal resolution requirements from satellite measurements and must be either corrected or
observations have been suggested as the minimum to avoided. These include ionosphere and atmosphere
study the following ocean processes: (1) barrier layer effects, cosmic and galactic background radiations,
effects on tropical Pacific heat flux: 0.2 (PSS-78), surface roughness from winds, sun glint, solar flux,
100 km, 30 days; (2) steric adjustment of heat storage and rain effects. Cosmic background and lower
from sea level: 0.2, 200 km, 7 days; (3) North Atlantic atmospheric adsorption are nearly constant biases
588 SATELLITE REMOTE SENSING: SALINITY MEASUREMENTS

TB as a function of S and T at zero incidence angle

91 91 90.5 90
36 91.5 91.5 91

Salinity
92 92
.5 92.5
34 91 92.5 92
.5 93
92 93.5 93
93

92
32 93.5
0 5 10 15 20 25 30
Temperature ( °C)

TB differential with respect to S (solid) and T (dashed)

0.
05
36
Salinity

0.1

0
34

32
0 5 10 15 20 25 30
Temperature ( °C)

V and H Pol TB vs. incidence angle

200

150 T = 20.0 °C, S = 35.00

TB (K)

100
Mean

50

0
0 10 20 30 40 50 60 70
Incidence angle (deg)

Figure 2 Brightness temperature (TB) properties as a function of S, T, and incidence angle for typical ocean surface conditions.
Top: TB contours. Middle: TB derivatives relative to S (solid curves) and T (dashed curves). Bottom: TB variation vs. incidence angle for
H and V polarization. Calculations based on formulas in Klein LA and Swift CT (1977). An improved model for the dielectric constant of
sea water at microwave frequencies. IEEE Transactions on Antennas and Propagation AP-25(1): 104–111.

easily corrected. An additional correction will be ranges between 0.1 and 0.4 K/(ms  1). Sea state
needed when the reflected radiation from the galactic conditions can change significantly within the few
core is in the field of view. The ionosphere and surface hours that may elapse until ancillary measurements
winds (roughness) have wide spatial and temporal are obtained from another satellite. Simultaneous
variations and require ancillary data and careful wind roughness measurement can be made with an
treatment to avoid TB errors of several kelvins. onboard radar backscatter sensor. A more accurate
The ionosphere affects the measurement though correction can be applied using a direct relationship
attenuation and through Faraday rotation of the H between the radar backscatter and the TB response
and V polarized signal. There is no Faraday effect rather than rely on wind or sea state information
when viewing at nadir (y ¼ 0) because H and V emis- from other sensors.
sivities are identical, whereas off-nadir corrections will Microwave attenuation by rain depends on rain
be needed to preserve the polarization signal. Cor- rate and the thickness of the rain layer in the at-
rection data can be obtained from ionosphere models mosphere. The effect is small at the intended
and analyses but may be limited by unpredictable microwave frequency, but for the required accuracy
short-term ionosphere variations. Onboard correction the effect must be either modeled and corrected with
techniques have been developed that require fully po- ancillary data, or the contaminated data discarded.
larimetric radiometer measurements from which the For the accumulation of all the errors described here,
Faraday rotation may be derived. Sun-synchronous it is anticipated that the root sum square salinity
orbits can be selected that minimize the daytime peak error will be reduced to less than 0.2 with adequate
in ionosphere activity as well as solar effects. radiometer engineering, correction models, onboard
The magnitude of the wind roughness correction measurements, ancillary data, and spatiotemporal
varies with incidence angle and polarization, and filtering with methods now in development.
 SATELLITE REMOTE SENSING: SALINITY MEASUREMENTS 589

See also Lagerloef G, Swift C, and LeVine D (1995) Sea surface

salinity: The next remote sensing challenge. Oceano-
Satellite Altimetry. Satellite Remote Sensing of Sea graphy 8: 44--50.
Surface Temperatures. Lagerloef GSE (2000) Recent progress toward satellite
measurements of the global sea surface salinity field.
Further Reading In: Halpern D (ed.) Elsevier Oceaongraphy Series, No.
63: Satellites, Oceanography and Society, pp. 309--319.
Blume HC and Kendall BNM (1982) Passive microwave Amsterdam: Elsevier.
measurements of temperature and salinity in coastal Lerner RM and Hollinger JP (1977) Analysis of 1.4 GHz
zones. IEEE Transactions on Geoscience and Remote radiometric measurements from Skylab. Remote
Sensing GE 20: 394--404. Sensing Environment 6: 251--269.
Blume H-JC, Kendall BM, and Fedors JC (1978) Le Vine DM, Kao M, Garvine RW, and Sanders T (1998)
Measurement of ocean temperature and salinity via Remote sensing of ocean salinity: Results from
microwave radiometry. Boundary-Layer Meteorology the Delaware coastal current experiment. Journal
13: 295--380. of Atmospheric and Ocean Technology 15:
Blume H-JC, Kendall BM, and Fedors JC (1981) 1478--1484.
Multifrequency radiometer detection of submarine Le Vine DM, Kao M, Tanner AB, Swift CT, and Griffis A
freshwater sources along the Puerto Rican coastline. (1990) Initial results in the development of a synthetic
Journal of Geophysical Research 86: 5283--5291. aperture radiometer. IEEE Transactions on Geoscience
Broecker WS (1991) The great ocean conveyer. and Remote Sensing 28(4): 614--619.
Oceanography 4: 79--89. Lewis EL (1980) The Practical Salinity Scale 1978 (PSS-78)
Delcroix T and Henin C (1991) Seasonal and interannual and its antecedents. IEEE Journal of Oceanic Engineer-
variations of sea surface salinity in the tropical Pacific ing OE-5: 3--8.
Ocean. Journal of Geophysical Research 96: 22135-- Miller J, Goodberlet M, and Zaitzeff J (1998) Airborne
22150. salinity mapper makes debut in coastal zone. EOS
Delworth T, Manabe S, and Stouffer RJ (1993) Transactions, American Geophysical Union 79(173):
Interdecadal variations of the thermohaline circulation 176--177.
in a coupled ocean–atmosphere model. Journal of Reynolds R, Ji M, and Leetmaa A (1998) Use of salinity to
Climate 6: 1993--2011. improve ocean modeling. Physics and Chemistry of the
Dickson RR, Meincke R, Malmberg S-A, and Lee JJ (1988) Earth 23: 545--555.
The ‘great salinity anomaly’ in the northern North Swift CT and McIntosh RE (1983) Considerations for
Atlantic, 1968–1982. Progress in Oceanography 20: microwave remote sensing of ocean-surface salinity.
103--151. IEEE Transactions on Geoscience and Remote Sensing
Droppelman JD, Mennella RA, and Evans DE (1970) An GE-21: 480--491.
airborne measurement of the salinity variations of the UNESCO (1981) The Practical Salinity Scale 1978 and the
Mississippi River outflow. Journal of Geophysical International Equation of State of Seawater 1980.
Research 75: 5909--5913. Technical Papers in Marine Science 36, 25pp.
Kendall BM and Blanton JO (1981) Microwave radiometer Webster P (1994) The role of hydrological processes in
measurement of tidally induced salinity changes off the ocean–atmosphere interactions. Reviews of Geophysics
Georgia coast. Journal of Geophysical Research 86: 32(4): 427--476.
6435--6441. Yueh SH, West R, Wilson WJ, Li FK, Njoku EG, and
Klein LA and Swift CT (1977) An improved model for the Rahmat-Samii Y (2001) Error sources and feasibility for
dielectric constant of sea water at microwave fre- microwave remote sensing of ocean surface salinity.
quencies. IEEE Transactions on Antennas and Propa- IEEE Transactions on Geoscience and Remote Sensing
gation AP-25(1): 104--111. 39: 1049--1060.
 SEA SURFACE EXCHANGES OF MOMENTUM,
HEAT, AND FRESH WATER DETERMINED BY
SATELLITE REMOTE SENSING
L. Yu, Woods Hole Oceanographic Institution, backbone for constructing the climatological state of
Woods Hole, MA, USA the global flux fields. Over the past two decades,
& 2009 Elsevier Ltd. All rights reserved. satellite remote sensing has become a mature tech-
nology for remotely sensing key air–sea variables.
With continuous global spatial coverage, consistent
quality, and high temporal sampling, satellite meas-
urements not only allow the construction of air–sea
Introduction fluxes at near-real time with unprecedented quality
but most importantly, also offer the unique oppor-
The ocean and the atmosphere communicate tunity to view the global ocean synoptically as an
through the interfacial exchanges of heat, fresh entity.
water, and momentum. While the transfer of the
momentum from the atmosphere to the ocean by
wind stress is the most important forcing of the
ocean circulation, the heat and water exchanges af- Flux Estimation Using Satellite
fect the horizontal and vertical temperature gradients Observations
of the lower atmosphere and the upper ocean, which,
Sea Surface Wind Stress
in turn, modify wind and ocean currents and main-
tain the equilibrium of the climate system. The sea The Seasat-A satellite scatterometer, launched in
surface exchanges are the fundamental processes of June 1978, was the first mission to demonstrate that
the coupled atmosphere–ocean system. An accurate ocean surface wind vectors (both speed and dir-
knowledge of the flux variability is critical to our ection) could be remotely sensed by active radar
understanding and prediction of the changes of glo- backscatter from measuring surface roughness.
bal weather and climate. Scatterometer detects the loss of intensity of trans-
The heat exchanges include four processes: the mitted microwave energy from that returned by the
short-wave radiation (QSW) from the sun, the out- ocean surface. Microwaves are scattered by wind-
going long-wave radiation (QLW) from the sea sur- driven capillary waves on the ocean surface, and the
face, the sensible heat transfer (QSH) resulting from fraction of energy returned to the satellite (back-
air–sea temperature differences, and the latent heat scatter) depends on both the magnitude of the wind
transfer (QLH) carried by evaporation of sea surface stress and the wind direction relative to the direction
water. Evaporation releases both energy and water of the radar beam (azimuth angle). By using a
vapor to the atmosphere, and thus links the global transfer function or an empirical algorithm, the
energy cycle to the global water cycle. The oceans are backscatter measurements are converted to wind
the key element of the water cycle, because the vectors. It is true that scatterometers measure the
oceans contain 96% of the Earth’s water, experience effects of small-scale roughness caused by surface
86% of planetary evaporation, and receive 78% of stress, but the retrieval algorithms produce surface
planetary precipitation. wind, not wind stress, because there are no adequate
The amount of air–sea exchange is called sea sur- surface-stress ‘ground truths’ to calibrate the algo-
face (or air–sea) flux. Direct flux measurements by rithms. The wind retrievals are calibrated to the
ships and buoys are very limited. Our present equivalent neutral-stability wind at a reference
knowledge of the global sea surface flux distribution height of 10 m above the local-mean sea surface. This
stems primarily from bulk parametrizations of the is the 10-m wind that would be associated with the
fluxes as functions of surface meteorological vari- observed surface stress if the atmospheric boundary
ables that can be more easily measured (e.g., wind layer were neutrally stratified. The 10-m equivalent
speed, temperature, humidity, cloud cover, precipi- neutral wind speeds differ from the 10-m wind
tation, etc.). Before the advent of satellite remote speeds measured by anemometers, and these differ-
sensing, marine surface weather reports collected ences are a function of atmospheric stratification and
from voluntary observing ships (VOSs) were the are normally in the order of 0.2 m s–1. To compute

590
 SEA SURFACE EXCHANGES 591

the surface wind stress, t, the conventional bulk attenuation by rain will reduce the signal causing an
formulation is then employed: underestimation of the wind speed.
Under rain-free conditions, scatterometer-derived
wind estimates are accurate within 1 m s  1 for speed
t ¼ ðtx ; ty Þ ¼ rcd Wðu; vÞ ½1 and 201 for direction. For low (less than 3 m s  1)
and high winds (greater than 20 m s  1), the un-
where tx and ty are the zonal and meridional com- certainties are generally larger. Most problems with
ponents of the wind stress; W, u, and v are the low wind retrievals are due to the weak backscatter
scatterometer-estimated wind speed at 10 m and its signal that is easily confounded by noise. The low
zonal component (eastward) and meridional com- signal/noise ratio complicates the ambiguity removal
ponent (northward), respectively. The density of processing in selecting the best wind vector from the
surface air is given by r and is approximately equal set of ambiguous wind vectors. Ambiguity removal is
to 1.225 kg m  3, and cd is a neutral 10-m drag over 99% effective for wind speed of 8 m s–1 and
coefficient. higher. Extreme high winds are mostly associated
Scatterometer instruments are typically deployed with storm events. Scatterometer-derived high winds
on sun-synchronous near-polar-orbiting satellites are found to be underestimated due largely to de-
that pass over the equator at approximately the same ficiencies of the empirical scatterometer algorithms.
local times each day. These satellites orbit at an al- These algorithms are calibrated against a subset of
titude of approximately 800 km and are commonly ocean buoys – although the buoy winds are accurate
known as Polar Orbiting Environmental Satellites and serve as surface wind truth, few of them have
(POES). There have been six scatterometer sensors high-wind observations.
aboard POES since the early 1990s. The major NSCAT worked flawlessly, but the spacecraft
characteristics of all scatterometers are summarized (ADEOS-I) that hosted it demised prematurely in
in Table 1. The first European Remote Sensing June 1997 after only 9 months of operation. A re-
(ERS-1) satellite was launched by the European placement mission called QuikSCAT was rapidly
Space Agency (ESA) in August 1991. An identical developed and launched in July 1999. To date,
instrument aboard the successor ERS-2 became op- QuikSCAT remains in operation, far outlasting the
erational in 1995, but failed in 2001. In August expected 2–3-year mission life expectancy. QuikSCAT
1996, the National Aeronautics and Space Ad- carries a Ku-band scatterometer named SeaWinds,
ministration (NASA) began a joint mission with the which has accuracy characteristics similar to NSCAT
National Space Development Agency (NASDA) of but with improved coverage. The instrument meas-
Japan to maintain continuous scatterometer missions ures vector winds over a swath of 1800 km with a
beyond ERS satellites. The joint effort led to the nominal spatial resolution of 25 km. The improved
launch of the NASA scatterometer (NSCAT) aboard sampling size allows approximately 93% of the ocean
the first Japanese Advanced Earth Observing Satellite surface to be sampled on a daily basis as opposed to 2
(ADEOS-I). The ERA scatterometers differ from the days by NSCAT and 4 days by the ERS instruments. A
NASA scatterometers in that the former operate on second similar-version SeaWinds instrument was
the C band (B5 GHz), while the latter use the Ku placed on the ADEOS-II mission in December 2002.
band (B14 GHz). For radio frequency band, rain However, after only a few months of operation, it
attenuation increases as the signal frequency in- followed the unfortunate path of NSCAT and failed in
creases. Compared to C-band satellites, the higher October 2003 due – once again – to power loss.
frequencies of Ku band are more vulnerable to signal The Advanced Scatterometer (ASCAT) launched by
quality problems caused by rainfall. However, ESA/EUMETSAT in March 2007 is the most recent
Ku-band satellites have the advantage of being more satellite designed primarily for the global measure-
sensitive to wind variation at low winds and of ment of sea surface wind vectors. ASCAT is flown on
covering more area. the first of three METOP satellites. Each METOP has
Rain has three effects on backscatter measure- a design lifetime of 5 years and thus, with overlap, the
ments. It attenuates the radar signal, introduces series has a planned duration of 14 years. ASCAT is
volume scattering, and changes the properties of the similar to ERS-1/2 in configuration except that it has
sea surface and consequently the properties of increased coverage, with two 500-km swaths (one on
microwave signals scattered from the sea surface. each side of the spacecraft nadir track).
When the backscatter from the sea surface is low, the The data collected by scatterometers on various
additional volume scattering from rain will lead to missions have constituted a record of ocean vector
an overestimation of the low wind speed actually winds for more than a decade, starting in August
present. Conversely, when the backscatter is high, 1992. These satellite winds provide synoptic global
 592

Table 1 Major characteristics of the spaceborne scatterometers

SEA SURFACE EXCHANGES

Characteristics Scatterometer

SeaSat-A ERS-1 ERS-2 NSCAT SeaWinds on SeaWinds on ADEOS II ASCAT

QuikSCAT

Operational Ku band C band C band Ku band Ku band Ku band C band

frequency
14.6 GHz 5.255 GHz 5.255 GHz 13.995 GHz 13.402 GHz 13.402 GHz 5.255 GHz
Spatial 50 km  50 km with 50 km  50 km 50 km  50 km 25 km  25 km 25 km  25 km 25 km  6 km 25 km  25 km
resolution 100-km spacing
Scan Two-sided, double One-sided, single One-sided, single Two-sided, double Conical scan, one Conical scan, one wide Two-sided, double
characteristics 500 km swaths 500-km swath 500-km swath 600-km swaths wide swath of swath of 1800 km 500-km swaths
separated by a separated by a 1800 km separated by a
450 km nadir gap 329-km nadir gap 700-km nadir
gap
Daily coverage Variable 41% 41% 77% 93% 93% 60%
Period in service Jul. 1978–Oct. 1978 Aug. 1991–May. 1997 May. 1995–Jan. 2001 Sep. 1996–Jun. 1997 Jun. 1999–current Dec. 2002–Oct. 2003 Mar. 2007–current
 SEA SURFACE EXCHANGES 593

view from the vantage point of space, and provide geostationary satellite measurements with polar or-
excellent coverage in regions, such as the southern biter measurements as supplemental when there are
oceans, that are poorly sampled by the conventional no geostationary measurements, the ISCCP cloud-
observing network. Scatterometers have been shown retrieval algorithm includes the conversion of radi-
to be the only means of delivering observations ance measurements to cloud scenes and the inference
at adequate ranges of temporal and spatial scales and of cloud properties from the radiance values. Radi-
at adequate accuracy for understanding ocean– ance thresholds are applied to obtain cloud fractions
atmosphere interactions and global climate changes, for low, middle, and high clouds based on radiance
and for improving climate predictions on synoptic, computed from models using observed temperature
seasonal, and interannual timescales. and climatological lapse rates.
In addition to the global cloud analysis, ISCCP
also produces radiative fluxes (up, down, and net) at
Surface Radiative Fluxes
the Earth’s surface that parallels the effort under-
Direct estimates of surface short-wave (SW) and taken by the Global Energy and Water Cycle Ex-
long-wave (LW) fluxes that resolve synoptic to re- periment – Surface Radiation Budget (GEWEX-SRB)
gional variability over the globe have only become project. The two projects use the same ISCCP cloud
possible with the advent of satellite in the past two information but different ancillary data sources and
decades. The surface radiation is a strong function of different radiative transfer codes. They both compute
clouds. Low, thick clouds reflect large amounts of the radiation fluxes for clear and cloudy skies to es-
solar radiation and tend to cool the surface of the timate the cloud effect on radiative energy transfer.
Earth. High, thin clouds transmit incoming solar Both have a 3-h resolution, but ISCCP fluxes are
radiation, but at the same time, they absorb the produced on a 280-km equal-area (EQ) global grid
outgoing LW radiation emitted by the Earth and ra- while GEWEX-SRB fluxes are on a 11  11 global
diate it back downward. The portion of radiation, grid. The two sets of fluxes have reasonable agree-
acting as an effective ‘greenhouse gas’, adds to the ment with each other on the long-term mean basis, as
SW energy from the sun and causes an additional suggested by the comparison of the global annual
warming of the surface of the Earth. For a given surface radiation budget in Table 2. The total net
cloud, its effect on the surface radiation depends radiation differs by about 5 W m  2, due mostly to
on several factors, including the cloud’s altitude, size, the SW component. However, when compared with
and the particles that form the cloud. At present, ground-based observations, the uncertainty of these
the radiative heat fluxes at the Earth’s surface are fluxes is about 10–15 W m  2. The main cause is the
estimated from top-of-the-atmosphere (TOA) SW uncertainties in surface and near-surface atmospheric
and LW radiance measurements in conjunction with properties such as surface skin temperature, surface
radiative transfer models. air and near-surface-layer temperatures and humid-
Satellite radiance measurements are provided by ity, aerosols, etc. Further improvement requires im-
two types of radiometers: scanning radiometers and proved retrievals of these properties.
nonscanning wide-field-of-view radiometers. Scan- In the late 1990s, the Clouds and the Earth’s Ra-
ning radiometers view radiance from a single dir- diant Energy System (CERES) experiment was
ection and must estimate the hemispheric emission or
reflection. Nonscanning radiometers view the entire
Table 2 Annual surface radiation budget (in W m  2) over
hemisphere of radiation with a roughly 1000-km global oceans. Uncertainty estimates are based on the standard
field of view. The first flight of an Earth Radiation error of monthly anomalies
Budget Experiment (ERBE) instrument in 1984 in-
cluded both a scanning radiometer and a set of Data 21-year Parameter
nonscanning radiometers. These instruments obtain mean 1984–
2004
good measurements of TOA radiative variables in-
SW LW SW þ LW
cluding insolation, albedo, and absorbed radiation. Net Net Net
To estimate surface radiation fluxes, however, more downward downward downward
accurate information on clouds is needed.
To determine the physical properties of clouds from ISCCP 173.279.2  46.979.2 126.3711.0
satellite measurements, the International Satellite (Zhang et al.,
2004)
Cloud Climatology Project (ISCCP) was established in GEWEX-SRB 167.2713.9  46.375.5 120.9711.9
1983. ISCCP pioneered the cross-calibration, analysis, (Gupta et al.,
and merger of measurements from the international 2006)
constellation of operational weather satellites. Using
594 SEA SURFACE EXCHANGES

developed by NASA’s Earth Observing System (EOS) above its surface and each pass of the satellite pro-
not only to measure TOA radiative fluxes but also to vides a 2399-km-wide swath, it usually takes 1 or 2
determine radiative fluxes within the atmosphere and weeks, depending on the actual cloud coverage, to
at the surface, by using simultaneous measurements obtain a complete global coverage. Clouds, on the
of complete cloud properties from other EOS in- other hand, have little effect on the microwave
struments such as the moderate-resolution imaging radiometers so that microwave Ts retrievals can be
spectroradiometer (MODIS). CERES instruments made under complete cloud cover except for raining
were launched aboard the Tropical Rainfall Meas- conditions. The TRMM microwave imager (TMI)
uring Mission (TRMM) in November 1997, on the launched in 1997 has a full suite of channels ranging
EOS Terra satellite in December 1999, and on the from 10.7 to 85 GHz and was the first satellite sensor
EOS Aqua spacecraft in 2002. There is no doubt that capable of accurately measuring SST through clouds.
the EOS era satellite observations will lead to great The low-inclination equitorial orbit, however, limits
improvement in estimating cloud properties and the TMI’s coverage only up to c. 381 latitude. Fol-
surface radiation budget with sufficient simultaneity lowing TMI, the first polar-orbiting microwave
and accuracy. radiometer capable of measuring global through-
cloud SST was made possible by the NASDA’s
Sea Surface Turbulent Heat Fluxes advanced microwave scanning radiometer (AMSR)
flown aboard the NASA’s EOS Aqua mission in
Latent and sensible heat fluxes are the primary
2002.
mechanism by which the ocean transfers much of the
While SST can be measured in both infrared and
absorbed solar radiation back to the atmosphere.
microwave regions, the near-surface wind speed can
The two fluxes cannot be directly observed by space
only be retrieved in the microwave region. The rea-
sensors, but can be estimated from wind speed and
son is that the emissivity of the ocean’s surface at
sea–air humidity/temperature differences using the
wavelengths of around 11 mm is so high that it is not
following bulk parametrizations:
sensitive to changes in the wind-induced sea surface
roughness or humidity fluctuations in the lower at-
QLH ¼ rLe ce Wðqs  qa Þ ½2 mosphere. Microwave wind speed retrievals are
provided by the special sensor microwave/imager
QSH ¼ rcp ch WðTs  Ta Þ ½3 (SSM/I) that has been flown on a series of polar-
orbiting operational spacecrafts of the Defense
where Le is the latent heat of vaporization and is a Meteorological Space Program (DMSP) since July
function of sea surface temperature (SST, Ts) ex- 1987. SSM/I has a wide swath (B 1400 km) and a
pressed as Le ¼ (2.501  0.002 37  Ts)  1.06. cp is coverage of 82% of the Earth’s surface within 1 day.
the specific heat capacity of air at constant pressure; But unlike scatterometers, SSM/I is a passive micro-
ce and ch are the stability- and height-dependent wave sensor and cannot provide information on the
turbulent exchange coefficients for latent and sens- wind direction. This is not a problem for the com-
ible heat, respectively. Ta/qa are the temperature/ putation in eqns [2] and [3] that requires only wind
specific humidity at a reference height of 2 m above speed observations. In fact, the high space-time
the sea surface. qs is the saturation humidity at Ts, resolution and good global coverage of SSM/I has
and is multiplied by 0.98 to take into account the made it serving as a primary database for computing
reduction in vapor pressure caused by salt water. the climate mean and variability of the oceanic latent
The two variables, Ts and W, in eqns [2] and [3] and sensible heat fluxes over the past B20-year
are retrieved from satellites, and so qs is known. The period. At present, wind speed measurements with
remote sensing of Ts is based on techniques by which good accuracy are also available from several NASA
spaceborne infrared and microwave radiometers de- satellite platforms, including TMI and AMSR.
tect thermally emitted radiation from the ocean sur- The most difficult problem for the satellite-based
face. Infrared radiometers like the five-channel flux estimation is the retrieval of the air humidity and
advanced very high resolution radiometer (AVHRR) temperature, qa and Ta, at a level of several meters
utilize the wavelength bands at 3.5–4 and 10–12 mm above the surface. This problem is inherent to all
that have a high transmission of the cloud-free at- spaceborne passive radiometers, because the meas-
mosphere. The disadvantage is that clouds are opa- ured radiation emanates from relatively thick at-
que to infrared radiation and can effectively mask mospheric layers rather than from single levels. One
radiation from the ocean surface, and this affects the common practice to extract satellite qa is to relate qa
temporal resolution. Although the AVHRR satellite to the observed column integrated water vapor (IWV,
orbits the Earth 14 times each day from 833 km also referred to as the total precipitable water) from
 SEA SURFACE EXCHANGES 595

SSM/I. Using IWV as a proxy for qa is based on the physically based system. The qa and Ta estimates
several observational findings that on monthly from the models contain less ambiguity associated
timescales the vertical distribution of water vapor is with the vertical integration and large spatial aver-
coherent throughout the entire atmospheric column. aging of the various parameters, though they are
The approach, however, produces large systematic subject to systematic bias due to model’s subgrid
biases of over 2 g kg  1 in the Tropics, as well as parametrizations. The third approach is to obtain a
in the mid- and high latitudes during summertime. better estimation of qa and Ta through an optimal
This is caused by the effect of the water vapor con- combination of satellite retrievals with the model
vergence that is difficult to assess in regions where outputs, which has been experimented by the Ob-
the surface air is nearly saturated but the total IWV is jectively Analyzed air–sea Fluxes (OAFlux) project at
small. Under such situations, the IWV cannot reflect the Woods Hole Oceanographic Institution (WHOI).
the actual vertical and horizontal humidity vari- The effort has led to improved daily estimates of
ations in the atmosphere. Various remedies have global air–sea latent and sensible fluxes.
been proposed to improve the qa–IWV relation and
to make it applicable on synoptic and shorter time-
Freshwater Flux
scales. There are methods of including additional
geophysical variables, replacing IWV with the IWV The freshwater flux is the difference between pre-
in the lower 500 m of the planetary layer, and/or cipitation (rain) and evaporation. Evaporation re-
using empirical orthogonal functions (EOFs). Al- leases both water vapor and latent heat to the
though overall improvements were achieved, the atmosphere. Once latent heat fluxes are estimated,
accuracy remains poor due to the lack of detailed the sea surface evaporation (E) can be computed
information on the atmospheric humidity profiles. using the following relation:
Retrieving Ta from satellite observations is even
more challenging. Unlike humidity, there is no co-
herent vertical structure of temperature in the at- E ¼ QLH =rw Le ½4
mosphere. Satellite temperature sounding radiometers
offer little help, as they generally are designed for where QLH denotes latent heat flux and rw is the
retrieval in broad vertical layers. The sounder’s low density of seawater.
information content in the lower atmosphere does not Spaceborne sensors cannot directly observe the
enable the retrieval of near-surface air temperature actual precipitation reaching the Earth’s surface, but
with sufficient accuracy. Different methods have been they can measure other variables that may be highly
tested to derive Ta from the inferred qa, but all showed correlated with surface rainfall. These include vari-
limited success. Because of the difficulties in deter- ations in infrared and microwave brightness tem-
mining qa and Ta, latent and sensible fluxes estimated peratures, as well as visible and near-infrared albedo.
from satellite measurements have large uncertainties. Infrared techniques are based on the premise that
Three methods have been tested for obtaining rainfall at the surface is related to cloud-top prop-
better qa and Ta to improve the estimates of latent erties observed from space. Visible/infrared obser-
and sensible fluxes. The first approach is to enhance vations supplement the infrared imagery with visible
the information on the temperature and moisture imagery during daytime to help eliminate thin cirrus
in the lower troposphere. This is achieved by com- clouds, which are cold in the infrared imagery and
bining SSM/I data with additional microwave are sometimes misinterpreted as raining using infra-
sounder data that come from the instruments like the red data alone. Visible/infrared sensors have the ad-
advanced microwave sounding unit (AMSU-A) and vantage of providing good space and time sampling,
microwave humidity sounder (MHS) flown aboard but have difficulty capturing the rain from warm-
the National Oceanic and Atmospheric Adminis- topped clouds. By comparison, microwave (MW)
tration (NOAA) polar-orbiting satellites, and the estimates are more physically based and more ac-
special sensor microwave temperature sounder curate although time and space resolutions are not as
(SSM/T) and (SSM/T-2) on the DMSP satellites. Al- good. The principle of MW techniques is that rainfall
though the sounders do not directly provide shallow at the surface is related to microwave emission from
surface measurements, detailed profile information rain drops (low-frequency channels) and microwave
provided by the sounders can help to remove vari- scattering from ice (high-frequency channels).
ability in total column measurements not associated While the primary visible/infrared data sources are
with the surface. The second approach is to capital- the operational geostationary satellites, microwave
ize the progress made in numerical weather predic- observations are available from SSM/I, the NOAA
tion models that assimilate sounder observations into AMSU-B, and the TRMM spacecraft.
596 SEA SURFACE EXCHANGES

TRMM opened up a new era of estimating not (GPCP) to produce global precipitation analyses
only surface rainfall but also rain profiles. TRMM is from 1979 and continuing.
equipped with the first spaceborne precipitation
radar (PR) along with a microwave radiometer
(TMI) and a visible/infrared radiometer (VIRS).
Summary and Applications
Coincident measurements from the three sensors The satellite sensor systems developed in the past two
are complementary. PR provides detailed vertical decades have provided unprecedented observations of
rain profiles across a 215-km-wide strip. TMI (a five- geophysical parameters in the lower atmosphere and
frequency conical scanning radiometer) though has upper oceans. The combination of measurements
less vertical and horizontal fidelity in rain-resolving from multiple satellite platforms has demonstrated
capability, and it features a swath width of 760 km. the capability of estimating sea surface heat, fresh
The VIRS on TRMM adds cloud-top temperatures water, and momentum fluxes with sufficient accuracy
and structures to complement the description of the and resolution. These air–sea flux data sets, together
two microwave sensors. While direct precipitation with satellite retrievals of ocean surface topography,
information from VIRS is less reliable than that ob- temperature, and salinity (Figure 1), establish a
tained by the microwave sensors, VIRS serves an complete satellite-based observational infrastructure
important role as a bridge between the high-quality for fully monitoring the ocean’s response to the
but infrequent observations from TMI and PR and changes in air–sea physical forcing.
the more available data and longer time series data Atmosphere and the ocean are nonlinear turbulent
available from the geostationary visible/infrared fluids, and their interactions are nonlinear scale-
satellite platforms. dependent, with processes at one scale affecting
The TRMM satellite focuses on the rain variability processes at other scales. The synergy of various
over the tropical and subtropical regions due to the satellite-based products makes it especially advan-
low inclination. An improved instrument, AMSR, tageous to study the complex scale interactions be-
has extended TRMM rainfall measurements to tween the atmosphere and the ocean. One clear
higher latitudes. AMSR is currently aboard the Aqua example is the satellite monitoring of the development
satellite and is planned by the Global Precipitation of the El Niño–Southern Oscillation (ENSO) in 1997–
Measurement (GPM) mission to be launched in 98. ENSO is the largest source of interannual vari-
2009. Combining rainfall estimates from visible/in- ability in the global climate system. The phenomenon
frared with microwave measurements is being is characterized by the appearance of extensive warm
undertaken by the Global Climatology Project surface water over the central and eastern tropical

Short- and Iong-wave Vector wind

radiation Seasat-A, ERS, NSCAT, SeaWinds
Rain Evaporation ERBE, CERES on QuikSCAT and ADEOS-II,
GPCP, TRMM SSMI, TMI, AMSR ISCCP, GEWEX-SRB ASCAT
Sensible heat
Water vapor and latent heat

Sea surface salinity Sea surface temperature Sea level

Aquarius, SMOS AVHRR, TMI, AMSR TOPEX/Poseidon, JASON

Figure 1 Schematic diagram of the physical exchange processes at the air–sea interface and the upper ocean responses, with
corresponding sensor names shown in red.
 SEA SURFACE EXCHANGES 597

Pacific Ocean at a frequency of c. 3–7 years. The WWB of December 1996 in the western Pacific. A
1997–98 El Niño was one of the most severe events series of subsequent WWB-induced Kelvin waves
experienced during the twentieth century. During the further enhanced the eastern warming, and fueled
peak of the event in December 1997 (Figure 2), the the El Niño development. The positive feedback
SST in the eastern equatorial Pacific was more than between synoptic WWB and the interannual SST
5 1C above normal, and the warming was accom- warming in making an El Niño is clearly indicated by
panied by excessive precipitation and large net heat satellite observations. On the other hand, the syn-
transfer from the ocean to the atmosphere. optic WWB events were the result of the develop-
The 1997–98 event was also the best observed ment of equatorial twin cyclones under the influence
thanks largely to the expanded satellite-observing of northerly cold surges from East Asia/western
capability. One of the major observational findings North Pacific. NSCAT made the first complete re-
was the role of synoptic westerly wind bursts cording of the compelling connection between near-
(WWBs) in the onset of El Niño. Figure 3 presents equatorial wind events and mid-latitude atmospheric
the evolution of zonal wind from NSCAT scatte- transient forcing.
rometer combined with SSM/I-derived wind product, Clearly, the synergy of various satellite products
sea surface height (SSH) from TOPEX altimetry, and offers consistent global patterns that facilitate the
SST from AVHRR imagery in 1996–98. The ap- mapping of the correlations between various pro-
pearance of the anomalous warming in the eastern cesses and the construction of the teleconnection
basin in February 1997 coincided with the arrival of pattern between weather and climate anomalies
the downwelling Kelvin waves generated by the in one region and those in another. The satellite

18 December 1996
40
30
TOPEX SSH
Degrees N

20 Wind (m s−1) anom (cm) SST anom (°C)

10 1/96 1/96
0
3/96 3/96
–10
–20 5/96 5/96
100 120 140 160 180 7/96 7/96
Degrees E
9/96 9/96
27 December 1996
40 11/96 11/96
30 1/97 1/97
Degrees N

20
3/97 3/97
10
0 5/97 5/97
–10 7/97
7/97
–20
100 120 140 160 180 9/97 9/97
Degrees E 11/97 11/97

1/98 1/98

3/98 3/98

5/98 5/98

120 160 –160 –120 –80 120 160 –160 –120 –80 120 160 –160 –120 –80

–12 –8 –4 0 4 8 12 –32 –24 –16 –8 0 8 16 24 32 –2 –1 0 1 2 3 4

Figure 2 (First column) An example of the scatterometer observations of the generation of the tropical cyclones in the western
tropical Pacific under the influence of northerly cold surges from East Asia/western North Pacific. The effect of westerly wind bursts on
the development of El Niño is illustrated in the evolution of the equatorial sea level observed from TOPEX/Poseidon altimetry and SST
from AVHRR. The second to fourth columns show longitude (horizontally) and time (vertically, increasing downwards). The series of
westerly wind bursts (second column, SSM/I wind analysis by Atlas et al. (1996)) excited a series of downwelling Kelvin waves that
propagated eastward along the equator (third column), suppressed the thermocline, and led to the sea surface warming in the eastern
equatorial Pacific (fourth column).
598 SEA SURFACE EXCHANGES

Le latent heat of vaporization

qa specific humidity at a reference height
above the sea surface
qs specific humidity at the sea surface
QLH latent heat flux
QSH sensible heat flux
Ta temperature at a reference height above
Ocean temp. (°C) the sea surface
Ts temperature at the sea surface
–3 –2 –1 0 1 2 3 4 5
u zonal component of the wind speed
v meridional component of the wind speed
W wind speed
r density of surface air
rw density of sea water
t wind stress
tx zonal component of the wind stress
ty meridional component of the wind stress

P − E (mm d–1)

–10 –5 0 5 10 15 See also

IR Radiometers. Satellite Altimetry. Satellite
Oceanography, History and Introductory
Concepts. Satellite Remote Sensing of Sea
Surface Temperatures. Satellite Remote Sensing:
Salinity Measurements.

