67

3 Geodesy, Datums, Map Projec-

tions and Coordinate Systems

Introduction
Geographic information systems are area, but the distortion is quite apparent on
different from other information systems large-area maps. Quantitative measure-
because they contain spatial data. These ments on these maps are affected by the
spatial data include coordinates that define distortion, so we must somehow reconcile
the location, shape, and extent of geo- the mapping of a curved surface onto a flat
graphic objects. To effectively use GIS, we surface.
must develop a clear understanding of how The second main problem when defin-
coordinate systems are established and how ing a coordinate system comes from the
coordinates are measured. This chapter irregular nature of the Earth’s shape. We
introduces geodesy, the science of measur- learn in our earliest geography lessons that
ing the shape of the Earth, and map projec- the Earth is shaped as a sphere. This is a
tions, the transformation of coordinate valid approximation for many uses, how-
locations from the Earth’s curved surface ever, it is only an approximation. Several
onto flat maps. forces in the present and past have resulted
Defining coordinates for the Earth’s in an irregularly shaped Earth. These defor-
surface is complicated by two main factors. mations affect how we best map the surface
First, most people best understand geogra- of the Earth, and how we define Cartesian
phy in a Cartesian coordinate system on a coordinate systems for mapping and GIS.
flat surface. Humans naturally perceive the
Earth’s surface as flat, because at human
scales the Earth’s curvature is barely per- Early Measurements
ceptible. Humans have been using flat Humans have long speculated on the
maps for more than 40 centuries, and shape and size of the Earth. Babylonians
although globes are quite useful for percep- believed the Earth was a flat disk floating
tion and visualization at extremely small in an endless ocean, a notion also adopted
scales, they are not practical for most pur- by Homer, one of the more widely known
poses. Greek writers. The Greeks were early
A flat map must distort geometry in champions of geometry, and they had many
some way because the Earth is curved. competing views of the shape of the Earth.
When we plot latitude and longitude coor- One early Greek, Anaximenes, believed the
dinates on a Cartesian system, “straight” Earth was a rectangular box, while Pythag-
lines will appear bent, and areas will be oras and later Aristotle reasoned that the
incorrect. This distortion may be difficult to Earth must be a sphere. This deduction was
detect on detailed maps that cover a small based on many lines of evidence, and also
on a belief of divine direction by the gods.
68 GIS Fundamentals

Pythagoras believed the sphere was the most that at exactly the same date and time a verti-
perfect geometric figure, and that the gods cal post cast a shadow when located at Alex-
would use this perfect shape for their great- andria, about 805 kilometers north. The
est creation. Aristotle also observed that shadow/post combination defined an angle
ships disappeared over the horizon, the which was about 7o12’, or about 1/50th of a
moon appeared to be a sphere, the stars circle (Figure 3-1).
moved in circular patterns, and noted reports Eratosthenes deduced that the Earth
from wandering fishermen on shifts in the must be 805 multiplied by 50 or about
constellations. These observations were all 40,250 kilometers in circumference. His cal-
consistent with a spherical Earth. culations were all in stadia, the unit of mea-
After scientific support for a spherical sure of the time, and have been converted
Earth became entrenched, Greek scientists here to the metric equivalent, given our best
turned toward estimating the size of the idea of the length of a stadia. Eratosthenes’
sphere. Eratosthenes, a Greek scholar in estimate is close to our modern measurement
Egypt, performed one of the earliest well- of the Earth circumference of 38,762 kilo-
founded measurements of the Earth’s cir- meters. The difference is less than 4%.
cumference. He noticed that on the summer The accuracy of Eratosthenes’ estimate
solstice the sun at noon shone to the bottom is quite remarkable, given the equipment for
of a deep well in Syene. He believed that the measuring distance and angles at that time,
well was located on the Tropic of Cancer, so and because a number of his assumptions
that the sun would be exactly overhead dur- were incorrect. The well at Syene was
ing the summer solstice. He also observed

Figure 3-1: Measurements made by Eratosthenes to determine the circumference of the Earth.
 Chapter 3: Geodesy, Projections, and Coordinate Systems 69

Figure 3-2: The Earth’s radius and circumference may be determined by simultaneous measurement of
zenith angles at two points. Two points are separated by an arc distance a measured on the Earth surface.
These points also span an angle θ defined at the Earth center. The local zenith angles α and γ are related
to θ, and the Earth radius is related to a and θ. Once the radius is calculated, the Earth circumference may
be determined.

located about 60 kilometers off the Tropic of between the zenith angles used to calculate
Cancer, so the sun was not directly over- the circumference of the Earth. In Figure 3-
head. The true distance between the well 2, the two zenith angles, α and γ, are mea-
location and Alexandria was about 729 kilo- sured to a star. The surface distance between
meters, not 805, and the well was 3o3’ east these two locations is also measured, and the
of the meridian of Alexandria, and not due measurements combined with geometric
north. However these errors either compen- relationships to calculate the Earth circum-
sated for or were offset by measurement ference (Figure 3-2). Posidonius calculated
errors to end up with an amazingly accurate the difference in the zenith angles to a
estimate. known star as about 1/48th of a circle when
Posidonius made an independent esti- measured at Rhodes and Alexandria. By
mate of the size of the Earth sphere by mea- estimating these two towns to be about 805
suring angles to a star near the horizon kilometers apart, he calculated the circum-
(Figure 3-2). Stars visible in the night sky ference of the Earth to be 38,600 kilometers.
define a uniform reference. The angle Again there were compensating errors,
between a local vertical line and a star loca- resulting in an accurate value. Another
tion is called a zenith angle. The zenith angle Greek scientist determined the circumfer-
may be measured simultaneously at two ence to be 28,960 kilometers, and unfortu-
locations on Earth, and the difference nately this shorter measurement was adopted
70 GIS Fundamentals

by Ptolemy for his world maps. This esti-

mate was widely accepted until the 1500s,
when Gerardus Mercator revised the figure
upward.
During the 17th and 18th centuries two
developments led to intense activity directed
at measuring the size and shape of the Earth.
Sir Isaac Newton and others reasoned the
Earth must be flattened somewhat due to
rotational forces. They argued that centrifu-
gal forces cause the equatorial regions of the
Earth to bulge as it spins on its axis. They
proposed the Earth would be better modeled
by an ellipsoid, a sphere that was slightly
flattened at the poles. Measurements by their
French contemporaries taken north and
south of Paris suggested the Earth was flat-
tened in an equatorial direction and not in a Figure 3-3: An ellipsoidal model of the Earth’s
shape.
polar direction. The controversy persisted
until expeditions by the French Royal Acad-
emy of Sciences between 1730 and 1745
measured the shape of the Earth near the
Equator in South America and in the high
northern latitudes of Europe. Complex, Specifying the Ellipsoid
repeated, highly accurate measurements Once the general shape of the Earth was
established the curvature of the Earth was determined, geodesists focused on precisely
greater at the Equator than the poles, and measuring the size of the ellipsoid. The
that an ellipsoid flattened at the poles was ellipsoid has two characteristic dimensions
the best geometric model of the Earth’s sur- (Figure 3-3). These are the semi-major axis,
face. the radius r1 in the Equatorial direction, and
Note that the words spheroid and ellip- the semi-minor axis, the radius r2 in the
soid are often used interchangeably. For polar direction. The equatorial radius is
example, the Clarke 1880 ellipsoid is often always greater than the polar radius for the
referred to as the Clarke 1880 spheroid, Earth ellipsoid. This difference in polar and
because Clarke provided parameters for an equatorial radii can also be described by the
ellipsoidal model of the Earth’s shape. GIS flattening factor, as shown in Figure 3-3.
software often prompts the user for a spher- Earth radii have been determined since
oid when defining a coordinate projection, the 18th century using a number of methods,
and then lists a set of ellipsoids for choices. and for many parts of the globe. The most
An ellipsoid is sometimes referred to as common methods until recently have
a special class of spheroid known as an involved astronomical observations similar
“oblate” spheroid. Thus, it is less precise but to the those performed by Posidonius. These
still correct to refer to an ellipsoid more gen- astronomical observations, also called celes-
erally as a spheroid. It would perhaps cause tial observations, are combined with long-
less confusion if the terms were used more distance surveys over large areas. Star and
consistently, but the usage is widespread. sun locations have been observed and cata-
Ellipsoids are almost always used to model loged for centuries, and combined with
the Earth’s shape, and they are often referred accurate clocks, the positions of these celes-
to as spheroids. tial bodies may be measured to precisely
establish the latitudes and longitudes of
 Chapter 3: Geodesy, Projections, and Coordinate Systems 71

points on the surface of the Earth. Measure- listed in Table 3-1? Different ellipsoids were
ments during the 18th, 19th and early 20th adopted in various parts of the world for var-
centuries used optical instruments for celes- ious reasons. Different ellipsoids were cho-
tial observations (Figure 3-4). sen primarily because there were different
Measurement efforts through the 19th sets of measurements used in each region or
and 20th centuries led to the establishment of continent.
a set of official ellipsoids (Table 3-1). Why Historically, geodetic surveys were iso-
not use the same ellipsoid everywhere on lated by large water bodies. For example,
Earth, instead of the different ellipsoids surveys in Australia did not span the Pacific

Table 3-1: Official ellipsoids. Radii may be specified more precisely than the 0.1 meter
shown here (from Snyder, 1987 and other sources).

