GEOG1022

Lecture 9: Remote sensing applications in global change

Peng Zhu, Department of Geography

Room 1005, Jockey Club Tower,
Centennial Campus
Email: [email protected]

1
Previous lecture review
 Cities are both contributors and
victims of climate change

Cities Cities are

CONTRIBUTE VULNERABLE to
over 70% of the the EFFECTS of
GHG emissions climate change
causing climate
change
How Do Heat Islands Form?
 UHI Effects

• Increased energy consumption; augment

the expenditure of the people and
government as well.

• Elevated emissions of air pollutants and

greenhouse gases
Course learning outcomes
• Comprehend basics of remote sensing
• Explain how remote sensing is utilized in global
change research
• Understand the advantage and disadvantage of
remote sensing in global change research

6
Essential Climate Variables
Climate observation system based on sensors

Climate observation
-the foundation of our understanding of the climate
system (Overpeck, 2011, Science)
Precision Sea Level Monitoring

Annual global sea level rise trend from

1993–2012 derived from satellite altimetry
Global surface water dynamics

https://glad.earthengine.app/view/surface-water-dynamics
 Groundwater Measurements
Using Gravitational Instruments
 Glaciers and Ice Sheets

• Change in Greenland ice sheet thickness per year

 Global Atmospheric
Composition Monitoring

Atmospheric carbon dioxide concentrations from 1 October 2014

through 11 November 2014, as recorded by NASA’s Orbiting
Carbon Observatory (OCO-2).
What is Remote Sensing

Remote sensing was formally defined by the

American Society for Photogrammetry and Remote
Sensing (ASPRS) as:
The measurement or acquisition of information of
some property of an object or phenomenon, by a
recording device that is not in physical or intimate
contact with the object or phenomenon under study
(Colwell, 1983).
What is Remote Sensing

Acquiring information about a subject with no

physical contact
through the use of electromagnetic radiation
(visible light, laser, radio wave…)

Two remote sensing systems you have all used

before:
Cameras
Your eyeballs!
Advantage of Remote Sensing

Global observation coverage

• systematic data collection for the entire planet and
storing data in long-term observation archives.
Spatially and temporally consistent observations
• basis of large scale synoptic observations and
change detection
Real-time observation
• allowing sensors to be targeted at any point on Earth
Multi-temporal and multi-scale observations
• form complimentary data sources for various
applications
Knowledge Points

Information acquisition mechanisms:

Sensor: Active vs Passive
Concept of resolution(s):
Spectral/Spatial/Temporal
•Passive sensors record electromagnetic radiation that is
reflected or emitted from the land surface
•Active sensors such as microwave (RADAR) and LIDAR
(laser signal) have their own source of light or illumination. In
particular, they actively send a pulse and measure the
backscatter reflected to the sensor.
Active sensors
•Light detection and ranging (LIDAR)
•Laser light
•Radio detection and ranging (RADAR)
• Radio waves

•Sound navigation ranging (SONAR)

• Sound propagation
 Concept of resolution

A measure of sensor ability to distinguish between signals

•Spectral resolution
The locations and widths of “bands”
•Spatial resolution
Size of smallest image element (pixel)
•Temporal resolution
How often an image can be collected
Spectral resolution

•Spectral resolution is the number and dimension

(size) of specific wavelength intervals (referred to
as bands or channels) in the electromagnetic
spectrum to which a remote sensing instrument is
sensitive.
•Multispectral remote sensing systems record
energy in multiple bands of the electromagnetic
spectrum.
Hyperspectral remote sensing
•A hyperspectral remote sensing instrument typically
acquires data in hundreds of spectral bands
Spatial resolution
Temporal resolution
•The temporal resolution of a remote sensing
system generally refers to how often the sensor
records imagery of a particular area.
•Ideally, the sensor obtains data repetitively to
capture unique discriminating characteristics of the
object under investigation (e.g., critical dates in the
phenological cycle) LAI

Time
 The problem with clouds…

Landsat 8 OLI
10 Jan 2016

26
 Microwave Radiation

10GHz 1GHz
Frequency & wavelengths for
microwave bands
0.2μm 1.0μm 10μm 1mm 10mm 10cm 1m
Wavelength Frequency
Middle Thermal Microwave Bands Band
Visible
infra-red infra-red range[mm] range[GHz]
UV KaKuX C S L P Ka 7.5 - 11.0 40.0 - 26.5
Near infra-red
K 11.0 - 16.7 26.5 - 18.0
The electromagnetic spectrum (after NASA/JPL, 1988) Ku 16.7 - 24.0 18.0 - 12.5
X 24.0 - 37.5 12.5 - 8.0
Transmittance (%)

100 C 37.5 - 75.0 8.0 - 4.0

S 75.0 - 150 4.0 - 2.0
50
L 150 - 300 2.0 - 1.0
P 300 - 1000 1.0 - 0.3
0 10μm
0.2 μm 1.0μm 1mm 10nm 10cm 1
m
Wavelength
X, C, S, L bands are used for
Characteristic of atmospheric spectral transmittance (after NASA/JPL, 1988) SAR on-board satellites

27
Surface scatter (Reflection) mechanism

forest grass lake

©METI/JAXA
Backscatter Change in A Paddy Field

Open water Planting Vegetated

weak strong
HH backscattering
29
 Seasonal Change
Seasonal changes in microwave backscatter provide useful information to detect
paddy field area.