Further Reading
Net heat loss (W m–2)
Adler RF, Huffman GJ, Chang A, et al. (2003) The Version
–80 –60 –40 –20 0 20 40 60 80 100 120 2 Global Precipitation Climatology Project (GPCP)
monthly precipitation analysis (1979–present). Journal
Figure 3 Satellite-derived global ocean temperature from
of Hydrometeorology 4: 1147--1167.
AVHRR (top), precipitation minus evaporation from GPCP and
Atlas R, Hoffman RN, Bloom SC, Jusem JC, and
WHOI OAFlux, respectively (middle), and net heat loss
(QLH þ QSH þ QLW  QSW) from the ocean (bottom) during the Ardizzone J (1996) A multiyear global surface
El Niño in Dec. 1997. The latent and sensible heat fluxes wind velocity dataset using SSM/I wind observations.
QLH þ QSH are provided by WHOI OAFlux, and the short- and Bulletin of the American Meteorological Society 77:
long-wave radiative fluxes are by ISCCP. 869--882.
Bentamy A, Katsaros KB, Mestas-Nuñez AM, et al. (2003)
Satellite estimates of wind speed and latent heat flux
observing system will complement the in situ ground over the global oceans. Journal of Climate 16: 637--656.
observations and play an increasingly important role Chou S-H, Nelkin E, Ardizzone J, Atlas RM, and Shie C-L
in understanding the cause of global climate changes (2003) Surface turbulent heat and momentum fluxes over
and in improving the model skills on predicting global oceans based on the Goddard satellite retrievals,
weather and climate variability. version 2 (GSSTF2). Journal of Climate 16: 3256--3273.
Gupta SK, Ritchey NA, Wilber AC, Whitlock CH, Gibson
GG, and Stackhouse RW, Jr. (1999) A climatology of
Nomenclature surface radiation budget derived from satellite data.
cd drag coefficient Journal of Climate 12: 2691--2710.
Liu WT and Katsaros KB (2001) Air–sea flux from satellite
ce turbulent exchange coefficient for latent
data. In: Siedler G, Church J, and Gould J (eds.) Ocean
heat Circulation and Climate, pp. 173--179. New York:
ch turbulent exchange coefficient for Academic Press.
sensible heat Kubota M, Iwasaka N, Kizu S, Konda M, and Kutsuwada
cp specific heat capacity of air at constant K (2002) Japanese Ocean Flux Data Sets with Use of
pressure Remote Sensing Observations (J-OFURO). Journal of
E evaporation Oceanography 58: 213--225.
 SEA SURFACE EXCHANGES 599

Wentz FJ, Gentemann C, Smith D, and Chelton D (2000) http://www.gewex.org

Satellite measurements of sea surface temperature – The Global Energy and Water Cycle Experiment
through clouds. Science 288: 847--850. (GEWEX).
Yu L and Weller RA (2007) Objectively analyzed air–sea http://precip.gsfc.nasa.gov
heat fluxes (OAFlux) for the global oceans. Bulletin of – The Global Precipitation Climatology Project.
the American Meteorological Society 88: 527--539. http://isccp.giss.nasa.gov
Zhang Y-C, Rossow WB, Lacis AA, Oinas V, and – The International Cloud Climatology Project.
Mishchenko MI (2004) Calculation of radiative http://oaflux.whoi.edu
fluxes from the surface to top of atmosphere based on – The Objectively Analyzed air–sea Fluxes project.
ISCCP and other global data sets: Refinements of the http://www.ssmi.com
radiative transfer model and the input data. Journal – The Remote Sensing Systems Research Company
of Geophysical Research 109: D19105 (doi:10.1029/ http://www.gfdi.fsu.edu
2003JD004457). – The SEAFLUX Project, Geophysical Fluid Dynamics
Institute.
Relevant Websites http://eosweb.larc.nasa.gov
– The Surface Radiation Budget Data, Atmospheric
http://winds.jpl.nasa.gov Science Data Center.
– Measuring Ocean Winds from Space.
 GRAVITY
M. McNutt, MBARI, Moss Landing, CA, USA Gravity anomalies (observed gravitational acceler-
Copyright & 2001 Elsevier Ltd.
ation minus an expected value) are typically much
smaller, about 0.5% of the total field. The SI-com-
patible unit for gravity anomaly is the gravity unit
(gu): 1 gu ¼ 106 m s2. However, the older c.g.s unit
for gravity anomaly, the milligal (mGal), is still in
Introduction very wide use: 1 mGal ¼ 10 gu. Typical small-scale
variations in gravity over the ocean range from a few
The gravity field varies over the oceans on account of tens to a few hundreds of gravity units. Lateral
lateral variations in density beneath the ocean sur- variations in gravitational acceleration (gravity gra-
face. The most prominent anomalies arise from un- dients) are measured in Eötvös units (E):
dulations on density interfaces, such as occur at the 1 E ¼ 109 s2. Another quantity useful in gravity
water–rock interface at the seafloor or at the crust– interpretation is the density of earth materials,
mantle interface, also known as the Moho dis- measured in kg m3. In the marine realm, relevant
continuity. Because marine gravity is relatively easy densities range from about 1000 kg m3 for water to
to measure, it serves as a remote sensing tool for more than 3300 kg m3 for mantle rocks.
exploring the earth beneath the oceans. The inter- A close relative of the marine gravity field is the
pretation of marine gravity anomalies in terms of the marine geoid. The geoid, measured in units of height,
Earth’s structure is highly nonunique, however, and is the elevation of the sea level equipotential surface.
thus requires simultaneous consideration of other Geoid anomalies are measured in meters and are the
geophysically observed quantities. The most useful departure of the true equipotential surface from that
auxiliary measurements include depth of the ocean predicted for an idealized spheroidal Earth whose
from echo sounders, the shape of buried reflectors density structure varies only with radius. Geoid
from marine seismic reflection data, and/or the anomalies range from 0 to more than 7100 m. The
density of ocean rocks as determined from dredge direction of the force of gravity is everywhere per-
samples or inferred from seismic velocities. pendicular to the geoid surface, and the magnitude of
Depending on the spatial wavelength of the ob- the gravitational attraction is the vertical derivative
served variation in the gravity field, marine gravity of the geopotential U (eqn [1]).
observations are applied to the solution of a number
of important problems in earth structure and dy- @U
namics. At the very longest wavelengths of 1000 to g¼ ½1
@z
10 000 km, the marine gravity field is usually com-
bined with anomalies over land to infer the dynamics Geoid height N is related to the same equipotential U
of the entire planet. At medium wavelengths of sev- via Brun’s formula eqn ([2]).
eral tens to hundreds of kilometers, the gravity field
contains important information on the thermal and U
N¼ ½2
mechanical properties of the lithospheric plates and g0
on the thickness of their sedimentary cover. At even
shorter wavelengths, the field reflects local ir- in which g0 is the acceleration of gravity on the
regularities in density, such as produced by seafloor spheroid.
bathymetric features, magma chambers, and buried For a ship sailing on the sea surface (the
ore bodies. On account of the large number of po- equipotential), it is easier to measure gravity. From a
tential contributions to the marine gravity field, satellite in free-fall orbit high over the Earth’s
modern methods of analysis include spectral filtering surface, radar altimeters can measure with centi-
to remove signals outside of the waveband of interest meter precision variations in the elevation of sea
and interpretation within the context of models that level, an excellent approximation to the true geoid
obey the laws of phyics. that would follow the surface of a motionless ocean.
Regardless of whether geoid or gravity is the quan-
tity measured directly, simple formulas in the wave-
Units
number domain allow gravity to be computed from
Gravity is an acceleration. The acceleration of geoid and vice versa. Given the same equipotential,
gravity on the earth’s surface is about 9.81 m s2. the gravity representation emphasizes the power in

600
 GRAVITY 601

the high-frequency (short-wavelength) part of the at other locations can be calculated by observing
spectrum, whereas the geoid representation em- how much more or less that same spring is stretched
phasizes the longer wavelengths. Therefore, for in- at other locations by the same mass. Although such
vestigations of high-frequency phenomena, gravity is systems are relatively inexpensive to build and easy
generally the quantity interpreted even if geoid is to deploy, they suffer from drift: in effect, the spring
what was measured. The opposite is true for long- constant changes with time because no physical
wavelength phenomena. spring is perfectly elastic. To first order, the drift can
be corrected by returning to the same or another base
station with the same instrument, and assuming that
Measurement of Marine Gravity the drift was linear with time in between. The ac-
Marine gravity measurements can be and have been curacy of this linear drift assumption improves with
acquired with several different sorts of sensors and more frequent visits to the base station, but this is
from a variety of platforms, including ships, sub- usually impractical for marine surveys. Through
marines, airplanes, and satellites. The ideal combin- clever design, the latest generation of marine grav-
ation of sensor system and platform depends upon imeters has greatly reduced the drift problem as
the needed accuracy, spatial coverage, and available compared with earlier instruments.
time and funds. The measurement of the gravity gradient tensor
was widely used early in the twentieth century for oil
Gravimeters exploration, but fell into disfavor in the 1930s as
scalar gravimeters became more reliable and easy to
The design for most marine gravimeters is borrowed
use. Gravity gradiometry at sea is currently making a
from their terrestrial counterparts and are either
comeback as the result of declassification of military
absolute or relative in their measurements. Absolute
gradiometer technology developed for use in sub-
gravimeters measure the full acceleration of gravity g
marines during the Cold War. Gravity gradiometers
at the survey site along the direction of the local
measure the three-dimensional gradient in the grav-
vertical. Modern marine absolute gravimeters meas-
ity vector using six pairs of aligned gravimeters, with
ure precisely the vertical position z of a falling mass
accuracies reaching better than 1 Eötvös. In com-
(e.g., a corner cube reflector) as a function of time t
parison with measurements of gravity, the gravity
in a vacuum cylinder using laser interferometry and
gradient has more sensitivity to variations at short
an atomic clock. The acceleration is then calculated
wavelenghts (B5 km or less), making it useful for
as the second derivative of the position of the falling
delimiting shallow structures buried beneath the
mass as a function of time (eqn [3]).
seafloor.
Geoid anomalies can be directly measured from
d2 zðtÞ orbiting satellites carrying radar altimeters. The al-
g¼ ½3
dt2 timeters measure the travel time of a radar pulse
Absolute gravimeters tend to be larger, more difficult from the satellite to the ocean surface, from which it
to deploy, costlier to build, and more expensive to is reflected and bounced back to the satellite.
run than relative gravimeters, and thus are only used Tracking stations on Earth solve for the position of
when relative gravimeters are inadequate for the the satellite with respect to the center of the Earth.
problem being addressed. These two types of information are then combined to
Most gravity measurements at sea are relative calculate the height of the sea level equipotential
measurements, Dg: the instrument measures the dif- surface above the center of the Earth. Because the
ference between gravity at the study site and at an- solid land surface does not follow an equipotential,
other site where absolute gravity is known (e.g., the altimeters cannot be used to constrain the terrestrial
dock where the expedition originates). Modern geoid. Furthermore, it is difficult to extract geoid
relative gravimeters are based on Hooke’s law for the from ocean areas covered by sea ice. However, in the
force F required to extend a spring a distance x (eqn near future, laser altimeters deployed from satellites
[4]), where ks is the spring constant, calculated by hold the promise of extracting geoid information
extending the spring under a known force. even over ocean surfaces marred with sea ice, on
account of their enhanced resolution.
F ¼ ks x ¼ mg ½4
Platforms
If a mass m is suspended from this spring at a site
where gravity g is known (e.g., by deploying an ab- Marine gravity data can be acquired either from
solute gravimeter at that base station), then gravity moving or from stationary platforms. Because the
602 GRAVITY

gravity field from variations in the depth of the sea A second correction that must be made if the
floor is such a large component of the observed sig- gravity is measured from a moving vehicle, such as a
nal, most marine gravity surveys have relied on ships ship or airplane, accounts for the effect on gravity of
or submarines that enable the simultaneous acqui- the motion of the vehicle with respect to the Earth’s
sition of depth observations. However, airborne spin. A ship steaming to the east is, in effect, rotating
gravity measurements have been acquired success- faster than the Earth. The centrifugal effect of this
fully over ice-covered areas of the polar oceans, and increased rate of rotation causes gravity to be less
orbiting satellites have measured the marine geoid than it would be if the ship were stationary. The
from space. opposite effect occurs for a ship steaming to the west.
A major challenge in acquiring gravity data from a This term, called the Eötvös correction, is largest near
floating platform at the sea surface is in separating the Equator and involves only the east–west com-
the acceleration of the platform in the dynamic ocean ponent vEW of the ship’s velocity vector (eqn [6]),
from the acceleration of gravity. This problem is in which o is the angular velocity of the Earth’s
overcome by mounting marine gravimeters deployed rotation.
from ships on inertially stabilized tables. These tables
employ gyroscopes to maintain a constant attitude gEOT ¼ 2ovEW cos Y ½6
despite the pitching and rolling of the ship beneath
the table. The nongravitational acceleration is The free air gravity correction, which accounts for
somewhat mitigated by mounting marine grav- the elevation of the measurement above the Earth’s
imeters deep in the hold and as close to the ship’s sea level equipotential surface, is obviously not nee-
center of motion as possible. Special damping ded if the measurement is made on the sea surface.
mechanisms also prevent the spring in the gravimeter The free air correction gFA is required if the meas-
from responding to extremely high-frequency chan- urement is made from a submersible or an airplane:
ges in the force on the suspended mass. eqn [7], where h is elevation above sea level in
Instruments deployed in submersibles resting on meters.
the bottom of the ocean or in instrument packages
lowered to the bottom of the ocean do not suffer gFA ¼ 3:1h gu ½7
from the dynamic accelerations of the moving ocean This correction is added to the observation if the
surface, but bottom currents can also be an import- sensor is deployed above the Earth’s surface, and
ant source of noise in submarine gravimetry. In- subtracted for stations below sea level.
stalling instruments in boreholes is the most effective For land surveys, the Bouguer correction accounts
way to counter this problem, but it is also an ex- for the extra mass of the topography between the
pensive solution. observation and sea level. For its marine equivalent, it
adds in the extra gravitational attraction that would
Reduction of Marine Data be present if rock rather than water existed between
sea level and the bottom of the ocean. Except in areas
A number of standard corrections must be applied to of rugged bathymetry, the Bouguer correction gB is
the raw gravity data (either g or Dg) prior to inter- calculated using the slab formula (eqn [8]).
pretation. In addition to any drift correction, as
mentioned above for relative gravity measurements, gB ¼ 2pDrGz ½8
a latitude correction is immediately applied to ac-
count for the large change in gravity between the Here Dr is the density difference between oceanic
poles and the Equator caused by Earth’s rotation. crust and sea water, G is Newton’s constant, and z is
Near the Equator, the centrifugal acceleration from the depth of the sea floor. This correction is seldom
the Earth’s spin is large, and gravity is about 50 000 used because it produces very large positive gravity
gu less, on average, than at the poles. Because this anomalies. Furthermore, there are more accurate
effect is 5000 times larger than typical regional corrections for the effect of bathymetry that do not
gravity signals of interest, it must be removed from make the unrealistic assumption that the expected
the data using a standard formula for the variation of state for the oceans should be that the entire depth is
gravity g0 on a spheroid of revolution best fitting the filled with crustal rocks displacing the water. The
shape of the Earth (eqn [5]; Y ¼ latitude). Bouguer correction is necessary, however, when
gravity measurements are made from a submarine, in
 order to combine those data with more conventional
g0 ðYÞ ¼ 9:7803185 1 þ 5:278895  103 sin2 Y
 ½5 observations from the sea surface. In this case, the
þ2:3462  105 sin4 Y ms2 Bouguer correction is applied twice: once to remove
 GRAVITY 603

the upward attraction of the layer of water above the time series of variations in gravity over the oceans
submarine, and once more to add in that layer’s from vessels under way. Without any need to stop the
gravitation field below the sensor. ship on station, a time series of gravity measurements
Satellite measurements of sea surface height go could be obtained at only small incremental cost to
through a different processing sequence to recover ship operations. With the advent of the new instru-
marine geoid anomalies. The most important step is mentation, the catalogue of marine gravity values
in calculating precise orbits. Information from has grown in the past 40 years to more than 2.5
tracking stations is supplemented with a ‘crossover million measurements.
analysis’ that removes long-wavelength bias in orbit A new era of precision in marine gravity began
elevation by forcing the height valves to agree with the advent of the Global Positioning System
wherever orbits cross. Corrections are then made for (GPS) in the late 1980s. Prior to this time, the largest
known physical oceanographic effects such as tides, source of uncertainty in marine gravity lay in the
and wave action is averaged out. The height of the Eötvös correction. Older navigation systems (dead
sea level geoid above the Earth’s center, assuming the reckoning, celestial, and even the TRANSIT satellite
standard spheroid, is subtracted from the data to system) were too imprecise in the absolute position
create geoid anomalies. of the ship and too infrequently available to allow
accurate velocity estimation from minute to minute,
especially if the ship was maneuvering. Typically,
History gravity data had to be discarded for an hour or so
A principal impediment to the acquisition of useful near the time of any change in course. The high
gravity observations at sea was the difficulty in sep- positioning accuracy and frequency of GPS fixes now
arating the desired acceleration of gravity from the allows such precise calculation of the Eötvös cor-
acceleration of the platform floating on the surface of rection that it is no longer the limiting factor in the
the moving ocean. For this reason, the first successful accuracy of marine gravity data.
gravity measurements to be acquired at sea were A breakthough in determining the global marine
taken from a submarine by the Dutch pioneer, gravity field was achieved with the launching of the
Vening Meinesz, in 1923. He used a pendulum GEOS-3 (1975–1977) and Seasat (1978) satellites,
gravimeter, which was the state of the art for meas- which carried radar altimeters. Altimeters were de-
uring absolute gravity at that time. By accurately ployed for the purpose of measuring dynamic sea
timing the period, T, of the swinging pendulum, the surface elevation associated with physical oceano-
acceleration of gravity, g, can be recovered according graphic effects. The Seasat satellite carried a new,
to eqn [9], in which l is the length of the pendulum high-precision altimeter that characterized the vari-
arm. ations in sea surface elevation with unprecedented
detail. The satellite failed prematurely, but not before
sffiffiffi it returned a wealth of data on the marine gravity
l field from its observations of the marine geoid. The
T ¼ 2p ½9
g geoid variations at mid- and short-wavelength were
so large that the dynamic oceanographic effects
By 1959, five thousand gravity measurements had motivating the mission could be considered a much
been acquired from submarines globally. These smaller noise term. The success of the Seasat mission
measurements were instrumental in revealing the led to the launch of Geosat, which measured the
large gravity anomalies associated with the great geoid at even higher precision and resolution. Un-
trenches along the western margin of the Pacific. fortunately, most of that data remained classified by
However, these gravity observations were very time- the US military until the results from a similar
consuming to acquire because of the long integration European mission were about to be released into the
times needed to achieve a high-precision estimate of public domain. The declassification of the Geosat
the pendulum’s period, and could not be adapted for data in 1995 fueled a major revolution in our
use on a surface ship. understanding of the deep seafloor (Figure 1).
Gravity measurements at sea became routine and The latest developments in marine gravity stem
reliable in the late 1950s with the development of from the desire to detect the shortest spatial wave-
gyroscopically stabilized platforms and heavily lengths of gravity variations by taking gravimeters to
damped mass-and-spring systems constrained to the bottom of the ocean. Gravity is one example of a
move only vertically. The new platforms compen- potential field, and as such the amplitude, A, of the
sated for the pitch and roll of the ship such that signal of interest decays with distance, z, between
simple mass-and-spring gravimeters could collect source and detector as in eqn [10], where k is the
604 GRAVITY

Figure 1 Gravity field over the North Pacific. This view is based on satellite altimetry data from the Geosat and other missions. (Data
from Sandwell and Smith (1997).)

modulus of the spatial wavenumber, the reciprocal of least ambiguous to interpret since they invariably are
the spatial wavelength. of shallow origin. The upward continuation factor
guarantees that any spatially localized anomalies
ABe2pkz ½10 with deep sources will be undetectable at the ocean
surface. Near-bottom gravity measurements are able
For sensors located on a ship at the sea surface in to improve somewhat the detection of concentrated
average ocean depths of 4.5 km, it is extremely dif- density anomalies buried at deeper levels, but most
ficult to detect short-wavelength variations in gravity are assumed to lie within the oceanic crust.
of a few kilometers or less. Even lowering the grav- One of the most useful applications of short-
imeter to the cruising depth of most submarines (a wavelength gravity anomalies has been to predict
few hundred meters) does little to overcome the up- ocean bathymetry (Figure 2). Radar altimeters de-
ward attenuation of the signal from localized sources ployed on the Seasat and Geosat missions measured
on and beneath the seafloor. The solution to this with centimeter accuracy the height of the underlying
problem recently has been to take gravimeters to the sea surface, an excellent approximation to the mar-
bottom of the ocean, either in a deep-diving sub- ine geoid, over all ice-free marine regions. The ac-
mersible such as Alvin, or as an instrument package curacy and spatial coverage was far better than had
lowered on a cable. Most gravity measurements at been provided from more than a century of marine
sea are relative measurements. However, recent ad- surveys from ships. At short wavelengths, undu-
vances in instrumentation now allow absolute grav- lations of the rock–water interface are the largest
imeters to be deployed on the bottom of the ocean, contribution to the short-wavelength portion of the
avoiding the problem of instrument drift that adds geoid spectrum, which opened up the possibility of
error to relative gravity measurements. However, predicting ocean depth from the excellent geoid data.
noise associated with the short baselines required for For example, an undersea volcano, or seamount,
operation in the deep sea remains problematic. represents a mass excess over the water it displaces.
The extra mass locally raises the equipotential sur-
face, such that positive geoid anomalies are seen over
Interpretation of Marine Gravity volcanoes and ridges while geoid lows are seen over
narrow deeps and trenches. The prediction of ba-
Short-Wavelength Anomalies
thymetry from marine geoid or gravity data is tricky:
The shortest-wavelength gravity anomalies over the the highest frequencies in the bathymetry cannot be
oceans (less than a few tens of kilometers) are the estimated because of the upward attenuation
 GRAVITY 605

(A)

(B)

Figure 2 Example of bathymetric prediction from gravity anomalies in a largely unexplored region of the South Pacific. (A) The best
available bathymetry from sparse echo soundings available in the early 1990s. (B) A diagram shows a dramatic improvement in
definition of the bathymetry when satellite gravity observations are used to constrain the short-wavelength component of the
bathymetry. (Adapted from McNutt and Bonneville (1996).)

problem, and the longer wavelengths are canceled methods of isostatic compensation, this mechanism
out in the geoid by their isostatic compensation (see predicts the smallest gravity anomalies over a given
following section). These longer wavelengths in the feature. From analysis of marine gravity, we now
bathymetry must be introduced into the solution know that this sort of compensation mechanism is
using traditional echo soundings from sparse ship only found where the oceanic crust is extremely
tracks. Nevertheless, the best map we currently have weak, such as on very young lithosphere near a
of the depth of the global ocean is courtesy of sat- midocean ridge. For example, large plateaus formed
ellite altimetry. when hotspots intersect midocean ridges are largely
supported by Airy-type isostasy. Elsewhere the
oceanic lithosphere is strong enough to exhibit some
Mid-Wavelength Anomalies
lateral strength in supporting superimposed vol-
The mid-wavelength part of the gravity spectrum canoes and other surface loads.
(tens to hundreds of kilometers) is dominated by the An extremely common form for support of bathy-
effects of isostatic compensation. Isostasy is the metric features in the oceans is elastic flexure. Oceanic
process by which the Earth supports variations in lithosphere has sufficient strength to bend elastically,
topography or bathymetry in order to bring about a thus distributing the weight of a topographic feature
condition of hydrostatic equilibrium at depth. The over an area broader than that of the feature itself
definition of isostasy can be extended to include both (Figure 3). Analysis of marine gravity has been in-
static and dynamic compensation mechanisms, but at strumental in establishing that the elastic strength of
these wavelengths the static mechanisms are most the oceanic lithosphere increases with increasing age.
important. There are a number of different types of Young lithosphere near the midocean ridge is quite
isostatic compensation at work in the oceans, and weak, in some cases hardly distinguished from Airy-
the details of the gravity field can be used to dis- type isostasy. The oldest oceanic lithosphere displays
tinguish them and to estimate the thermomechanical an effective thickness equivalent to that of a perfectly
behavior of oceanic plates. elastic plate 40 km thick. The fact that this thickness is
One of the simplest mechanisms for isostatic less than that of the commonly accepted value for the
compensation is Airy isostasy: the oceanic crust is thickness of the mechanical plate that drifts over the
thickened beneath areas of shallow bathymetry. The asthenosphere indicates that the base of the oceanic
thick crustal roots displace denser mantle material, lithosphere is not capable of sustaining large devia-
such that the elevated features float on the mantle toric stresses (of the order of 100 MPa or more) over
much like icebergs float in the ocean. Of the various million-year timescales.
606 GRAVITY

Net gravity anomaly

Contribution from volcano

Contribution from
flexure of plate

Volcano

Light crust

Elastic plate
Dense mantle

Figure 3 Cartoon showing how the seafloor is warped as an elastic plate under the weight of a small volcano. The gravity anomaly
that would be detected by a ship sailing along the sea surface over this feature is the net difference between the positive gravity
perturbation from the extra mass of the volcano and the negative gravity perturbation produced when the elastically flexed light crust
replaces denser mantle. (Adapted from McNutt and Bonneville (1996).)