Equatorial Polar Flat-

Name Year Radius, r1 Radius, r2 tening Users
meters meters Factor,
f

Airy 1830 6,377,563.4 6,356,256.9 1/ Great

299.32 Britain

Bessel 1841 6,377,397.2 6,356,079.0 1/ Central

299.15 Europe,
Chile,
Indonesia

Clarke 1866 6,378,206.4 6,356,583.8 1/ North

294.98 America;
Philippines

Clarke 1880 6,378,249.1 6,356,514.9 1/ Most of

293.46 Africa;
France

Inter- 1924 6,378,388.0 6,356,911.9 1/ Much of

national 297.00 the World

Austra- 1965 6,378,160.0 6,356,774.7 1/ Australia

lian 298.25

WGS72 1972 6,378,135.0 6,356,750.5 1/ NASA, US

298.26 Defense
Dept.

GRS80 1980 6,378,137.0 6,356,752.3 1/ Worldwide

298.26

WGS84 1984 - 6,378,137.0 6,356,752.3 1/ Worldwide

cur- 298.26
rent
72 GIS Fundamentals

One ellipsoid could not be fit to all the

World’s survey data because during the 18th
and 19th centuries there was no clear way to
combine a global set of measurements.
The differences in the best fitting ellip-
soid were due to a number of factors. There
are real differences in the shape of the Earth
across the globe (described in the next sec-
tion). The Earth’s true shape diverges from
the shape of an ellipsoid, giving different
estimates of the shape when only a portion
of the Earth is surveyed. Differences in the
ellipsoids were also due to differences in
survey methods and data analyses employed.
Computational resources, the sheer number
of survey points, and the scarcity of survey
points for many areas were barriers to the
development of global ellipsoids. Methods
for computing positions, removing errors, or
adjusting point locations were not the same
worldwide, and may have led to differences
in ellipsoidal estimates. Different methods
applied to the same data may cause differ-
ences, and it took time for the best methods
Figure 3-4: An instrument used in the early 1900s to be developed, widely recognized, and
for measuring the position of celestial bodies. adopted.
More recently, data derived from satel-
lites, lasers, and broadcast timing signals
have been used for extremely precise mea-
Ocean to reach Asia. Geodetic surveys relied surements of relative positions across conti-
primarily on optical instruments prior to the nents and oceans. A global set of
early 20th century. These instruments were measurements became possible, and with it
essentially precise telescopes, and thus the ability to combine and compare measure-
sighting distances were limited by the ments on all continents simultaneously. This
Earth’s curvature. Individual survey legs has allowed the calculation of globally appli-
greater than 50 kilometers (30 miles) were cable ellipsoids. Internationally recognized
rare, so during this period there were no ellipsoids such as WGS72, GRS80, or
good ways to connect surveys between con- WGS84 have become widely used as more
tinents. precise measurements have been developed
Because continental surveys were iso- over a larger portion of the globe. Continu-
lated, ellipsoidal parameters were fit for ing measurements may lead to further ellip-
each continent or comparably large survey soid refinements.
area. These ellipsoids represented continen-
tal measurements and conditions. Measure- The Geoid
ments based on Australian surveys yielded a
different “best” ellipsoid than those in As noted in the previous section, the true
Europe. Europe’s best ellipsoidal estimate shape of the Earth varies slightly from the
was different from Asia’s, and from South mathematically smooth surface of an ellip-
America, North America, or other regions. soid. Differences in the density of the Earth
 Chapter 3: Geodesy, Projections, and Coordinate Systems 73

cause variation in the strength of the gravita-

tional pull, in turn causing regions to dip or
bulge above or below a reference ellipsoid.
This undulating shape is called a geoid.
Geodesists have defined the geoid as the
three-dimensional surface along which the
pull of gravity is a specified constant. The
geoidal surface may be thought of as the
level of an imaginary sea, were it to cover
the entire Earth and not be affected by wind,
waves, the moon, or forces other than grav-
ity. The surface of the geoid is in this way
related to mean sea level, or other references
against which heights are measured. Geode-
sists may measure surface heights relative to
the geoid, and at any point on Earth there are
three important surfaces, the ellipsoid, the
geoid, and the Earth surface (Figure 3-5).
Figure 3-6: Locally fit ellipsoids may provide
Because we have two reference sur- poor fits in other portions of the Earth’s surface.
faces, a geiod and an ellipsoid, we also have
two bases from which to measure height.
Elevation is typically defined as the vertical
distance above a geoid. This height above a orthometric height labeled H. The differ-
geoid is also called the orthometric height. ence, shown in the figure as N, has various
Orthometric heights are the basis for eleva- names, including geoidal height and geoidal
tion measurements. Heights above the ellip- separation.
soid are often referred to as an ellipsoidal Geoidal variation in the Earth’s shape is
height. These are illustrated in Figure 3-5, the main cause for different ellipsoids being
with the ellipsoidal height labeled h, and employed in different parts of the world. The
best local fit of an ellipsoid to the geoidal
surface in one portion of the globe may not
be the best fit in another portion. This is
illustrated in Figure 3-6. Ellipsoid A fits well
over one portion of the geoid, ellipsoid B in
another, but both provide a poor fit in many
other areas of the Earth’s surface. Initial
measurements were often localized, so the
deviations in other parts of the Earth were up
to several hundred meters. A global array of
measurements allows the estimation of a
globally optimal ellipsoid.
The geoid varies by less than 100 meters
from the ellipsoid over most of the Earth
(Figure 3-7), and these differences between
the geoid and the ellipsoid are small relative
to the differences between an ellipsoidal and
Figure 3-5: Ellipsoidal, orthometric, and geoidal
height are interrelated. Note that values for N a spherical model. The separation between
are highly exagerated in this figure - undulations the geoid and ellipsoid represented by the
are typically much less than differences in H. geoidal heights is also quite small relative to
the polar and equatorial radii. As noted in
74 GIS Fundamentals

Figure 3-7: Geoidal heights vary across the globe. This figure depicts positive geoidal heights in lighter
tones (geoid above the ellipsoid) and negative geoidal heights in darker tones. Note that geoidal heights are
positive for large areas near Iceland and the Philippines (A and B, respectively), while large negative values
are found south of India (C). Continental and country borders are shown in white.

Table 3-1, the Earth’s equatorial radius is the geoid is estimated by measuring the
about 6,780,000 meters, or about 32,000 change in gravity with elevation. Gravimet-
times the range of the highest to lowest geoi- ric surveys have been conducted for the
dal heights. This small difference would be entire globe, although measurements are
imperceptible in an object at human scales. more frequent over the continents and in the
For example, this is less than the relative developed countries. Satellite-based mea-
thickness of a coat of paint on a ball three surements in the late 20th century substan-
meters in diameter. While relatively small, tially improved the coverage and quality of
the geoid must still be considered for accu- the network across the globe. Geoidal mod-
rate vertical and horizontal mapping over els have been estimated which allow accu-
continental or global scales, and therefore rate prediction of the geoidal height for the
must be precisely defined. entire globe.
It is important to note that the geoid is a The network of measurements show that
measured and interpolated surface, and not a the geoidal height, defined above as the dif-
mathematically defined surface. The geoid’s ference between the ellipsoid and geoidal
surface is measured using a number of meth- surfaces, varies between 85 and -105 meters
ods, initially by a combination of plumb bob over most of the globe (Figure 3-7).
and surface measurements, and later with Although it may at first seem difficult to
various types of gravimeters. These instru- believe, the “average” ocean surface near
ments measure the gravitational pull as they Iceland is more than 150 meters “higher”
are towed, flown, or driven on or above the than the ocean surface northeast of Jamaica.
Earth’s surface. Variation in the orbits of This height difference is measured relative
artificial satellites may also be used to esti- to the ellipsoid. Since gravity pulls in a
mate the shape of the geoid. The height to direction that is perpendicular to the geoidal
the reference gravitational surface defining surface, the force is at a right angle to the
 Chapter 3: Geodesy, Projections, and Coordinate Systems 75

surface of the ocean, resulting in permanent the ellipsoid, consisting of latitude, which
bulges and dips in the mean ocean surface. varies from north to south, and longitude,
Variation in ocean heights due to swells and which varies from east to west (Figure 3-
wind-driven waves are more apparent at 8). Lines of constant longitude are called
local scales, but are much smaller than the meridians, and lines of constant latitude are
long-distance geoidal undulations. called parallels. Parallels run parallel to
each other in an east-west direction around
the Earth. The meridians are geographic
Geographic Coordinates, Lati- north/south lines, and converge at the
tude, and Longitude poles.
Once a size and shape of the reference By convention, the Equator is taken as
ellipsoid has been determined, the poles and zero degrees latitude, and latitudes increase
Equator are also defined. The poles are in absolute value to the north and south (Fig-
defined by the axis of revolution of the ellip- ure 3-8). Latitudes are thus designated by
soid, and the Equator is defined as the circle their magnitude and direction, for example
mid-way between the two poles, at a right 35oN or 72oS. When signed values are
angle to the polar axis, and spanning the required, northern latitudes are designated
widest dimension of the ellipsoid. We now positive and southern latitudes designated
must go about defining a coordinate grid, a negative. An international meeting in 1884
reference system by which we may specify established a longitudinal origin interscting
the position of features on the ellipsoidal the Royal Greenwich Observatory in
surface. England. Known as the Prime or Greenwich
A geographic coordinate system has Meridian, this north to south line is the ori-
been defined for the Earth with reference to gin, or zero value, for longitudes. East or
west longitudes are specified as angles of

Figure 3-8: The geographic coordinate system.

76 GIS Fundamentals

rotation away from the Prime Meridian. defined on a curved surface, and the longitu-
When required, west is considered negative dinal lines converge at the poles. Therefore
and east positive. there are no Cartesian parallels in the north-
There is often confusion between mag- south direction, and the Earth surface length
netic north and geographic north. Magnetic of a degree of longitude varies from approxi-
north is the point towards which a compass mately 111.3 kilometers at the Equator to 0
points. The geographic North Pole is located kilometers at the poles. This distortion is
at one of the poles of the Earth’s axis of rota-
tion. Magnetic north and the geographic
North Pole do not coincide (Figure 3-9). If
you were standing on the geographic North
Pole with a compass, it would point approxi-
mately in the direction of northern Canada,
towards magnetic north some 600 kilometers
away.
Because magnetic north and the geo-
graphic North Pole do not coincide, mag-
netic north and the geographic North Pole
are in slightly different directions most
places on Earth. This angular difference is
called the magnetic declination and varies
across the globe. The specification of map
projections and coordinate systems is always
in reference to the geographic North Pole,
not magnetic north.
Geographic coordinates do not form a
Cartesian system (Figure 3-10). As previ- Figure 3-9: Magnetic north and the geographic
North Pole.
ously described in Chapter 2, a Cartesian
system defines lines of equal value to form a
right-angle grid. Geographic coordinates are

Figure 3-10: Geographic coordinates on a spherical (left) and Cartesian (right) representation. Notice the cir-
cles with a 5 degree radius appear distorted on the spherical representation, illustrating the change in surface
distance represented by a degree of longitude from the Equator to near the poles.
 Chapter 3: Geodesy, Projections, and Coordinate Systems 77

seen on the left in Figure 3-10. Circles with a maintained by the National Geodetic Survey
fixed 5 degree radius appear distorted near (NGS). This unit of the U.S. federal govern-
the poles when drawn on a globe. The circles ment establishes geodetic latitudes and lon-
become “flattened” in the east-west direc- gitudes of known, monumented points.
tion. In contrast, circles appear as circles These points, taken together, form the basis
when the geographic coordinates are plotted for a geodetic datum, upon which most sub-
in a Cartesian system, as at the right of Fig- sequent surveys and positional measure-
ure 3-10, but the underlying geography is ments are based.
distorted; note the erroneous size and shape A datum is a reference surface. A geo-
of Antarctica. While the ground distance detic datum consists of two major compo-
spanned by a degree of longitude changes nents. The first component is the previously
markedly across the globe, ground distance described specification of an ellipsoid with a
for a degree of latitude varies only slightly, spherical coordinate system and an origin.
from 110.6 kilometers at the Equator to The second part of a datum consists of a set
111.7 kilometers at the poles. of points and lines that have been painstak-
Spherical coordinates are most often ingly surveyed using the best methods and
recorded in a degrees-minutes-seconds equipment. Different datums are specified
(DMS) notation, e.g. N43o 35’ 20”, signify- through time because new points are added
ing 43 degrees, 35 minutes, and 20 seconds and survey methods improve. We periodi-
of latitude. Minutes and seconds range from cally update our datum when a sufficiently
large number of new survey points has been
zero to sixty. Alternatively, spherical coordi-
measured. We do this by re-estimating the
nates may be expressed as decimal degrees coordinates of our datum points after includ-
(DD). Examples for conversion DMS to DD
and back are shown in Figure 2-7.

Horizontal Datums
We have defined a geographic coordi-
nate system that provides for specifying
locations on the Earth. This gives us the
exact longitude of only one location, the
Greenwich Observatory. By definition the
longitude of the Greenwich meridian is 0
degrees. We must estimate the longitudes
and latitudes of all other locations through
measurements. We establish a set of points
for which the horizontal and vertical posi-
tions have been accurately determined. All
other coordinate locations we use are mea-
sured with reference to this set of precisely
surveyed points, including the coordinates
we enter in our GIS to represent spatial fea-
tures.
Many countries have a government
body charged with making precise geodetic
surveys. For example, nearly all horizontal
locations in the United States are traceable
Figure 3-11: An early geodetic survey, used
or related to a national network of highly in fixing datum locations. (courtesy NCGS)
accurate survey points established and/or
78 GIS Fundamentals

tion from specific local features are

recorded. For example, in areas of deep soil,
the NGS may bore a hole one to three meters
deep at the point location. Successive con-
crete posts may be placed one atop another,
with a brass monument affixed to each post,
with the top of the final post resting flush
with the ground surface. If the surface mon-
ument is lost, buried, or destroyed, e.g.,
sheared with a bulldozer or covered during
road improvements or construction, the
point may still be recovered, albeit with
some effort. Control survey points are often
identified with a number of nearby signs, to
aid in identification and recovery (Figure 3-
13).
Early geodetic surveys combined hori-
Figure 3-12: A brass disk used to monument a sur- zontal measurements with repeated, excruci-
vey benchmark.
atingly precise astronomical observations to
determine latitude and longitude of a subset
of points. Only a few datum points were
determined using astronomical observations.
ing these newer measurements, thereby Astronomical observations were typically
improving our estimate of the position of used at the starting point, a few intermediate
each point. points, and near the end of geodetic surveys.
Historically, the relative positions of a This is because star positions required
set of datum points were determined using repeated measurements over several nights.
celestial measurements in combination with Clouds, haze, or a full moon often length-
high-accuracy ground measurements. Most ened the measurement times. In addition,
early measurements involved precise field
surveys with optical instruments (Figure 3-
11). These methods have been replaced in
recent years by sophisticated electronic and
satellite-based surveying systems. Measure-
ments from an extensive set of surveys were
combined, errors and inconsistencies identi-
fied and removed, and the geographic loca-
tion for each datum point estimated.
Precisely surveyed points used in devel-
oping a datum are often monumented, and
these monumented points are known as
benchmarks. Benchmarks usually consist of
a brass disk embedded in rock or concrete
(Figure 3-12), although they also may con-
sist of squares or circles chiseled in rocks,
iron posts, or other long-term marks.
Because of the considerable effort and cost
expended in establishing the coordinates for Figure 3-13: Signs are often placed near control
points to warn of their presence and aid in their
each benchmark, they are often redundantly location.
monumented, and their distance and direc-
 Chapter 3: Geodesy, Projections, and Coordinate Systems 79

celestial measurements required correction zontal surface measurements. Triangulation

for atmospheric refraction, a process which surveys utilize a network of interlocking tri-
bends light and changes the apparent posi- angles to determine positions at survey sta-
tion of stars. Refraction depends on how tions. Because there are multiple
high the star is in the sky at the time of mea- measurements to each survey station, the
surement, as well as temperature, atmo- location at each station may be computed by
spheric humidity, and other factors. various paths. The survey accuracy can be
Horizontal measurements were as pre- field checked, because large divergence in a
cise and much faster than astronomical mea- calculated station location via different paths
surements when surveys originated at known may quickly be identified. There are always
locations. Thus, most campaigns involved some differences in the measured locations
the establishment of a few starting, interme- when traversing different paths. An accept-
diate, and ending points using astronomical able error limit was set, usually as a propor-
measurements. Horizontal surface measure- tion, e.g., differences in the measured
ments were then used to connect these astro- location of more than 1 part in 100,000
nomically surveyed points and thereby would be considered unacceptable. When
create an expanded, well-distributed set of unacceptable errors were found, survey lines
known datum points. Figure 3-14 shows an were re-measured.
example survey, where open circles signify Triangulation surveys were also adopted
points established by astronomical measure- because they required few surface distance
ments and filled circles denote points estab- measurements. This was an advantage
lished by surface measurements. between the late 18th and early 20th century,
Triangulation surveys (Figure 3-14) the period during which the widespread net-
were used to establish datum points via hori- works of datum points were established.
During that time it was much easier to accu-
rately measure horizontal and vertical angles
than to measure horizontal distances. Direct
measurements of surface distances were less
certain because they relied on metal tapes,
chains, or rods. Tapes and chains sagged by
varying amounts. Metal changed length with
temperature and stretched with use. Inge-
nious bi-metallic compensation bars were
developed that combined two metals along
their length. These metals changed length at
different rates with temperature. A scale
engraved in the bars indicated the amount of
expansion and facilitated a compensation to
offset the temperature-caused error. None-
theless, laying the bars end-to-end over great
distances was a painstaking, slow process,
and so surface measurements were to be
avoided.
Triangulation surveying requires the
surface measurement of an initial baseline,
and then relies on geometric relationships
and angle measurements to calculate all sub-
Figure 3-14: A triangulation survey network. Sta- sequent distances. Figure 3-15 shows a
tions may be measured using astronomical (open sequence of measurements for a typical tri-
circles) or surface surveys (filled circles).
angulation survey conducted in the 19th or
80 GIS Fundamentals

early 20th centuries. The positions of sta- Survey stations were typically placed on
tions S1 and S2 (Figure 3-15, top) were high vantage points to lengthen the distance
determined by celestial observations, and the between triangulation stations. Longer dis-
length A measured using compensation bars tances meant fewer stations were required to
or equally precise methods. A surveying cover a given area. Since the triangulation
instrument was placed at stations S1, S2, networks spanned continents, there was a
and S3 and used to measure angles a, b, and strong impetus to keep costs down. The sur-
c. The law of sines allows the calculation of vey stations were quite far apart, up to 10’s
lengths B and C, and combined with the of kilometers, and measurements were often
angle measurements allows precise esti- made from tall objects such as mountain-
mates of the location of station S3. Angles tops, church steeples, or specially con-
e, f, and g (Figure 3-15, bottom) were then structed Bilby towers to give long sight lines
measured, and the precise location of station and hence long survey legs (Figure 3-16).
S4 estimated. The geodetic survey could Besides use in developing datums, these
then be extended indefinitely by repeated monumented points may be used as starting
application of angle measurements and the points for precise local surveys.
law of sines, with only occasional surface The positions of all points in a reference
distance measurements as a check on posi- datum are estimated in a network-wide
tions. datum adjustment. “Highest accuracy” mea-

Figure 3-15: Early geodetic surveys used triangulation networks to reduce the number of surface dis-
tance measurements.
 Chapter 3: Geodesy, Projections, and Coordinate Systems 81

surements are collected for each angle or points still exist, even though the best avail-
distance in the triangulation surveys of able methods have been used. The survey
datum points. However, some errors are network must be adjusted, that is, the error
inevitable, and small errors in each measure- reduced and distributed across the network
ment are allowed. For example, distance of a in some “optimum” fashion. A datum adjust-
30 kilometer (20 mile) survey leg may differ ment resolves and distributes these errors
by 30 centimeters (1 foot) when measured across the survey network.
using two independent methods. While quite Periodic datum adjustments result in
small, these small discrepancies in the dis- multiple regional or global reference
tances or angles for redundantly measured datums. For example, there are several refer-
ence datums for North America. Several
datums exist because new surveys are added
to regions that had been poorly measured
before. In addition, survey instruments are
continuously improving, and computational
and survey adjustment methods are also get-
ting better. When enough new survey points
have been collected, a new datum is esti-
mated. This means new coordinate locations
are estimated for all the datum points. The
datum points don’t move, but our best esti-
mates of the datum point coordinates will
change. Differences between the datums
reflect differences in the control points, sur-
vey methods, and mathematical models and
assumptions used in the datum adjustment.
The calculation of a new datum requires
that all surveys must be simultaneously
adjusted to reflect our current “best” esti-
mate of the true positions of each datum
point. Generally a statistical least-squares
adjustment is performed, but this is not a
trivial exercise, considering the adjustment
may include survey data for tens of thou-
sands of old and newly surveyed points from
across the continent, or even the globe.
Because of their complexity, these conti-
nent-wide or global datum calculations have
historically been quite infrequent. Computa-
tional barriers to datum adjustments have
diminished in the past few decades, and so
datum adjustments and new versions of
datums are becoming more common.
Three main horizontal datums have been
used widely in North America. The first of
these is the North American Datum of 1927
Figure 3-16: A Bilby tower near Bozeman,
Montana, USA, used as a platform for a trian- (NAD27). NAD27 is a general least-squares
gulation survey. Towers or other high vantage adjustment that included all horizontal geo-
points increased the distance between survey detic surveys completed at that time. The
stations. (courtesy NCGS)
geodesists used the Clarke Ellipsoid of 1866
82 GIS Fundamentals