Another very important method of isostatic com- The principal signal at these wavelengths arises
pensation in the ocean is Pratt isostasy. This method from the subduction of lithospheric slabs and other
of support supposes that the height of a vertical sorts of convective overturn within the mantle. Three
column of bathymetry is inversely proportional to its sorts of gravitational contributions must be con-
density. Low-density columns can be higher because sidered: (1) the direct effect of mass anomalies within
they are lighter, whereas heavy columns must be the mantle, either buoyant risers or dense sinkers
short in order to produce the same integrated mass at which drive convection; (2) the warping of the sur-
some assumed depth of compensation. In the oceans, face caused by viscous coupling of the risers or
variations in the temperature of the lithosphere sinkers to the earth’s surface; and (3) the warping of
produce elevation changes in the manner of Pratt any deeper density discontinuities (such as the core–
isostasy. For example, ridges stand 4 km above the mantle boundary) also caused by viscous coupling.
deep ocean basins because the underlying lithosphere In the 1980s, estimates of the locations and dens-
is hotter when the plate is young. The bathymetric ities of mass anomalies in the mantle responsible for
swells around young hotspot volcanoes may also be the first contribution above began to become avail-
supported by Pratt-type isostasy, although some able courtesy of seismic tomography. Travel times of
combination of crustal thickening and dynamic earthquake waves constrained the locations of seis-
isostasy may be operating as well. Again, gravity and mically fast and slow regions in the mantle. By as-
geoid anomalies have been principal constraints in suming that the seismic velocity variations were
arguing for the mechanism of support for bathy- caused by temperature differences between hot, ris-
metric swells. ing material and cold, sinking material, it was pos-
sible to convert velocity to density using standard
relations. Knowing the locations of the mass anom-
alies driving convection inside the Earth led to a
Long-Wavelength Anomalies
breakthrough in understanding the long-wavelength
At wavelengths from 1000 to several thousand gravity and geoid fields.
kilometers, gravity anomalies are usually derived The amount of deformation on density interfaces
from satellite observations and interpreted using above and below the mass anomalies inferred from
equations appropriate for a spherical earth. Geoid is tomography (contributions (2) and (3) above) de-
interpreted more commonly than gravity directly, as pends upon the viscosity structure of Earth’s mantle.
it emphasizes the longer wavelengths in the geopo- Coupling is more efficient with a more viscous
tential field. Isostatic compensation for smaller-scale mantle, whereas a weaker mantle is able to soften the
bathymetric features, such as seamounts, can usually transmission of the viscous stresses from the risers
be ignored in that the gravity anomaly from and sinkers. Therefore, one of the principal uses of
bathymetry is canceled out by that from its com- marine gravity anomalies at long wavelengths has
pensation when spatially averaged over longer been to calibrate the viscosity structure of the
wavelengths. oceanic upper mantle. This interpretation must be
 GRAVITY 607

constrained by estimates of the dynamic surface McNutt MK and Bonneville A (1996) Mapping the
topography over the oceans, which is actually easier seafloor from space. Endeavour 20: 157--161.
to estimate than over the continents because of the McNutt MK and Bonneville A (2000) Chemical origin for
relatively uniform thickness of oceanic crust. the Marquesas swell. Geochemistry, Geophysics and
A fairly common result from this sort of analysis is Geosystems 1.
Sandwell DT and Smith WHF (1997) Marine gravity for
that the oceanic upper mantle must be relatively invis-
Geosat and ERS-1 altimetry. Journal of Geophysical
cid. The geoid shows that there are large mass anom- Research 102: 10039--10054.
alies within the mantle driving convection that are Smith WHF and Sandwell DT (1997) Global seafloor
poorly coupled to variations in the depth of the seafloor. topography from satellite altimetry and ship depth
If the upper mantle were more viscous, there should be soundings. Science 277: 1956--1962.
a stronger positive correlation between marine geoid Turcotte DL and Schubert G (1982) Geodynamics:
and depth of the seafloor at long wavelengths. Applications of Continuum Physics to Geological
Problems. New York: Wiley.
Watts AB, Bodine JH, and Ribe NM (1980) Observations
See also of flexure and the geological evolution of the Pacific
Ocean basin. Nature 283: 532--537.
Manned Submersibles, Deep Water. Satellite Wessel P and Watts AB (1988) On the accuracy of marine
Altimetry. gravity measurements. Journal of Geophysical Research
93: 393--413.
Zumberg MA, Hildebrand JA, Stevenson JM, et al. (1991)
Further Reading Submarine measurement of the Newtonian
gravitational constant. Physical Review Letters 67:
Bell RE, Anderson R, and Pratson L (1997) Gravity 3051--3054.
gradiometry resurfaces. The Leading Edge 16: 55--59. Zumberg MA, Ridgeway JR, and Hildebrand JA (1997) A
Garland GD (1965) The Earth’s Shape and Gravity. New towed marine gravity meter for near-bottom surveys.
York: Pergamon Press. Geophysics 62: 1386--1393.
APPENDICES
 APPENDIX 1. SI UNITS AND SOME EQUIVALENCES

Wherever possible the units used are those of the International System of Units (SI). Other ‘‘conventional’’
units (such as the liter or calorie) are frequently used, especially in reporting data from earlier work.
Recommendations on standardized scientific terminology and units are published periodically by inter-
national committees, but adherence to these remains poor in practice. Conversion between units often
requires great care.
612 APPENDIX 1. SI UNITS AND SOME EQUIVALENCES

SI base units and derived units may be used with multiplying prefixes (with the exception of kg, though
prefixes may be applied to gram ¼ 103kg; for example, 1 Mg ¼ 106 g ¼ 106kg)
 APPENDIX 1. SI UNITS AND SOME EQUIVALENCES 613

The SI units for pressure is the pascal (1 Pa ¼ 1 N m2). Although the bar (1 bar ¼ 105 Pa) is also retained
for the time being, it does not belong to the SI system. Various texts and scientific papers still refer to gas
pressure in units of the torr (symbol: Torr), the bar, the conventional millimetre of mercury (symbol: mmHg),
atmospheres (symbol: atm), and pounds per square inch (symbol: psi) – although these units will gradually
disappear (see Conversions between Pressure Units).

Irradiance is also measured in W m2. Note: 1 mol photons ¼ 6.02  1023 photons.

The SI unit used for the amount of substance is the mole (symbol: mol), and for volume the SI unit is the
cubic metre (symbol: m3). It is technically correct, therefore, to refer to concentration in units of molm3.
However, because of the volumetric change that sea water experiences with depth, marine chemists prefer to
express sea water concentrations in molal units, mol kg1.
 APPENDIX 5. PROPERTIES OF SEAWATER

A5.1 The Equation of State The value at one standard atmosphere (p ¼ 0) is

given by
It is necessary to know the equation of state for the
ocean very accurately to determine stability prop-
KðS; t; 0Þ ¼ Kw þ Sð54:674 6  0:603 459t
erties, particularly in the deep ocean. The equation of
state defined by the Joint Panel on Oceanographic þ 1:099 87  102 t2  6:167 0
Tables and Standards fits available measurements  105 t3 Þ þ S3=2 ð7:944  102
with a standard error of 3.5 ppm for pressure up to
1000 bar, for temperatures between freezing and þ 1:648 3  102 t
40 1C, and for salinities between 0 and 42. The  5:300 9  104 t2 Þ ½A5:5
density r (kg m 3) is expressed in terms of pressure p
(bar), temperature t (1C), and practical salinity S. and the value at pressure p by
The last quantity is defined in such a way that its
value (in practical salinity units or psu) is very close KðS; t; pÞ ¼ KðS; t; 0Þ þ pð3:239 908 þ 1:437 13
to the old value expressed in parts per thousand (%  103 t þ 1:160 92  104 t2
or ppt). Its relation to previously defined mea-
sures of salinity is given by Lewis and Perkin.  5:779 05  107 t3 Þ
The equation for r is obtained in a sequence of þ pSð2:283 8  103  1:098 1
steps. First, the density rw of pure water (S ¼ 0) is
 105 t  1:607 8  106 t2 Þ
given by
þ 1:910 75  104 pS3=2
rw ¼ 999:842 594 þ 6:793 952  102 t þ p2 ð8:509 35  105  6:122 93
 9:095 290  103 t2 þ 1:001 685  104 t3  106 t þ 5:278 7  108 t2 Þ
6 4
 1:120 083  10 t þ p2 Sð9:934 8  107 þ 2:081 6
9 5
þ 6:536 332  10 t ½A5:1  108 t þ 9:169 7  1010 t2 Þ ½A5:6

Second, the density at one standard atmosphere Values for checking the formula are r(0, 5, 0) ¼
(effectively p ¼ 0) is given by 999.966 75, r(35, 5, 0) ¼ 1027.675 47, and r(35, 25,
1000) ¼ 1062.538 17.
rðS; t; 0Þ ¼ rw þ Sð0:824 493  4:089 9  103 t Since r is always close to 1000 kg m 3, values
þ 7:643 8  105 t2  8:246 7  107 t3 quoted are usually those of the difference (r  1000)
in kg m 3 as is done in Table A5.1. The table is
þ 5:387 5  109 t4 Þ þ S3=2 ð5:724 66 constructed so that values can be calculated for 98%
 103 þ 1:022 7  104 t  1:654 6 of the ocean (see Figure A5.1). The maximum errors
in density on straight linear interpolation are
 106 t2 Þ þ 4:831 4  104 S2 ½A5:2
0.013 kg m 3 for both temperature and pressure in-
Finally, the density at pressure p is given by terpolation and only 0.006 for salinity interpolation
in the range of salinities between 30 and 40. The
rðS; t; 0Þ error when combining all types of interpolation for
rðS; t; pÞ ¼ ½A5:3 the 98% range of values is less than 0.03 kg m 3.
1  p=KðS; t; pÞ

where K is the secant bulk modulus. The pure water

value Kw is given by
A5.2 Other Quantities Related to
Density
Kw ¼ 19 652:21 þ 148:420 6t  2:327 105t2 Older versions of the equation of state usually gave
2 3
þ 1:360 477  10 t  5:155 288 formulas not for calculating the absolute density r,
but for the ‘specific gravity’ r/rm, where rm is the
 105 t4 ½A5:4 maximum density of pure water. Since this is always
Table A5.1

p S t r  1000 @r/@S a @a/@S cp @cp/@S y @y/@S cs @cs /@S

(bar) (1C) (kg m  3) (10  7 K  1) (J kg  1 K  1) (10  31C) (m s  1)

0 35 2 28.187 0.814 254 33 3989  6.2  2000 0 1439.7 1.37

0 35 0 28.106 0.808 526 31 3987  6.1 0 0 1449.1 1.34
0 35 2 27.972 0.801 781 28 3985  5.9 2000 0 1458.1 1.31
0 35 4 27.786 0.796 1021 26 3985  5.8 4000 0 1466.6 1.29
0 35 7 27.419 0.788 1357 23 3985  5.6 7000 0 1478.7 1.25
0 35 10 26.952 0.781 1668 20 3986  5.5 10 000 0 1489.8 1.22
0 35 13 26.394 0.775 1958 17 3988  5.3 13 000 0 1500.2 1.19
0 35 16 25.748 0.769 2230 15 3991  5.2 16 000 0 1509.8 1.16
0 35 19 25.022 0.764 2489 14 3993  5.1 19 000 0 1518.7 1.13
0 35 22 24.219 0.760 2734 12 3996  4.9 22 000 0 1526.8 1.10
0 35 25 23.343 0.756 2970 11 3998  4.9 25 000 0 1534.4 1.08
0 35 28 22.397 0.752 3196 9 4000  4.8 28 000 0 1541.3 1.06
0 35 31 21.384 0.749 3413 8 4002  4.7 31 000 0 1547.6 1.03
100 35 2 32.958 0.805 552 31 3953  5.8  2029 2 1456.1 1.38
100 35 0 32.818 0.799 799 28 3953  5.7  45 2 1465.5 1.35
100 35 2 32.629 0.793 1031 26 3954  5.6 1939 2 1474.5 1.33
100 35 4 32.393 0.788 1251 24 3955  5.5 3923 2 1483.1 1.30
100 35 7 31.958 0.781 1559 21 3957  5.3 6901 1 1495.1 1.26
100 35 10 31.431 0.774 1844 18 3960  5.2 9879 1 1506.3 1.22
100 35 13 30.818 0.769 2111 16 3963  5.1 12 858 1 1516.7 1.19
100 35 16 30.126 0.763 2363 14 3967  5.0 15 838 1 1526.4 1.16
100 35 19 29.359 0.759 2603 13 3970  4.9 18 819 1 1535.3 1.13
200 35 2 37.626 0.797 834 28 3922  5.5  2076 3 1472.8 1.39
200 35 0 37.429 0.791 1058 26 3923  5.4  107 3 1482.3 1.36
200 35 2 37.187 0.786 1269 24 3925  5.3 1862 3 1491.2 1.33
200 35 4 36.903 0.781 1469 22 3927  5.2 3832 3 1499.8 1.30
200 35 7 36.402 0.774 1750 19 3931  5.1 6789 3 1511.8 1.26
300 35 2 42.191 0.789 1101 26 3893  5.2  2140 5 1489.9 1.39
300 35 0 41.941 0.783 1303 24 3896  5.1  186 5 1499.3 1.36
300 35 2 41.649 0.778 1494 22 3899  5.0 1771 5 1508.2 1.33
300 35 4 41.319 0.774 1676 20 3903  5.0 3728 5 1516.6 1.30
400 35 2 46.658 0.781 1351 24 3867  4.9  2221 7 1507.2 1.39
400 35 0 46.356 0.776 1534 22 3871  4.8  279 6 1516.5 1.36
400 35 2 46.017 0.771 1707 20 3876  4.8 1665 6 1525.3 1.33
400 35 4 45.643 0.767 1872 19 3880  4.7 3610 6 1533.7 1.30
500 35 2 51.029 0.773 1587 22 3844  4.7  2316 8 1524.8 1.38
500 35 !0 50.678 0.769 1751 20 3849  4.6  386 8 1534.0 1.35
500 35 2 50.293 0.764 1907 19 3854  4.6 1546 7 1542.7 1.32
APPENDIX 5. PROPERTIES OF SEAWATER

600 35 2 55.305 0.766 1807 20 3824  4.4  2426 9 1542.6 1.37

600 35 0 54.908 0.762 1954 18 3829  4.4  506 9 1551.6 1.34
615

600 35 2 54.481 0.758 2094 17 3835  4.4 1416 9 1560.2 1.31

616 APPENDIX 5. PROPERTIES OF SEAWATER

0° 10° 20° 30° 34 36 30 40 24 26 28 30

(28.5) (17.5)

50

2 km 2 km 2 km

4 km 4 km 4 km

t S



Figure A5.1 The ranges of temperature t (in 1C) and salinity S for 98% of the ocean as a function of depth and the corresponding
ranges of density s and potential density sy. From Bryan K and Cox MD (1972) An approximate equation of state for numerical models
of ocean circulation. Journal of Physical Oceanography 2: 510–514.

close to unity, a quantity called s was defined by The salinity expansion coefficient b can be calculated
by using the given values of qr/qS.


r 1000
1000 1 ¼ ðr  rm Þ ½A5:7
rm rm
A5.4 Specific Heat
Since The specific heat at surface pressure is given by
Millero et al. and can be calculated in two stages.
rm ¼ 999:975 kg m3 ½A5:8 First, the value in J kg  1  K  1 for fresh water is
given by
it follows that s, as defined above, is related to the
(r  1000) values by cp ð0; t; 0Þ ¼ 4217:4  3:720 283t þ 0:141 285 5t2
 2:654 387  103 t3
s ¼ ðr  1000Þ þ 0:025 ½A5:9
þ 2:093 236  105 t4 ½A5:11
that is, 0.025 must be added to the values of
Second,
(r  1000) on the table to obtain the old s value. The
notation s, (sigma tau) was used for the value of s
cp ðS; t; 0Þ ¼ cp ð0; t; 0Þ þ Sð7:644 4
calculated at zero pressure, and sy (sigma theta) for
the quantity corresponding to potential density. An- þ 0:107 276t  1:383 9  103 t2 Þ
other quantity commonly used in oceanography is þ S3=2 ð0:177 09  4:077 2  103 t
the specific volume (or steric) ‘anomaly’ d defined by
þ 5:353 9  105 t2 Þ ½A5:12
d ¼ vs ðS; t; pÞ  vs ð35; 0; pÞ ½A5:10 The formula can be checked against the result cp(40,
8 3 1 40, 0) ¼ 3981.050. The standard deviation of the
and usually reported in units of 10 m kg . algorithm fit is 0.074. Values at nonzero pressures
can be calculated by using eqn [A5.13] and the
equation of state:
A5.3 Expansion Coefficients



The thermal expansion coefficient a is given in Table @cp @ 2 vs
A5.1 in units of 10  7 K  1 along with its S deriva- ¼ T ½A5:13
@p T @T 2 p
tive. The maximum error from pressure interpolation
is 2 units, that from temperature interpolation is 3 The values in Table A5.1 are based on the above
units, and that for salinity interpolation (30oSo40) formula and a polynomial fit for higher pressures
is 2 units plus a possible round-off error of 2 units. derived from the equation of state by N.P. Fofonoff.
 APPENDIX 5. PROPERTIES OF SEAWATER 617

The intrinsic interpolation errors in the table are 0.4, Potential temperature (°C/dbar) × 10−1
0.1, and 0.3 J kg  1 K  1 for pressure, temperature, −0.10 −0.05 0 0.05 0.10 0.15 0.20 0.25 0.30

and salinity interpolation, respectively, and there are

Brunt-Väisälä frequency (cph)
additional obvious round-off errors. 0 1 2 3 4 5 6 7 8 9 10
0

25
A5.5 Potential Temperature
50
The ‘adiabatic lapse rate’ G is given by N
75
gaT
G¼ ½A5:14 100 θz
cp
125
and therefore can be calculated from the above for-
mulas. The definition of potential temperature 150

k 175
y pr
¼ ½A5:15
T p 200

Pressure (dbar) × 101

where pr is a reference pressure level (usually 1 bar) 225
and k ¼ (g  1)/g, where g is the ratio of specific heats
250
at constant pressure and at constant volume, can
then be used to obtain y. The following algorithm, 275
however, was derived by Bryden, using experimental
300
compressibility data, to give y (1C) as a function of
salinity S, temperature t (1C), and pressure p (bar) for 325
30oSo40, 2oto30, and 0opo1000:
350
4 5
yðS; t; pÞ ¼ t  pð3:650 4  10 þ 8:319 8  10 t 375
 5:406 5  10 t þ 4:027 4  109 t3 Þ
7 2
400
 pðS  35Þð1:743 9  105  2:977 8
425
 107 tÞ  p2 ð8:930 9  107  3:162 8
450
 108 t þ 2:198 7  1010 t2 Þ þ 4:105 7
475
 109 ðS  35Þp2  p3 ð1:605 6 1010
þ 5:048 4  1012 tÞ ½A5:16 500

Figure A5.2 A profile of buoyancy frequency N in the ocean.

A check value is y(25, 10, 1000) ¼ 8.467 851 6, and
From the North Atlantic near 281 N, 701 W, courtesy of Dr. R.C.
the standard deviation of Bryden’s polynomial fit was Millard.
0.001 K. Values in Table A5.1 are given in milli-
degrees, the intrinsic interpolation errors being 2, 0oto40, 0opo1000 with a standard deviation of
0.3, and 0 millidegrees for pressure, temperature, 0.19 ms  1. Values in the table are given in meters per
and salinity interpolation, respectively (Figure A5.2). second, the intrinsic interpolation errors being 0.05,
0.10, and 0.04 ms  1 for pressure, temperature, and
salinity interpolation, respectively.
A5.6 Speed of Sound
The speed of sound cs can be calculated from the
equation of state, using eqn [A5.17] A5.7 Freezing Point of Sea Water

 The freezing point tf of sea water (1C) is given by
@p
c2s ¼ ½A5:17
@r y;S
tf ðS; pÞ ¼  0:057 5S þ 1:710 523
Values given in Table A5.1 use algorithms derived  103 S3=2  2:154 996
by Chen and Millero on the basis of direct meas-
urements. The formula applies for 0oSo40,  104 S2  7:53  103 p ½A5:18
618 APPENDIX 5. PROPERTIES OF SEAWATER

The formula fits measurements to an accuracy of Kraus EB (1972) Atmosphere–Ocean Interaction. London:
70.004 K. Oxford University Press.
Lewis EL and Perkin RG (1981) The Practical Salinity
Scale 1978: Conversion of existing data. Deep Sea
Research 28A: 307--328.
Further Reading Millero FJ (1978) Freezing point of seawater. In: Eighth
Report of the Joint Panel on Oceanographic Tables and
Bryan K and Cox MD (1972) An approximate equation of Standards, UNESCO Technical Papers in Marine
state for numerical models of ocean circulation. Journal Science No. 28, Annex 6. Paris: UNESCO.
of Physical Oceanography 2: 510--514. Millero FJ, Chen C-T, Bradshaw A, and Schleicher K
Bryden HL (1973) New polynomials for thermal expansion, (1980) A new high pressure equation of state for
adiabatic temperature gradient and potential tempera- seawater. Deep Sea Research 27A: 255--264.
ture gradient of sea water. Deep Sea Research 20: Millero FJ and Poisson A (1981) International one-
401--408. atmosphere equation of state for seawater. Deep Sea
Chen C-T and Millero FJ (1977) Speed of sound in sea- Research 28A: 625--629.
water at high pressures. Journal of the Acoustical Millero FJ, Perron G, and Desnoyers JE (1973) Heat
Society of America 62: 1129--1135. capacity of seawater solutions from 5 to 25 1C and 0.5
Dauphinee TM (1980) Introduction to the special issue on to 22% chlorinity. Journal of Geophysical Research 78:
the Practical Salinity Scale 1978. IEEE, Journal of 4499--4507.
Oceanic Engineering OE 5: 1--2. UNESCO (1981) Tenth Report of the Joint Panel on
Gill AE (1982) Atmosphere–Ocean Dynamics, Inter- Oceanographic Tables and Standards, UNESCO
national Geophysics Series Volume 30. San Diego, CA: Technical Papers in Marine Science No. 36. Paris:
Academic Press. UNESCO.
 INDEX

Notes

Cross-reference terms in italics are general cross-references, or refer to subentry terms within the main entry
(the main entry is not repeated to save space). Readers are also advised to refer to the end of each article for
additional cross-references - not all of these cross-references have been included in the index cross-references.

The index is arranged in set-out style with a maximum of three levels of heading. Major discussion of a
subject is indicated by bold page numbers. Page numbers suffixed by T and F refer to Tables and Figures
respectively. vs. indicates a comparison.

This index is in letter-by-letter order, whereby hyphens and spaces within index headings are ignored in the
alphabetization. For example, ‘oceanography’ is alphabetized before ‘ocean optics.’ Prefixes and terms in
parentheses are excluded from the initial alphabetization.

Where index subentries and sub-subentries pertaining to a subject have the same page number, they have
been listed to indicate the comprehensiveness of the text.

Abbreviations used in subentries

AUV - autonomous underwater vehicle

CPR - continuous plankton recorder
DOC - dissolved organic carbon
ENSO - El Niño Southern Oscillation
MOR - mid-ocean ridge
ROV - remotely operated vehicle
SAR - synthetic aperture radar
SST - sea surface temperature

Additional abbreviations are to be found within the index.

A response characteristics, 242–243 Accelerometer, expendable bottom

sensor design, 243 penetrometer probe (XBP), 46
Aanderaa RCM4 deep ocean rotor-vane see also Absorption spectra; Optical Accessory pigments, 258
instrument, 116F fibers; Reflectance spectroscopy Acetylcholine esterase, pesticide sensors
Aanderaa RCM8 current meter, 117T Absorption (optical) coefficient, 272–273, and, 245
Aanderaa RCM9 current meter, 116F, 323–324, 493 Ac-meters, 274
117T components, 494 Acoustical systems see Acoustic systems
ABE see Autonomous Benthic Explorer definition, 315–316 Acoustic backscatter, diffuse seafloor
ABE Autonomous Benthic Explorer detritus, 273 flows, 376–377
(AUV), 129F, 137, 138F, 163F gelbstoff, 273 temporal correlation, 377–378
ABS see Acoustic backscatter system Hydrolight simulation, 319–320, Acoustic backscatter system (ABS), 344,
(ABS) 319F 344F
Absolute surface temperature, 584 oceanic water, 251F, 253 STABLE II, 351, 352
Absorbance (of EM radiation by a as bio-optical model quantity, 248T, Acoustic Doppler current meters,
compound), 239–240 249 116–118, 116F, 117T, 234
see also Absorbance spectroscopy phytoplankton, 273 Acoustic Doppler current profiler
Absorbance spectroscopy, chemical pure sea water, 273 (ADCP), 204, 347
sensors, 239–246 sewage plume waters, 276–279, 279F see also Moorings; Single Point Current
background subtraction, 244 see also Ocean optics Meters
basic principles, 239–240 Absorption spectra Acoustic Doppler velocimeter (ADV), 347
molecular recognition element, 242–243 chlorophyll, 573F Acoustic floats, 92, 94–95
optical fibers for, 240–241 compared with fluorescence spectra, Acoustic imaging, high-frequency, 433
optoelectronics, 241–242 258F Acoustic measurement, sediment
reagent support material, 242 gelbstoff, 494, 573F transport, 343–356

619
620 INDEX

Acoustic modems, and moorings, 105, refracted waves, 389F, 392–394 porosity, 381–383, 386, 386F, 387F
105F Herglotz-Bateman-Wieckert inte- sediment transport, 380
Acoustic navigation gration, 393–394, 394F sediment type
autonomous underwater vehicles (AUV), Acoustic ripple profiler (ARP), 344, 344F, clay, 395F
167 345, 345F, 351, 355F mud, 388T
see also Sonar 3-D, 345, 346F, 355 sand, 387F, 388T, 395F
Acoustic navigation systems, manned bed profile, 352F seismo-acoustic waves see
submersibles (deep water), 152 Acoustic ripple scanner (ARS), 343–344, Seismo-acoustic waves
Acoustic noise, 357–366 344F, 345, 346F shear modulus, 383, 383T, 389–391
ambient noise, 357 Acoustics, Arctic, 397–405 water-sediment interface, seismo-
active sonar systems, 443 acoustic environment, 406 acoustic waves in, 383–384, 384F
measurement data, 357, 358F acoustic modes, 401–402, 402–403 see also Acoustics, deep ocean;
hydrophone flow noise, 360 current research, 404 Acoustics, shallow water; Benthic
microseism noise, 359 history, 397 boundary layer (BBL)
molecular noise, 358F, 365 modal dispersion, 403F Acoustics, shallow water, 417–424
prevailing noise, definition, 357 noise, 400–403 acoustic environment, 417
reflection, 363 sound propagation, 399–400, 403F low-frequency cutoff, 422
scattering, 361, 363 sound speed structure, 397–399 modeling, 423, 424F
sources, 357 transmission loss, 401F ray paths, 417–418, 418F, 423F
atmospheric turbulence, 359, 360 variability, 400 time domain, 422–423
distant shipping see Ship(s) Acoustics, deep ocean, 406–416 geometrical dispersion, 423
molecular agitation, 358F, 365 noise, 410–412 measured pulse arrivals, 423, 423F
spatially discrete, 357 noise spectra, 413F time dispersion, 423
temporally intermittent, 357 reflection, 410F transmission loss, 417–418
whale vocalizations, 360, 361–362 refraction, 408 bottom reflection, 419, 420F
swell, 359 scattering, 409–410 boundary scattering, 419–420
topographic influences, 361 sonar equation, 412–414 data, 420–421
transmission loss, 360–361, 361F sound propagation, 406–407 frequency dependence, 421–423,
ultralow frequency band (ULF) see bottom loss, 408–409 422F
Ultralow frequency band (ULF) Lloyd mirror effect, 407, 407F geometrical spreading, 418–419
Urick, R J, 357 long-range, 407 seasonal variability, 417–418
vertical line array (VLA) data, 362F, models, 409, 411F seawater attenuation, 419
363–364, 364 paths, 408F variability, 422F
waveguide, 359, 362–363, 363, short-range, 407 Acoustic scattering (marine organisms),
363–364 spreading loss, 407–408 367–375
Wenz, G M, 357, 358F volume attenuation, 408, 409F area backscattering coefficient, 369
see also Acoustics, Arctic; Acoustics, units, 406 bandwidth, 375
deep ocean; Acoustics, shallow Acoustics, marine sediments, 380–396 bistatic cross-section, 368–369
water; Ship(s) attenuation, 381–383, 385, 387, 395 challenges, 374–375
Acoustic rays bulk density, 383, 383T, 385 clams, 374
diffraction, 452–453 bulk modulus, 383, 383T, 385 extinction cross-section, 369, 372
refracted-refracted, 453 forward modeling, 395, 395F fish, 370
refracted-surface-refracted, 453 measurement of geoacoustic parameters, groups of organisms, 369
sampling properties, 455–456, 457F 385–386 historical overview, 367
travel time, 452–454, 454 laboratory, sediment core samples see imaging, 375
vertical travel-time ambiguity, 464F Sediment core samples jellyfish, 373–374
Acoustic release, moorings, 98 remote sensing see Acoustic remote mammals, 374
Acoustic remote sensing, 388–391 sensing organism classification, 368
characteristic shallow water signal, 389, transmission loss technique, 394–395, orientational sensitivity and frequency,
389F 395F 370
data inversion, 392–393 relationship between geoacoustic parameters, 368
experiment setup, 392F parameters, 386 extrinsic, 368
geophone arrays, 392–393, 392F, 393F porosity and relative compressional intrinsic, 368
interface waves, 389F, 394 wave velocity, 386, 387F physical basis, 367–368
dispersion diagram, 394, 394F relative density and porosity, 386, quantification, 368
reflected waves, 389–391 386F measurement, 368–369
angle of intromission case, 391–392 SAFARI, 395, 395F modeling, 369–370
critical angle case, 390F, 391 seafloor roughness, 385 nomenclature, 368–369
grazing angle, 380, 389–391, 389F, sedimentary acoustic medium, 380, 383 single organisms, 368–369
390F sediment structure, 380, 380–381 squid, 373–374
half-space seafloor, 389–391, 389F composition, 380, 381F target strength, 369
layered seafloor, 391F, 392, 392F density, 380–381, 381F, 386, 386F volume scattering coefficient, 369
phase shift, 391–392, 391F deposition, 380 zooplankton, 372
reflection coefficient, 389–391, fluid saturation, 381–383 gas-filled bodies, 373
391–392, 391F layering, 380, 380–381, 382F, hard-shelled, 373
reflection loss, 391F, 392, 392F 383–384, 384, 391F, 392 liquid-like, 372–373
 INDEX 621

Acoustic scintillation thermography Air see also Satellite oceanography

(AST), 376–379 temperature Airfoil
application, 377–379 satellite remote sensing, 594–595, 595 probes, 232–233
error, 379 Airborne laser fluorosensing, 471–472 Air-sea gas exchange
method of operation, 376–377, 376F Airborne Oceanographic Lidar (AOL), satellite remote sensing of SST
noise, 378 472 application, 558–559
temporal correlation, 377–378 Airborne Topographic Mapper (ATM), Air-sea heat flux
flow characteristics and, 378 473, 474F transfer processes, 590
Acoustic survey work, sensors, towed Airborne Visible/Infrared Imaging see also Radiative fluxes
vehicles, 68–69 Spectrometer (AVIRIS), 475–477 Air-segmented continuous flow analyzers,
Acoustic systems accomplishments, 477 335
high-frequency, zooplankton sampling, design based on satellite system, 475 applications, shipborne laboratories,
83 high altitude and high speed flights, 335–336, 336
oceanographic research vessels, 6 475–476 disadvantages, 335–336
Acoustic Thermometry of Ocean Climate see also Aircraft for remote sensing methods to overcome, 336
(ATOC), 465–466 Air conditioning flow injection analysis vs., 336
Acoustic transponder navigation, oceanographic research vessels, 6 technique, 335
204–205, 207 Aircraft for remote sensing, 471–479 total oxidised nitrogen determinations,
Acoustic travel time (ATT) current active systems/sensors, 471 335, 335F
meters, 117T, 119–120, 121F airborne laser fluorosensing, 471–472 Airy isostasy, 605
applications, 120 Airborne Oceanographic Lidar AIW see Arctic Intermediate Water (AIW)
disadvantages, 120 (AOL), 472 ALACE see Autonomous Lagrangian
techniques, 120 examples of data from AOL, 472, Circulation Explorer; Autonomous
ADCP see Acoustic Doppler Current 473F Lagrangian Circulation Explorer
Profiler (ADCP) laser-induced fluorescence, 472 (ALACE)
ADEOS-I see Advanced Earth Observing airborne lidar coastal mapping, 472 Alaska
Satellite Airborne Topographic Mapper oil spills, satellite remote sensing,
ADEOS II (Advanced Earth Observing (ATM), 473, 474F 561F
Satellite II), 526 GPS kinematic differential, 472 Albatross,
Adiabatic lapse rate, seawater, 617 other airborne lidar systems, 473–474 US Fish Commission, 4–5
Advanced ATSR (AATSR), 559 system accuracy, 472–473 Albedo, 272
Advanced Earth Observing Satellite pump and probe fluorometry, single-scattering, 316–317
(ADEOS), 591 472–474 Albert I of Monaco, 90
ocean color sensors, 575T determining phytoplankton Algae
Advanced Earth Observing Satellite II photosynthesis, 474 acoustic scattering, 374
(ADEOS II), 526 advantages over satellites, 471 productivity measures, 260–262
Advanced Land Observing Satellite applications to oceanography, 471 see also Algal blooms; Cyanobacteria
(ALOS), 560 disadvantages, 471 Algal blooms
Advanced Microwave Scanning hyper-spectral systems, 475 harmful, monitoring, CPR survey, 23
Radiometer (AMSR), 526, 594 Airborne Visible/Infrared Imaging monitoring, CPR survey, 23
Advanced microwave sounding unit Spectrometer (AVIRIS), 475–477 satellite remote sensing, 499
(AMSU-A), 595 AVIRIS see Airborne Visible/Infrared see also Red tides
Advanced piston corer (APC), 38, 39F Imaging Spectrometer (AVIRIS) Algorithms
Advanced Scatterometer (ASCAT), 591, Compact Airborne Spectographic bio-optical, 495, 577
592T Imager (CASI), 475–477 chlorophyll a, ocean color by satellite
Advanced Very High Resolution Portable Hyper-spectral Imager for remote sensing, 577, 578F
Radiometer (AVHRR), 216, 549, Low-Light Spectroscopy (PHILLS), infrared atmospheric correction, SST
552–554 477 measurement, 550
channels, 551 instrumentation for studying ocean NASA Team, 527, 532F
day-time contamination by reflected properties, 471 ocean color, coastal waters, 495
sunlight, 551 passive systems/sensors, 474 optimization for coastal water remote
non-linear SST and Pathfinder SST, Multichannel Ocean Color Sensor sensing, 496, 496F
549–550 (MOCS), 474–475 SeaWiFS, 322
operating method, 551 Ocean Data Acquisition System Almaz-1, 560
post-process data, 549–550 (ODAS) and SeaWiFS Aircraft Along-Track Scanning Radiometer
scan geometry, 551–552, 552F Simulator (SAS), 474–475 (ATSR), 554
Advection long-time series aircraft studies, 475 brightness temperature calibration, 553
through surface sediments, and benthic phytoplankton cycle productivity, infrared rotating scan mirror, 552–553,
flux measurements, 174 475, 476F 552F
Advection-diffusion balance, 492–494 SAS replaces ODAS, 475 relatively narrow swath width, 553–554
Aerosols radiometers measuring reflected two brightness temperature
atmospheric, infrared atmospheric sunlight, 474–475 measurements, 552
correction algorithms, SST radar altimetry, 477 Altimeters
measurement, 550 sensors used, 471 satellite radar see Satellite radar
spectral range for remote sensing, 495T Synthetic Aperture Radar (SAR) altimeters
Aglantha, 202T systems, 477–478 satellite sea level data, 586–587
622 INDEX