and held fixed the latitude and longitude of a Coordinates across different datums
survey station in Kansas. NAD27 yielded may differ slightly, because of the differ-
adjusted latitudes and longitudes for approx- ences in the ellipsoid and in the measure-
imately 26,000 survey stations in the United ments and methods used to estimate the
States and Canada. locations of datum points. Differences may
The North American Datum of 1983 be small, e.g., the shift in coordinate loca-
(NAD83) is the immediate successor datum tions from WGS84 to NAD83 is often less
to NAD27. It was undertaken by the than a meter, but shifts may be quite large.
National Coast and Geodetic Survey to Care should be taken to not mix spatial data
include the large number of geodetic survey across datums unless the magnitude for the
points established between the mid-1920s datum shift has been established for the area
and the early 1980s. Approximately 250,000 of interest, and this magnitude is considered
stations and 2,000,000 distance measure- small relative to overall data use and accu-
ments were included in the adjustment. The racy specifications.
GRS80 ellipsoid was used as reference. Datum shifts between NAD27 and
NAD83 uses an Earth-centered reference, NAD83 are often quite large, up to 100's of
rather than the fixed station selected for meters. Figure 3-17 indicates the relative
NAD27. size of shifts across the country between
The World Geodetic System of 1984 NAD27 and NAD83, based on estimates
(WGS84) is also commonly used, and was provided by the National Geodetic Survey.
developed by the U.S. Department of Notice that datum shifts between NAD27
Defense (DOD). It was developed in 1987 and NAD83 are approximately -20 to 40
based on Doppler satellite measurements of meters (70 to 140 feet) northward and
the Earth, and is the base system used in between -40 and 100 meters (140 to 330
most DOD maps and positional data. feet) east-west.
WGS84 has been updated based on more You should note that there are several
recent satellite measurements and is speci- versions of NAD83. Geodesists at the U.S.
fied using a version designator, e.g., the National Geodetic Survey find they have
update based on data collected up to January enough improved measurements of datum
1994 is designated as WGS84 (G370). The points to perform a complete network re-
WGS84 ellipsoid is very similar to the adjustment every five to ten years. Each re-
GRS80 ellipsoid. adjustment combines the new measurements
Since different datums are based on dif- with older measurements in a least squares
ferent sets of measurements and ellipsoids, model, and distributes new, smaller esti-
the coordinates for benchmark datum points mates of error across the entire network of
typically differ between datums. That is to datum points. Therefore, there is typically a
say, the latitude and longitude location of a datum year and version associated with a set
given benchmark in the NAD27 datum will of benchmark coordinates, e.g., coordinates
likely be different from the latitude and lon- may be provided for a point with reference
gitude of that same benchmark in the to the datum NAD83 (1986), or with refer-
NAD83 or WGS84 datums. This is ence to NAD83 (1996). Distances of the
described as a datum shift. The monumented shifts due to re-adjustments are typically less
points do not move. Physically, the points than one-half meter, and therefore they are
remain in the same location; our estimates of often small relative to other positional uncer-
the point locations change. As survey mea- tainties inherent in much spatial data. How-
surements improve through time, and there ever, these distances can be significant in
are more of them, we obtain better estimates many projects, and even if not, they may be
of the true locations of the monumented cause for confusion unless their source is
datum points. understood. Because of datum shifts due to
the re-adjustments, for example, from
 Chapter 3: Geodesy, Projections, and Coordinate Systems 83

Figure 3-17: Datum shift in meters between the North American Datum of 1927 (NAD27) and the
North American Datum of 1983 (NAD83).
84 GIS Fundamentals

NAD27 to NAD83, or due to differences in liquid level to establish horizontal lines. As

subsequent versions of NAD83, it is impor- illustrated in Figure 3-18, horizontal rods
tant to note the datum underpinning a GIS were placed between succeeding leveling
development effort. If source data or maps posts across the landscape to physically
are referenced to different datums, then they measure height differences.
must be converted before they will overlay The number, accuracy, and extent of lev-
correctly. eling surveys increased substantially in the
18th and 19th centuries. Colonizing powers
Vertical Datums had a strong interest in measuring and monu-
menting the extent of their domain and to dig
Just as there are networks of well-mea- canals for navigation and develop other pub-
sured points to define horizontal position, lic works. This same period also saw a
there are networks of points to define verti- marked increase in craftsman and merchant
cal datums. Vertical datums are used as a ref- classes, combined with the industrial revolu-
erence for specifying heights. Much like tion that provided the means to more rapidly
horizontal datums, they are established and accurately measure over longer dis-
through a set of painstakingly surveyed con- tances. High quality optical instruments
trol points. These point elevations are pre- could be reliably produced, and advances in
cisely measured, initially through a set of mathematics and statistics eased the simulta-
optical surface measurements, but more neous adjustment of many survey measure-
recently using GPS, laser, satellite, and other ments. Grand and epic surveys that lasted
measurement systems. decades were commissioned, such as the
Leveling surveys, or leveling, involve Great Arc, from southern India to the Hima-
the precise measurement of horizontal and layas. These surveys were performed at sub-
vertical distance, and are among the oldest stantial capital and human expense; in one
surveying techniques.(Figure 3-18). Dis- portion of the Great Arc more than 60% of
tances and elevation differences are pre- the field crews died over a six year period,
cisely measured from an initial point. Early due to illness and mishaps. Surface leveling
leveling surveys were performed with the provided most height measurements for ver-
simplest of instruments, including a plumb tical datums until the mid to late 20th cen-
bob to establish leveling posts, and a simple tury, with the introduction of a variety of
satellite-based methods.
Until the 20th century most leveling sur-
veys employed optical instruments and trig-
onometry to measure changes in height, as
shown in Figure 3-19. Surface distance
along the slope was measured to avoid the
tedious process of establishing vertical posts
and leveling rods.The vertical angle was also
measured from a known station to an
unknown station. The angle was typically
measured with a small telescope fitted with a
precisely scribed angle gauge. The gauge
could be referenced to zero at a horizontal
position, usually with an integrated bubble
level, or later, with an electronic level. Sur-
face distance would then be combined with
Figure 3-18: Ancient surveys often used the measured vertical angle to calculate the
“level” bars placed vertical posts, hence the horizontal and vertical distances. Early sur-
name leveling surveys.
veys measured surface distance along the
 Chapter 3: Geodesy, Projections, and Coordinate Systems 85

Figure 3-19:Leveling surveys often employ optical measurements of vertical angle (α) with measure-
ments of surface distance (d) and knowledge of trigonometric relationships to calculate horizontal
distance (h) and vertical distance (v).

slope with ropes, metal chains, and steel lines by 1900. Similar increases in leveling
tapes, but these physical devices have surveys occurred in the rest of North Amer-
largely been replaced by improved optical ica, most of Europe, and many parts of Asia
methods, or by laser-based methods. during the late 1800s and early 1900s. Lev-
Leveling surveys for geodetic reference eling has continued through to the present
may be thought of as beginning at a station time, with hundreds of thousands of kilome-
near the ocean. This ocean station has been ters of leveling lines measured to date. How-
vertically referenced to a mean ocean height ever, optical methods are little used at
nearby. Leveling surveys may then be used present, as satellite-based positioning meth-
to establish the heights of points that are far ods now predominate vertical surveying.
from the coast, but relative to the standard As with horizontal datums, the primary
coastal benchmark or station. Leveling sur- vertical datums in use have changed through
veys often share stations with horizontal time as the number, distribution, and accu-
datum surveys because the stations are racy of vertical survey points have
already occupied and the horizontal posi- increased. The two most common vertical
tions are precisely known. Independent lev- datums in North America are the National
eling surveys may be conducted, but they Geodetic Vertical Datum of 1929
must also include precise horizontal distance (NGVD29) and the North American Vertical
and direction measurements to ensure high Datum of 1988 (NAVD88). The 1929 Verti-
mapping accuracy. cal Datum was derived from a best fit of the
National leveling surveys typically average elevation of 26 stations in North
started near population and commerce cen- America. The 1988 datum is based on over
ters, and grew to encompass entire countries. 600,000 kilometers (360,000 miles) of con-
National leveling surveys began in the trol leveling performed since 1929, and also
United State in the 1850s, and grew to reflects geologic crustal movements or sub-
include over 21,000 kilometers of leveling sidence that may have changed benchmark
86 GIS Fundamentals

elevation. NAVD88 adjusted the estimate of inches). Accuracies are specified by Class
mean sea level slightly. Differences are and Orders, down to a Class III, 2nd order
observed when comparing NGVD29 and point with an error of no more than 1 part in
NAVD88 elevations for the same bench- 5,000.
marks. Federal, state, provincial, and county
surveyors typically have a set of identified
Control Accuracy and Mainte- points that have been precisely surveyed to
nance augment the local control network. The
accuracy of these points is usually available,
In most cases the datum control points and point description and location may be
are too sparse to be sufficient for all needs in obtained from the appropriate surveying
GIS data development. For example, precise authority.
point locations may be required when setting
up a GPS receiving station, to georegister a The U.S. NGS maintains and dissemi-
scanned photograph or other imagery, or as nates a list of control points, including those
the basis for a detailed subdivision or high- points used in datum definitions and adjust-
way survey. It is unlikely there will be more ment. Point descriptions are provided in
than one or two datum points within the hardcopy and digital formats, including
work area for each of these activities. In access via the world wide web (http://
many instances a denser network of known www.ngs.noaa.gov). Stations may be found
points is desirable. The monumented, pre- based on a state and county name, a type of
cisely determined point locations that are station (horizontal or vertical), by survey
part of a datum may be used as a starting order or accuracy, date, or coordinate loca-
point for additional surveying. These smaller tion.
area surveys increase the density of precisely Figure 3-20 lists a partial description of
known points. The quality of the point loca- a control point data sheet for a station named
tions depends on the quality of the interven- Guernsey, located in Kings County, Califor-
ing survey, and a set of standards has been nia. This station was first surveyed in 1944,
established for reporting survey quality. and is a first-order horizontal and vertical
The Federal Geodetic Control Commit- control point. Only a portion of the four-
tee of the United States (FGCC) has pub- page description is shown here. Datasheets
lished a detailed set of survey accuracy for each point include the name, history,
specifications. These specifications set a physical description, directions for location,
minimum acceptable accuracy for surveys coordinates, relevant datums, and adjust-
and establish procedures and protocols to ment history.
ensure the advertised accuracy has been Lists of precisely surveyed points may
obtained. The FGCC specifications establish also be obtained from state, county, city, or
a hierarchy of accuracy. First order survey other surveyors, along with a physical
measurements are accurate to within 1 part description of the points and their location
in 100,000. This means the error of the sur- relative to nearby features. These control
vey is no larger than one unit of measure for points may then be used as starting locations
each 100,000 units of distance surveyed. The for local surveys to develop a denser net-
maximum horizontal measurement error of a work of control points, and as a basis for the
5,000 meter baseline (about 3 miles) would development of spatial data.
be no larger than 5 centimeters (about 2
 Chapter 3: Geodesy, Projections, and Coordinate Systems 87