Altimeters (continued) Ross Sea, 530, 531F ARIES (Autosampling and Recording
Topex/Poseidon see Topex/Poseidon Southern Hemisphere, 530, 531–532, Instrumental Environmental
satellite altimeter system 531F Sampler), 78–79, 78F
Altimetry, 514 Weddell Sea, 530, 531F ARS see Acoustic ripple scanner
radar see Radar altimetry Antennas, 585–586 Artificial neural nets (ANN), in analytical
satellite see Satellite altimetry Antisubmarine Detection Investigation flow cytometry, 296
tomography and, 466F Committee (ASDIC), 442 ASCAT (Advanced Scatterometer), 591
Altitude, atmospheric pressure changes Anti-submarine warfare (ASW), 442 Ascent, glider, 143
with, 211 AO see Arctic Oscillation ASDIC (Antisubmarine Detection
Aluminum AOP see Apparent optical properties Investigation Committee), 442
fluorescent sensing, 270, 270T (AOPs) AST see Acoustic scintillation
Alvin deep-water manned submersible, Aperture synthesis, 586 thermography (AST)
148–150, 153, 153F, 154, 199T, Apparent optical properties (AOPs), 275, Aster X, 205T
201F, 207 317, 323, 493 ASW (anti-submarine warfare), 442
deep submergence studies, 125–126, chlorophyll concentration and, 250, Atlantic see Atlantic Ocean
126F, 127F, 130F, 133–136 253–255, 254T, 255F Atlantic Ocean
key discoveries by, 126–127, 135F definition, 313 nepheloid layers, 288F
Ambient noise, 357 inherent optical properties and, 275 North see North Atlantic
active sonar systems, 443 measurements, 275, 275–276, 277F sound speed contours, 407F
measurement data, 357, 358F models, 247 subtropical salinities, 585F
see also Acoustic noise see also Bio-optical models see also North Atlantic; North
Ammonia quantities, 314T Atlantic Oscillation
absorptiometric sensor, 245 see also Inherent optical properties Atlantic Water (AW)
AMSR see Advanced microwave scanning (IOPs); Ocean optics; Radiative seismic reflection water-column
radiometer (AMSR) transfer profiling, 434F, 435
Analogue-digital (A/D) converters, Aqua AMSR-E, 527–528, 533 Atmosphere
seawater profiling, 32 Aquarius/SAC-D mission, 586, 587 ocean surface interactions, 548
Analytical flow cytometry (AFC), Aqua satellite, 526, 533, 575T Atmospheric aerosols, infrared
293–295 Aquashuttle towed vehicle, 60, 67 atmospheric correction algorithms,
applications, 293–295 Arctic (region) SST measurement, 550
bacteria, 295–296, 296F acoustics see Acoustics, Arctic Atmospheric correction, ocean color, 495,
larval fish, 297 climate, 530 576–577
phytoplankton, 295–296, 295T, ecosystems, 530 Atmospheric pressure
296F sea ice see Arctic sea ice changes with altitude, 211
protozoa, 296–297, 297F warming, 529, 530 measurements, 211
viruses, 297 Arctic Climate Observations using Atmospheric turbulence, acoustic noise,
zooplankton, 297 Underwater Sound (ACOUS) 359, 360
artificial neural nets in, 296 experiment, 465 Atmospheric windows, 548
cell sorting, 295 Arctic Intermediate Water (AIW) ATSR see Along-Track Scanning
commercial instruments, 294–295 acoustics, 399–400 Radiometer (ATSR)
future trends, 298 Arctic Ocean Attack angle, gliders, 144
in-situ, 297–298 acoustics see Acoustics, Arctic Attenuation
photomultiplier tubes in, 294 bathymetric map, 398F acoustics in marine sediments, 381–383,
technique, 293–295, 294F sound speed profile, 399F, 400 385, 387, 395
Anchor moorings see also Arctic (region); Arctic Attenuation coefficient, as bio-optical
deployment, 106–107 Oscillation (AO); Arctic sea ice model quantity, 248T
release, 98 Arctic Oscillation (AO), 533 ATV, 202T
Anemometers Arctic sea ice Aurelia aurita, 374
calibration, 211–212 Arctic Ocean, 529–530, 529F Autonomous Benthic Explorer (ABE), 85,
cone, 221 decrease, 529 163F, 205T, 207
cup, wind speed, 211, 212F extents, 529, 533 appearance, 129F
hot-film, 221 Kara and Barents Seas, 529–530, development at Woods Hole, 137,
hot-wire, 220 529F 204–205
K-Gill, 221–222, 221F Northern Hemisphere, 529–530, 529F, tracklines , Juan de Fuca Ridge dives,
propellor, 211, 212F 531–532 138F
sonic, 212F, 213, 222, 222F oscillatory pattern, 529, 530 Autonomous Lagrangian Circulation
thrust, 221 Seas of Okhotsk and Japan, 529–530, Explorer (ALACE), 85–86, 93F,
Anemones, fluorescence, 258 529F 140
Angle of intromission, 391–392 see also Arctic Ocean; North Atlantic Autonomous Ocean Sampling Network
ANN (artificial neural nets), in analytical Oscillation (NAO) (AOSN) program, 141
flow cytometry, 296 Argentine Basin, turbidity, 290 Autonomous Ocean Sampling Network II
Antarctic Peninsula, 530, 531–532 Argo, 519 (AOSN) program, 172, 172F
Antarctic sea ice Argo II optical/acoustic mapping systems, Autonomous underwater vehicles (AUVs),
Bellingshausen and Amundsen Seas, 125–126, 130F 155, 162, 202–207, 203–204,
530, 531F Argo program, 93 162–173
extents, 529, 530 Argos satellite system, 92 advantages, 162–164, 202–203
 INDEX 623

applications, 168–169, 202 Backscattering sensors, 274, 329 Beer’s law

mapping, 168–169 see also Nephelometry transmittance of radiation, 224–225
under-ice operations, 170–171 Backup recovery, moorings, 99 Bé multiple plankton sampler, 79, 80F
water-column profiling, 169–170 Bacteria Benthic boundary layer (BBL)
Arctic, acoustic research, 404 analytical flow cytometry, 295–296, processes, application of
bathymetric survey, 169F 296F transmissometry/nephelometry,
compared with alternatives, 168 as optical constituents of sea water, 318 331T
cost, 168, 168F see also Cyanobacteria see also Bottom water; Turbulence
deep submergence and, 129F Baleen whales (Mysticeti) Benthic flux
definition, 162–164 sound production, 426 calculation
disadvantages, 202–203 see also Cetaceans from chamber measurements,
drifters, 203–204 Ballasting, RAFOS float, 95 176–177
failure, 166 Ballast systems, manned submersibles from sensor measurements, 180
future prospects, 205–207 (shallow water), 158 eddies and, 182
hull shape, 162 Baltic Sea flotation, 175
instrumentation, 203–204 cyanobacterial bloom, 499F measurement, 174–175
mission types, 166 Barents Sea chamber incubation, 174
multiple thruster type, 162 Arctic sea ice, 529–530, 529F concentration gradient sampling,
navigation, 166–168 sound propagation losses, 422, 422F 174–175
acoustic, 167 Barometers flow near seabed and, 178, 181
dead-reckoning, 167 aneroid, 211 Benthic flux chambers, 175
geophysical parameter-based, mercury, 211 data examples, 177F
167–168 types, 211 limitations, 177–178
observation systems/observatories and, Barotropic tides, tomography, 461–463 Benthic flux landers, 174–182
171–172 Barrier layer chamber landers, 175–177
performance modeling, 164 formation, 586–587, 587 general design, 175–177
propulsion, 162, 165F Bary catcher, 74, 75F limitations and design variations,
range, 164 Basalt(s) 177–178
seafloor imaging, 205 geoacoustic properties, 421T flux measurement strategies, 174–175
software, 166 Batfish towed vehicle, 60, 64 integrated sediment disturbing, 181–182
sonar, 168–169 Bathymetric map oxystat, 181–182
systems, 164–166 Arctic ocean, 398F recovery, 175
thermal gradient engine, 163–164 Bathymetry sediment disturbance, 177–178,
types, 162, 163F autonomous underwater vehicles (AUV), 180–181, 182
water-column profiling, 169 169F, 204–205 sensor landers, 178–180
zooplankton sampling, 85 gravimetry and, 604–605 design and operation, 178–180, 179F
see also Gliders (subaquatic) surveys, deep submergence science instrumentation, 179
Autonomous Vertically Profiling Plankton studies, 137, 138F limitations and variations, 180–181
Observatory (AVPPO), 81 see also Seafloor topography microelectrode deployment apparatus,
Autosampling and Recording Bathyscaph, history, 148 178, 179F
Instrumental Environmental Bathysphere, 150 results, 179, 180F
Sampler (ARIES), 78–79, 78F history, 148, 150 specialised, 181–182
Autosub 3, 205T Bathythermograph, 26 a-Benzoin oxime, 245
Autosub 6000, 205T expendable see Expendable Bering Sea
AUVs see Autonomous underwater bathythermograph (XBT) coccolithophore bloom, 582–583,
vehicles (AUVs) mechanical, 43 582F
AVIRIS see Airborne Visible/Infrared Batteries fish abundance and yearly ice extent,
Imaging Spectrometer (AVIRIS) autonomous underwater vehicles (AUV), satellite remote sensing, 566
AVPPO (Autonomous Vertically Profiling 164, 202–203 front near ice edge, SAR, 566, 566F
Plankton Observatory), 81 gliders, 143 polar low SAR image, 564–565
human-operated vehicles (HOV), Bermuda Testbed Mooring (BTM),
B 197–199 optical systems, 277F
BBL see Benthic boundary layer (BBL) data, 279–280, 280F
Backscatter Beamforming Big-eye tuna (Thunnus obesus)
acoustic see Acoustic backscatter active sonars, 443, 444, 445F, 446–447 acoustic scattering, 371
microwave, 590–591 advanced, 480 Bin averages, drifter velocity
see also Scatterometry matched field processing (MFP), 450 measurements, 89
from ocean surfaces, satellite remote plane wave model, 450 BIOMAPER II (BIo-Optical
sensing, 561–562 wavefront curvature, 450 Multi-frequency Acoustical and
Backscattering (optical) coefficient, 316, Beaufort scale, 414T Physical Environmental Recorder),
493 Beaufort Sea 84, 85F
as bio-optical model quantity, 248T, sound speed, 400 BIONESS (Bedford Institute of
249–250, 254–255 Bedford Institute of Oceanography Net Oceanography Net and
components, 494 and Environmental Sensing System Environmental Sensing System),
Backscattering efficiency, as bio-optical (BIONESS), 72T, 79, 80F 72T, 79, 80F
model quantity, 248T, 249–250 Beer-Lambert law, 239–240 Bio-optical algorithms, 495, 577
624 INDEX

Bio-optical models, 247–256 see also Acoustic remote sensing whitecaps see Whitecaps
Case 1 waters, 247, 248T, 252T, 253F, Boltzmann’s constant, 548 see also Bubble clouds
254T, 255F, 256 Bongo net, 70, 71F Bulk density, seafloor sediments, 383,
Case 2 waters, 247, 248T, 256 Boric acid (B(OH)3) 383T, 385
chlorophyll concentration sound volume attenuation, 409, 409F Bulk modulus, seafloor sediments, 383,
apparent optical properties and, 250, Bottles, as drifters, 90 383T, 385
253–255, 254T, 255F Bottom currents Buoyancy
inherent optical properties and, 250, benthic flux measurement, 174 remotely operated vehicles (ROVs), 183
251–252, 251F, 252T, 253F benthic flux measurement and, 174 Buoyancy flux(es)
coastal waters, 494 Bottoming point, 432 downward, 219–220
concepts, 248T Bottom mixed nepheloid layer (BMNL), surface, 219
limitations, 256 282 Buoyancy frequency
oceanic waters related to biological sediment concentrations, 285–286 pressure and, 617F
state, 247–250, 255–256 Bottom nepheloid layer (BNL), thickness, Buoys
for particles (individual/populations), 285 data from, 532
247, 249F, 255–256 Bottom topography moorings, 98, 101
quantities, 248T maps, satellite altimetry, 503, 504F shape, 98–99
semianalytical, 255, 255F shallow water, satellite remote sensing Burt, P J, acoustic noise, 359–360, 360
spectral domain, 247 application, 562F, 564 Bushnell’s Turtle submarine, 156
theory, 248–249 see also Seafloor topography
see also Apparent optical properties Bottom water formation C
(AOPs); Inherent optical properties sea ice, 524
(IOPs); Ocean color; Ocean optics; Bouguer gravimetric correction, 602–603 Cable drag, tow cables see Tow cable
Optical particle characterization; Boundary drag
Radiative transfer benthic see Benthic boundary layer Caicos drift, 289F
BIo-Optical Multi-frequency Acoustical (BBL) Calanoida copepods
and Physical Environmental surface water masses, satellite remote biodiversity, 21–22, 22F
Recorder (BIOMAPER II), 84, 85F sensing of SST, 556, 557F Calanus finmarchicus (zooplankton),
Bio-optics, 272 Boundary-element modeling, acoustic 373F
research, 273 scattering, fish, 371 abundance, North Atlantic Ocean index
see also Ocean optics Bowhead whale (Balaena mysticetus) and, 19–20, 21F
BIOPS optical system, 277F acoustic noise, 404 distribution
data, 278F see also Baleen whales Continuous Plankton Recorder
see also Ocean optics Bradshaw, A, 29, 30 survey, 18, 19F
Biot-Stoll model, 381–383 Bragg resonant wave back scattering, Calanus hegolandicus
input parameters, 381–383, 383T 561–562 abundance, North Atlantic Ocean index
Bit depth, seawater profile measurements, satellite remote sensing, 561–562, 561T and, 19–20
32 BRIDGET, 198T distribution, Continuous Plankton
Black-body Brightness temperature, 584, 585, 586, Recorder survey, 18, 19F
calibration, satellite remote sensing, 588F Calibration
548, 551 errors, 587–588 absorptiometric chemical sensors, 244
emittance, 481, 481–483 model function, 587 anemometers, 211–212
radiance/radiation, 481, 481F sensitivity, 587 black-body, satellite remote sensing,
Black Sea SST measurement by satellite, 548 548, 551
light attenuation, 498F temperature deficit relationship, 549 narrow beam filter radiometers see
‘Black smokers’ Broadband (BB) ocean bottom Narrow beam filter radiometers
deep submergence science and, 125, seismometers, 191 Optical Plankton Counter (OPC), 299
126F characteristics, 191T, 193 salinometers, 30F
Bloom cycles, transmissometers in LDEO-BB, 191T, 194F sensors
tracking of, 331T ORI-BB, 191T ocean color, satellite remote sensing,
Bluefin, 205T SIO/IGPP-BB, 191T, 192F 575–576
Blue-green ratio algorithms, 495, 577 SIO/ONR, 191T, 194F wind speed measurements, 211–212
Blue whale WHOI-BB, 191T single point current meters, 120–121
sound production, 426 Broadband pyrgeometers, 485–487, 485F spacecraft instruments, 551
tracking, 426, 426F Brown, Neil, 30, 30–31, 31–33, 33F thermistors, 27
see also Baleen whales Browning, D G, acoustic noise, 364 thermometers, 26–27, 27
Blue whiting (Micromesistius poutassou) Brun’s formula, 600 transmissometers, 326–327
acoustic scattering, 371 BT see Bathythermograph Canadian Scientific Submersible Facility,
Boat(s) BTM see Bermuda Testbed Mooring remotely-operated vehicles (ROV),
workboats, oceanographic research (BTM) 202T
vessels, 6 Bubble clouds Canadian Space Agency, Radarsat-1 and
see also Ship(s) acoustic scattering, 412 Radasat-2, 560
Body waves, 383, 383–384, 384 Bubbles Canyons
compressional waves see Compressional acoustic noise, 362, 363, 365 nepheloid layers and, 290–291
waves acoustic scattering, 421 Capacitance change hygrometers, 215
shear waves see Shear waves resonance frequency, 370 Capacitance sensors, humidity, 214–215
 INDEX 625

Carbamate pesticides, sensors, 245 basic principles, 239–240 Climate

Carbon dioxide (CO2) calibration and error correction, 244 change see Climate change
dissolved design, 243–244, 243F predictions, seasonal to interannual,
sensors, 245–246 molecular recognition element, 586–587
see also Total inorganic carbon 242–243 warming, 530, 533
fluorescent sensing, 270, 270T optical fibers, 240–241, 240F Climate change
photochemical flux, spectral dependence optoelectronics, 241–242 ‘global thermometer,’ SST, 556
of, 310F reagent support material, 242 satellite oceanography, 520–521
photochemical production, 305F, 306 response characteristics, 242–243 see also El Niño Southern Oscillation
Carbon disulfide (CS2) typical configuration, 243F (ENSO); Greenhouse gas; Sea level
photochemical oxidation, 307 wavelength selector, 241 changes/variations
Carbon monoxide (CO) evanescent wave, 243–244, 244F Climate change forecast models, 556–558
photochemical flux, spectral dependence reflectometric, 240, 243 Climate models
of, 310F Chemical transduction, chemical sensors, climate change forecast, 556–558
photochemical production, 305F, 306, 242, 243 Climate prediction models, 586–587, 587
309F see also Molecular recognition element Closing cod-end systems, zooplankton
Carbonyl sulfide (COS) Cherokee, 202T sampling, 76, 78F
photochemical flux, spectral dependence Chimney Cloud(s)
of, 310F ‘black smokers,’ deep submergence contamination, infrared atmospheric
photochemical production, 305, 305F science and, 125, 126F corrected algorithms, SST measures,
Carey, W M, acoustic noise, 364 convective see Convective chimney 550
Cartilaginous fish, acoustic scattering, China Ocean Mineral Resources R&D penetrated by satellite remote sensing,
372 Association 560
‘Case 1’ waters, 492 autonomous underwater vehicles, 205T satellite radiometry, 593
bio-optical modeling see Bio-optical deep-towed vehicles, 198T Clouds and Earth’s Radiant Energy
models human-operated vehicles (HOV), 199, System (CERES) experiment,
definition, 248T, 313 199T 593–594
Hydrolight simulations, 319, 319F, Chlorophyll Clupeoids
320F, 321F, 322F autofluorescence, 295, 296F acoustic scattering, 370
‘Case 2’ waters, 492 concentration ([Chl]) CMP see Common midpoint
bio-optical modeling see Bio-optical apparent optical properties and, CO2 see Carbon dioxide (CO2)
models 250–251, 253–255, 254T, 255F Coastal ocean dynamics application radar
definition, 248T, 313 as index of bio-optical state, 250 (CODAR), 172F
Hydrolight simulations, 319, 319F, inherent optical properties and, Coastal Ocean Dynamics Experiment
320F, 321F, 322F 250–251, 251–252, 251F, 252T, (CODE), 91–92
CASI (Compact Airborne Spectographic 253F Coastal region/seas
Imager), 475–477 variation, 250 weather forecasting for, satellite remote
Catch-dividing bucket, 77–78 fluorescence, 494 sensing, 569–570
CDB (catch-dividing bucket), 77–78 spectral range for remote sensing, Coastal waters
CDOM see Gelbstoff 495T classification, 492, 493F
CDVP see Coherent Doppler velocity see also Chlorophyll a colored dissolved organic matter
profiler (CDVP) spectral range for remote sensing, 495T influence, 303
CEAREX (Coordinated East Arctic Chlorophyll a optical properties, 492–494
Experiment), 397–398 algorithm, ocean color by satellite bottom and, 494
Cetaceans remote sensing, 577, 578F Coastal wind vectors, from calibrated
bioacoustics, 425 El Niño event, 580F SAR images, 560
suborders fluorescence, 257, 259 Coastal Zone Color Scanner (CZCS),
Odontoceti see Odontocetes quantification, 263 571, 575T
see also Marine mammals; Sperm whales ocean color and, 571, 582 sensor calibration stability, 575–576
(Physeteriidae and Kogiidae) bio-optical algorithms, 577, 578F Cobalt
Chaetoceros muelleri, 258F SeaWiFS data, 579F, 580F sensors, absorptiometric, 245
Chalk Chlorophyll-specific absorption coefficient Coccolithophores
geoacoustic properties, 421T of phytoplankton, 248T, 251 blooms, 499
Challenger, HMS (survey ship1872-1876), Chukchi Sea ocean color and, 582–583, 582F
4 sound speed, 400 see also Phytoplankton
Challenger Expedition Circulation models Cod (Gadus), 368
drifter schematic, 90F tomography and, 458–460 CODAR see Coastal ocean dynamics
Chamber incubation (of benthic flux), Circulation Obviation Retrofit Kit application radar
174 (CORK), 41F, 42, 127, 134F CODE (Coastal Ocean Dynamics
Chapman, N R, acoustic noise, 363F, 364 Clams Experiment), 91–92
Chapman-Harris curves, 411 acoustic scattering, 374 Coherent Doppler velocity profiler
Chemical sensing, 265 Clarke-Bumpus net, 71–72, 73F (CDVP), 344, 344F, 347, 347F
Chemical sensors Classical sling psychrometer, 213–214 single-axis vs. three-axis, 355
absorptiometric, 239–246 Clay(s) turbulent flow visualization, 348F
applications, 244–245 acoustics in marine sediments, 395F Colored dissolved organic matter
background subtraction, 244–245 geoacoustic properties, 421T (CDOM), 318
626 INDEX

photochemical production, 302, 304F database, 16–17 Corrosion-resistant materials, remotely

absorbed photons, effects of, 305 future directions, 24 operated vehicles (ROVs), 186
absorption spectra, 302, 303F history, 15, 16F Coscinodiscus wailessii, 23
photolysis, 306 methods, 17–18 Cosmic radiation, 587
spectral absorption, 324, 324F instrumentation, 16F, 18, 72T, 74–76, Cost(s)
see also Gelbstoff 75F AUV mission, 168, 168F
Common midpoint (CMP) records, 432 early, 15, 16F gliders (subaquatic), 140
Communications, internal/external, open access data policy, 16 RAFOS float, 95
oceanographic research vessels, 6 results/data applications, 18, 24 WOCE float, 93
Compact Airborne Spectographic Imager biodiversity, 22–23, 22F Cousteau, Jacques
(CASI), 475–477 Calanus and North Atlantic Soucoupe, 156
Compasses, 167 Oscillation, 20–21, 21F CPR see Continuous Plankton Recorder
Compliance monitoring, application of exceptional events, 23 (CPR) survey
nephelometry, 331T Gulf Stream North Wall index and CR-01, 205T
Compressee, floats, 95 copepod numbers, 21–22, 22F CR-02, 205T
Compressional waves, acoustics in marine harmful algal blooms, 23–24, 23F Cranes, oceanographic research vessels,
sediments, 384, 384–385, 384F, herring, 19–20, 20F 6
388, 392–393 kittiwake breeding success, 19–20, Critical angle, acoustic remote sensing,
arrival times, 385, 385F, 393 20F 390F, 391
first kind, 383 marine plankton biogeography, Critical depth (acoustic), 407–408
refracted, 392–393, 393, 393F 18–19, 19F Crossover wind speed, 359–360, 364
second kind, 383 nonindigenous species, 23–24 Crouch, W W, acoustic noise, 359–360,
velocity, 383T, 385, 386, 387, 391, 393, North Sea ecosystem regime shift, 21, 360
394, 394F 21F CSIRO modified SeaSoar, 63F, 68F, 69F
and porosity, 386, 387F phytoplankton, 19–20, 20F CSIRO multi-frequency vehicle, 65F
velocity-depth curve, 393–394, 394F weather, 19–20, 20F CTD recorders, towed vehicles, 63F,
see also Acoustic remote sensing zooplankton, 19–20, 20F 67–68, 68F
Computer control, remotely operated routes, 15, 17F CUFES (continuous underway fish egg
vehicles (ROVs), 184, 185–186 see also Plankton sampling system), 83
Computer hardware, autonomous Continuous underway fish egg sampling Cup anemometers, 211, 212F
underwater vehicles (AUV), 166 system (CUFES), 83 Current(s)
Computer modeling, 530 Convection AUV navigation and, 167
COMRA see China Ocean Mineral double-diffusive see Double-diffusive bottom see Bottom currents
Resources R&D Association convection contour see Bottom currents
Conductivity (sea water) subpolar regions, 587 internal tides see Internal tides
sensors, 234 thermohaline see Thermohaline internal waves see Internal wave(s)
Conductivity (electrical), measurement, convection Labrador Sea, 146F
31–33, 32 tomography, 460–461 satellite remote sensing, 562–563
digital, 32 Convective chimney, 460–461 sediment transport, instrumentation for
salinity and, 33 Greenland Sea, 460F, 461F measuring, 347, 348F
Conductivity-temperature-depth (CTD) Cook, Captain James, 4 Current meters
package, 140 Coordinated East Arctic Experiment Aanderaa RCM9/RCM8, 116F, 117T
Conductivity-temperature-depth (CTD) (CEAREX), 397–398 acoustic travel time see Acoustic travel
profilers, 30, 32F, 25–34 Copenhagen standard seawater, 29 time (ATT) current meters
conductivity measurement, 31–33 Copepod(s) electromagnetic see Electromagnetic
conductivity cell, 31, 32F abundance, Gulf Stream North Wall current meters (ECMs)
extended deployment, 33 index and, 21–22, 22F single point see Single point current
lowering speeds, 31 acoustic scattering, 372–373 meters
pressure measurement, 30–31 calanoid, biodiversity, 21–22, 22F vector averaging, 116–118, 116F, 117T,
salinity measurement, 28–30, 30F prosome and oil sac, 373F 119
calibration of salinometer, 30F Copper (Cu) CV (wireline reentry system), 202, 202T,
resolution, 30 sensors, absorptiometric, 245 204F
seismic reflection profiling and, 435, Coral(s) Cyanobacteria
435–439, 435F, 438F bleaching blooms, 499F
temperature measurement, 25–28 satellite remote sensing of sea surface Cyanophytes, fluorescence, 258
Conductivity-temperature-depth (CTD) temperatures, 556–558 Cylindrical spreading (sound), 419, 419F
recorders, towed vehicles, 63F, see also Coral reef(s) loss, 408
67–68, 68F fluorescence, 258 see also Geometrical spreading
Cone anemometer, 221 Coral reef(s) CytoBuoy, 298
Conservation remote sensing, 500F CZCS see Coastal Zone Color Scanner
satellite remote imaging and, 499 threatened, time-series SST predicting,
Continental boundaries, internal waves 556 D
and, 440, 440F CORK (circulation obviation retrofit kit),
Continuous Plankton Recorder (CPR) 41F, 42 D’Asaro, Eric, Eric, 95
survey, 15, 15–24 Cornish, J W, acoustic noise, 363F, 364 Data
atlas, 18 Correlation velocity log, 167 assimilation see Data assimilation
 INDEX 627

collection, satellite remote sensing see time-series observations, 127, 135F, range for towed vehicles/cable system,
Satellite remote sensing 137 65F, 66, 66F
storage, solid-state, CTD profilers, 33 ocean phenomena and processes for shallow vs deep-water manned
transmission, moorings, 104 discoveries, submersibles, 156, 158
Data assimilation technologies enabling, 125–127 Depth sensors, remotely operated vehicles
tomography, 458–460 ABE Autonomous Benthic Explorer, (ROVs), 184
Davis, Russ, 140 129F, 137, 138F Detritus
DDPS (Discrete Depth Plankton Sampler), Alvin, 125–126, 126F, 127F, 130F, as optical constituents of sea water, 318
77–78, 78F 133–136 spectral absorption, 273
Decibel, 406–407, 418 Argo II optical/acoustic mapping De Voorst Laboratory, 351
Deep Flight (high speed submersible), systems, 125–126, 130F Dew point hygrometers, 215
151–152 AUVs, 129F Differential Global Positioning System
Deep Jeep (USN submersible), 156–157 DSL-120 towed multibeam sonar, (DGPS), manned submersibles
Deep ocean 130F (deep water), 152
acoustic noise, 360 Jason ROV, 125–126, 127F, 128F, Diffuse attenuation coefficient, 317
biological research, deep submergence 130F, 133–136, 136F Diffuse flow, Juan de Fuca ridge, 377F,
science, 132, 135F, 136F key discoveries by Alvin, 126–127, 378F
Deep ocean work boat, 156 135F Diffuse reflection, 240
Deep scattering layer (DSL), 412 manned submersibles, 125, 127F Diffusion
Deep-sea drilling oceanographic technology and mixing and, 492
methodology, 35–42 instrumentation, 125 in sediments, 180
in situ measurement tools, 39–42 ROVs, 125, 127F, 129F, 130F see also Double-diffusive convection;
sampling tools, 37–39 R/V Atlantis, support ship, 127F, 130F Mixing; Turbulent diffusion
seafloor observatories, 42 Tiburon remotely operated vessel, Dimethylsulfide (DMS)
technology, 35–37 129F photochemical processes/production,
Deep Sea Drilling Project (DSDP), 35 time-series and observatory-based 305, 307
drilling sites, 36F research, 125 Discovery, HMS (1794-5), 4
Deep-sea exploration vehicles, 197–208 Ventana remotely operated vessel, Discovery Expedition, Continuous
see also Autonomous underwater 129F Plankton Recorder, 15, 16F
vehicles (AUVs); Human-operated see also Autonomous Underwater Discrete Depth Plankton Sampler (DDPS),
vehicles (HOV); Remotely operated Vehicles (AUVs); Manned 77–78, 78F
vehicles (ROVs) submersibles (deep water); Dispersion
Deep-sea observatories Remotely Operated Vehicles deep ocean see Deep ocean
long-term, deep submergence science (ROVs) shallow water, 423
studies, 127, 133–136, 136F, 137 utilization of ODP boreholes, 127, 132F, Dispersion diagram, interface waves, 394,
Deep-sea vehicles see Manned 134F, 135F 394F
submersibles (deep water) Deep submergence vehicles, 125–127, Dissolved organic carbon (DOC)
Deep sound channel, 406, 407–408, 407F 136–137 fluorometry, 269, 269T
floats and, 94 Deep Tow 4KC, 198F, 198T Dissolved organic matter (DOM)
noise directionality, 415F Deep Tow 4KS, 198F, 198T colored see Colored dissolved organic
ray paths, 453F Deep Tow 6KC, 197, 198F, 198T matter (CDOM)
tomography, 451–454 Deep-towed vehicle systems, 198F, fluorometry, 269, 269T
Deep submergence science, 125–139 198T as optical constituents of sea water,
achievements/topics studied, 127–137 Deep-Tow survey System, 198T 318
biodiversity of marine communities, Deep water Distributed observation systems, 172F
131–132 investigations Dive cycles, gliders (subaquatic), 143–144
CORK, 127, 134F difficulties, 148 Diving
deep ocean biological research, 132, severe environment, 148 elephant seals see Elephant seals
135F, 136F see also Deep submergence science; gliders, 143
deep ocean processes research, 129, Manned submersibles (deep pinnipeds (seals) see Pinnipeds (seals)
133–136 water) Diving bell, Halley, Dr Edmund, 156
geochemical studies, 132–133 Deep Western Boundary Current DMSP (United States Defense
global MOR discoveries, 127 (DWBC), 590 Meteorological Satellite Program),
hydrothermal communities, 127, Defense Meteorological Space Program 524, 525–526, 528F, 529F, 532F,
129–131, 135F (DMSP) satellites, 594, 595 533
long-term deep seafloor observatories, Delta Flume, 346F, 351, 351F DNA, fluorescent tagging, 259
127, 133–136, 136F, 137 Dense surface waters, 584 DOC see Dissolved organic carbon
Ocean Drilling Program, 127, 132F, Density (DOC)
133–136, 134F of seawater, 614 Docking, autonomous underwater
provision of deep submergence calculation, 30 vehicles (AUV), 166, 205–207
vehicles, 136–137 quantities related to, 614–616 Doliolum nationalis, North Sea, 23–24
Scripps Institution’s re-entry vehicle, see also Double-diffusive Dolphin 3K ROV, 187T
127, 132F convection Dolphins (Delphinidae)
seafloor bathymetric surveys, 137, Depth acoustic scattering, 374
138F current strength and, 143 oceanic see Oceanic dolphins
subduction zone processes, 132–133 measurement, by hydrowire casts, 26 see also Odontocetes (toothed whales)
628 INDEX

DOM see Dissolved organic matter EcoSCOPE, 83 definition, 480, 481F

(DOM) Eddies Endurance, gliders, 143
Doppler principle, marine animals and, benthic flux and, 182 Energetically optimum speed,
367 gliders and, 140 autonomous underwater vehicles
Doppler shift/shifting Mediterranean see Meddies (AUV), 164
and float tracking, 96 satellite remote sensing application, Engine(s), gliders, efficiency, 144
Doppler velocity log (DVL), 167 564F, 565 English Channel
Dorado -class AUVs, 165F, 204, 205T turbulent see Turbulent eddies sound propagation losses, 422F
Double-diffusive convection, 434–435 EEZ see Exclusive economic zone (EEZ) ENSO see El Niño Southern Oscillation
see also Intrusions; Mixing; Turbulence Elastic flexure, 605 (ENSO)
DOWB (deep ocean work boat), 156 Electrical components, remotely operated Environmental monitoring, by satellite
Downwelling vehicles (ROVs), 186 remote sensing, 569
irradiance, 255–256, 274–275, 571 Electrical conductivity see Conductivity Envisat-1, 560
Drag Electrically scanning microwave Envisat-1 MERIS, 575T
gliders, 144 radiometer (ESMR), 525, 526, 528 EOLE satellite tracking, 91
and endurance, 143 Electrically scanning thinned array Eötvös correction, 602, 603
Drag coefficient radiometer (ESTAR), 586 Eötvös unit, 600
induced, towed vehicles, 64 Electrodes ‘Equation of state’ (of sea water), 614
normal, tow cable drag, 61, 61F, 66, 66F for benthic flux measurement, 178–179, ERS-1 satellite, 591, 592T
tangential, tow cable drag, 61–62, 61F 181 see also European Remote Sensing (ERS)
tow cables see Tow cable drag Electromagnetic current meters (ECMs), satellites
wind on sea surface, 590–591 116F, 117T, 118–119, 119F, 351 ERS-2 satellite, 591, 592T
see also Wind stress Faraday effect as basis of, 119, 119F global sea level variability detection,
Drebel, Cornelius van, manned shallow- Electromagnetic radiation, absorption, 505–506, 506F
water submersibles, 156 239–240 see also European Remote Sensing
Drifter Data Assembly Center, 91 Electromagnetic velocity sensors (ERS) satellites
Drifters, 89–90, 89–96, 203–204 measurements in Earth’s magnetic field, ESA (European Space Agency), 518
data interpretation, 89 122 ESMR (electrically scanning microwave
drogues, depth, 91–92 measurements with artificial magnetic radiometer), 525, 526, 528
flotsam and jetsam, 90 fields, 122 ESTAR (electrically scanning thinned
other, 90 Electromechanical (EM) cable, in array radiometer), 586
problems, 91 moorings, 105 Estuaries
sea bottom, 90 Elephant seals colored dissolved organic matter
ships, 90 diving influence, 303
tracking, 90–91 profile, 430F, 431 Euphausia superba (Antarctic krill)
see also Float(s) telemetry, 430F, 431 acoustic scattering, 372–373
Drilling, 35 El Niño see also Krill (Euphausiacea)
deep-sea technology see Deep-sea global sea level variability, contribution Euphausiids
drilling to, 505–506, 506F acoustic scattering, 372–373
open hole system see Nonriser drilling satellite remote sensing of SST, 556 see also Krill (Euphausiacea)
Drill pipe, 35 coupled ocean-atmosphere system European Center for Medium Range
Drogues, 89–90, 90F perturbations, 554 Weather Forecasts (ECMWF),
depth, 91–92 normal Pacific SST distribution, 463–465
RAFOS float, 95 554–555 European Remote Sensing (ERS)
WOCE drifter, 91 seasonal predictions of disturbed satellites, 591
DSDP see Deep Sea Drilling Project patterns, 555–556, 555F meteorological measurements, 216
(DSDP) Warm Pool movement, 554–555 revisit times, 499–500
DSL-120A, 198F, 198T see also El Niño Southern see also ERS-1 satellite; ERS-2 satellite
DSL-120 towed multibeam sonar, 130F Oscillation (ENSO) European Space Agency (ESA), 518, 586
Ducted waves, acoustics in marine El Niño forecasting, 586–587 Evanescent waves
sediments, 384, 384F see also El Niño Southern Oscillation chemical sensors, 243–244, 244F
DVL (Doppler velocity log), 167 (ENSO) optical fibers, 241, 241F, 243–244
El Niño Southern Oscillation (ENSO), Evaporation, 585F, 587
E 554 satellite remote sensing, 595
1997-1998 event see also Humidity; Precipitation
Earth sea level variability measured by Excitation-emission matrix spectroscopy,
mantle see Mantle satellite altimetry, 506–507, 506F 266, 267F
origins of life on, 137–139 chlorophyll a concentrations, 580F Exclusive economic zone (EEZ), 11
Earth Observing System (EOS), 526, 554 ocean color and, 582 oceanographic research vessel
Earthquakes satellite remote sensing, 596–597, 597F, operations, 11
acoustic noise, 404 598F survey vessels, 9
East China Sea see also Climate change Expendable bathythermograph (XBT), 27,
ERS-1 SAR image, 568–569, 569F Emiliania huxleyi coccolithophore, 582F 28F, 43, 43–44
ECMWF see European Center for Emissivity, calculation, 481–483, 482F air-launched, 47
Medium Range Weather Forecasts Emittance JJYY data exchange format, 47–48
(ECMWF) black body, 481 launch procedure, 45
 INDEX 629

limited depth capability, and expendable probe capabilities, 43 Finite difference acoustic modeling,
extrapolation from data, 44 measurement precision, 48 shallow-water, 424, 424F
operational effects of finite depth, 44–45 calibration affects climate change Fish(es)
probe accuracy, 44 data, 48 acoustic scattering, 370, 412, 421
probe design, 45 depth data checking, 48 high frequency, 370–372
connecting wire, 44 fall-rate equation errors, 48 low frequency, 370
electrical contacts, 44 within specified tolerances, 48 medium frequency, 370
probe body, 44–45 temperature errors, 48 swimbladder-bearing, 370
protective shell, 44 mechanical bathythermograph (MBT), swimbladderless, 372
thermistor element, 44 43 cartilaginous, acoustic scattering, 372
two-strand wire, 44 normal (surface ship) deployment, larvae see Fish larvae
probe operation, 45 46–47 larval, analytical flow cytometry, 297
role/use for ocean surveying, 48 limited if towing equipment, 46 satellite remote sensing, 499
submarine-launched, 47–48 reducing wind effects, 46 FISH (fluorescence in situ hybridization),
vertical temperature profile, 43–44 removal of data from top 3-5 m, 259
Expendable bottom penetrometer probe 46 Fisheries
(XBP), 46 standard ship launchers, 46 recruitment to, SAR images and, 564F,
accelerometer, 46 requirements of physical oceanography, 565
seabed properties, 46 43 Fisheries research vessels, 9–10
see also Acoustics, shallow water survey with small spatial scales, 43 design characteristics, 10
Expendable conductivity-temperature- types, 43–44 fisheries research, 9–10
depth probe (XCTD), 45–46 use at speed, 43 government agencies, 10
pre-calibrated with two-wire use for ocean surveying, 48 Fish larvae
connection, 45 budget-dependence, 48 analytical flow cytometry, 297
see also CTD deductions made from data archives, see also Plankton; Zooplankton
Expendable current profiler (XCP), 46 48 sampling
capability and resolution, 45–46 only when rapidity is essential, 48 Fletcher’s Ice Island, 397
measures current velocity and direction, uses in support of surveys, 48 FLIM (fluorescence lifetime imaging),
45 XBT role in ocean science, 48 259–260
Expendable optical irradiance probe Expendable sound velocity probe (XSV), FLM (fluorescence line height), 263
(XKT), 46 45 Float(s), 89–96, 92–94
optical diffuse attenuation K, 46 air-launched, 47 acoustic, 92, 94–95
see also Inherent optical properties precision and depth options, 45 data interpretation, 93
(IOPs); Irradiance sing around speed sensor, 45 instrumentation, 92–93
suspended particle measurement, 46 submarine-launched, 47–48 modification to glider, 94
Expendable sensors, 43, 43–49 use by operational submarines, 45 sampling schemes, 92
data recording and handling, 48 Experiment, US Coast Survey (1831), 4 schematic, 93F
data exchange format JJYY, 47–48 Extended core barrel, 38, 39F vertical water velocity measurement, 95
PC with dedicated electronic interface, Extreme environments, operation of see also Drifters
47 remotely operated vehicles, 186 Flow
deployment variations, 46–47 Reynolds decomposition, 492
air-launched or submarine-launched F Flow cytometers
options, 46 underwater, development of, 331–332
air-launched XBT and XSV, 47 FAMOUS (French-American Mid-Ocean see also Analytical flow cytometry
buoyant electronics package with rf Undersea Study), 154–155 (AFC)
transmitter, 46–47 Faraday rotation, 588 Flow cytometry (FCM), 258–259, 262T
large-area surveys of dynamic regions, Faraday’s Law of Electromagnetic Flow injection analysis (FIA), 335–336
47 Induction, as basis of advantages, 336, 337, 338
resolution and accuracy, 47 electromagnetic current meters, air-segmented continuous flow analysis
simultaneous monitoring of several 119, 119F vs., 336
probes, 47 Fast field program model (acoustic), 410 applications, 336
submarine-launched XBT and ASV, Fast repetition rate (FRR) fluorometry, water industry, 336
47–48 259, 260F in-situ, 338
assesses sonar propagation, 47 Fathometers submersible sensors, 338–339
needs expensive technology, 47 history, 442 manifold arrangement, 337F
XCTD equivalent is imminent, 47 Fatigue, moorings, 103 principles, 337, 337–338
depth variation of temperature, 43 FCM see Flow cytometry simultaneous nitrate and nitrite
see also Acoustics, deep ocean Fermat’s principle, 452 determinations, 336
expendable bathythermograph see FGVR (free vehicle grab respirometer), system, 337, 337F
Expendable bathythermograph 175, 176F periodic mixing, 336
(XBT) FIA see Flow injection analysis (FIA) Fluorescence, 265–266, 273
expendable bottom penetrometer probe Fiber optics see Optical fibers accessory pigments, 258
see Expendable bottom Fickian diffusion, 492 chlorophyll, 295, 296F, 494
penetrometer probe (XBP) Fick’s First Law of Diffusion colored dissolved organic matter, 305
expendable current profiler see benthic flux and, 180 decay times, 265, 268–269
Expendable current profiler (XCP) Figure of merit, sonar, 414–415 definition, 257
630 INDEX

Fluorescence (continued) Foams Gelatinous animals, observed using

emission spectra, 265 syntactic, 98 human-operated vehicle, 199
absorption spectrum and, 265, 265F Focal plane array (FPA) detector Gelatinous zooplankton
compared with absorption spectra, technology, 490 observed using human-operated vehicle,
258F problems with, 490 199
gelbstoff, 494–495 Forecasting/forecasts Gelbstoff
phycobiliproteins, 295 climate change, 556–558 absorption spectra, 494, 573F
polarization, 266 weather, for coastal regions, satellite coastal areas, 494
measurement, 267 remote sensing, 569–570 fluorescence, 494–495
quenching, 268 Fourier transform infrared spectrometers spectral absorption, 273
dynamic, 268 (FTIR), 488–489, 489F, 490–491 spectral range for remote sensing, 495T
static, 268 FRAM II ice camp, 400F see also Colored dissolved organic
Stern-Volmer equation, 268, 268F Fram Strait, 397 matter (CDOM)
reflectance (optical) spectra and, 494 Frauenhofer laser diffraction instruments, General purpose vessels, 9
variable induction, 259 274 coastal vessels, 9
water-column, Massachusetts Bay, 170F FRAZ displays, passive sonar flexible scientific facilities, 9
yield, 257 beamforming, 450 increased size, 9
see also Fluorometry Frazil ice multidiscipline vessels, 9
Fluorescence in situ hybridization (FISH), formation, transmissometers in study of, Geoacoustic models, 420
259 331T Geochemical Ocean Sections Study
Fluorescence lifetime imaging (FLIM), Free vehicle grab respirometer (FVGR), (GEOSECS) nephelometer, 282
259–260 175, 176F Geochemical studies, deep submergence
Fluorescence line height (FLM), 263 Freezing point, seawater, 617–618 science studies, 132–133
Fluorescent dyes, near infrared, 259 French Research Institute for Exploitation Geoid, 502–503, 600
Fluorometers, biological sensing, 257, of the Sea (IFREMER), 520 long-wavelength anomalies, 605–606
258, 258F French Research Institute for Exploration models, extraordinary gravity field data,
scales of sampling, 262T of the Sea (IFREMER) 505
Fluorometry autonomous underwater vehicles, 205T see also Gravitational potential energy
biological sensing, 257–264 deep-towed vehicles, 198T GEOMAR-SP ocean bottom seismometer,
advantages, 257 human-operated vehicles (HOV), 190T, 193F
algal and aquatic plant productivity, 199T Geometric acoustic spreading, power loss,
259–260 remotely-operated vehicles (ROV), 202T 407–408
estimating abundance, 257–258 Frequency-azimuth (FRAZ) displays, Geometrical spreading (acoustic), 423
fluorescence defined, 257 passive sonar beamforming, 450 see also Spherical spreading
phylogenetic discrimination, Freshwater flux, 586–587, 587 Geometric dispersion, 423
258–259 satellite remote sensing, 595–596, 598F Geophone arrays, 392–393, 392F, 393F
physiological applications, 259–260 see also Precipitation; Salinity Geophysical measurement systems, deep-
scales of productivity measurements, Fresnel reflection, 585 towed, 197
259–260 incidence angle, 585, 587, 588, 588F Geophysical research vessels, 10
chemical sensing, 265–271 Fronts MCS (multichannel seismic surveys), 10
applications, 267–269, 269–270, positions, satellite remote sensing of SST, purpose, 10
270T 556 specialized design, 10
direct sensing, 266–267 satellite remote sensing application, Geos-3 satellite, 512, 603
applications, 269–270, 269T 565–566, 565F, 566F Geosat, 511F, 517–518
excitation-emission matrix FRR fluorometry (fast repetition rate gravimetry, 603
spectroscopy, 266, 267F fluorometry), 259, 260F GEOSECS see Geochemical Ocean
Raman standard, 266, 266F FSI 3D ACM acoustic travel time, 117T Sections Study (GEOSECS)
time-resolved, 266–267, 268F Fulton, Robert, copper submarines, 156 GEOSS (Global Earth Observation
X-ray, 269T, 270 System of Systems), 521
indicators and sensors, 266–267, G Geostationary orbit, ocean color sensing
270T and, 575
applications, 269–270, 270T Gadoids Geostrophic circulation, global, dynamic
device components, 267 acoustic scattering, 370 sea surface topography, 503–505,
measurement of fluorescent species in Galactic radiation, 587 505F
sea water, 269–270, 269T Galapagos Islands Geosynchronous orbit measurements of
techniques, 265–266, 266 chlorophyll a concentrations, 581F SST, GOES Imager, 554
FRR, 259 global sea level variability, satellite GFP (green fluorescence like proteins),
pump and probe, aircraft for remote altimetry, 506–507, 506F 258
sensing, 472–474 Galapagos Rift Gilvin see Colored dissolved organic
see also Fluorescence survey of hydrothermal venting, 207 matter (CDOM); Gelbstoff
Flux Gallium triple point cell, 27 Glass, in optical fibers, 240
definition, 218 Garrett-Munk spectrum, 439–440 Glass spheres, on benthic flux landers,
measurements see Micrometeorological Gas(es) 178–180
flux measurements chemical sensors, absorptiometric, 245 Gliders (subaquatic), 108, 142F, 203–204,
Flyaway mode, remotely-operated Gas hydrates, 39 140–147
vehicles (ROV), 201 Gauss-Markov estimation, 455–456 advantages, 140
 INDEX 631

cost, 140 suction samplers, 55–56 Gravimetry, 600–608

design, 141–143 conventional grabs, 50–51 bathymetric features and, 605F, 606F
disadvantages, 140 Baird grab, 51, 52F data interpretation, 603–604
dive cycles, 143–144 Day grab, 51, 51F long-wavelength anomalies, 605–606
diving and ascent, 143 digging profiles, 52F mid-wavelength anomalies, 604–605
history, 140–141 Hunter grab, 51, 51F short-wavelength anomalies, 604–605
hydrodynamics, 144, 145F Petersen grab, 50, 50F data reduction, 601–602
instrumentation, 144–145 Smith-McIntyre grab, 51, 51F free air gravity correction, 602
longest deployment, 141 van Veen grab, 50–51, 50F latitude correction, 602
missions, 145–147 present technology, 57–58 vehicular motion, 602
size, 141 disadvantages of box corers, 58 field gradient, 600, 601
specialized versions, 141 disadvantages of warp activation, 58 measurement, 600–601
speed, 143, 144 popularity of grabs, 58 auxiliary measurements, 600
thermal stratification engine, 143 problems, requirements and future history, 602–603
volume displacement, 141 developments, 58 instrumentation, 601
wings, 141 chemical studies, 58 platforms, 601
WOCE floats as, 94 macrobenthic studies, 58–59 North Pacific, 604F
Gliding, 143 cost of analysis, 59 observations, applications, 600
Global Drifter Program, 91 need for new instrument, 59 South Pacific, 605F
Global Earth Observation System of precision required, 59 uncertainty, 603
Systems (GEOSS), 521 spatial/temporal population units, 600
Global geostrophic circulation, dynamic variations, 59 Gravitational potential energy, 502–503
sea surface topography, 503–505, meiofauna studies, 58–59 Gravity, 502–503
505F sample volume and capture numbers, anomalies, gravitational sea surface
Global mean surface salinity, 584, 585F 54F topography, 502–503, 504F
Global mid-ocean ridge (MOR) sampling different sediments, 57 Gravity gradiometers, 601
discoveries, deep submergence science Hamon grab, 57, 58F see also Gravimeters
studies, 127 Holme grab, 57, 57F Gravity models, extraordinary gravity
observation using human-operated self-activated bottom samplers, 54 field data, 505
vehicles (HOV), 199 Bedford Institute of Oceanography’s Gravity unit, 600
Global Ocean Data Assimilation sampler, 55 Gravity waves
Experiment, 217 compressed-air-powered samplers, internal, phytoplankton interactions,
Global Ocean Observing System (GOOS), 54–55 171F
217 Heriot-Watt University’s grab, 55, 55F Grazing angle, 380, 389–391, 389F, 390F
Global Positioning System (GPS), 167 history of hydrostatic machines, 55 Green fluorescence like proteins (GFP),
drifter tracking, 91 hydraulically-powered samplers, 55 258
oceanographic research vessels, 6 hydrostatically-powered samplers, 55 Greenhouse gas
shallow-water manned submersibles modified Petersen grab, 54–55 SST modulates exchanges of heat and,
using, 159–160 Shipek grab, 54, 54F 558
Global Precipitation Measurement (GPM) solution to warp-activation problems, see also Climate change
mission, 596 54 Greenland Sea
Global sea level spring-powered samplers, 54 mixed layer depth profile, 461F
variability, satellite altimetry, 505–506, uses, 50 tomographic experiments, 460–461,
506F warp activation problems, 51–52 460F
see also Sea level changes/variations; contact with vessel, 52 GSNW (Gulf Stream North Wall), index,
Sea level rise drift, 53 copepod numbers and, 21–22, 22F
‘Global thermometer,’ satellite remote effects of wind/current conditions, 53 Guildline Inc., 32–33
sensing of SST, 558 grab bounce, 52–53 Gulf of Mexico (GOM)
climate change forecast models, effect of ship’s roll, 52–53 fronts, satellite remote sensing and,
556–558 effect of weather conditions, 53, 53F 565–566, 565F
global climate change problems, 556 initial penetration, 53–54 ‘Gulf’ series of high-speed samplers, 74,
thermal inertia of the ocean, 556–558 final sample volume, 53, 53F 75F
GOES Imager, 554 influence of sampler weight, 53 Gulf Stream
GOOS (Global Ocean Observing System), minimal penetration required, 53–54 identification, satellite remote sensing of
217 pressure wave effect, 54 SST, 556
GPS see Global Positioning System (GPS) effects of downwash, 54 Gulf Stream North Wall (GSNW), index,
Grab samplers see Grabs for shelf benthic warp heave, 52 copepod numbers and, 21–22, 22F
sampling effect of rigging, 51, 52 Gulf stream rings, 89
Grabs for shelf benthic sampling, 50–59 effect on bite depth, 52
alternatives to grabs, 55–56 Gravel H
chemical sampling, 57 geoacoustic properties, 421T
ideal sediment sample, 55 Gravimeters, 601 Half-space seafloor, 389–391, 389F
Knudsen sampler, 55–56, 56F absolute, 601 Halley, Dr Edmund, diving bell, 156
precision corers, 57 platforms, 601 Harbor Branch Oceanographic
Reineck sampler, 56–57, 56F relative, 601 Institution, human-operated
spade box samplers, 56–57, 57F see also Gravity gradiometers vehicles (HOV), 199T
632 INDEX

Hardware (computer), autonomous Hooke’s law, gravimetry and, 601 Hydrographic departments, history, 4
underwater vehicles (AUV), 166 HOPLASA (Horizontal Plankton Hydrolight, 319
Hardy, Sir Alister Sampler), 76, 77F simulations, 318–319, 319, 319F, 320F,
Continuous Plankton Recorder, 15, 16F, Horizontal Plankton Sampler 321F, 322F
74–76, 75F (HOPLASA), 76, 77F Hydrologic budget, 587
plankton indicator, 74, 75F Hotel load, 164 Hydrophone arrays, bioacoustic research,
Hardy Continuous Plankton Recorder, 67 Hot film sensor, 233 426F, 430, 431
Hatshepsut, Queen of Egypt, 3 disadvantages, 233 Hydrophone flow noise, 360
Hawaii-2 Observatory, deployment, 136F HOV see Human-operated vehicles Hydrophones
Hawaiian Ridge, internal tides, 463F HROV (hybrid remotely-operated moorings, 108, 108F
Hawaii Mapping Research Group, deep- vehicle), 202 Hydrothermal communities, deep
towed vehicles, 198T Hugin AUV, 163F submergence science studies, 127,
Hawaii Undersea Research Laboratory, Hulk vent, 377F 129–131, 135F
human-operated vehicles (HOV), Human-operated vehicles (HOV), Hydrothermal systems
199T 197–199 diffuse flow mapping, 376
HBOI (Harbor Branch Oceanographic advantages, 197–199 Hydrothermal vent(s)
Institution), 199T dive duration, 197–199 deep manned submersibles for study,
Head wave, acoustics in marine equipment, 201F, 207 154–155
sediments, 385, 385F, 393 future prospects, 207–208 mid-ocean ridge (MOR) crest, discovery,
Hearing Humic acid(s), fluorometry, 269T, 270 125
marine mammals, 426–427, 428F Humidity salinity and, 28
terrestrial mammals, 427–428, 428 measurement see Humidity scattering sensors in monitoring of,
Heat flux measurement 331T
latent, 219 satellite remote sensing, 594–595 Hydroxyl radical (OH), 304–305, 305
measurement, 220, 223–225 see also Evaporation Hygrometers, 224
resistance wires, 223 Humidity Exchange Over the Sea capacitance change, 215
sensor mounts, 223–224, 223F (HEXOS) experiment, 216, dew point, 215
sonic thermometry, 224 475–476 infrared, 224–225, 224F, 225F
thermistors, 223 Humidity measurement, 213–214 krypton, 224
thermocouples, 223, 223F capacitance sensors, 214–215 Lyman a, 224
sensible, 219 capacitance change hygrometers, 215 Hyper Dolphin, 202T
Heating classical sling psychrometer, 213–214 HYPER-DOLPHIN ROV, 187T
oceanographic research vessels, 6 psychrometric method, 213–214, Hyperspectral sensors, 497–498
Heat transfer processes, sea surface, 590 214F Hyper-spectral systems
Heavy metals see also Evaporation; Humidity used by aircraft for remote sensing see
fluorometry, 269T, 270 exposure to salt, 215 Aircraft for remote sensing
Helium dew point hygrometers, 215
ideal gas and, 27 protective devices, 215 I
Henry the Navigator, Prince, 4 resistance thermometer psychrometer,
Hensen net, 70, 71F 214 IAPSO standard seawater, 29
Herglotz-Bateman-Wieckert integration, error due to salt build-up, 214 Ice
393–394, 394F housing arrangements, 214 sea see Sea ice
Herring (Clupea), 368, 370F spray-removal devices, 215–216 strengthening, oceanographic research
North Sea abundance, time-series, device performance, 215–216 vessels, 6
19–20, 20F quantitative effectiveness, 216 see also Sea ice
HEXMAX (HEXOS Main Experiment), spray flinger, 215, 215F Ice, Cloud and land Elevation Satellite
216, 475–476 Hurricane(s), modeling see hurricane (ICE-SAT), 528
HEXOS (Humidity Exchange Over the models Icebreakers, acoustic noise, 404
Sea) experiment, 216, 475–476 1999 season (Atlantic), 556 Icebreaking, polar research vessels’
HF see High frequency band (HF) intensification capability, 10
High-frequency acoustics, zooplankton prediction of development of storms, Iceland-Faroes front, 261F
sampling, 83 SST, 556 Ideal gas law, and temperature scale, 27
High frequency band (HF), 365 satellite remote sensing of SST Idronaut,S.r.l, 32–33
definition, 357 application, 556 IFREMER see French Research Institute
High-speed plankton samplers, 70–71, see also Heat flux for Exploration of the Sea
72T, 75F Hurricane Edouard, 276, 278F (IFREMER)
Hikino submersible, 157–158 Hybrid remotely-operated vehicle IGOS (Integrated Global Observing
HiROS optical system, 277F (HROV), 202 Strategy), 521
see also Ocean optics Hydrocasts, 25–26, 26F IGW (internal gravity waves),
History Hydrodynamics phytoplankton interactions, 171F
shallow-water manned submersibles, gliders, 144, 145F IKMT (Isaacs-Kidd midwater trawl), 70
156–158 Hydrogen (H2) Imaging systems, deep-towed, 197
see also Satellite oceanography absorption band, 584, 585 IMI-30, 198T
HMRG (Hawaii Mapping Research Hydrogen sulfide, 245 IMO see International Maritime
Group), 198T Hydrographic cast (hydrocast), 25–26, Organization (IMO)
Hoki, acoustic scattering, 371 26F Indian Ocean
 INDEX 633

nepheloid zone turbidity, 287F new technologies, 330 Puerto Rico, 464F
Indian Ocean Standard net, 70, 71F volume scattering function see see also Acoustics, marine sediments;
Industrial inflows, impact on water Nephelometry Internal wave(s); Satellite altimetry
quality, monitoring with scattering models, 247 Internal wave(s)
sensors, 331T see also Bio-optical models energy spectra, 439–440, 440F
Inelastic optical scattering, 494–495 quantities, 314T see also Garrett-Munk spectrum
Inertial navigation system, 165F, 167 see also Apparent optical properties seismic reflection profiling and,
Infrared atmospheric algorithms, satellite (AOPs); Infrared (IR) radiometers; 439–440
remote sensing of sea surface Ocean color; Ocean optics; Optical tidal frequency, satellite remote sensing
temperatures, 548–550 particle characterization application, 567–568, 568F
Infrared hygrometers, 224–225, 224F, INL (intermediate nepheloid layers), 282, tomography, 461–463
225, 225F 290–291 see also Internal tides
Infrared (IR) radiation, 272, 277F, 480 Inorganic particles International Antarctic Treaty (1959),
measurement theory, 480 as optical constituents of sea water, 318 10
emittance, 480, 481F INS (inertial navigation system), 165F, International Geophysical Year, 5
Planck’s law, 481 167 International Maritime Organization
radiance, 480, 481F In situ measurements, 528, 529 (IMO), 6
radiant energy, 480 In situ measurement techniques, International Satellite Cloud Climatology
radiant flux, 480 geoacoustic parameters, 386–387, Project, 593
radiant intensity, 480, 481F 388T International Ship Operators Meeting,
spectral radiance, 481, 481F cross-hole method, 387–388, 388F 12
see also Solar radiation ISSAMS (in situ Sediment Acoustic InterOcean S4 electromagnetic current
Infrared (IR) radiometers, 483F, 480–491 Measurement System), 387, 387F meter, 116F, 117T
application, 485 near-surface method, 387 Interrogation Recording and Locating
broadband pyrgeometers, 485–487, In situ zooplankton detecting device, 82F, System (IRLS), 91
485F 83 Intrusions
multichannel radiometers see Instabilities North Atlantic Front, 434–435
Multichannel radiometers in Pacific Ocean, 556, 558F Inverse catenary moorings, 102–103,
narrow beam filter see Narrow beam Instrumentation 102F, 103T
filter radiometers aircraft for remote sensing, 471 Inversions
spectroradiometers, 489F, 490–491 benthic flux landers, 179 intermediate nepheloid layers and, 282,
cooled detectors, 551 Continuous Plankton Recorder (CPR) 290–291
design, 480–483 survey, 16F, 18, 72T, 74–76, 75F IODP see Integrated Ocean Drilling
calibration system to quantify output, floats, 92–93 Program (IODP)
484–485 gliders (subaquatic), 144–145 Ionic conductivity see Conductivity (sea
detector and electronics system, 483 gravimetry, 601 water)
environmental system, 483–484 oceanographic, deep submergence Ionosphere activity, 588
antireflection coatings, 484 science, 125 IOP see Inherent optical properties (IOPs)
thermal shock considerations, 484 ocean optics see Ocean optics IRLS (Interrogation Recording and
window materials, 484, 484F wet chemical analyzers, 338–339 Location System), 91
fore-optics system, 483 Integrated Global Observing Strategy Iron (Fe)
mirrors, 483 (IGOS), 521 photochemical reaction, effects of,
spectral filter windows, 484, 484F Integrated Ocean Drilling Program 308
detector types, 483 (IODP), 35 Irradiance, 272–281
future directions, 490 drilling sites, 36F angular distribution, 572
microwave radiometers vs, satellite logging tools, 39–40 definition, 274–275
remote sensing of SST, 550–551 Integrated water vapor (IWV), 594–595 downwelling, 255–256, 274–275
new generation/future improvements, Interface waves, seismo-acoustic, measurements, 274–275, 275–276,
558 384–385, 389F, 392–393 314–315
satellite-borne, 490–491, 549F, 551, acoustic remote sensing, 389F, 394 scalar, 274–275
551T arrival times, 385, 385F, 394 solar spectrum, 573F
see also Air-sea gas exchange; dispersion diagram, 394, 394F upwelling, 253, 274–275
Radiative transfer; Satellite remote Rayleigh wave, 384–385 see also Infrared (IR) radiometers;
sensing of sea surface temperatures Scholte wave, 383T, 384–385, 384F Ocean color; Ocean optics;
Inherent optical properties (IOPs), Stoneley wave, 384–385 Radiative transfer
272–273, 272–281, 313–315, 323, see also Acoustics, marine sediments; Irradiance reflectance, 274–275
493 Seismo-acoustic waves bio-optical modeling, 253, 255F
apparent optical properties and, 275 Intermediate nepheloid layers, 282, definition, 253–254, 317
chlorophyll concentration and, 250, 290–291 Isaacs-Kidd midwater trawl (IKMT), 70
251–252, 251F, 252T, 253F Internal gravity waves, phytoplankton ISCCP (International Satellite Cloud
coefficients, 272–273, 323 interactions, 171F Climatology Project), 593
definition, 313, 316F Internal tides, 461–463 Isis, 202T
mathematical operations, 273 Hawaiian Ridge, 463F Isostatic compensation, 605
measurement observations ISSAMS (in situ Sediment Acoustic
beam attenuation coefficient see satellite altimetry see Satellite Measurement System), 387, 387F
Transmissometry altimetry IWV (integrated water vapor), 594–595
634 INDEX