DATABASE = Sybase, PROGRAM = datasheet, VERSION = 6.98

1 National Geodetic Survey, Retrieval Date = APRIL 11, 2004
GT0651 ***********************************************************************
GT0651 DESIGNATION - GUERNSEY
GT0651 PID - GT0651
GT0651 STATE/COUNTY- CA/KINGS
GT0651 USGS QUAD - GUERNSEY (1954)
GT0651
GT0651 *CURRENT SURVEY CONTROL
GT0651
___________________________________________________________________
GT0651* NAD 83(1992)- 36 12 41.51519(N) 119 38 24.98749(W) ADJUSTED
GT0651* NAVD 88 - 65.61 (+/-2cm) 215.3 (feet) VERTCON
GT0651
___________________________________________________________________
GT0651 EPOCH DATE - 1991.35
GT0651 LAPLACE CORR- 4.10 (seconds) DEFLEC99
GT0651 GEOID HEIGHT- -33.14 (meters) GEOID03
GT0651
GT0651 HORZ ORDER - FIRST
GT0651 VERT ORDER - FIRST CLASS II (See Below)
GT0651
GT0651.The horizontal coordinates were established by classical geodetic methods
GT0651.and adjusted by the National Geodetic Survey in March 1994.
GT0651.The horizontal coordinates are valid at the epoch date displayed above.
GT0651.The epoch date for horizontal control is a decimal equivalence
GT0651.of Year/Month/Day.
GT0651
.
.
.
.GT0651 STATION DESCRIPTION
GT0651
GT0651'DESCRIBED BY COAST AND GEODETIC SURVEY 1944 (JCS)
GT0651'STATION IS LOCATED ABOUT 8 MILES S OF HANFORD, AND 0.1 MILE
GT0651'SE OF GUERNSEY RAILROAD STATION, IN THE SE CORNER OF
GT0651'SECTION 1, T. 20 S., R. 21 E., ON LAND OWNED BY THE ATCHISON,
GT0651'TOPEKA AND SANTA FE RAILWAY.
GT0651'
GT0651'UNDERGROUND MARK IS A BRONZE STATION DISK SET IN THE TOP OF
GT0651'A CONCRETE POST.
GT0651'
GT0651'STATION MARK, STAMPED GUERNSEY 1944, IS A BRONZE STATION
GT0651'DISK SET IN A 10-INCH SQUARE CONCRETE POST PROJECTING ABOUT
GT0651'6 INCHES ABOVE THE GROUND. IT IS 31 PACES N OF THE CENTER
GT0651'LINE OF THE ROAD, 130 FEET SW OF SEMAPHORE SIGNAL, 28.2
GT0651'FEET NNW OF SIGN (NOTICE THIS IS NOT A PUBLIC HIGHWAY), 67.3
GT0651'FEET NE OF A RAIL STUCK UPRIGHT TO IDENTIFY RAILROAD
GT0651'RIGHT-OF-WAY.
GT0651'
GT0651'REFERENCE MARK 1, STAMPED GUERNSEY NO 1 1944, IS A BRONZE
GT0651'REFERENCE DISK SET IN A 10-INCH SQUARE CONCRETE POST PROJECTING
GT0651'ABOUT 8 INCHES ABOVE THE GROUND. IT IS 10 FEET S OF W END
GT0651'OF CULVERT, 42 FEET N OF THE CENTER LINE OF THE ROAD..........

Figure 3-20 A portion of a National Geodetic Survey control point data sheet.
88 GIS Fundamentals

Map Projections and Coordinate Systems

Datums tell us the latitudes and longi-
tudes of a set of points on an ellipsoid. We
need to transfer the locations of features
measured with reference to these datum
points from the curved ellipsoid to a flat
map. A map projection is a systematic ren-
dering of locations from the curved Earth
surface onto a flat map surface. Points are
“projected” from the Earth surface and onto
the map surface.
Many map projections may be envi-
sioned as a process by which a light source
shines rays from a projection center onto a
map surface (Figure 3-21). Light rays radiate
out from a source to intersect both the ellip- Figure 3-22: Distortion during map projection.
soid surface and the map surface. The rays
specify where each point on the ellipsoid
surface is to be placed on the map surface.
Other projection methods exist, for example
in some projections the source is not a single
point, however the basic process involves projected from a complexly curved Earth
the systematic transfer of points from the surface to a flat or simply curved map sur-
curved ellipsoidal surface to a flat map sur- face. Portions of the rendered Earth surface
face. must be compressed or stretched to fit onto
the map. This is illustrated in a side view of
Distortions are unavoidable when mak- a projection from an ellipsoid onto a plane
ing flat maps. This is because locations are (Figure 3-22). The map surface intersects the
Earth at two locations, I1 and I2. Points
toward the edge of the map surface, such as
D and E, are stretched apart. The scaled map
distance between d and e is greater than the
distance from D to E measured on the sur-
face of the Earth. More simply put, the dis-
tance along the line (map) is greater than the
corresponding distance along the curve
(Earth surface). Conversely, points such as A
and B that lie in between I1 and I2 would
appear compressed together. The scaled map
distance from a to b would be less than the
surface-measured distance from A to B. Dis-
tortions at I1 and I2 are zero.
Figure 3-22 demonstrates a few impor-
tant facts. Distortion may take different
forms in different portions of the map. In
one portion of the map features may be com-
Figure 3-21: A conceptual view of a map
pressed and exhibit reduced areas or dis-
projection. tances relative to Earth surface
measurements, while in another portion of
 Chapter 3: Geodesy, Projections, and Coordinate Systems 89

the map areas or distances may be expanded. points in the grid system, and the
Second, there are often a few points or lines Pythagorean formula used to calculate dis-
where distortions are zero, where length, tance between the two points. This distance
direction, or some other geometric property will be expressed in the grid coordinate sys-
is preserved. Finally, distortion is usually tem, and hence will include the projection
less near the points or lines of intersection, distortion. The distance may also be calcu-
where the map surface intersects the imagi- lated for a great circle route along the spher-
nary globe. Distortion usually increases with oid surface. This will approximate the
increasing distance from the intersection unprojected distance, measured on the sur-
points or lines. face of the Earth. This is only an approxima-
Different map projections may distort tion, as we know from the previous section,
the globe in different ways. The projection because the Earth is shaped more like an
source, represented by the point at the mid- ellipsoid, and has geoidal undulations. How-
dle of the circle in Figure 3-22, may change ever, the approximation is quite accurate,
locations. The surface onto which we are generally off by less than a few parts per
projecting may change in shape, and we may thousand over several hundred kilometers.
place the projection surface at different loca- The great circle and grid coordinate distance
tions at or near the globe. If we change any may then be compared to estimate the dis-
of these three factors, we will change how or tance distortion (Figure 3-23).
where our map is distorted. Many map projections are based on a
Formulas exist for most projections to developable surface, a geometric shape onto
calculate the magnitude of these distortions. which the Earth surface locations are pro-
These estimates of projection distortion may jected. Cones, cylinders, and planes are the
be used to select the appropriate projection most common types of developable surfaces.
or to limit the area projected such that the A plane is already flat, and cones and cylin-
needs of the mapping project are met. For ders may be mathematically “cut” and
example, the formulas may be used to ensure “unrolled” to develop a flat surface (Figure
distance distortions are smaller than a set 3-24). Projections may be characterized by
threshold. the developable surface, e.g., as conic
(cone), cylindrical (cylinder), and azimuthal
An approximation of the distance distor- (plane). The orientation of the developable
tion may be obtained for any projection by surface may also change among projections,
comparing grid coordinate distances to great e.g., the axis of a cylinder may coincide with
circle distances. A great circle distance is a the poles (equatorial) or the axis may pass
distance measured on the surface of the through the Equator (transverse).
spheroid that is contained in a plane that
intersects the spheroid or ellipsoid center. It Many map projections are not based on
may be visualized as a planar surface that developable surfaces. Projections with
intersects the two points on the Earth’s sur- names such as pseudocylindrical, Moll-
face and also splits the spheroid into two weide, sinusoidal, and Goode homolosine
equal halves (Figure 3-23). The smallest are examples. These projections often spec-
great circle distance is the shortest path ify a direct mathematical projection from an
between two points on the surface of the ellipsoid onto a flat surface. They use mathe-
spheroid. matical forms not related to cones, cylinders,
planes, or other three-dimensional figures,
As described in the previous chapter, and may change the projection surface for
most maps may be characterized by a Carte- different parts of the globe. For example,
sian grid that defines an X and Y coordinate projections such as the Goode homolosine
system. This coordinate system is usually projection may be formed by fusing other
the projected coordinate system of the map. projections, and “interrupting” segments of
Coordinates may be identified for any two the Earth to give a global view. These pro-
90 GIS Fundamentals

Figure 3-23: Example calculation of the distance distortion due to a map projection. The great circle and
grid distances are compared for two points on the Earth’s surface, the first measuring along the curved
surface, the second on the projected surface. The difference in these two measures is the distance distor-
tion due to the map projection. Calculations of the great circle distances are approximate, due to the
assumption of a spheroidal rather than ellipsoidal Earth, but are very close.
 Chapter 3: Geodesy, Projections, and Coordinate Systems 91

Figure 3-24: Projection surfaces are derived from curved “developable” surfaces that may be mathemati-
cally “unrolled” to a flat surface.

jections use complex rules and breaks to Common Map Projections in GIS
reduce distortion for many continents, and
are not based on developable surfaces. There are hundreds of map projections
used throughout the world, however most
We typically have to specify several spatial data in GIS are specified using pro-
characteristics when we specify a map pro- jections from a relatively small number of
jection. For example, for an azimuthal pro- projection types.
jection we must specify the location of the
projection center (Figure 3-25) and the loca- The Lambert conformal conic and trans-
tion and orientation of the plane onto which verse Mercator are among the most common
the globe is projected. Azimuthal projections projection types used for spatial data in
are often tangent to (just touch) the ellipsoid North America and much of the World (Fig-
at one point, and we must specify the loca- ure 3-26). Standard sets of projections have
tion of this point. A projection center been established from these two basic types.
(“light” source location) must also be speci- The Lambert conformal conic projection
fied, most often placed at one of three loca- may be conceptualized as a cone intersecting
tions. The projection center may be at the the surface of the Earth, with points on the
center of the ellipsoid (a gnomonic projec- Earth’s surface projected onto the cone. The
tion), at the antipodal surface of the ellipsoid cone in the Lambert conformal conic inter-
(diametrically opposite the tangent point, a sects the ellipsoid along two arcs, typically
stereographic projection), or at infinity (an parallels of latitude, as shown in Figure 3-26
orthographic projection). Scale factors, the (top left). These lines of intersection are
location of the origin of the coordinate sys- known as standard parallels.
tem, and other projection parameters may be Distortion in a Lambert conformal conic
required. Defining characteristics must be projection is typically smallest near the stan-
specified for all projections, e.g., we must dard parallels, where the developable sur-
specify the size and orientation of a cone in a face intersects the Earth. Distortion
conic projection, or the size, intersection increases in a complex fashion as distance to
properties, and orientation of a cylinder in a these intersection lines increases. This char-
cylindrical projection. acteristic is illustrated at the top of Figure 3-
26. Circles of a constant 5 degree radius are
92 GIS Fundamentals

Figure 3-25: As with all map projections, we must specify a suite of characteristics
when defining an azimuthal map projection. One characteristic is the location of
the projection center, most often placed at the center of the ellipsoid, or the anti-
pode, or at infinity.