J Kittiwake (Rissa tridactyla), Layered seafloor, acoustic remote sensing,

breeding success 391F, 392, 392F
Jaguar/Puma, 205T time-series, 19–20, 20F LBL (long-baseline navigation), 207
JAMSTEC see Japan Marine Science and Knudsen Sea State, 403 LDEO-BB ocean bottom seismometer,
Technology Center (JAMSTEC) Krill (Euphausiacea) 191T, 194F
JAMSTEC-ORI ocean bottom acoustic scattering, 372–373 Leavitt net, 71–72, 73F
seismometer, 190T, 191 Krypton hygrometers, 224 LF see Low frequency band (LF)
Japanese Marine Observation satellites, Kubelka-Munk function, 240 LHPR see Longhurst-Hardy Plankton
525–526 Kumukahi, acrylic plastic sphere, 158 Recorder (LHPR)
Japanese National Space Development Kuril-Japan trench, 291 LIDAR see Light detection and ranging
Agency, 526 Kuroshio Current Light, 272
Japan Marine Science and Technology frontal position, satellite remote sensing sources, absorptiometric chemical
Center (JAMSTEC) of SST, 556 sensors, 241–242
autonomous underwater vehicles, 205T region upstream of Tokra Strait see also Ocean optics; Solar radiation
deep-towed vehicles, 198T East China Sea see East China Sea Light attenuation coefficient, Black Sea,
human-operated vehicles (HOV), 199T seismic reflection water-column 498F
remotely-operated vehicles (ROV), profiling, 435, 436F Light detection and ranging (LIDAR),
202T see also East China Sea 262–263, 262T
Jason-1 satellite, 519 Kuroshio Extension Ligurian Sea
Jason II, 202T, 203F, 204F front, seismic reflection profiling, 437F acoustics in marine sediments, 388,
Jason and the Argonauts, 3 388F, 389F
Jason/Medea ROV, 187T L Limacina retroversa, 373
Jason remotely operated vessel (ROV), Limestone
deep submergence studies, LAA (large amorphous aggregates) geoacoustic properties, 421T
125–126, 127F, 128F, 130F, camera, 83 Link, Edwin, 159
133–136, 136F Laboratories LISST (laser in situ scattering
JCOMM (Joint Technical Commission for oceanographic research vessels, 6 transmissometry), 353–355, 355F
Oceanography and Marine heating, ventilation and air Lloyd mirror effect, 407, 407F
Meteorology), 518 conditioning, 6 LOCHNESS (Large Opening Closing
JERS-1 satellite, 560 Labrador Current (LC), 434–435 High Speed Net and
Johnson-Sea-Link, 158, 159 Labrador Sea Environmental Sampling System),
corrosion immunity, 159 Seaglider mission, 146F 79
diver operations, 159 tomography, 460 LOFAR (Low Frequency Acoustic
multi-science dives, 160 Lake, Simon, Argonaut I, 156 Recording and Analysis), passive
search and recovery operation, 160 Lambert’s scattering law, 411 sonar beamforming, 449–450
state of the art vehicles, 159–160 Laminaria saccarina, acoustic scattering, Logging
thicker plexiglas, 158, 159, 159F 374 definition, 40
unique features, 159 Lamont nephelometer, 282 tools, 39–40
Johnson Sea-Link submersibles, 199T La Niño Logging-while-drilling (LWD), 41–42
JOIDES Resolution, R/V chlorophyll a concentrations, 581F, 582 Loligo opalescens, acoustic scattering,
facts, 37T see also El Niño Southern Oscillation 373–374
Joint Oceanographic Institutions (ENSO) Loligo vulgaris reynardii, acoustic
Incorporated (JOI), 514 Large amorphous aggregates (LAA) scattering, 373–374
Joint Panel on Oceanographic Tables and camera, 83 Lomonosov Ridge
Standards, seawater equation of Large Opening Closing High Speed Net acoustic reflection, 402
state, 614 and Environmental Sampling Long-baseline (LBL) navigation, 207
Joint Technical Commission for System (LOCHNESS), 79 Longhurst-Hardy Plankton Recorder
Oceanography and Marine Large-scale salinity events and (LHPR), 72T, 78–79, 78F
Meteorology (JCOMM), 518 thermohaline convection, 587 Optical Plankton Counter vs., 299–300
Juan de Fuca Ridge LARS (launch and recovery system), Loran (LOng RAnge Navigation), 5
diffuse flow, 377F, 378F 185–186 Love waves, 383T, 384, 384F
Juday net, 70, 71F Larval fish arrival times, 385, 385F
Junge distribution, 248 analytical flow cytometry, 297 Lower atmospheric adsorption, 587
see also Fish larvae Lowestoft sampler, 74, 75F
K Laser diffraction instruments, 274 Low frequency band (LF), 359–360
Laser-induced fluorescence, aircraft definition, 357
Kaiko (Japanese ROV), 151, 187T remote sensing, 472 distant shipping noise see Ship(s)
Kaula report, 510–511 Laser in situ scattering transmissometry Low molecular weight (LMW) organic
Kelvin-Helmholtz shear instabilities (LISST), 353–355, 355F compounds, photochemical
modeling, 492, 493F Latent heat flux production, 304–305
Kennedy, R M, acoustic noise, 364, 364F satellite remote sensing, 594–595 Low-noise microwave radiometers, 584
Kerman, B R, 364 Launch and recovery system (LARS), Lulu, R/V (support vessel), 148–150,
Kewley, D J, 364 185–186 152–153
K-Gill anemometer, 221–222, 221F Launchers, normal (surface ship) LWD (logging-while-drilling), 41–42
Kirchoff approximation, marine deployment of expendable sensors, Lyman a hygrometers, 224
organisms, 371 46 Lyon, Waldo, 398–399
 INDEX 635

M surfacing, 150 vocalizations, 425, 427–428

second generation, 148–150 songs, 428–429
Mackerel (Scomber scombrus), shallow-water submersibles vs, 156, 158 sensory systems
acoustic scattering, 372 surface support, 152 hearing, 426–427, 428F
Macro Flow Planktometer, 297–298 support ships, 152–153, 153, taxonomy
Magellan, Ferdinand, 4 153–154 Order Cetacea see Cetaceans
Magnetic compasses, 167 third generation, 150 Marine Optical Buoy (MOBY), 576
Makakai submersible, 158 typical components, 148 Marine sediments see Seafloor sediments;
Manganese oxides, photochemical water pressure, 150 Sediment(s)
production, 308 ambient pressure, 150–151, 150F Marine snow
Manipulators pressure compensated housings, 151, underwater video profiling, 83
manned submersibles, 152 151F MARVOR floats, 95
master-slave system, 152 pressure hull material, 151 Massachusetts Bay
remotely operated vehicles (ROVs), 184 size and shape, 151 water-column profiles, 170F
Manned submersibles (deep water), Manned submersibles (shallow water), Master-slave system, manipulators,
148–155 156–161 manned submersibles, 152
advantages of ROVs as scientific achievements, 160 MASZP (moored, automated, serial
research vehicles, 186–187 acrylic plastic pressure hulls, 157–158 zooplankton pump), 79, 81F
buoyancy, 150–151 acrylic sphere limits, 158 Matlab, 25
extra buoyancy, 151 diver ‘lock-out’ capability, 157, 159 MBARI see Monterey Bay Aquarium
syntactic foam, 151 drop weight methods, 158 Research Institute
cameras, 152, 152F enabling underwater activity, 156 MBT (mechanical bathythermograph), 43
deep submergence science, 125, 127F history, 156–158 MCS (multichannel seismic surveys), 6,
energy, 151 marine science in situ, 160 10
batteries, 151–152 new smaller vehicles, 160 Mean meteorological variables
examples of deep submersibles operations, 160 measurement methods by land and sea,
Alvin, US submersible, 148–150, 153, multi-science dives, 160 211
153F, 154 search and recovery operation, 160 sensors for measurement see Sensors for
Deep Flight (high speed submersible), Space Shuttle Challenger disaster, 160 mean meteorology
150 present day, 158–160 Mechanical bathythermograph (MBT), 43
Lulu, R/V (support vessel), 148–150, replacement by ROVs, 160 Meddies, 89
152–153 scientific uses, 159 Mediterranean Sea
Mir I and Mir II (Russian support vessel, 160 acoustic time dispersion, 423, 423F
submersibles), 148–150, 153, transition point (1000m), 156, 158 bottom layering, 418F
153–154, 154F underwater missions, 156 sound speed profile, shallow water, 417
Nautile (French submersible), unique sample collection, 160 tomography, 460, 463–465, 465F
148–150, 153, 154F variable ballast systems, 158 Medium resolution imaging spectrometer
Sea Cliff (US Navy submersible), viewing windows, 158 (MERIS), 574–575
148–150, 153 Mantle Meinesz, Vening, 603
Shinkai (Japanese submersible), gravimetric anomalies, 605–606 Melt ponds, 532
148–150, 151, 153–154, 154–155, see also Melt Metal(s)
155F Mapping benthic exchange, 181–182
world-wide, 153 AUVs, 204–205 chemical sensors, 245
history, 148 undersea see Bathymetry electrodes see Electrodes
instrumentation, 151–152 Mapping and charting vessels, 9 heavy see Heavy metals
cameras, 152, 152F hydrographic survey ships, 9 Meteor Expedition, 5, 11
manned submersibles (deep water), Marginal ice zone (MIZ) Meteorological models, use in satellite
151–152 acoustic propagation, 404 altimetry, 502
pressure resistant housing, 152F Marianas Trench, sea surface topography, Meteorological satellites, 519
life support, 151 502–503, 504F Meteorological sensors, moorings,
extra life support, 151 Marine-Atmospheric Emitted Radiance 100–101
pressure hull, 151, 152F Interferometer (M-AERI), 488–489 Meteorological variables, 211
major contributions, 151 Marine biodiversity Methane (CH4)
manipulators, 152 deep submergence science studies, absorptiometric sensor, 245
master-slave system, 152 131–132 Methyl iodide, photochemical production,
navigation, 152 zooplankton, 22–23, 22F 308
acoustic navigation systems, 152 Marine mammals Methyl nitrate, photochemical
Differential Global Positioning System bioacoustics, 426–427, 425–431 production, 308
(DGPS), 152 basics of, 425 METOP satellites, 591
pressure resistant housing, 152F research methods, 429–430, 430F MF see Midfrequency band (MF)
principles, 150 acoustic location, 426F, 430 Microbead thermistors, 223
Deep Flight (high speed submersible), telemetry, 430F, 431 Microelectrodes, sensor landers, 178,
151–152 sound production mechanisms, 427F, 179F
descent and ascent, 149F, 150 428–429, 429F Micrometeorological flux measurements,
negative buoyancy, 150 structure and biological function of 218, 219F
neutral buoyancy, 150 calls, 425–426, 426F, 427F heat see Heat flux
636 INDEX

Micrometeorological flux measurements Mixing components, 97–100

(continued) nepheloid layers, 286 data transmission, 104, 106
moisture, 218, 224 see also Diffusion; Double-diffusive deployment, 106–108
see also Hygrometers convection; Mixed layer anchor-first, 107–108
momentum see Momentum flux MIZ see Marginal ice zone anchor-last, 106, 107F
measurements MLP (multilayer perceptron neural designs, comparison of shapes,
sensors, 220, 218–226 networks), 496F early designs, 97
frequency response, 220, 225 Mnemiopsis leidyi (a comb jelly) fatigue, 103, 105
performance degradation, 220, 220F acoustic scattering, 373–374 hybrid, 103–104, 104F
state-of-the-art, 225, 225F MOBY (Marine Optical Buoy), 576 hydrophone array, 108, 108F
see also Sensors for mean MOCNESS (Multiple Opening/Closing instrumentation, 99
meteorology; Turbulence Net and Environmental Sensing inverse catenary, 102–103, 102F, 103T
sensors System), 72T, 79, 80F, 81T combination with subsurface design,
see also Air-sea gas exchange dual-beam acoustic system, 84, 85F lines
Microseism noise, 359 MOCS (Multichannel Ocean Color bite damage, 97
Microstructure Sensor), 474–475 breakage, 98
salinity, 29F MODE see Mid-Ocean Dynamics compliance, 101
temperature, 28 Experiment (MODE) telemetry via, 105
Microwave backscattering, 590–591 Moderate Resolution Imaging termination, 98
see also Scatterometry Spectroradiometer (MODIS), 554, weight, 97
Microwave data, atmospheric 593–594 meteorological sensors, 100–101
interference, 525, 527 36-band imaging radiometer, 554 modeling of performance, 108–109
Microwave humidity sounder (MHS), 595 narrower swath width than AVHRR, observational disadvantages, 140
Microwave measurements, satellite 554 plankton sampling systems, 76–79,
remote sensing of SST, 550–551 ocean color sensing, 574–575 81F
Microwave optical depth, 585, 587 ocean color sensor, 575T recovery, 97, 98, 99
Microwave radar, 502 polarization errors, 578 remote repair, 104
Microwave radiometers, 477, 550 MODIS see Moderate Resolution Imaging semi-taut, 101–102, 102F, 103F, 103T
infrared radiometers vs, satellite remote Spectroradiometer (MODIS) subsurface
sensing of SST, 550–551 Moho (Mohorovic discontinuity), 600 advantages and disadvantages, 97, 98
low-noise, 584 Molecular noise, 358F, 365 design, 97–100, 99F, 100F
Microwave radiometry, 514, 584, 586 Molecular recognition element, surface
Microwave salinometers, 478–479 absorptiometric chemical sensors, advantages and disadvantages, 97
proposed space instrument for SSS, 242–243 applications, 100–101
478–479 Momentum flux measurements, 218, depth-measurement problems, 103
sea surface salinity measures, 478 220–222, 221F, 222F design, 100–106
tracer of water mass movement, 478, direct covariance method, 219 two-dimensional array, 109F
478F eddy correlation, 219 U-shaped, 108, 108F
Microwave scanning radiometer, 526, 594 longitudinal velocity fluctuations, 219, see also Current meters
Microwave scattering, sea surface, 219F, 220F MOR see Mid-ocean ridge(s) (MOR)
590–591 sensors, 220–222 Moraine, geoacoustic properties, 421T
Microwave sensing, freshwater flux, cone anemometers, 221 MORS optical system, 277F
595–596 gyro-stabilized systems, 223 see also Ocean optics
Mid-Arctic Ridge, earthquakes and hot-film anemometers, 221 MOSS optical system, 277F
acoustic noise, 404 hot-wire anemometers, 220 see also Ocean optics
Midfrequency band (MF), 362–364 K-Gill anemometer, 221–222, 221F Motors, autonomous underwater vehicles
definition, 357 motion correction, 220, 222–223 (AUV), 164
Mid-Ocean Dynamics Experiment sonic anemometers, 222, 222F Moving Vessel Profiler towed vehicle,
(MODE) strapped-down systems, 223 60
floats, 94–95 vertical velocity fluctuations, 218–219, MT (multiple turnover fluorescence
Mid-ocean ridge(s) (MOR) 219F induction), 259
crest, hydrothermal vents, discovery, see also Turbulence sensors Mud
125 Monterey Bay acoustics in marine sediments, 388T
Mid-ocean ridge crest, hydrothermal autonomous underwater vehicles (AUV) Multibeam bathymetric mapping, active
vents, discovery, 125 observations, 171F sonar images, 447, 448F
Mie scattering, 283–284 ocean field decorrelation, 171–172 Multibeam sonar, 168–169
Mie theory, spectral scattering, 324–325 Monterey Bay Aquarium Research Multichannel Ocean Color Sensor
Milligal (mGal) (unit for gravity Institute (MOCS), 474–475
anomaly), 600 autonomous underwater vehicles, 205T Multichannel radiometers, 488F, 489–490
MIR submersibles, 199T remotely-operated vehicles (ROV), 202T in situ, 487
dive duration, 207 Moored, automated, serial zooplankton Multichannel seismics, oceanographic
Mir I and Mir II (Russian submersibles), pump (MASZP), 79, 81F research vessels, 6
148–150, 153, 153–154, 154F Moorings, 97–109 Multichannel seismic surveys (MCS),
Mixed later modeling, 587 anchor release, 98 geophysical research vessels, 10
Mixed layer buoys, 98, 101 Multilayer perceptron (MLP) neural
Arctic, acoustics, 400 compliance, 101 networks, 496F
 INDEX 637

Multiple Opening/Closing Net and properties of seafloor sediments, measurements, 323–324, 324
Environmental Sensing System see measurement of, 386 optical rejection, 327
MOCNESS (Multiple Opening/ sediment cores, 386, 387F principles, 328
Closing Net and Environmental Nautile (French submersible), 148–150, transmissometry vs., 323, 332
Sensing System) 153, 154F, 199T, 200F see also Inherent optical properties
Multiple turnover (MT) fluorescence Naval Electronics Laboratory (NEL), (IOPs); Ocean optics; Optical
induction, 259 398–399 particle characterization;
Naval research, optical sensor Turbulence sensors; Volume
N development and, 329 scattering function (VSF)
Navicula planamembranacea, 18 Nereus (ROV), 202T, 205F
Nansen, Fridtjof, 29 Navier-Stokes equation Nested survey strategy, 207
Nansen bottles, 29 turbulence, 229 Neural networks
salinity measurements from, 30 Navigation, 152 algorithm optimization for coastal water
Nansen closing net, 71–72, 73F acoustic, autonomous underwater remote sensing, 496, 496F
NAO see North Atlantic Oscillation vehicles (AUV), 167 Neuston, net samplers, 71–74, 76F
(NAO) acoustic systems, 152 Nichols, R H, 360
Narrow beam filter radiometers, 487–488 autonomous underwater vehicles (AUV), Nimbus 5 satellite, 524, 525
calibration, 486 166–168 Nimbus 6 satellite, 91, 525–526
‘black-body’ strategy, 486, 486–487, using geophysical parameters, Nimbus 7 satellite, 511F, 512, 524,
487F, 488F 167–168 525–526, 529F
chopper systems, 485–486 dead-reckoning, 167 Coastal Zone Color Scanner (CZCS),
advantages, 486 deep-sea vehicles, 207 571, 575T
dynamic detector bias compensation, Differential Global Positioning System Nisken bottles, 32F
486 (DGPS), 152 Nitrate (NO-3)
rotary, 485–486, 486F hyperbolic, 167 benthic flux, 175–176, 177F
tuning fork, 485–486, 486F inertial system, 165F, 167 nitrite and, simultaneous determination
polarization of sea surface reflection as spherical, 167 by flow injection analysis, 336
function of view angle, 486, 487F Navy Remote Ocean Observing Satellite photolysis, 307–308
NASA (United States National (NROSS), 511F, 515 see also Nitrogen
Aeronautics and Space NDSF see National Deep Submergence Nitric oxide (NO)
Administration), 525 Facility photochemical production, 307–308
early data, 509–510, 509F, 510F Necho II, Pharaoh of Egypt, 3–4 Nitrite (NO2-)
Earth Observing System (EOS), 554 Neil Brown Instrument Systems, Mark III nitrate and, simultaneous determination
missions see Satellite oceanography CTD Profiler, 31–32 by flow injection analysis, 336
partnership with field centers, 513 operating conditions, 32 Nitrogen dioxide (NO2)
satellites see Satellite oceanography Nekton submersibles, 157 absorptiometric sensor, 245
SIR-B and SIR-C experiments, 561 Nemo, plexiglas hull, 158, 159 photochemical production, 307–308
Space Shuttle, 561 Nepers, 409 NOAA see National Oceanographic and
NASA scatterometer (NSCAT), 515, 591, Nepheloid layers, 282-292 Atmospheric Administration
592T Atlantic Ocean, 288F (NOAA)
NASA Skylab, 586 bottom, thickness, 285 NOAA satellites, 551
NASA Team algorithm, 527, 532F bottom mixed, 282, 285–286 NOAA/TIROS, 511F
National Deep Submergence Facility chemical scavenging, 287–289 NOC see National Oceanographic
(NDSF), 125–126 features, 285–286 Centre, Southampton
autonomous underwater vehicles (AUV), intermediate, 282, 290–291 Nodularia spumigena, 499F
205T mixing, 286 Noise (acoustic)
human-operated vehicles (HOV), 199T particle concentration decay, 286–287 acoustic scintillation thermography, 378
remotely-operated vehicles (ROV), 202T separated mixed-layer model, 286 Arctic, 400–403
vehicles operated by, 125–126, 130F settling, 287 anthropogenic, 404
National Ice Center sources, 289 spectral profiles, 402F
satellite remote sensing, 563–564 surface, 282 deep ocean, 410–412, 414F
National Oceanographic and Atmospheric trenches and, 291 microwave scatterometry, 591
Administration (NOAA), 91 Nephelometers, 282–285 ocean color data, 578–582
satellites, 551 Nephelometric turbidity units (NTU), 325 tomography and, 455
National Oceanographic Centre, Nephelometry, 323–332 Nonlinear SST (NLSST), 549–550
Southampton definition, 323 Nonmethane hydrocarbons (NMHC)
autonomous underwater vehicles, 205T depth profiles, Rockall Trough, 284F photochemical production, 307
deep-towed vehicles, 198T new technologies, 330–332 Non-opening/closing nets, zooplankton
remotely-operated vehicles (ROV), 202T scattering sensors, 325–327, 327F sampling, 70, 71F
National Office of Ocean Exploration, angular configurations, 328, 328F Nonriser drilling, 35–37, 35F
remotely-operated vehicles (ROV), applications, 323, 327–329, 329, Normal (surface ship) deployment,
202T 331T standard ship launchers, 46
National Space Development Agency of naval, 329 Normal mode model (acoustic), 410
Japan (NASDA), 560 backscatter, 274, 329 Normal reflectivity (p), pure water,
NATO Undersea Research Centre calibration, 328–329, 329 481–483, 482F
(SACLANTCEN), geoacoustic deployment, 330 Nortek Aquadopp current meter, 116F
638 INDEX

North Atlantic exchanges, 526 categories, 3

current see North Atlantic Current see also Air-sea gas exchange characteristics, 3
(NAC) Ocean bottom communications (internal/external), 6
current harmonic constants, 462F sound reflection and absorption, control of ship, 6
tidal currents, 462F 408–409 deck working area, 6
North Atlantic Current (NAC), 434–435 topography, satellite altimetry, 503, definition, 6
North Atlantic Front, seismic reflection 504F design characteristics, 8–9
profiling, 434–435 Ocean bottom seismograph (OBS), 189 design factors, 9
North Atlantic Oscillation (NAO), 533 broadband see Broadband (BB) ocean design priorities, 9
North Atlantic Oscillation Index bottom seismometers endurance, 6
Calanus abundance and, 20–21, 21F pool (OBSIP), 189 fisheries research vessels, 9–10
increased, 20–21 short period see Short period (SP) ocean future ships, 12
Northern blue whiting, acoustic bottom seismometers growth in interest, 12
scattering, 371 synthetic larger sizes, 12
Northern hemisphere pressure, 195F new types, 12
geostrophic circulation, dynamic sea vertical motion, 195F SWATH (Small Waterplane-Area Twin
surface topography, 503–505, 505F Ocean color Hull) ships, 12
Northern Sea Route atmospheric correction, 495, 576–577 general purpose vessels, 9
acoustics research and, 397 coastal waters, 492–494 geophysical research vessels, 10
North Pacific algorithms, 495 heating, ventilation and air conditioning,
gravity field, 604F data validation, 496–497 6
North Sea data processing requirements, 574F history, 3–6
Doliolum nationalis, 23–24 definition, 571 larger improved fleet, 5–6
ecosystem regime shift, 21 instruments, 571 ice strengthening, 6
phytoplankton color, 20–21, 21F satellite remote sensing see Satellite laboratories, 6
Norwegian Sea remote sensing mapping/charting vessels, 9
seismic reflection water-column sensor design, 573–575, 575T miscellaneous classes of vessels, 10–11
profiling, 434F, 435, 439F theory, 571–573 mission requirements, 6–8
NOSS (National Oceanographic Satellite see also Reflectance multichannel seismics, 6
system), 513 Ocean currents see Current(s) nature of, 6–8
NROSS (Navy Remote Ocean Observing Ocean Data Acquisition System (ODAS), acoustical systems, 6
Satellite), 511F, 515 474–475 equipment and facilities, 6
NSCAT (NASA scatterometer), 515 Ocean Drilling Program (ODP), 35, 127, IMO category, 6
NTU (nephelometric turbidity units), 325 132F, 133–136, 134F multi-disciplinary role, 6, 12
Nuclear submarines, history, 442 drilling sites, 36F primary requirement, 6
Ocean floor science mission requirements, 6
O record of Earth’s history, 125 scientific personnel, 6
Earth and ocean history, 125, 126F specific purposes, 6
OASIS (Optical-Acoustical Submersible hydrothermal vents, 125, 126–127, navigation/positioning, 6
Imaging System), 83–84 126F, 128F operations, 11
Objectively Analyzed air-sea Fluxes study of seafloor terrain, 125 cruise planning, 11
(OAFlux) project, 595 see also Seafloor EEZ (exclusive economic zone), 11
OBS see Ocean bottom seismograph Oceanic convection international cooperation, 11
(OBS) subpolar regions, 587 measurements and observations, 11,
Observatories see also Convection 11T
AUV docking stations, 205–207 Oceanic dolphins Meteor, German Atlantic Expedition,
AUVs and, 171–172 bioacoustics, 425 11
deep-sea, 207–208 echolocation, 426–427, 427F other classes, 11
and AUVs, 205–207 mechanism, 429, 429F commercial expediency, 11
and benthic flux rovers, 182 optimal frequency, 427 occasional employment, 11
undersea, AUV and, 164 hearing, 428, 428F overside handling, 6
Ocean(s) see also Odontocetes (toothed whales) polar research vessels, 10
color see Ocean color Oceanic precipitation, 587 satellite monitoring, 6
deep see Deep ocean Oceanic processes remote sensing seakeeping and station keeping, 6
deposition of particulate material see program, 586 shipbuilding boom, 5
Aerosols Ocean mixing see Mixing size, 6
features Oceanographic campaigns, remote sensing speed, 6
satellite remote sensing of SST data validation, 496–497 storage facilities, 6
application, 554 Oceanographic expeditions, history, 4 support vessels, 10–11
rainfall estimates and global hydrologic Oceanographic institutions towing, 6
budget, 587 history, 4–5 vans (portable), 6
satellite remote sensing see Satellite Oceanographic instrumentation, deep winches, 6
remote sensing submergence science studies, 125 workboats, 6
trajectories, velocity measurement, 89, Oceanographic research vessels, 3 world fleet, 11–12
91–92 accommodation, 6 International Ship Operators Meeting,
Ocean-atmosphere interactions acoustical systems, 6 12
 INDEX 639

numbers, 12, 12T Odyssey class AUVs, 205T Optical systems, zooplankton sampling
varying status and categories, 12 Office of Naval Research (ONR), 141 see Optical Plankton Counter
see also Fisheries research vessels; acoustics research, 397–398 (OPC); Zooplankton sampling
General purpose vessels; O’Hara plankton sampler, 79, 81F Optodes (optrodes), 239F, 262, 262T,
Geophysical research vessels; Oil industry, surveying, 162–164 239–246
Mapping and charting vessels; Oil slicks Orange roughy (Hoplostethus atlanticus),
Support vessels dampening effect on short surface 368
Oceanographic research voyages, history, waves, satellite remote sensing, acoustic scattering, 371
5 561F, 563 Orbit (satellites), ocean color sensing,
Oceanographic technology, deep remote sensing, 499–500 574–575
submergence science studies, 125 see also Oil spills see also Geostationary orbit
Oceanography from Space (report), 510, Oil spills Organic particles
514 tracking, satellite remote sensing as optical constituents of sea water, 318
Ocean optics, 272 application, 561F, 564 Organophosphate pesticides
apparent optical properties see Apparent see also Oil slicks sensors, 245
optical properties (AOPs) OML see Mixed layer ORI-BB ocean bottom seismometer, 191T
experiment data sets, 276–280 One-dimensional models Origin of life, 137–139
Bermuda Testbed Mooring, 279–280, sediment entrainment, 352–353 Otariinae (sea lions)
280F OPC see Optical Plankton Counter (OPC) hearing, 428F
BIOPS system, 276, 278F Optic(s), ocean see Ocean optics Overshot, 38
sewage plume waters, 276–279, 279F Optical-Acoustical Submersible Imaging Overside handling, oceanographic
ship-based profiling system, 276–279 System (OASIS), 83–84 research vessels, 6
fundamentals, 272 Optical backscatter, water-column, Oxygen (O2)
inherent optical properties see Inherent Massachusetts Bay, 170F benthic flux, 177F
optical properties (IOPs) Optical backscattering sensors, 274, 329 concentration gradient, in sediments,
instrumentation, 273–276 see also Nephelometry 179
absorption meter, 330 Optical fibers deep-sea-floor microelectrode results,
ac-meters, 274 advantages, 241 178, 179F
antifouling methods, 274 chemical sensors, 240–241 fluorescent sensing, 270, 270T
biofouling issue, 274 reflectometric, 244F from photosynthesis, productivity
future goals, 276, 280–281 typical instrumentation system, 242F measures, 260–261
laser (Frauenhofer) diffraction construction, 240 Oxystat benthic lander, 181–182
instruments, 274 evanescent waves, 241, 241F Ozone (O3)
mooring applications, 276, 277F remotely-operated vehicles (ROV), 201 ocean color sensing and, 577
novel, 330 structure, 240, 240F UV-B radiation level, concentration
photosynthetically available radiation transmission characteristics, 240–241, effects to, 302
(PAR) sensor, 275, 277F 241F Ozone Monitoring Instrument (OMI),
scattering sensors see Nephelometry types, 241 577
measurements and fundamental values, Optically active components (AOC), 572
323 Optical particle characterization, 293– P
quantities, radiometric, 313, 314T 301
commonly used, 314T techniques, 293 Pacific Ocean
quasi-inherent optical property, 275 analytical flow cytometry see frontal position instabilities, satellite
terminology, 313 Analytical flow cytometry (AFC) remote sensing of SST, 556, 558F
see also Irradiance; Ocean color; imaging, 300 North see North Pacific
Optical particle characterization; photographic, 300 tropical, SST distribution, satellite
Radiance; Radiative transfer using remotely operated vehicle, 300 remote sensing, 554–555
(oceanic) video, 300 see also North Pacific; South Pacific
Ocean Sensors, Inc, 32–33 optical plankton counting see Optical Pack ice
Ocean surface, 548 Plankton Counter (OPC) Arctic Ocean, 400F
atmosphere interactions, 548 rapid, 293 see also Arctic Ocean
current velocity data, 587 see also Bio-optical models; Paine Instruments, Inc., 30–31
high-resolution global coverage of Fluorometry; Inherent optical PALACE (profiling autonomous
satellites, 548 properties (IOPs); Irradiance; Lagrangian circulation explorer),
see also Satellite oceanography Ocean color; Ocean optics 85–86
ocean-atmosphere interactions, 548 Optical Plankton Counter (OPC), 72T, Parabolic equation acoustic modeling,
see also Sea surface 82F, 83, 298 shallow water, 424
ODAS (Ocean Data Acquisition System), applications, 298, 299F Parabolic equation model (acoustic), 410,
474–475 limitations, 300 411F
Odontocetes (toothed whales) Longhurst-Hardy Plankton Recorder Pargo (USS), 398–399
sound production, 429, 429F vs., 299–300 Particle(s)
taxonomy operation bio-optical models, 247, 249F
family Phocenidae see Porpoises calibration, 299 characterization, 293
see also Marine mammals considerations, 298–299 nonoptical techniques, 293
ODP see Ocean Drilling Program (ODP) technique, 298, 298F optical techniques see Optical particle
Odyssey AUV, data obtained, 170F, 171F see also Zooplankton sampling characterization
640 INDEX