drawn on the projected surface. These cir- The transverse Mercator is another com-
cles are distorted, depending on the location mon map projection. This map projection
on the projection surface. Circles nearer the may be conceptualized as enveloping the
standard parallels are less distorted. Those Earth in a horizontal cylinder, and projecting
farther away tend to be more distorted. Dis- the Earth surface onto the cylinder (Figure 3-
tortions can be quite severe, as illustrated by 26, bottom). The transverse Mercator com-
the apparent expansion of southern South monly intersects the Earth ellipsoid along a
America. single north-south tangent, or along two
Note that sets of circles in an east-west secant lines where the cylinder intersects the
row show similar distortion properties in the ellipsoid. A line parallel to and midway
Lambert conformal conic projection (Figure between the secants is often known as the
3-26, top right). Those circles that fall central meridian. The central meridian
between the standard parallels exhibit a uni- extends north and south through transverse
formly lower distortion than those in other Mercator projections.
portions of the projected map. One property As with the Lambert conformal conic,
of the Lambert conformal conic projection is the transverse Mercator projection has a
a low-distortion band running in an east- band of low distortion, but this band runs in
west direction between the standard paral- a north-south direction. Distortion is least
lels. Thus, the Lambert conformal conic pro- near the line(s) of intersection. At the bottom
jection may be used for areas which are of Figure 3-26 the intersection occurs at zero
larger in an east-west than a north-south degrees longitude, traversing western Africa,
direction, as there is little added distortion eastern Spain, and England. Distortion is
when extending the mapped area in the east- lowest near this line and increases markedly
west direction. with distance east or west away from this
 Chapter 3: Geodesy, Projections, and Coordinate Systems 93

Figure 3-26: Lambert conformal conic (LCC, top) and transverse Mercator (TM, bottom) projections.
The LCC is derived from a cone intersecting the ellipsoid along two standard parallels (top left). The
“developed” map surface is mathematically unrolled from the cone (top right). In a similar manner the
TM typically intersects the ellipsoid near a central meridian (bottom left), and the flat map comes from
“unrolling” the developable surface (bottom right). Distortion is illustrated in the developed surfaces by
the deformation of the 5-degree diameter geographic circles and by the bent shape of the latitude/longi-
tude lines.
94 GIS Fundamentals

line, e.g., the shape of South America is part in 10,000. State Plane coordinate sys-
severely distorted in the bottom right of Fig- tems are used in many types of work, includ-
ure 3-26. For this reason transverse Mercator ing property surveys, property subdivisions,
projections are often used for areas that large-scale construction projects, and photo-
extend in a north-south direction, as there is grammetric mapping, and are often adopted
little added distortion extending in that for GIS.
direction. One State Plane projection zone may
Different projection parameters may be suffice for small states. Larger states com-
used to specify an appropriate coordinate monly require several zones, each with a dif-
system for a region of interest. Specific stan- ferent projection, for each of several
dard parallels or central meridians are used geographic zones of the state. For example
to minimize distortion over a mapping area. Delaware has one State Plane coordinate
An origin location, measurement units, x zone, while California has 7, and Alaska has
and y (or northing and easting) offsets, a 10 State Plane coordinate zones, each corre-
scale factor, and other parameters may also sponding to a different projection within the
be required to define a specific projection. state. Zones are added to a state to ensure
Once a projection is defined then the coordi- acceptable projection distortion within all
nates of every point on the surface of the zones (Figure 3-27, left). Within each zone,
Earth may be determined, usually by a the distance on the curving Earth surface dif-
closed-form or approximate mathematical fers by less than one part in 10,000 from the
formula. distance on the flat projection surface. Zones
are defined by county, parish, or other
municipal boundaries. For example, the
The State Plane Coordinate Sys- Minnesota south/central zone boundary runs
tem approximately east-west through the state
The State Plane Coordinate System is a along defined county boundaries (Figure 3-
standard set of projections for the United 27, left).
States. The State Plane coordinate system The State Plane coordinate system is
specifies positions in a Cartesian coordinate based on two basic types of map projections:
system over several county to whole-state the Lambert conformal conic and the trans-
areas. There are typically one or more State verse Mercator projections. Different projec-
Plane coordinate systems defined for each tions are used for each state or sub-area
state in the United States. More than one within a state, and are chosen to minimize
State Plane zone may be required to limit distortion within a given state and zone.
distortion errors due to the map projection. Because distortion in a transverse Mercator
State Plane systems greatly facilitate increases with distance from the central
surveying, mapping, and spatial data devel- meridian, this projection type is most often
opment in a GIS, particularly when county used with states that have a long north-south
or larger areas are involved. Over relatively axis (e.g., Illinois or New Hampshire). Con-
small areas the surface of the Earth can be versely, a Lambert conformal conic projec-
assumed to be flat without introducing much tion is most often used when the long axis of
distortion. However, Earth curvature causes a state is in the east-west direction (e.g.
area, distance, and shape distortion to North Carolina and Virginia). When comput-
increase when we assume a flat surface over ing the State Plane coordinates, points are
larger areas. The State Plane system pro- projected from their geodetic latitudes and
vides a common coordinate reference for longitudes to x and y coordinates in the State
horizontal coordinates over large areas while Plane systems.
limiting error to specified maximum values. The Lambert conformal conic projection
Zones are specified in each state such that is specified in part by two standard parallels
projection distortions are kept below one that run in an east-west direction. A different
 Chapter 3: Geodesy, Projections, and Coordinate Systems 95

Figure 3-27: The State Plane zones of Minnesota, and details of the standard parallel placement for the
Minnesota central State Plane zone.

set of standard parallels is defined for each meridian defines grid north in the projection.
State Plane zone. These parallels are placed A line along the central meridian points to
at one-sixth of the zone width from the north geographic north, and specifies the Cartesian
and south limits of the zone (Figure 3-27, grid direction for the map projection. All
right). The zone projection is defined by parallels of latitude and all meridians except
specifying the standard parallels and a cen- the central meridian are curved for a trans-
tral meridian that has a longitude near the verse Mercator projection, and hence these
center of the zone. This central meridian lines do not parallel the grid x or y direc-
points in the direction of geographic north, tions.
however all other meridians converge to this Finally, note that more than one version
central meridian, so they do not point to geo- of the State Plane coordinate system has
graphic north. The Lambert conformal conic been defined. Changes were introduced with
is used to specify projections for State Plane the adoption of the North American Datum
zones for 31 states. of 1983. Prior to 1983 the State Plane pro-
The transverse Mercator projection is jections were based on the NAD27 datum.
used to define State Plane coordinates for Changes were minor in some cases, and
states or zones where the long axis is north- major in others, depending on the state and
south, such as Illinois. Distortion increases State Plane zone. Some states, such as South
in an east-west direction when using a trans- Carolina, Nebraska, and California dropped
verse Mercator. By centering the projection zones between NAD27 and NAD83 versions
on “tall”, narrow zones, we may avoid of the State Plane coordinate system (Figure
excessive distortion while requiring few 3-28). Others maintained the same number
zones. The transverse Mercator is used for of State Plane zones, but changed the projec-
22 State Plane systems (the sum of states is tion by the placement of the meridians, or by
greater than 50 because both the transverse switching to a metric coordinate system
Mercator and Lambert conformal conic are rather than one using feet, or by shifting the
used in some states, e.g., Florida). origin. State Plane zones are sometimes
As noted earlier, the transverse Merca- identified by the Federal Information Pro-
tor specifies a central meridian. This central cessing System (FIPS) codes, and most
96 GIS Fundamentals

Figure 3-28: State Plane coordinate system zones and FIPS codes for California based on the NAD27 and
NAD83 datums. Note the inclusion of zone 407 from NAD27 into zone 405 in NAD83.

codes are similar across NAD27 and NAD83 nation. The conversion was defined in the
versions of the State Plane coordinate sys- U.S. as one meter equal to exactly 39.97
tem. Care must be taken when using older inches. This yields a conversion for a U.S.
data to identify the version of the State Plane survey foot of:
coordinate system used because the FIPS
and State Plane zone designators may be the
same, but the projection parameters may 1 foot = 0.3048006096012 meters
have changed between NAD27 and NAD83.
Conversion among State Plane projec- Unfortunately, revolutionary tumult,
tions may be additionally confused by the national competition, and scientific differ-
various definitions used to translate from ences led to the eventual adoption of differ-
feet to meters. The metric system was first ent conversion factor in Europe, and most of
developed during the French Revolution in the rest of the world, for an international
the late 1700s, and adopted as the official foot of:
unit of distance in the United States, by the
initiative of Thomas Jefferson. President Jef-
ferson became enamored of the metric sys- 1 foot = 0.3048 meters
tem, both due to improved scientific
measurements based on a well-defined, inte-
grated system, and to reduce commercial
fraud and improve trade within the new
 Chapter 3: Geodesy, Projections, and Coordinate Systems 97

The U.S. definition of a foot is slightly Transverse Mercator (UTM) coordinate sys-
longer than the European definition, by tem. The State Plane system is defined only
about one part in five million. Both systems for the United States. The UTM is a global
were used in measuring distance, the U.S. coordinate system. It is widely used in the
conversion of surveys in the U.S., and the U.S.A. and other parts of North America,
international conversion of surveys or mea- and is also used in many other countries
surements elsewhere. The European conver- worldwide.
sion was adopted as the standard for all The UTM system divides the Earth into
measures under an international agreement zones that are six degrees in longitude wide,
in the 1950s. However, there was a long his- and extend from 80 degrees south latitude to
tory of the use of the U.S. conversion in U.S. 84 degrees north latitude. UTM zones are
geodetic and land surveys. Therefore, the numbered from 1 to 60 in an easterly direc-
U.S. conversion was deemed the U.S. survey tion, starting at longitude 180 degrees West
foot. This slightly longer metric to foot con- (Figure 3-29). Zones are further split north
version factor should be used for all conver- and south of the Equator. Therefore, the zone
sions among geodetic coordinate systems containing most of England is identified as
within the U.S., for example, when convert- UTM Zone 30 North, while the zones con-
ing from a State Plane coordinate system taining most of New Zealand are designated
specified in feet to one specified in meters. UTM Zones 59 South and 60 South. Direc-
tional designations are here abbreviated,
Universal Transverse Mercator e.g., 30N in place of 30 North.
Coordinate System The UTM system also defines a subzone
Another standard coordinate system has numbering system called the Military Grid
been based on the transverse Mercator pro- Reference System (MGRS). The MGRS is
jection, distinct from the State Plane system. rarely used in civilian settings, but is
This system is known as the Universal included here for completeness. The MGRS