Particle(s) (continued) processes, 302–312 Pigments

detection systems, zooplankton abiotic constituents, optical accessory, 258
sampling, 79–80, 82F properties, 302 Pinnipeds (seals)
inorganic see Inorganic particles calculations, 308 hearing, 428, 428F
organic, as optical constituents of sea CDOM see Colored dissolved organic see also Marine mammals
water, 318 matter (CDOM) Pisces IV, 199T
properties, 293 microbial activity, 306 Pisces V, 199T
size, distribution function, 248, 249F PDM see Particulate detrital matter Pitch attitude, gliders, 143–144
suspended see Suspended particles (POM) and speed, 144
see also Particulate matter production Pitot tubes
Particulate detrital matter (POM) low molecular weight organic momentum flux measurements, 233
absorption spectra, 303F compounds, 304–305 shear stress measurements, 221
definition, 303–304 reactive oxygen species, 302–304 Planck’s constant, 548
photochemical processes, 302 trace gases, 304–305 Planck’s equation, 548
Particulate inorganic carbon (PIC) trace metal, 305–308 Planck’s function, linearized, 549
concentrations, determination by Photodetector, absorptiometric chemical Planktobenthos sampling nets, 74–76,
underwater transmissometry, sensor, 241–242 77F
331–332, 331T Photographic systems Plankton
Particulate matter particle imaging, 300 acoustic scattering, 412
coastal waters, absorption (optical) zooplankton sampling, 79 biogeography, Continuous Plankton
spectra, 494 Photolysis, colored dissolved organic Recorder survey, 18–19
concentration depth profile, 286F matter, effects of, 306 krill see Krill (Euphausiacea)
suspended, optical scattering, 282 Photons recording see Continuous Plankton
see also Particle(s); Particulate colored dissolved organic matter, 305 Recorder (CPR) survey
inorganic carbon (PIC); Suspended Photo-oxidation, 304–305 see also Copepod(s); Fish larvae
particles Photoprotective pigments, 258 Plankton collectors, towed vehicles, 67
Passive systems Photosynthesis, 302 Plants
aircraft for remote sensing see Aircraft phytoplankton see Phytoplankton aquatic, productivity measures, 260–262
for remote sensing Photosynthetically available radiation Plate tectonics
Multichannel Ocean Color Sensor (PAR), 253, 274–275, 315 deep manned submersibles for study,
(MOCS), 474–475 sensor, 275, 277F 154–155
see also Ocean color Photosynthetic electron transfer (PET), Platinum resistance thermometer, 26–27
Ocean Data Acquisition System 260 Pliny the Elder, 4
(ODAS), 474–475 Photosynthetic unit size, photosystem II, PMMA (poly(methyl methacrylate)), 240
see also Satellite remote sensing of sea 260–261 Polar Experiment (POLEX), 397–398
surface temperatures Photosystem II Polar lows, satellite remote sensing
Pathfinder SST (PSST), 549–550 photosynthetic unit size, 260–261 application, 563F, 564–565
Perry Submarine Builders, 156 physiological measurement conditions, Polar research vessels, 10
Clelia, 158F 260 ice capability, 10
Deep Diver, 157F Phycobiliproteins limited icebreaking capability, 10
Personnel sphere, 197–199 autofluorescence, 295 special requirements, 10
Peru trench, nepheloid layers, 290F, 291 Phycocyanin, 258F, 259–260 Polar Water, acoustics, 399–400
Pesticides Physoclists, 368 POLDER ocean color sensor, 575T
optical sensors, 245 Physostomes, 368 POLEX (Polar Experiment), 397–398
PET (photosynthetic electron transfer), Phytoplankton Pollack (Pollachius pollachius),
260 absorption spectra, 303F swimbladder, 371F
pH analytical flow cytometry, 295–296, Pollutants
deep-sea-floor microelectrode results, 295T, 296F fluorometry, 269–270, 269T
178, 179F color, 17–18, 20–21, 21F Poly(methyl methacrylate), in optical
fluorescent sensing, 270, 270T North Sea, 20–21, 21F fibers, 240
gradient in sediments, 179 see also Continuous Plankton Polyaniline, 245
optical sensors, 245 Recorder survey Polycyclic aromatic hydrocarbons (PAHs),
Phaeocystis, CPR survey data, 23, 23F Continuous Plankton Recorder survey, fluorometry, 269–270, 269T
Phenol red, 245 19–20, 20F Pop-down nets, 73–74
PHILLS (Portable Hyper-spectral Imager fluorescence imaging, 261F Porcupine Abyssal Plain, thorium flux,
for Low-Light Spectroscopy), 477 laboratory grown, 262 287–289
Phosphate (PO43-) as optical constituents of sea water, 318 Porosity, 179
benthic flux, 177F photosynthesis Porpoises
Photochemical efficiency, 259–260 determination by pump and probe hearing, 428, 428F
Photochemistry fluorometry, 474 see also Odontocetes (toothed whales)
consumption spectral absorption, 273 Portable Hyper-spectral Imager for Low-
low molecular weight organic see also Coccolithophores Light Spectroscopy (PHILLS), 477
compounds, 304–305 Phytoplankton cycle productivity, passive Potassium chromate, 28–29
trace gases, 304–305 sensors used by aircraft for remote Potential temperature
photochemical fluxes, global estimation, sensing, 475, 476F seawater, 615T, 617
309–311 Piccard, Jacques, 197 Power consumption
 INDEX 641

autonomous underwater vehicles (AUV), Q Radiative transfer equation (RTE), 247,

164, 171–172 254, 315–317
sonar and, 168–169 Quantum yield of fluorescence, 257 Radio direction finding, drifter tracking,
speed and, 164 Quartz 90–91
gliders, 143 in optical fibers, 240 Radiometer antennas, 584, 586
torpedo AUVs, 162 Quasi-inherent optical property, 275 beam width, 585–586
P.P. Shirshov Institute of Oceanology, see also Ocean optics footprint, 585–586
human-operated vehicles (HOV), Quest, 202T Radiometers, 548
199T QuikSCAT satellite, 519, 591, 592T fixed platform, 497, 497F
Practical salinity, 614 infrared see Infrared (IR) radiometers
Practical Salinity Scale (PSS-78), 584, R in-water, 497
584–585 microwave see Microwave radiometers
Practical salinity units (psu), 584 Radar multichannel see Multichannel
Pratt isostasy, 606 drifter tracking, 90–91 radiometers
Precipitation synthetic aperture see Synthetic aperture narrow beam filter see Narrow beam
estimates, satellite measurements, 216 radar (SAR) filter radiometers
measurement, 471 Radar altimetry, 477 reflected sunlight measurement,
see also Evaporation; Rain basic measurements, 502 474–475
Precipitation radar (PR), 596 ocean applications of airborne systems, satellite, 593
Pressure 477 Radiometric quantities, 313, 314T
buoyancy frequency and, 617F Topex/Poseidon satellite altimeter RAFOS (Ranging and Fixing of Sound),
sea, measurement, 30–31 system, 502, 503F 397
Pressure-compensated housings, remotely Radarsat-1 and Radarsat-2, 560 RAFOS floats, 94–95, 95–96
operated vehicles (ROVs), 186 Gulf of Mexico image, 565–566, 565F ballasting, 95
Pressure measurements, 211 oil spills off Point Barrow (Alaska), 561F diagram of structure, 95F
atmospheric pressure and altitude, 211 South China Sea image, 567–568, 568F operating depth, 95
barometer types, 211 Radiance, 481, 571 Rain
Pressure-proof housings, remotely black-body, 481, 481F attenuation, 585, 588
operated vehicles (ROVs), 186 definition, 274–275, 313–314, 314F, see also Precipitation; Rainfall
Pressure sensors, strain gauges, 31F 480, 481F Rainfall
Pressure surfaces, floats and, 92 Hydrolight simulation, 320–321, 320F, heavy, eastern tropical Pacific Ocean and
Pressure transducer (PT), 355F 321F El Niño, 554–555
Prevailing noise, definition, 357 measurement, 314–315, 483 see also Precipitation; Rain
Prochlorophytes, 258–259 see also Infrared (IR) radiometers Raman scattering, fluorescence and, 266,
Profilers, moorings, 99–100 reflectance, 275 266F
Profiling autonomous Lagrangian upwelling, diffuse attenuation RAMS (Random Access Measurement
circulation explorer (PALACE), coefficient, 572 System), 91
85–86 water-leaving, 279–280, 280F, 329, 571, Random Access Measurement System
Propellor anemometers, 211, 212F 572 (RAMS), 91
Propulsion see also Ocean optics Range
efficiency, autonomous underwater Radiant energy, definition, 480 autonomous underwater vehicles (AUV),
vehicles (AUV), 164 Radiant flux, definition, 480 164
remotely operated vehicles (ROVs), Radiant intensity, definition, 480, 481F gliders, 143
183–184 Radiation Ranging and Fixing of Sound (RAFOS),
Proteinaceous material, fluorometry, 269, absorption, 239–240, 240F 397
269T black-body, 481, 481F Rapid backscatter ripple profiler
Protozoa cosmic see Cosmic radiation (BSARP), 344, 344F
analytical flow cytometry, 296–297, electromagnetic, absorption, 239–240 Rayleigh region, active sonars, 445
297F galactic, 587 Rayleigh roughness parameter, 421
Psychrometer infrared see Infrared (IR) radiation Rayleigh scattering, 324–325
classical sling, 213–214 solar see Solar radiation solar irradiance, 577
resistance thermometer, 214 transmittance, Beer’s law, 224–225 Rayleigh waves
Psychrometric method, classical sling ultraviolet, 272 acoustics in marine sediments,
psychrometer, 213–214, 214F see also Solar radiation 384–385
PT (pressure transducer), 355F Radiation shields, for temperature Ray theory (acoustic), 410, 411F
Ptolemy, ship reports used for world map sensors, 212F, 213 shallow water, 424
compilation, 4 Radiative fluxes, 593–594 see also Acoustic rays
Puerto Rico global budget, 593T Razor clam, acoustic scattering, 374
internal tides, 464F see also Air-sea heat flux Reactive oxygen species (ROS), 302–304
internal waves, 461–463 Radiative transfer (oceanic), 313-322 Recovery
Puerto Rico Trench anomaly, 512 studies, 254, 319, 331T benthic flux landers, 175
Pulse repetition frequency (PRF), active see also Hydrolight moorings, 97
sonars, 443 theory, 275, 313 Rectangular mouth opening trawl (RMT),
Pyrgeometer see also Bio-optical models; Infrared 72T, 79, 80F
broadband, 485, 485–487, 485F (IR) radiometers; Ocean color; Red Sea
Pyrogallol red, 245 Ocean optics salinity, 584
642 INDEX

Red tides in particle imaging, 300 ROVER benthic flux lander, 177–178,
spectral range for remote sensing, 495T portable design, 186–187 181F, 182
see also Algal blooms; Red tides propulsion, 183–184 rRNA, fluorescent tagging, 259
Redundancy scanning sonars, 184 RR rays (refracted-refracted rays), 453
mooring instrumentation, 100 scientific research see Scientific research RSR rays (refracted-surface-reflected
Reentry cone, 38–39, 40F vehicles rays), 453
Reeve net, 70 seafloor diffuse flow mapping, Russia
Reflectance, 240, 493 sensors, 184 Arctic research, 397–398
definition, 571 shallow-water manned submersibles Russian Cosmos satellite, 525
irradiance, 571 replaced by, 160 R/V Atlantis, support ship, 127F, 130F
remote sensing, 571, 572 size range, 183
sea bottom, 494 support ships, 201 S
substance concentrations and, 572 tether management see Tether
total surface, 572 management system (TMS) SACLANTCEN, geoacoustic properties of
Reflectance spectra, water types, 493F types, 183 seafloor sediments, measurement
Reflectance spectroscopy, 240 see also Towed vehicles of, 386
chemical sensors, sensor design, 243 video cameras, 184 sediment cores, 386, 387F
see also Absorbance spectroscopy vision, 184 SAFARI, acoustics in marine sediments,
Reflection (of electromagnetic radiation), zooplankton sampling, 84, 85F 395, 395F
240 see also Manned submersibles (deep SAFIRE optical system, 277F
Reflection coefficient, acoustic remote water); Manned submersibles see also Ocean optics
sensing, 389–391, 391–392, 391F (shallow water) SAHFOS see Sir Alister Hardy
Reflection loss, acoustic remote sensing, Remote repair, moorings, 104 Foundation for Ocean Science
391F, 392, 392F Remote sensing (SAHFOS)
Reflectivity (acoustic) aircraft for see Aircraft for remote Salinity
critical angle, 420 sensing conservative nature, 28
sea bottom, 419–420 fluorescence, 263 definition, 584–585
Refracted-refracted (RR) rays, 453 satellite see Satellite remote sensing global mean surface, 584, 585F
Refracted-surface-reflected (RSR) rays, single point current meters, 116F, 120 intrusions see Intrusions
453 validation, application of Labrador Sea see Labrador Sea
Remote environmental measuring units transmissometry and nephelometry, measurement(s), 584–585
(REMUS), 85 331T airborne, 586
Remotely operated vehicles (ROVs), 155, see also Satellite remote sensing all-weather observations, 585, 588
183–188, 199–202, 203F Remote-sensing reflectance, 274, 572 automated electronic in situ, 584
advantages and disadvantages, 201–202 definition, 317 CTD profiler resolution, 30
basic design characteristics, 183 Hydrolight simulation, 322, 322F Practical Salinity Scale (PSS-78),
common designs, 183 Remote sensing theory, 584–585 584
buoyancy, 183 REMUS (remote environmental practical salinity units (psu), 584
challenges and solutions, 186 measuring units), 85 radio frequency, 585
corrosion-resistant materials, 186 REMUS (Remote Environmental titration, 28–29
electrical components, 186 Monitoring Unit S), 204, 205T measuring devices, 28–30
extreme operating environment, 186 Resistance thermometer psychrometer, mid-ocean elevation, 585F
human operators, 186 214 practical, 614
pressure-compensated housings, 186 Resistance wires, heat flux measurement, ranges, 584, 616F
pressure-proof housings, 186 223 seawater, 616F
surface components on deck, 186 Resistivity, deep-sea-floor microelectrode Red Sea, 584
components, 183 results, 178, 179F satellite remote sensing, 584, 585, 586,
control, 184 Response time, absorptiometric chemical 584–589
computer control, 184, 185–186 sensors, 242–243 history of, 586
hardwire control, 184, 185–186 Reverberation, 409–410, 420–421 requirements, 586–587
deep submergence science studies, 125, sea ice, 402 resolution and error sources, 587
127F, 129F, 130F Reversible reactions, in chemical sensors, science requirements, 587
depth sensors, 184 242–243 scientific issues, 586–587
description, 183 Riser drilling, 37, 38F theory, 584–585
development, 183 RMT (rectangular mouth opening trawl), Tyrrhenian Sea, 29F
flyaway mode, 201 72T, 79, 80F see also Freshwater flux
frame, 183 Robotic arms, human-operated vehicles Salinity, temperature, depth (STD)
future improvement (HOV), 197–199 profiler, 30
efficiency, 186 Rockall Trough, nephelometric profile, Salinometers, 29
intervention technology, 187–188 284F calibration, 30F
Kaiko (Japanese ROV), 151 Ropes, wire, in moorings, 97–98 laboratory-based, 29
manipulators, 184 ROPOS ROV, 187T, 202T microwave see Microwave
operated by Scripps Institute of ROS see Reactive oxygen species salinometers
Oceanography, 127, 132F Rotary Core Barrel (RCB), 38, 39F ocean-going, 30
operated by UNOLS NDSF, 125–126, ROV see Remotely operated vehicles Salt exposure, effect on humidity
130F (ROVs) measurement, 214
 INDEX 643

Sand Voluntary Observing Ships (VOS), 216 sensor design, 573–575

acoustics in marine sediments, 387F, see also Satellite remote sensing oceanic observation techniques, 521F
388T, 395F Satellite monitoring, oceanographic partnership with oceanography, 513
geoacoustic properties, 421T research vessels, 6 passive sensors, 522, 522F
SAS (SeaWiFS Aircraft Simulator), Satellite oceanography, 509–523 salinity, 584–589
474–475 challenges ahead, 518 see also Salinity
Satellite(s) data policy, 518 suspended sediments, 499
aircraft for remote sensing advantages in situ observations, 518 turbulent heat fluxes, 594–595
over, 471 integrated observations, 518–519 wind stress, 590–593
infrared data, 532 international integration, 521 see also Satellite radiometers; Satellite
laser altimetry, 528 oceanographic institutional issue, scatterometers
ocean color sensing, 574–575 519–520 Satellite remote sensing of sea surface
passive-microwave instruments, transition from research to operations, temperatures, 548–559, 594, 598F
525–526 519 accuracy, 551
see also Satellite passive-microwave first generation, 511–512 advantages, 548
measurements of sea ice; Satellite missions, 516T applications, 554–556
remote sensing SAR origins, 509–511 air-sea exchanges, 558–559
polar-orbiting, sun-synchronous, 216 second generation, 512–513 coral bleaching, 556–558
sea surface temperatures (SST), 586 third generation, 513 coral reefs, threatened, time-series SST
see also Satellite remote sensing of sea implementation, 515 predicting, 556
surface temperatures space policy, 515–517 frontal positions, 556, 557F
Satellite altimetry, 502–508 studies, 514 ‘global thermometer’, 558
accuracy, 502, 508 understanding and consensus, 515 hurricane intensification, 556
dynamic sea surface topography, joint missions, 517 instabilities in Pacific Ocean, 556,
503–505, 508 mission launch dates, 515 558F
dedicated gravity field missions, 505 new starts, 515 oceanographic features resolution,
dynamic topography, 503–505 partnerships in implementation, 554
global geostrophic circulation, 513–514 storm prediction, 556
503–505, 505F promotion and advocacy, 514 black-body calibration, 548, 551
global sea level rise, 505 research strategy for the decade, 514 characteristics, 551–552
long global sea level time-series, 505 Satellite passive-microwave measurements coverage, 551
extraordinary gravity field information, of sea ice, 524, 528, 529F, 531F, global, 551
505 532, 532F, 524–534 limited to sea surface, 550–551
geoid and, 600–601, 601 background, 524 future developments, 558
data correction, 603 rationale, 524 Advanced ATSR, 559
gravitational sea surface topography, theory, 524–525 atmospheric correction algorithms
502–503, 508 see also Satellite oceanography improvement, 558
geoid or equipotential surface, Satellite radar altimeters, 502 infrared radiometers, 558
502–503, 503F sea level spatial variation, 502 SST retrieval algorithm validation,
global maps of, 503, 504F see also Satellite oceanography 558
gravity anomalies, 502–503, 504F Satellite radiometers, 593 infrared atmospheric corrected
maps of bottom topography, 503, Satellite remote sensing, 522 algorithms, 548–550, 558
504F active sensors, 522, 522F atmospheric aerosols, 550
measurement method, 502 air temperature, 594–595, 595 AVHRR see Advanced Very High
observation accuracy, 502 applications, 596–598 Resolution Radiometer (AVHRR)
Topex/Poseidon satellite altimeter coastal waters, 492–501 constant temperature deficit, 548–549
system, 502, 503F applications, 497–500 contamination by clouds, 550
operational ocean applications, 508 non-optical, 499–500 longer wavelength window, 548–549,
sea level data, 586–587 data sets, 582–583 549F
sea level variability see Sea level changes/ data validation, data set Pathfinder SST (PSST), 549–550
variations intercomparison, 578 radiative transfer linearized, 549
sea surface topography profiles, 502, electromagnetic signals showing water vapor correction algorithm,
503F atmospheric transmittance, 520F 549, 549F
tomography and, 461 fishing and, 499 measurement principle, 548–550
wave height and wind speed, 507–508, fluxes, 596F, 590–599 brightness temperature, 548
508 history, 509 infrared atmospheric windows, 548,
Satellite measurements, meteorological, humidity, 594–595 549F
216–217 ocean color, 571–583 microwave measurements, 550–551
polar-orbiting, sun-synchronous applications, 582–583 radiation attenuation, 548
satellites, 216 atmospheric correction, 576–577 temperature deficit, 548
precipitation estimates, 216 bio-optical algorithms, 577, 578F thermal emission from sea, 548
solar radiation by radiative transfer methodology, 573–575 see also Infrared (IR) radiometers;
models, 216 postlaunch sensor calibration stability, Radiative transfer (oceanic)
surface-measured SST for calibration, 575–576 microwave measurements, 550–551
216–217 product validation, 577–582 infrared radiometers vs, 550, 550T
surface wind speed, 216 sensor calibration, 575–576 penetration depth, 550–551
644 INDEX

Satellite remote sensing of sea surface Scattering (optical), 240 roughness, acoustics in marine
temperatures (continued) suspended particles, 282, 285F sediments, 385
temperature characteristic of ‘skin of the Scattering (spectral) sediments see Seafloor sediments
ocean’, 550–551 coefficient see Scattering coefficient topography see Seafloor topography
time-series of global SST, 559 Mie theory, 324–325 see also Seabed
see also Sea surface temperature Rayleigh, 324–325 Seafloor mapping, active sonar images,
(SST) sensors see Nephelometry 447, 447F
Satellite remote sensing SAR, 560–570 see also Volume scattering function Seafloor sediments
data collecting, 569 (VSF) acoustics see Acoustic remote sensing;
definition, 560 Scattering coefficient, 272–273, 323–324 Acoustics, marine sediments
detecting change on decadal timescales, definition, 315–316, 316 bulk density, 383, 383T, 385
569–570 Hydrolight simulation, 319–320, 320F bulk modulus, 383, 383T, 385
examples of ocean features from, oceanic waters, 253, 253F clay see Clay
569–570 as bio-optical model quantity, 248T, core samples see Sediment core samples
bottom topography in shallow water, 249 mud see Mud
562F, 564 time series, 278F sand see Sand
eddies, 564F, 565 see also Ocean optics water-sediment interface, seismo-
fronts, 565–566, 565F, 566F Scatterometers, 216, 514 acoustic waves in, 383–384, 384F
important marine applications, Advanced Scatterometer (ASCAT), 591, see also Acoustics, marine sediments
563–564 592T Seafloor topography
internal waves of tidal frequency, NSCAT (NASA scatterometer), 515 acoustic noise, 361
567–568, 568F satellite see Satellite scatterometers gravitational, satellite altimetry, maps,
long surface waves or swell, 564F, Scatterometry, 514, 590–591 503, 504F
566–567, 567F microwave, 591 internal tides see Internal tides
polar lows, 563F, 564–565 rain and, 591 internal waves see Internal wave(s)
sea ice edges, 566, 566F tropical cyclones, 597F see also Bathymetry; Bottom topography
surface ship wakes, 568–569, 569F Schleicher, K E, 29, 30 Seaglider, 141, 142F, 143, 143–144, 205T
tracking oil spills, 561F, 564 Schlumberger, logging tools, 39–40 missions, 145–147, 146F
history Scholte wave, 383T, 384–385, 384F Sea ice
major spaceborne SAR missions, 560, Science acoustic noise, 403–404
561T deep submergence see Deep acoustic reflection, 400
shuttle imaging radar (SIR) submergence science acoustic reverberation, 402
experiments, 561 manned submersible (shallow water) Arctic, extent, 398F
imaging mechanism of ocean features, use, 159 bottom water formation, 524
563–569 Science Working Groups (SWGs), 513 brightness temperatures, 524, 525F, 526,
Bragg resonant wave back scattering, Scientific research vehicles, 186 527, 532, 532F, 533
561–562, 561T advantages over manned submersibles, concentration maps, 528, 528F
frontal features change wind stress, 186–187 concentrations, 524, 525, 526, 526–528,
563 current ROVs, 187, 187T 527, 532, 533
ocean currents and wind-driven manned submersibles use with, 186 cover
surface waves, 562–563 Scripps Institute of Oceanography annual cycles, 524, 525
oil slicks dampening effect, 561F, 563 autonomous underwater vehicles, 205T extents, 528, 533
mapping features in coastal regions, 560 remotely-operated vehicles (ROV), 202T extents and trends, 528–530, 533
monitoring changes in the marine Scripps Institution’s re-entry vehicle, 127, future monitoring, 532–533
environment, 569 132F global, 525
weather forecasting for coastal regions, Seabed interannual differences, 525, 528
569–570 flow regime, benthic flux measurement monitoring, 524, 525F, 533
see also Synthetic aperture radar and, 178, 180–181 seasonal cycle, 528, 529–530
(SAR) optical properties, 494, 495T trends, 528, 533
Satellite scatterometers, 592T satellite remote sensing, 499, 500F data, global, 524
deployment, 591 topography, acoustic profiling, 344 determinations from satellite passive-
Satellite tracking, drifters, 91 see also Seafloor microwave data, 526–528, 528,
SBL (short-baselinetracking systems), 167 SeaBED, 205, 205T, 206F 529F, 531F, 532F
Scalar irradiance, Hydrolight simulation, Seabird CTD instrument, towed vehicles, edges, satellite remote sensing
321F, 322 63F, 68F application, 566, 566F
Scampi, 197, 198T Sea-Bird Electronics, Inc., 31 geographical extent, 528, 533
Scanfish towed vehicle, 60, 64 Sea bottom trends, 528–530, 533
Scanning low frequency microwave acoustic reflection, 402 microwave emissions, 524
radiometer (SLFMR), 586 see also Seabed; Seafloor operations under, autonomous
Scanning multichannel microwave Sea Cliff (US Navy submersible), underwater vehicles (AUV), 162,
radiometer (SMMR), 524, 148–150, 153 170–171
525–526, 527, 528, 529F Seafloor polar climate state, 524
Scanning sonars, remotely operated acoustic reflection/scattering, 402, 418, satellite passive-microwave
vehicles (ROVs), 184 419, 420–421, 420F measurements, 524–534
Scattering (acoustic), 412F drifters, 90 see also Satellite passive-microwave
acoustic noise, 361, 363 imaging by AUVs, 205 measurements of sea ice
 INDEX 645

solar radiation reflection, 524, 526 Sea surface skin temperature (SSST), SeaWiFS Aircraft Simulator (SAS),
surveillance by SAR, 563–564 measurement, 481–483 474–475
thickness, 528, 529 see also Infrared (IR) radiometers SeaWiFS algorithm, 322
trends, 528–530, 533 Sea surface temperature (SST) SeaWinds scatterometer, 592T
types, 524, 526, 530–531 heat stored in the ocean indicated by, see also ADEOS II; QuikSCAT
first year ice, 530, 532F 556 Sediment(s)
frazil ice, 530 measurements, 213 layering, Mediterranean, 418F
grease ice, 530 infrared radiometers see Infrared (IR) layers see Sediment layers
multiyear ice, 530, 532F radiometers seafloor see Seafloor sediments
nilas, 530 satellite remote sensing see Satellite suspended see Suspended sediments
pancake ice, 530 remote sensing of sea surface transport see Sediment transport
slush ice, 530 temperatures Sedimentary acoustic medium, 380, 383
variables affecting, 531–532 sound speed and, 433F Sediment cores
season length, 531–532 spectral range for remote sensing, 495T benthic flux measurement and, 175
summer melt, 531–532, 532 surface-measured for calibration of Sediment core samples, geoacoustic
temperature, 524, 525–526, 531–532 satellite measurements, 216–217 parameters, measurement of, 380,
velocity, 531–532, 532 Sea surface topography 385–386, 388T
see also Bottom Water Formation gravitational see Satellite altimetry SACLANTCEN samples, 386, 387F
Sea Ice Mechanics Initiative (SIMI), Marianas Trench, 502–503, 504F sample degradation, 386
397–398 satellite altimetry, 502, 503F X-ray tomography, 380, 381F
Sea level SeaTech transmissometer, 282, 283F, 285 Sediment layers
anomaly, tomography vs. satellite Sea water analysis of deep-sea samples at surface,
altimetry, 466F adiabatic lapse rate, 617 limitations, 174
changes see Sea level changes/variations complex dielectric constant, 584–585, concentration gradients, recovery time
rise see Sea level rise 586, 588F after disturbance, 178
variations see Sea level changes/ conductivity, measurement in situ, 27 disturbance by landers, 177–178,
variations constituents 180–181, 182
Sea level changes/variations dielectric properties, 584 oxygen concentration profile, 179
proportional to heat anomaly, 506–507, electrical conductivity, 584 see also Benthic flux
507F emissivity coefficient, 584 Sediment transport
see also El Niño Southern Oscillation equation of state see ‘Equation of state’ acoustic measurement, 343–356
(ENSO) (of sea water) bed topography, 345–347
rises see Sea level rise fluorescent species, direct measurement, case study, 349–353
satellite altimetry, 505–507 269–270, 269T flow, 347
1997-98 El Niño event, 506–507, freezing point, 617–618 instrumentation, 344, 344F
506F optical constituents, 317–318 limitations, 355
global sea level variability, 505–506, dissolved organic compounds, 318 overview, 343–344
506F inorganic particles, 318 suspended sediments, 347–349
global tide models, 505 organic particles, 318 processes, 343, 343F
sea level proportional to heat bacteria, 318 Seismic observatories, 42
anomaly, 506–507, 507F detritus, 318 Seismic reflections, sound speed gradients
sea level signal record, 506–507, phytoplankton, 318 and, 433–434
506F water, 318 Seismic reflection water-column profiling,
see also Climate change; Sea level physical properties, 615T, 614–618 432–441
rise potential temperature, 617 compared with other acoustic
Sea level rise pure, spectral absorption, 273 technology, 432–434
dynamic sea surface topography, 505 salinity range, 616F conductivity-temperature-depth profiler
satellite altimetry, 507 see also Salinity measurements and, 435
dense coverage of satellite altimeter sound speed, 615T, 617 internal waves and, 439–440
data, 507 specific heat, 615T, 616–617 method, 432
global hydrologic system, 507 specific volume anomaly, 616 observations, 434–440, 434F
see also Climate change; Sea level standard, salinity, 29 Kuroshio Current, 435
changes/variations surface emissions, 584 ocean temperature and, 434F, 435
Sea Link, 200F temperature resolution, 433–434
Seals see Pinnipeds (seals) conservativity, 25–26 Seismo-acoustic waves, 383
Seasat-A satellite, 590–591, 592T range, 616F arrival structure, 385, 385F
Seasat satellite, 511F, 512, 525–526, 603 see also Sea surface temperature body waves see Body waves
global sea surface topography, 517F (SST) compressional waves see Compressional
SeaSoar towed vehicle, 60, 64, 69F thermal expansion coefficient, 615T, 616 waves
Sea surface see also Water ducted waves, 384, 384F
acoustic reflection, 418 SeaWiFS, 575T head wave, 385, 385F, 393
heat fluxes see Heat flux data, chlorophyll a, 579F, 580F interface waves see Interface waves
heat transfer processes, 590 data validation, 577–578 Love waves, 383T, 384, 384F
temperature see Sea surface temperature global area coverage, 574–575, 575F arrival times, 385, 385F
(SST) polarization errors, 578 mixed wave types, 385
see also Ocean surface sensor degradation, 573, 576F shear waves see Shear waves
646 INDEX