Figure 3-29: UTM zone boundaries and zone designators. Zones are six degrees wide and numbered from 1
to 60 from the International Date Line, 180oW. Zones are also identified by their position north and south of
the Equator, e.g., Zone 7 North, Zone 16 South.
98 GIS Fundamentals

subdivides each zone into subzones that span known, standard system. Because the UTM
eight degrees of latitude. These subzones are zones are 6 degrees wide there are many
designated by a letter. Subzone squares 100 large-area projects that would not fit in one
km on a side are then designated with repeat- State Plane zone but do fit in one UTM zone
ing sequences. The MGRS combines the (Figure 3-30). Utah, Indiana, Alabama, and
zone, subzone, and grid identifiers with other states fit entirely into one UTM zone.
coordinate numbers to identify locations on Many other states are predominantly in one
the Earth’s surface. The MGRS was never UTM zone, for example, most of Colorado is
widely used outside of the military, where included in UTM zone 13 (Figure 3-30).
there was a strong need to rapidly and unam- As indicated before, all regions for an
biguously define location for continental to analysis area must be in the same coordinate
global military operations, and also provide system if they are to be analyzed together. If
some information on the precision of the not, the data will not co-occur as they
coordinate measurement. should. The large width of the UTM zones
The UTM coordinate system is common accommodates many large-area analyses,
for data and study areas spanning large and many states, national forests, or multi-
regions, e.g., several State Plane zones. county agencies have adopted the dominant
Many data from U.S. federal government UTM coordinate system as a standard.
sources are in a UTM coordinate system We must note that the UTM coordinate
because many agencies manage land span- system is not always compatible with
ning large areas, and the UTM is a well- regional analyses. Because coordinate val-

Figure 3-30: UTM zones for the lower 48 contiguous states of the United States of America. Each UTM
zone is 6 degrees wide. All zones in the Northern Hemisphere are north zones, e.g., Zone 10 North, 11
North,...19 North.
 Chapter 3: Geodesy, Projections, and Coordinate Systems 99

ues are discontinuous across UTM zone up with values increasing to the north. UTM
boundaries, analyses are difficult across south zone projections are defined with a
these boundaries. UTM zone 15 is a different false northing equal to 10,000,000 meters.
coordinate system than UTM zone 16. The This false northing may be considered the
state of Wisconsin approximately straddles northing (N) coordinate value at the Equator.
these two zones, and the state of Georgia The false northing is added to all computed
straddles zones 16 and 17. If a uniform state- values during the projection. This ensures
wide coordinate system is required, the that only non-negative coordinate values are
choice of zone is not clear, and either one observed for all northings in UTM south
zone, or another, or some compromise zone zones.
must be chosen. For example, statewide
analyses in Georgia and in Wisconsin are
often conducted using UTM-like systems
that involve projections with central meridi-
ans running through the middle of each
State, three degrees offset from the standard
UTM system.
The UTM system specifies a unique
transverse Mercator projection for each
UTM zone. Each projection is centered on
the zone central meridian and extends east
and west to cover the six degree wide region
(Figure 3-31). Distances in the UTM system
are specified in meters north and east of a
zone origin. The y values increase in a north-
erly direction (northings) and the x values
increase in an easterly direction (eastings).
The origins of the UTM coordinate sys-
tem are defined differently depending on
whether the zone is north or south of the
Equator. In either case the UTM coordinate
system is defined so that all coordinates are
positive within the zone. Zone easting coor-
dinates are all greater than zero because the
central meridian for each zone is assigned an
easting value of 500,000 meters. This effec-
tively places the origin (E = 0) at a point
500,000 meters west of the central meridian.
All zones are less than 1,000,000 meters
wide, ensuring that all eastings will be posi-
tive.
The Equator is used as the northing ori-
gin for all north zones. Thus, the Equator is
assigned a northing value of zero for north
zones. This avoids negative coordinates,
because all of the UTM north zones are
defined to be north of the Equator. However, Figure 3-31: UTM zone 11N.
this will not work for UTM south zones. A
zero value at the Equator would yield nega-
tive northings, because the projection is set
100 GIS Fundamentals

Continental and Global Projec- jection, often used in maps of the World.
tions This projection is similar to a Mercator pro-
jection, and is based on a cylinder that inter-
There are map projections that are com- sects the Earth at the Equator. Distortion
monly used when depicting maps of conti- increases towards the poles, although not as
nents, hemispheres, or regions. Just as with much as with the Mercator.
smaller areas, map projections for continen-
tal or larger areas may be selected based on Distortion in world maps may be
the distortion properties of the resultant map. reduced by using a cut or interrupted sur-
Sizeable projection distortion in area, dis- face. Different projection parameters or sur-
tance, and angle are observed in most large- faces may be specified for different parts of
area projections. Angles, distances, and the globe. Projections may be mathemati-
areas are typically not measured or com- cally constrained to be continuous across the
puted from these projections, as the differ- area mapped.
ences between the map-derived and surface- Figure 3-33 illustrates an interrupted
measured values are too great for most uses. projection in the form of a Goode homo-
Large-area maps are most often used to dis- losine. This projection is based on a sinusoi-
play or communicate data for continental or dal projection and a Mollweide projection.
global areas. These two projection types are merged at the
There are a number of projections that parallels of identical scale. The parallel of
have been or are widely used for the World. identical scale is set near the mid-northern
These include variants of the Mercator, latitude of 44o 40’ N.
Goode, Mollweide, and Miller projections, Continental projections may also be
among others. There is a trade-off that must established. Generally, the projections are
be made in global projections, between a chosen to minimize area or shape distortion
continuous map surface and distortion. If a for the region to be mapped. Lambert con-
single, uncut surface is mapped, then there is formal conic or other conic projections are
severe distortion in some portion of the map. often chosen for areas with a long east-west
Figure 3-32 shows a Miller cylindrical pro- dimension, for example when mapping the

Figure 3-32: A Miller cylindrical projection, commonly used for maps of the World. This is an example
of an uninterrupted map surface.
 Chapter 3: Geodesy, Projections, and Coordinate Systems 101

Figure 3-33: A Goode homolosine projection. This is an example of an interrupted projection,

often used to reduce some forms of distortion when displaying the entire Earth surface. (from
Snyder and Voxland, 1989)

contiguous 48 United States of America, or

North America (Figure 3-34). Standard par-
allels are placed near the top and bottom of
the continental area to reduce distortion
across the region mapped. Transverse cylin-
drical projections are often used for large
north-south continents.
None of these worldwide or continental
projections are commonly used in a GIS for
data storage or analysis. Uninterrupted coor-
dinate systems show too much distortion to
be of use in measuring most spatial quanti-
ties, and interrupted projections do not spec-
ify a Cartesian coordinate system that
defines positions for all points on the Earth
surface. Worldwide data are typically stored
in geographic coordinates (latitudes and lon-
gitudes). These data may then be projected
to a specific coordinate system for display or
document preparation.

Conversion Among Coordinate

Systems
You might ask, how do I convert Figure 3-34: A Lambert conformal conic pro-
between geographic and projected coordi- jection of North America. Standard parallels are
placed to reduce distortion within the projected
nate systems? Exact or approximate mathe- continent.
matical formulas have been developed to
102 GIS Fundamentals

convert to and from geographic (latitude and Conversion among newer datums, e.g.,
longitude) to all commonly used coordinate from early to later versions of WGS84 is
projections (Figure 3-35). These formulas often much easier and less error prone
are incorporated into “coordinate calculator” because many newer datums are based on
software packages, and are also integrated much more accurate methods, particularly
into most GIS software. For example, given the Global Positioning System, described in
a coordinate pair in the State Plane system, Chapter 5. These and other new surveying
you may calculate the corresponding geo- techniques allow the measurement of wide-
graphic coordinates. You may then apply a spread points with high accuracy, and hence
formula that converts geographic coordi- little local distortion or variation. Relatively
nates to UTM coordinates for a specific zone simple equations may be used to convert
using another set of equations. Since the locations from one datum to another, e.g., X,
backward and forward projections from geo- Y, and Z shifts, three rotation parameters,
graphic to projected coordinate systems are one for each axis, and perhaps one to a few
known, we may convert among most coordi- additional parameters, depending on the
nate systems by passing through the geo- method used.
graphic system (Figure 3-36, a). Converting between datums is often
Care must be taken when converting more complicated when older datums are
among projections that use different datums. involved. As described earlier in this chap-
We must be sure to insert a datum transfor- ter, datums are defined by both an ellipsoid
mation to convert from one geographic coor- and a set of measured points on the Earth.
dinate system to another (Figure 3-36, b). A The amount of shift between one datum and
datum transformation is a calculation of the another often varies across the globe because
change in geographic coordinates when the errors in measurements may be distrib-
moving from one datum to another. uted idiosyncratically. Measurements in one
area or period may have been particularly
accurate, while another area or time may
exhibit particularly large errors. These errors
are an historical artifact, contingent on the
unique methods and events associated with
the geodetic measurements. If they are
included in the datum adjustment, they affect
the local and global differences among
datums in their own unique way. Simple for-
mulas, such as in Figure 3-35 for projec-
tions, do not exist for datums. For most
datum transformation there are no simple
formula that give the shift in geographic
coordinates, for example from NAD27 to
NAD83. This is particularly true for older
datums developed primarily from surface
survey measurements.
While shifts among newer datums may
be purely mathematical, there are no exact
calculations of the shifts among many older
datums or among many older and newer
Figure 3-35: Formulas are known for most projec- datums. Geographic coordinates are differ-
tions that provide exact projected coordinates, if ent between these datums, but not in a regu-
the latitude and longitudes are known. This exam-
ple shows the formulas defining the Mercator pro- lar, well-behaved way. To solve this
jection. problem, we have developed empirical
 Chapter 3: Geodesy, Projections, and Coordinate Systems 103

Figure 3-36: We may project between most coordinate systems via the back (or inverse) and forward pro-
jection equations. These calculate exact geographic coordinates from projected coordinates (a), and then
new projected coordinates from the geographic coordinates. We must insert an extra step when a projection
conversion includes a datum change. A datum transformation must be used to convert from one geodetic
datum to another (b).

datum transformations that provide very pre- methods for their countries, continents, or
cise approximations for the geodetic coordi- regions. For example, in the U.S., software
nate shift at any latitude and longitude. such as NADCON have been developed by
These datum transformations account for the U.S. NGS to apply datum transforma-
changes in the origin, rotation, and scale of tions, and NADCON and the underlying
the different datums and also estimate local methods are widely and freely available.
shifts due to the local magnitude and direc- These datum transformations have been
tion of surveying errors. These datum trans- incorporated into most GIS coordinate pro-
formations allow fairly accurate estimates jection software, and are often the basis for
for the changes in latitude/longitude coordi- the datum transformation shown in Figure 3-
nates due to datum shifts. National govern- 36.
ments have developed datum transformation
104 GIS Fundamentals