Seismo-acoustic waves (continued) Sharks Signal-to-noise ratio, sonar, 414–415

water-sediment interface, 383–384, liver size, acoustic scattering and, 372 Silicate (Si[OH]4)
384F Shear modulus, seafloor sediments, 383, benthic flux, 175–176, 177F
wave generation, 383–384, 384 383T, 389–391 Silt, geoacoustic properties, 421T
wave properties, 381–383, 383, 383T, Shear (airfoil) probes, 232–233 Silver nitrate, 28–29
385 Shear stress, 219 SIMBIOS (Sensor Intercomparison and
Seismology sensors, 189–196 measurements see Momentum flux Merger for Biological and
considerations, 191–193, 195F measurements Interdisciplinary Oceanic Studies),
hydrophones, 195–196 Shear waves 577–578
seafloor, 195 acoustics in marine sediments, 383, 384, SIMI (Sea Ice Mechanics Initiative),
Self-contained imaging profiler (SCIMP), 384–385, 384F, 388, 388F, 397–398
29F 392–393 Simple opening/closing nets, zooplankton
Self Propelled Underwater Research horizontally polarized (SH), 384 sampling, 70–71, 73F
Vehicle (SPURV), 162 velocity, 383T, 385, 387, 389F, 393, Simultaneous localization and mapping
Semi-taut moorings, 101–102, 102F, 394, 395F (SLAM), 167–168
103T velocity-depth profile, 394 Single point current meters, 115–122
Sensible heat flux vertically polarized (SV), 384 acoustic Doppler see Acoustic Doppler
satellite remote sensing, 594–595 see also Acoustic remote sensing current meters
Sensor Intercomparison and Merger for Shelikof Strait, SAR image, 564F, 565, acoustic travel time see Acoustic travel
Biological and Interdisciplinary 566–567, 567F time (ATT) current meters
Oceanic Studies (SIMBIOS), Shinkai (Japanese submersible), 148–150, calibration, 120–121
577–578 151, 153–154, 154–155, 155F characteristics, 117T
Sensors Ship(s), 3–12 design, 115–116, 116F
acoustic survey work, 68–69 acoustic noise, 358F, 359–360 directional measurement, 120
see also Ship(s) data measurement, 360 bar magnet, 120–121
aircraft for remote sensing see Aircraft directional spectra, 360–361 fluxgate compass, 120–121
for remote sensing deep water, 361, 362F electromagnetic see Electromagnetic
backscattering, 274, 329 pedestal of high noise, 361, 363F current meters
chemical see Chemical sensors shallow water, 361 evaluation, 120–121
depth, ROVs, 184 frequency spectra, 360–361, 361F evolutionary trends, 116F, 117T, 121F,
electromagnetic velocity, 122 summation, 360–361, 361 122
expendable see Expendable sensors very low frequency band (VLF), 360, historical aspects, 115
hot film, 233 362–364 intercomparison, 120–121
hyperspectral, 497–498 as drifter, 90 measurement problems, 115
meteorological, moorings, 100–101 general purpose vessels see General near surface, 115
pressure, strain gauges, 31F purpose vessels surface wave zone, 115–116
remotely operated vehicles (ROVs), 184 ice-breaking, 170–171 mechanical, 116F, 118
satellite see Satellite remote sensing observational disadvantages, 140 platforms, 115
seismology see Seismology sensors oceanographic research vessels see see also Moorings
towed vehicles, 67–69 Oceanographic research vessels remote sensing, 116F, 120
turbulence see Turbulence sensors support, for manned submersibles (deep vector averaging current meter,
Sensors for mean meteorology, 211–217 water), 152–153, 153, 153–154 116–118, 116F, 117T, 119
future developments, 217 wakes, satellite remote sensing vector measuring current meter, 119
Global Ocean Observing System application, 568–569, 569F see also Turbulence sensors
(GOOS), 217 see also Shipping Single-scattering albedo, 316–317
humidity, 213–214 Shipbuilding Single turnover fluorescence induction,
pressure, 211 boom, oceanographic research vessels, 5 259
satellite measurements see Satellite Shipping SIO see Scripps Institute of Oceanography
measurements, meteorological future developments, 558 SIO/IGPP-BB ocean bottom seismometer,
temperature, 213 see also Ship(s) 191T, 192F
wind speed and direction, 211–213 Ship wakes, satellite remote sensing SIO/IGPP-SP ocean bottom seismometer,
see also Humidity; Pressure application, 568–569, 569F 190T
measurements; Temperature; Wind Short-baseline tracking systems, 167 SIO/ONR ocean bottom seismometer,
speed Short period (SP) ocean bottom 191T, 194F
Sentry, 205T seismometers, 189–191 SIPPER (shadowed image particle
Shadowed image particle profiling and characteristics, 190T, 191 profiling and evaluation recorder),
evaluation recorder (SIPPER), GEOMAR-SP, 190T, 193F 81–83
81–83 JAMSTEC-ORI, 190T, 191 Sir Alister Hardy Foundation for Ocean
Shadow regions, 407–408 SIO-IGPP-SP, 190T Science (SAHFOS), 15
Shallow water UTIG, 189F, 190T open access data policy, 16
acoustic noise, 361 WHOI-SP, 190T, 192F training provided by, 16
acoustic remote sensing, characteristic Shot peening, 103 see also Continuous Plankton
signal, 389, 389F SHRIMP, 198F, 198T Recorder (CPR) survey
acoustics see Acoustics, shallow water Shuttle imaging radar (SIR), history, 561 Skate (USS), 397
bottom topography, satellite remote Side-entry sub technology, 40–41, 40F Skin of the ocean, satellite remote sensing
sensing application, 562F, 564 Signal excess, sonar, 414–415 of SST, 550–551
 INDEX 647

Skylab, 512 multibeam bathymetric mapping, 447, Southern hemisphere

SLAM (simultaneous localization and 448F sparse salinity data, 584
mapping), 167–168 seafloor mapping, 447, 447F Southern Oscillation (SO), 533
Slocum glider, 141, 142F, 143 unmanned underwater vehicle (UUV), South Pacific
Slush ice, 530 447 gravity anomaly, 605F
Smithsonian Tables, 214 advanced beamforming, 480 Soviet Union
SMMR see Scanning multichannel plane wave model, 450 Arctic research, 397–398
microwave radiometer wavefront curvature, 450 see also Russia
SMOS (soil moisture and ocean salinity), components, 445–446 Spacecraft instruments, 551–552
586 beamforming, 446–447 Along-Track Scanning Radiometer see
Snell’s law, 407, 409, 410F, 413–414, electronic array processing, 447 Along-Track Scanning Radiometer
453 spatial resolution, 446 (ATSR)
SO see Southern Oscillation (SO) reverberation-limited environment, black-body calibration, 551
SOFAR float, 94 443F, 445–446 cooled detectors for infrared
Software, autonomous underwater waveform design, 442 radiometers, 551
vehicles (AUV), 166 Doppler gating, 446 geosynchronous orbit measurements of
Soil moisture and ocean salinity (SMOS), optimum design, 446 SST, 554
586 range gating, 446 GOES Imager, 554
Solar irradiance spectrum, 573F definition, 442 Moderate Resolution Imaging
Solar radiation, 272 history, 442–443 Spectroradiometer see Moderate
clouds and, 593 passive sonar, 447–448 Resolution Imaging
infrared range, 272 applications, 447 Spectroradiometer (MODIS)
satellite measurements by radiative equation, 447–448 TRMM Microwave Imager see TRMM
transfer models, 216 history, 442 Microwave Imager (TMI)
ultraviolet range, 272 passive sonar beamforming, 448–449 Space Shuttle Challenger disaster,
visible range (light), 272 bearing-time recording, 449–450 160
see also Radiative fluxes bearing-time processing, 449 Special Sensor Microwave/Imager
Sonar, 414, 442 directional components, 449 (SSM/I), 216, 524, 525–526, 528,
autonomous underwater vehicles (AUV), display formats, 449 529F, 532F, 533, 594, 595
168–169 FRAZ displays, 450 Special sensor microwave temperature
echo integration, 367 frequency domain beamforming, sounder (SSM/T), 595
figure of merit, 414–415, 415 449 Specific gravity, 614–616
fish, 367 LOFAR (Low Frequency Acoustic Specific volume anomaly, seawater, 616
marine animals, development, 367 Recording and Analysis), 449–450 Spectra
signal excess, 414–415 time-series display formats, 449, 449F absorption see Absorption spectra
see also Acoustic navigation trackers, 450 infrared, 548, 549F
Sonar equation, 412–414 shallow-water, 417 reflectance, water types, 493F
active, 414–415 Songs, marine mammals, 428–429 Spectral attenuation coefficient, as bio-
passive, 414–415 Sonic anemometers, 212F, 213, 222, 222F optical model quantity, 248T
Sonar (Sound Navigation and Ranging) Sonic thermometers, 224 Spectral beam attenuation coefficient,
systems, 442–450 Sontek Argonaut-ADV acoustic travel 272–273, 273–274, 323–324
active sonar, 443–446 time, 117T definition, 315–316
absorption limits range, 445 Soucoupe, Cousteau, Jacques, 156 measurement see Transmissometry
absorption loss, 444–445, 445F Sound(s) see also Ocean optics
active sonar equation, 444 physics, 425 Spectral diffuse attenuation coefficient,
ambient noise, 443 Sound propagation research, sonar 317
beamformers, 443, 444, 445F systems, 442 Spectral interval, 548–549
echoes, 443 Sound speed, 406, 452–453, 467 Spectral radiance see Radiance
history, 442 Arctic, 397–399 Spectral scattering coefficient see
maximum operating range, 444–445 depth profile, 406F Scattering coefficient
noise-limited, 444 Atlantic, 407F Spectroradiometers, 275–276, 489F,
operating frequency, 444 North Atlantic, 453F 490–491
pulse repetition frequency (PRF), 443 sea surface temperature and, 433F Specular reflection, 240
Rayleigh region, 445 seawater, 615T, 617 Speed
receivers, 444, 445F shallow water, 418F autonomous underwater vehicles (AUV),
resolution, 444–445 temperature and, 417 164
reverberation, 443 variability, 417 gliders, 143
scattering losses, 445 temperature and, perturbation, 467, of sound see Sound speed
scattering strength, 444 467F Sperm whales (Physeteriidae and
side-scan, 444 temperature conversion, 466–467 Kogiidae)
synthetic aperture, 444 see also Deep sound channel sound production, 429–430, 429F
target cross-section, 445 South China Sea see also Odontocetes (toothed whales)
transmitters, 443, 443F Radarsat ScanSAR image, 567–568, Spherical navigation, 167
typical operating frequencies, 445, 568F Spherical spreading (sound), 407, 419,
446T Southern blue whiting, acoustic 419F
active sonar image examples, 447 scattering, 371 Spillhaus, Athelstan, 26
648 INDEX

Spray (glider), 141, 142F, 143, 143–144, see also Reflectance TAP (Transarctic Acoustic Propagation
163F, 205T, 206F Surface salinity, 584, 587 Experiment), 397–398, 465
Spray-removal devices, humidity global mean, 584, 585F Target strength (TS), measurement,
measurement, 215–216 interannual variations, 587 zooplankton sampling, 84
SPURV (Self Propelled Underwater Surface waves Telemetry, marine mammal bioacoustic
Research Vehicle), 162 long, satellite remote sensing research, 430F, 431
Squid (Teuthida) application, 564F, 566–567, 567F Temperature
acoustic scattering, 373–374 wind-driven, SAR imaging mechanism, brightness see Brightness temperature
SSM/I see Special Sensor Microwave/ 562–563 deficit
Imager (SSM/I) see also Wave(s) brightness temperatures, 549
SST see Sea surface temperature (SST) Surf clam, acoustic scattering, 374 definition, 548
STABLE II instrument platform, 351, Surveying, AUVs, 204–205 water vapor as contributor, 548–549
351F Suspended particles measurements, 213
Standard seawater, salinity, 29 areas of high concentration, 289–290 digital, 32
State estimation, tomography and, scattering (optical), 282, 285F positioning of sensors, 213
458–460 see also Particle(s) radiation shields for sensors, 212F,
STD (salinity, temperature, depth) Suspended sediments 213
profiler, 30 acoustic backscatter, 343–344, 347–349 sea surface temperature (SST), 213
Steel, in mooring lines, 97–98 concentration, 352–353 skin temperature, 213
Steric height, 586–587, 587 profiles, 349F, 350F thermometer types, 213, 214T
Stern-Volmer equation, fluorescence entrainment mechanisms, 349–351, perturbation rifle, 467
quenching, 268, 268F 350F potential see Potential temperature
Stommel, Henry, 141 modeling, 352–353, 354F profile
Stoneley wave, 384–385 waves, 353F Kuroshio extension front, 437F
Strain gauges, pressure sensors, 31F particle size, profiles, 349F, 350F range of seawater, 616F
Subduction zone processes, deep remote sensing applications, 499 scale, definition, 27
submergence science studies, spectral range for remote sensing, 495T sea surface see Sea surface temperature
132–133 Swaged mooring terminations, 98 (SST)
SUBICEX (Submarine Ice Exercises), Swallow, John, 94, 140 skin
398–399 Swallow floats, 94 measurement, 213
Submarine(s) SWATH (Small Waterplane-Area Twin see also Radiative transfer (oceanic)
acoustics research and, 398–399 Hull) ships, oceanographic from sound speed, 466–467
sonar system history, 442 research vessels, 12 sound speed and, perturbation, 467,
see also Human-operated vehicles Swell 467F
(HOV); Submersibles acoustic noise, 359 water-column, Massachusetts Bay, 170F
Submarine Ice Exercises (SUBICEX), satellite remote sensing application, see also Heat; Sea surface temperature
398–399 564F, 566–567, 567F (SST)
Submarine measurements, 528, 529 Swimbladder, 370F Temperature cutoff scale (LT), turbulent
Submarine Science Ice Expeditions acoustic scattering, 370 dissipation, 230–231, 231, 232F
(SCICEX), 398–399 boundary element model, 371F Terrestrial mammals
Submersibles Synechococcus cyanobacteria, 258F hearing, 427–428, 428
gravimetry, 602 Synopticity, 272 sound production, 429
Nekton, 157 Syntactic foam, 98 Terrestrial runoff, monitoring, using
see also Human-operated vehicles; on benthic flux landers, 175 scattering sensors, 331T
Submarine(s); Vehicles Synthetic aperture radar (SAR), 477–478, Tether management system (TMS), 184,
Subsurface floats see Float(s) 499–500, 514, 560 185F
Summer melt all-weather, day/night imaging, 560 computer control, 184, 185–186
onset, 524, 532 description, 560 control systems, 186
process, 530 ocean features detected, 560 deck winch, 184–185
see also Marine mammals; Pinnipeds ocean surface roughness measured, hardwire control, 184, 185–186
(seals) 477–478 launch and recovery system (LARS) and
Sunlight, near sea surface temperatures satellite remote sensing see Satellite winch, 185–186
and, 550–551 remote sensing SAR surface control stations, 185, 186
Sun-synchronous orbits, ocean color side-looking imaging radar, 560 tether cable, 184–185
remote sensing and, 574–575 see also Aircraft for remote sensing; see also Autonomous underwater
Supercooled ice, formation, Satellite altimetry vehicles (AUVs)
transmissometers in study of, 331T synthesized long antenna, 477 umbilical cable, 184–185, 185
Support ships, manned submersibles (deep see also Bio-optical models; Inherent vehicle tether, 184
water), 152–153, 153, 153–154 optical properties (IOPs); Irradiance Thermal expansion coefficient, 615T
Support vessels, 10–11 Systeme Acoustique Remorqué (SAR), seawater, 616
development, 10–11 198T Thermal gradient engine, gliders,
Surface circulation, 584 163–164
Surface nepheloid layer (SNL), 282 T Thermal imagers, 489, 490F
Surface radiation budget (GEWEX-SRB) focal plane array detector technology,
project, 593 Taiwan Tan, shallow water bathymetry, 490
Surface reflectance, 572 SAR image, 562F, 564 problems with, 490
 INDEX 649

Thermistors, 27, 223, 233–234 reference states, 454 rigid fairing tow-off, 62
microbead, 223 time-dependent, 457–458 wrap-round rigid fairing, 62, 62F
thin film sensors vs., 233–234 uncertainty, 456 tangential, 61, 61–62
Thermocline vertical slice, 454–455 vortex-induced oscillation, 61
sound speed and, 406 range-dependent, 456–457 TowCam, 197, 198F, 198T
Thermocouples, 223, 223F range-independent, 456–457 Towed deep-sea craft, 197, 199T
Thermohaline convection, 584, 587 noise, 451 Towed vehicles, 60–69
Thermometers observables, 452 active depth control (with), 60
calibration, 26–27 principle, 451 advantages, 60
CTD profilers, 31 resolution (spatial), 458F methods, 64, 67
response time, 31–32 sampling density and, 457 types, 60
electronic, 26–27 resolution matrix, 455 advantages, 60
reversing, 25–26, 25F results, 458–460 aerodynamics, 63–65
sonic, 224 sampling geometry, 457–458, 459F airfoil stall, 63, 63F, 64
subaquatic, 25–28 satellite altimetry and, 461, 462F angle of attack, 63F, 64, 64F
thin film, 233–234 temperature conversion, 466–467 aspect ratio, 64
types, 213, 214T transceiver arrays cambered airfoil, 63, 63F, 64F
see also Thermistors ATOC, 466F delta wing, 64
THETIS-2 experiment, 463–465, 465F Greenland Sea, 460F depth control methods, 64, 67
Thin film thermometers, 233–234 Puerto Rico, 464F forces and moment, 63, 63F, 67T
Thorium (Th) THETIS-2 experiment, 465F induced drag coefficient, 64
nepheloid layer flux, 287–289 travel-time perturbation depth profile, passive towed, 65, 65F
Thorp’s formula, 419 457F roll stability, 64–65, 65F
Three-dimensional (3D) fluorescence, 266, TOMS (Total Ozone Mapping vehicles generating lift, 63–65
267F Spectrometer), 577 vehicles not generating lift, 65
Thresher, USS, 156 TOPEX/Poseidon satellite, 510–511 winged vehicles, 63, 63F
Thrust anemometer, momentum flux Topex/Poseidon satellite altimeter system, compared with autonomous underwater
measurements, 221 502, 503F vehicles (AUV), 168
Thunnus albacares (yellowfin tuna) global sea level variability detection, components, 60
acoustic scattering, 371 505–506, 506F controllable wings, 60
Thunnus obesus (big-eye tuna) Torpedo AUVs, 162 drag, 66, 67T
acoustic scattering, 371 Tortuosity, 179 flight control, 67
Tiburon ROV, 129F, 187T, 202T, 203F Total inorganic carbon (TIC) tow speeds, 67
TIC see Total inorganic carbon (TIC) benthic flux, 175–176 undulating mode, 67
Time-resolved fluorometry, 266–267, Total oxidised nitrogen (TON) passive, 65, 65F, 68–69, 68F
268F determinations, air-segmented computer-controlled winch, 60
Titration, salinity measurement, 28–29 continuous flow analyzers, 335, performance of vehicle/cable system,
Titration alkalinity (TA), benthic flux, 335F 65–66
175–176, 177F flow injection analysis manifold computer-controlled winches, 60
TL see Transmission loss arrangement, 337F depth range, 65F, 66, 66F
TMI see TRMM Microwave Imager Total Ozone Mapping Spectrometer equilibrium angle, 66, 66F
(TMI) (TOMS), 577 equilibrium depth, 66, 66F, 67T
TMS see Tether management system Total Ozone Vertical Sounder (TOVS), normal drag coefficient, 66, 66F
(TMS) 577 vehicle drag, 66, 67T
TOBI, 198T Total precipitable water, 594–595 sensors, 67–69
Todores pacifica, acoustic scattering, TOVS (Total Ozone Vertical Sounder), acoustic survey work, 68–69
373–374 577 CTD recorders, 63F, 67–68, 68F
TOGA (Tropical Ocean - Global Tow cable, types, 61 plankton collectors, 67
Atmosphere) Study, 91 Tow cable drag, 60–61 Seabird CTD instrument, 63F, 68F
Tomography, 451–468 depressing forces, 61 types, 60
advantages, 451 drag coefficient wave-induced ship motion effects, 66–67
barotropic and baroclinic tides, 461–463 normal, 61, 61F, 66, 66F minimizing ship motion, 67
convection, 460–461 tangential, 61–62, 61F towing in rough seas, 66
data, perturbation field models, 454 induced drag coefficient, 64 without active depth control, 60
data assimilation, 458–460 normal, 61 Towing, oceanographic research vessels, 6
development, 451 reduction, 62 TPXO.2 tidal model, 461
example experiment, 452F faired cable winch, 69, 69F Trace gases
forward modeling, 451 flexible ribbon and hair fairing, 62–63 photochemical production/reactions,
deep sound channel, 451 ‘hair’ fairing, 62–63, 62F 304–305
ray travel time, 452–454 of normal drag, 62 Trace metals
heat content, 463–466 ribbon and hair replacement, 63 photochemical production/reactions,
inverse modeling, 454 ribbon fairing, 62–63, 62F, 65F 305–308
data, 454 rigid airfoil-shaped fairing, 62 Trackers, passive sonar beamforming, 450
horizontal slice, 457 rigid clip-on fairing, 62, 62F Tracking
ray properties, 455–456 rigid compressive load, 62 drifters, 90–91
ray travel times, 454 rigid fairing expense, 62 RAFOS floats, 95–96
650 INDEX

Tracking (continued) diffuse seafloor flows, acoustic Underwater Sound Laboratory, 397
errors, 96 backscatter and, 376–377 Underwater video profiler (UVP), 83
Transarctic Acoustic Propagation (TAP) dissipation, 229, 230F Undulating oceanographic recorder
experiment, 397–398, 465 rate, 229–230, 230F (UOR), 74–76
Transduction temperature cutoff scale, 230–231, United Nations Convention for the Law
chemical, chemical sensors, 242, 243 231, 232F of the Sea (UNCLOS), 9, 11
see also Molecular recognition velocity spectra, 229, 230F United States Defense Meteorological
element viscous cutoff scale, 229–230, 231, Satellite Program (DMSP), 524,
Transfer efficiency (of Photosystem II), 232F 525–526, 528F, 529F, 532F, 533
258 dissipation measurement, 229–230, 231, United States of America (USA)
Transmission loss 231F, 234 National Aeronautics and Space
acoustics, 360–361, 361F free-fall profilers, 232 Administration (NASA) see NASA
geoacoustic parameters, 394–395, horizontal, 232–233 Navy, acoustics research, 397
395F sampling frequency, 231, 232F University National Oceanographic
Transmissometry, 273–274, 323–332 sensors see Turbulence sensors Laboratories System (UNOLS),
definition, 323 temperature fluctuations, 230–231, 125–126, 130F
nephelometry vs., 323, 332 231F, 233–234 University of Washington
new technologies, 330–332 units, 231 autonomous underwater vehicles, 205T
sensors (transmissometers), 325–327, velocity fluctuations, 229–230, 230F, AUVs, 162
325F, 327F 231, 234 Unmanned underwater vehicles (UUV),
applications, 323, 327–329, 331T vertical profilers, 231, 231F, 232 164
beam size and, 326 eddy formation see Turbulent eddies active sonar images, 447
calibration, 326–327 Navier-Stokes equation, 229 see also Autonomous underwater
components, 325, 325F see also Kelvin-Helmholtz shear vehicles (AUVs)
deployment, 330 instabilities; Turbulent diffusion UOR (undulating oceanographic
instrument transmittance, 325–326 Turbulence sensors, 231–233, 229–235 recorder), 74–76
measurements, 323–324, 324F acoustic current meters, 234 Upward-looking profiling sonar, 398–399
particle size and, 326 airfoil probes, 232–233 Upwelling irradiance, 253, 274–275
resolution, 326 thin film sensors vs., 233 Upwelling radiance, 572
sensitivity, 326 Pitot tubes, 233 Urashima, 205T, 206F
see also Inherent optical properties temperature microstructure sensors, Urick, R J, 357
(IOPs); Ocean optics; Optical 233–234 USBL (ultrashort baseline acoustic
particle characterization; of conductivity, 234 navigation multibeam), 167
Turbulence sensors thermistors see Thermistors US National Data Buoy Center, 101
Transponders thin film thermometers, 233–234 US National Weather Service, SST record
acoustic, in LBL, 207 thin film sensors, 233 provided by, 216–217
autonomous underwater vehicles (AUV) airfoil probe vs., 233 USS Thresher, 156
navigation and, 167 disadvantages, 233 UTIG ocean bottom seismometer, 189F,
Trenches thermistors vs., 233–234 190T
nepheloid layers and, 291 see also Momentum flux U-Tow, 67
Trieste, sediment property measurements, measurements UUV see Unmanned underwater vehicles
386–387 Turbulent diffusion, 492 (UUV)
Trieste bathyscaph, history, 148, 154 molecular diffusion and, 492 UV see Ultraviolet radiation
Trieste submersible, 197, 197F Turbulent eddies, 218–219 UV-B radiation, ozone, concentration
Triple point, water, thermometer Turtle submersible, Bushnell’s, 156 effects, 302
calibration and, 26–27 Tyrrhenian Sea UVP (underwater video profiler), 83
Triple point cells, 26–27, 27, 27F salinity microstructure, 29F
TRMM Microwave Imager (TMI), V
554–556, 594 U
measurements restricted to tropical Van Drebel, Cornelius, shallow-water
regions, 554 Ultralow frequency band (ULF), 357 submersibles, 156
provides SST in the tropics, 554 definition, 357 Variable fluorescence induction, 259
see also Tropical Rainfall Ultrashort baseline (USBL) acoustic Vector averaging current meter (VACM),
Measurement Mission (TRMM) navigation, towed vehicles, 167 116–118, 116F, 117T, 119
Tropical Ocean - Global Atmosphere Ultraviolet radiation, 272 Vehicles
(TOGA) Study, 91 UV-B radiation, ozone, concentration autonomous underwater see
Tropical Rainfall Measurement Mission effects, 302 Autonomous underwater vehicles
(TRMM), 216, 596 see also Solar radiation (AUVs)
see also TRMM Microwave Imager Underwater flow cytometers, development deep-sea, 197–208
(TMI) of, 331–332 autonomous, 202–207
Tucker trawl, 71, 71F Underwater light fields future prospects, 207–208
Tuna (Thunnus), examples, 318–319 human-operated (submersibles),
acoustic scattering, 371 Hydrolight simulations, 319F, 320F, 197–199
Turbidity minimum, 289 321F, 322F navigation, 207
Turbulence, 229–231 see also Ocean optics; Radiative remotely-operated, 199–202
definition, 229 transfer (oceanic) towed, 197, 198T
 INDEX 651

ROVs see Remotely operated vehicles VOS see Voluntary observing ships (VOS) very low frequency band (VLF), 360
(ROVs) VPR see Video plankton recorder (VPR) wind-speed, 357, 359, 359F
towed see Towed vehicles VSF see Volume scattering function (VSF) Weather
Velocity logs, 167 forecasting, for coastal regions, satellite
Ventana remotely operated vessel, 129F, W remote sensing, 569–570
187T, 202T westerly, frequency time-series, 19–20,
Ventilation Walsh, Don, 197 20F
oceanographic research vessels, 6 Warm core rings (WCR), 556 Webb, Doug, 94, 140
Vertical line array (VLA), acoustic noise ‘Warm Pool,’ El Niño satellite remote Wenz, G M, 357, 358F
data, 362F, 363–364, 364 sensing of SST, 554–555 Westerly weather, frequency, time-series,
Very low frequency band (VLF), 359 Watch circle, 101 19–20, 20F
Crouch/Burt data, 359–360, 360 Water Westerly wind bursts
deep ocean noise, 360 triple point, thermometer calibration satellite remote imaging, 597
definition, 357 and, 26–27 Western Boundary Undercurrent,
Nichols data, 360 see also Sea water sediment interflow, 289F
noise mechanisms, 360 Water-column profiles/profiling Wet chemical analyzers, 335–339
shallow water noise, 360 autonomous underwater vehicles (AUV), air-segmented continuous flow see Air-
shipping noise, 360, 362–364 164, 166 segmented continuous flow
spectral density, 359–360 Massachusetts Bay, 170F analyzers
wave-wave interactions, 360 Water-leaving radiance, 571, 572 flow injection analyzers see Flow
wind speed, 359–360 satellite vs.in situ measurements, injection analysis (FIA)
Victor ROV, 187T 498F historical aspects, 335
Video cameras, remotely operated SeaWiFS data, 579F in-situ monitoring, 338–339
vehicles (ROVs), 184 Water transparency, 498–499 constraints, 338–339
Video plankton recorder (VPR), 72T, see also Light attenuation coefficient instrumentation, 338–339
80–81, 82F, 86, 300 Water vapor WetLabs transmissometer, 282
Viruses temperature deficit and, 548–549 Whales
analytical flow cytometry, 297 variability requiring atmospheric acoustic scattering, 374
Viscous cutoff scale (Lv), turbulent correction algorithm, 549 baleen see Baleen whales (Mysticeti)
dissipation, 229–230, 231, 232F Water velocity toothed see Odontocetes (toothed
Visible and infrared scanning radiometer fluctuations see Turbulence whales)
(VISR), 512 vertical, float measurement, 95 vocalizations, 360, 361–362
Visible light, absorption of, 302 Wave(s) Whiptail (Coryphaneodes subserrulatus),
Visible radiometry, 514 gravity see Gravity waves 368
Vision height see Wave height acoustic scattering, 371
remotely operated vehicles (ROVs), 184 internal see Internal wave(s) Whitecaps
VISR (visible and infrared scanning long surface, satellite remote sensing ocean color sensing and, 577
radiometer), 512 application, 564F, 566–567, 567F WHOI see Woods Hole Oceanographic
VLA (vertical line array), acoustic noise Wave breaking Institution
data, 362F, 363–364, 364 acoustic noise, 362, 363, 365 WHOI-BB ocean bottom seismometer,
VLF see Very low frequency band (VLF) dissipation, 365 191T
Volume attenuation (sound in seawater), individual breaking events, 365 WHOI-SP ocean bottom seismometer,
408, 409F Kennedy data, 364, 364F 190T, 192F
Volume backscattering, measurement, Kerman data, 364 Winch, 69F
zooplankton sampling, 84 see also Acoustic noise; Wind- computer-controlled
Volume displacement, gliders, 141 driven sea surface processes passive towed vehicles, 60
Volume scattering coefficient, marine Waveguides towed vehicles, 60
organisms, 369 acoustic noise, 359, 362–363, 363, deck, tether management system (TMS),
Volume scattering function (VSF), 273, 363–364 184–185
324 Wave height launch and recovery system (LARS),
bio-optical models, 249–250 sea state and, 414T 185–186
measurement(s), 274 wind speed and oceanographic research vessels, 6
constraints, 325, 331 satellite altimetry, 507–508 purpose-built, towed vehicles, 69, 69F
sea water, 316, 316F see also Satellite scatterometers Wind(s)
see also Nephelometry Wavelength, optical fibers, evanescent direction
particle size and, 324–325 wave penetration and, 241 measurement, wind vanes, 212–213,
Rayleigh scattering and, 324–325 Wavelength selector, absorptiometric 212F
Volume scattering phase function, 317 chemical sensors, 241, 242 problems for sensors, 213
Voluntary observing ships (VOS), 590 Wavelet analysis, 532 sonic anemometers, 212F, 213
satellite meteorological measurements, Wavenumber integration acoustic speeds see Wind speed
216 modeling, 424 stress see Wind stress
Vortices Wave-wave interactions Wind-driven sea surface processes,
seabed, sediment entrainment and, acoustic noise, 357 acoustic noise, 358F, 362–364
349–351, 350F causes, 358–359 bubbles, 362, 363, 365
Vorticity microseism spectral density, 358–359 peak frequency, 364, 365
tomography, 451 pressure spectral density, 357–359 source spectral density, 363F, 364–365
652 INDEX

Wind-driven sea surface processes, acoustic autonomous underwater vehicles, 205T analytical flow cytometry, 297
noise (continued) deep-towed vehicles, 198T imaging techniques, 300
Chapman/Cornish data, 363F, 364 remotely-operated vehicles (ROV), 202 photographic, 300
Kennedy data, 364–365, 364F Workboats, oceanographic research using remotely operated vehicles,
Kewley et al. data, 364 vessels, 6 300–301
wind speed, 364 World Ocean Circulation Experiment video, 300
vertical directional spectrum, 364 (WOCE), 91 optical plankton counting see Optical
reflection, 363 autonomous float, 92–94, 95–96 Plankton Counter (OPC)
refractive shadow zone, 363 cost, 93 Zooplankton sampling
scattering, 363 Hydrographic survey, 91 autonomous underwater vehicles
wave breaking see Wave breaking (AUVs), 85
Wind-driven surface waves, satellite X see also Autonomous underwater
remote sensing, 562–563 vehicles (AUVs)
Wind roughness correction, 588 XBT see Expendable bathythermograph gear types, 70, 72T
radar backscatter sensor, 588 (XBT) with nets and trawls, 70–86
Wind speed XCP see Expendable current profiler current status of systems, 83–84,
acoustic noise, 357, 359, 359–360, (XCP) 85F
359F, 364 XKT see Expendable optical irradiance future developments, 84
measurements, 211–213 probe (XKT) high-frequency acoustics, 72T, 83
calibration sensors, 211–212 X-ray fluorescence, 270 history, 70
cup anemometers, 211, 212F X-ray tomography, sediment core net systems, 72T
problems for sensors, 213 samples, 380, 381F closing cod-end systems, 76, 78F
propellor anemometers, 211, 212F XSV see Expendable sound velocity probe high-speed samplers, 70–71, 72T,
by satellite, 216 (XSV) 75F
sonic anemometers, 212F, 213 moored plankton collection systems,
ocean surface and, 414T Y 76–79, 81F
Wind stress multiple, 72T, 76, 80F, 81T
ocean frontal features, satellite remote Yo-yo pattern, autonomous underwater neuston samplers, 71–74, 76F
sensing, 563 vehicles (AUV), 166 non-opening/closing nets, 70, 71F
satellite remote sensing, 590–593 planktobenthos plankton nets, 76,
accuracy, 591 Z 77F
empirical function, 590–591 simple opening/closing nets, 70–71,
see also Drag coefficient Zooplankton 73F
Wind vanes, 212–213, 212F abundance, CPR survey data, 19–20, optical systems, 72T, 79
Wings, gliders, 140, 141 20F image-forming systems mounted on
Wire ropes, 97–98 acoustic scattering, 372 non-opening/closing nets, 79
‘3 x 19’, 98, 105 gas-bearing bodies, 373 optical instruments for
WOCE see World Ocean Circulation hard-shelled bodies, 373 nonquantitative studies, 80–83
Experiment (WOCE) liquid-like bodies, 372–373 particle detection systems, 79–80, 82F
WOCE autonomous float, 92–94, 95–96 biodiversity, CPR survey data, 22–23, stand-alone image-forming systems,
cost, 93 22F 79–80, 82F
WOCE drifter, 91, 92F gelatinous see Gelatinous zooplankton see also Continuous Plankton
Woods Hole Oceanographic Institution see also Copepod(s) Recorder (CPR) survey
(WHOI), 148–150, 510 Zooplankton analysis