The Public Land Survey System

For GIS practitioners in the United
States we must cover one final land designa-
tion system, known as the Public Land Sur-
vey System, or PLSS. The PLSS is not a
coordinate system, but PLSS points are often
used as reference points in the United States,
so the PLSS should be well understood for
work there. The PLSS is a standardized
method for designating and describing the
location of land parcels. It was used for the
initial surveys over most of the United States
after the early 1800s, therefore nearly all
land outside the original thirteen colonies
Figure 3-37: Typical layout and section numbering
uses the PLSS. An approximately uniform of a PLSS township
grid system was established across the land-
scape, with periodic adjustments incorpo-
rated to account for the anticipated error.
Parcels were designated by their location scape features. Subdivided parcels were
within this grid system. often poorly described, and hence the source
The PLSS was developed for a number of much litigation, ill will, and many ques-
of reasons. First, it was seen as a method to tionable real estate transactions.
remedy many of the shortcomings of metes The U.S. government needed a system
and bounds surveying, the most common that would provide unambiguous descrip-
method for surveying prior to the adoption tions of parcels in unsettled territories west
of the PLSS. Metes and bounds describe a and south of the original colonies. The fed-
parcel relative to features on the landscape, eral government saw public land sales as a
sometimes supplemented with angle or dis- way to generate revenue, to pay revolution-
tance measurements. In colonial times a par- ary war veterans, to expand the country, and
cel description might state “beginning at the to protect against encroachment by Euro-
joining of Shope Fork and Coweeta Creek, pean powers. Parcels could not be sold until
downstream along the creek approximately they were surveyed, hence the PLSS was
280 feet to a large rock on the right bank, created. Land surveyed under the PLSS can
thence approximately northwest 420 feet to a be found in thirty states, including Alaska
large chestnut oak blazed with an S, thence and most of the midwestern and western
west 800 feet to Mill Creek, thence down United States. Lands in the original 13 colo-
Mill Creek to Shope Fork Road, thence east nies, as well as West Virginia, Tennessee,
on Shope Fork Road to the confluence of Texas, and Kentucky were not surveyed
Shope Fork and Coweeta Creek”. under the PLSS system.
While metes and bounds descriptions The PLSS divided lands by north-south
require a minimum of surveying measure- lines running parallel to a principal merid-
ments, and these measurements might be ian. New north-south lines were surveyed at
crude, it was a less than ideal system for six mile intervals. Additional lines were sur-
describing locations or parcels. As you veyed that were perpendicular to these
might surmise, these metes and bounds north-south lines, in approximately east-
descriptions could be vague, the features in west directions, and crossing meridian lines.
the landscape might be moved or change, These east-west lines were also run at six-
and it was difficult to describe parcels when mile intervals. These lines form townships
there were few readily distinguished land- that were six miles square. Each township
 Chapter 3: Geodesy, Projections, and Coordinate Systems 105

was further subdivided into 36 sections, each areas and number of different survey parties
section approximately a mile on a side. Each involved. Rather than invite endless dispute
section was subdivided further, to quarter- and re-adjustment, the PLSS specifies that
sections (one-half mile on a side), or six- boundaries established by the appointed
teenth sections, (one-quarter mile on a side, PLSS surveyors are unchangeable, and that
commonly referred to as quarter-quarter sec- township and section corners must be
tions). Sections were numbered in a zig-zag accepted as true. The typical section con-
pattern from one to 36, beginning in the tains approximately 640 acres, but due in
northeast corner (Figure 3-37). part to errors in surveying, sections larger
Surveyors typically marked the section than 1200 acres and smaller than 20 acres
corners and quarter-corners while running were established (Figure 3-38).
survey lines. Points were marked by a num- The PLSS is important today for several
ber of methods, including stone piles, pits, reasons. First, since PLSS lines are often
blaze marks chiseled in trees, and pipes or property boundaries, they form natural corri-
posts sunk in the ground. dors in which to place roads, powerlines,
Because the primary purpose of the etc., so they are often evident on the land-
PLSS survey was to identify parcels, lines scape. Many road intersections occur at
and corner locations were considered static PLSS corner points, and these can be viewed
on completion of the survey, even if the cor- and referenced on many maps or imagery
ners were far from their intended location. used for GIS database development efforts.
Survey errors were inevitable given the large Thus the PLSS often forms a convenient sys-

Figure 3-38: Example of variation in the size and shape of PLSS sections. Most sections are approxi-
mately one mile square with section lines parallel or perpendicular to the primary meridian, as illus-
trated by the township in the upper left of this figure. However, adjustments due to different primary
meridians, different survey parties, and errors result in irregular section sizes and shapes.
106 GIS Fundamentals

tem to co-register GIS data layers. PLSS through both the Earth surface and the devel-
corners and lines are often plotted on gov- opable surface. Points on the Earth are pro-
ernment maps (e.g. 1:24,000 quads) or avail- jected along the rays and onto the
able as digital data (e.g. National developable surface. This surface is then
Cartographic Information Center Digital mathematically unrolled to form a flat map.
Line Graphs). Further, PLSS corners are Standard sets of projections are com-
sometimes re-surveyed using high precision monly used for spatial data in a GIS. In the
methods to provide cadastral control, partic- United States, the UTM and State Plane
ularly when a GIS is to be developed. Thus coordinate systems define a standard set of
these points may be useful to register spatial map projections that are widely used. Other
data layers to the Earth's surface. map projections are commonly used for con-
tinental or global maps, and for smaller
Summary maps in other regions of the World.
In order to enter coordinates in a GIS, A datum transformation is often
we need to uniquely define the location of all required when performing map projections.
points on Earth. We must develop a refer- This accounts for differences in geographic
ence frame for our coordinate system, and coordinates due to changes in the shape or
locate positions on this system. Since the origin of the spheroid, and in some cases to
Earth is a curved surface and we work with datum adjustments. Models may be analyti-
flat maps, we must somehow reconcile these cal or empirical, and are often applied as a
two views of the World. We define positions step in the map projection process when
on the globe via geodesy and surveying. We input and output projection datums differ.
convert these locations to flat surfaces via A system of land division was estab-
map projections. lished in the United States known as the
We begin by modeling the Earth’s shape Public Land Survey System (PLSS). This is
with an ellipsoid. An ellipsoid differs from not a coordinate system but rather a method
the geoid, a gravitationally-defined Earth for unambiguously and systematically defin-
surface, and these differences caused some ing parcels of land based on regularly spaced
early confusion in the adoption of standard survey lines in approximately north-south
global ellipsoids. There is a long history of and east-west directions. Intersection coordi-
ellipsoidal measurement, and we have nates have been precisely measured for
arrived at our best estimates of global and many of these survey lines, and are often
regional ellipsoids after collecting large, used as a reference grid for further surveys
painstakingly-developed sets of precise sur- or land subdivision.
face and astronomical measurements. These
measurements are combined into datums,
and these datums are used to specify the
coordinate locations of points on the surface
of the Earth.
Map projections are a systematic render-
ing of points from the curved Earth surface
onto a flat map surface. While there are
many purely mathematical or purely empiri-
cal map projections, the most common map
projections used in GIS are based on devel-
opable surfaces. Cones, cylinders, and
planes are the most common developable
surfaces. A map projection is constructed by
passing rays from a projection center
 Chapter 3: Geodesy, Projections, and Coordinate Systems 107

Suggested Reading

Bossler, J.D., Datums and geodetic systems, in Manual of Geospatial Technology, J. Bossler,
Ed., Taylor and Francis, London, 2002.

Brandenburger, A. J. and Gosh, S. K., The world’s topographic and cadastral mapping opera-
tions, Photogrammetric Engineering and Remote Sensing, 1985, 51:437-444.

Burkholder, E.F., Computation of horizontal/level distances, Journal of Surveying Engineer-

ing, 1993, 117:104-119.

Colvocoresses, A.P., The gridded map, Photogrammetric Engineering and Remote Sensing,
1997, 63:371-376.

Doyle, F.J., Map conversion and the UTM Grid, Photogrammetric Engineering and Remote
Sensing, 1997 63:367-370.

Habib, A., Coordinate transformation, in Manual of Geospatial Technology, J. Bossler, Ed.,

Taylor and Francis, London, 2002.

Iliffe, J. C., Datums and Map Projections for Remote Sensing, GIS, and Surveying, CRC
Press, Boca Raton, 2000.

Keay, J., The Great Arc, Harper Collins, London, 2000.

Leick, A., Accuracy standards for modern three-dimensional geodetic networks, Surveying
and Land Information Systems 53:111-127, 1993.

Maling, D. H., Coordinate Systems and Map Projections, George Phillip, London, 1992.

Schwartz, C.R., North American Datum of 1983, NOAA Professional Paper NOS 2, National
Geodetic Survey, Rockville, 1989.

Smith, J., Introduction to Geodesy: The History and Concepts of Modern Geodesy, Wiley,
New York, 1997.

Sobel, D., Longitude, Penguin Books, New York, 1995.

Snyder, J., Flattening the Earth: Two Thousand Years of Map Projections, University of Chi-
cago Press, Chicago, 1993.

Snyder, J. P., Map Projections, A Working Manual, USGS Professional Paper No. 1396,
United States Government Printing Office, Washington D.C., 1987.

Snyder, J.P., and Voxland, P.M., An Album of Map Projections, USGS Professional Paper No.
1453, United States Government Printing Office, Washington D.C., 1989.
108 GIS Fundamentals

Tobler, W.R., A classification of map projections, Annals of the Association of American

Geographers, 1962, 52:167-175.

Welch, R., and Homsey, Datum shifts for UTM coordinates, Photogrammetric Engineering
and Remote Sensing, 1997, 63:371-376.

Wolf, P. R., and C.D. Ghilani, Elementary Surveying, 10th Ed., Prentice-Hall, Upper Saddle
River, New Jersey, 2002.

Zilkoski, D., J. Richards, and G. Young, Results of the general adjustment of the North Amer-
ican Vertical Datum of 1988, Surveying and Land Information Systems, 1992, 53:133-149
 Chapter 3: Geodesy, Projections, and Coordinate Systems 109

Study Questions

Can you describe how Eratosthenes estimated the circumference of the Earth? What
value did he obtain?

How did the method of Eratosthenes and Posidonius differ? Were their estimates sim-
ilar, and how accurate were their estimates relative to current measurements?

What is an ellipsoid? How does an ellipse differ from a sphere? What is the equation
for the flattening factor?

Why do different ellipsoids have different radii? Can you provide three reasons?

Can you define the geoid? How does it differ from the ellipsoid, or the surface of the
Earth? How do we measure the position of the geoid?

Can you define a parallel or meridian in a geographic coordinate system? Where do

the “horizontal” and “vertical” zero lines occur?

How does magnetic north differ from the geographic North Pole?

Can you define a datum? Can you describe how datums are developed?

Why are there multiple datums, even for the same place on Earth? Can you define
what we mean when we say there is a datum shift?

What is a triangulation survey, a Bilby tower, and a benchmark?

What is a map projection? Can you define a gnomonic, stereographic, or ortho-

graphic projection?

What is a developable surface? What are the most common shapes for a developable
surface?

Can you describe the State Plane coordinate system? What type of projections are
used in a State Plane coordinate system?

Can you define and describe the Universal Transverse Mercator coordinate system?
What type of developable surface is used with a UTM projection? What are UTM
zones, where is the origin of a zone, and how are negative coordinates avoided?

What is a datum transformation? How does it differ from a map projection?

Can you describe the Public Land Survey System? Is it a coordinate system? What is
its main purpose?
110 GIS Fundamentals