SAR Backscatter
Low High
Ecological Applications

◾Land cover change: Example-Forest

cover/deforestation
◾Ecosystem production (GPP, NPP)
◾Ecosystem phenology
◾Ecosystem disturbance: drought, fire etc.
◾Agro-ecosystem: management, precision
agriculture
Unleash power of satellite data

32
Land cover change: Example-
Forest cover/deforestation

1 Unclassified
2 Alfalfa
3 Bare soil
4 Bare soil with wheat stubb.
4 Bare soil with green veg.
5 Corn
6 Legume
7 Outcrops
8 Pasture grazed
9 Pasture ungrazed
10Quarries and sand bars
11 Shrubs
12Trees
13Urban
14Water
15Wheat stubble
Land cover change: Example-
Forest cover/deforestation
Land cover change: Example-
Forest cover/deforestation

(Hansen et al., Science, 2013)

 Classification
• Supervised classification
– Target outputs are provided along with input data

• Unsupervised classification
– Discover patterns in input data for the purposes of
• clustering,
• density estimation,
• dimensionality reduction,
• visualization
36
 Unsupervised
• Goals
– to discover groups of similar
examples within the data, where
it is called clustering

– to determine the distribution of

data within the input space,
known as density estimation

– to project the data from a high-

dimensional space down to two
or three dimensions for the
purpose of visualization

37
Supervised vs unsupervised

38
 Supervised classification:
essential steps
• Class definition
– General land cover classes: impervious, cropland, grassland, forest, water, etc.
– Specific crops: wheat, corn, soybean, alfalfa, etc.

• Features and algorithm selection

• Generation of training (ground truth) data

– In situ (ground) measurements
– Drones
– Photo-interpretation

• Accuracy assessment

39
 Features for classification
• Spectral signature
Spectral profiles

40
Lillesand and Kiefer
Figure 7-39
 Lillesand and Kiefer
Figure 7-11

When the differences between features of interest are high,

it is possible to use a simple threshold to discriminate between
the features (water vs. land surface)
 Water The range in digital values for
these two surfaces do not
overlap, so you can use a level
slice to classify your image into
two categories

> 40 = Land
< 40 = water
Land

Lillesand and Kiefer

Figure 7-11
Global surface water dynamics

https://glad.earthengine.app/view/surface-water-dynamics
Dynamic World on GEE

https://www.dynamicworld.app/

45
 Vegetation Indices: Normalized
Difference Vegetation Index (NDVI)
• Make use of the red vs. NIR
reflectance differences for
green vegetation
• Veg indices are associated
with canopy characteristics
such as biomass, leaf area
index and percentage of
vegetation cover
 Vegetation Indices: Normalized
Difference Vegetation Index (NDVI)

NDVI = NIR − red

NIR + red

red = Reflectance in red channel

NIR = Reflectance in NIR channel
Healthy, dense vegetation has high NDVI
Stressed, or sparse vegetation produces lower NDVI
Bare rock, soil have NDVI near zero Snow
produces negative values of NDVI
Clouds produce low to negative values of NDVI
NDVI as an indicator of drought
Driver of Earth Greening
Ecosystem production
 2 Observation of the terrestrial ecosystem
Ecosystem phenology
Phenological shifts
• Start of season (SOS) method
– Find time breaks on Normalized difference vegetation
index(NDVI) curves: first upturn and midpoint
Ecosystem disturbance:
drought, fire
https://www.nasa.gov/topics/earth/features/plant-stress.html
Ecosystem disturbance:
drought, fire
Ecosystem disturbance:
drought, fire

https://youtu.be/iUZcoUvtHkY
 Ecosystem disturbance:
drought, fire

(188) GOES Satellites Wildfire Detection and Monitoring - YouTube

Monitoring agroecosystem
 Monitoring agricultural
production with satellite
 Deriving agricultural management features

⚫ Irrigation timing, intensity and distribution

• Soil moisture or vegetation index
• Land surface temperature
LST detected irrigation intensity
Anomalous soil moisture informed irrigation event through LST cooling effect

Source: USDA
High irrigation intensity

High cooling effect

Precision management in
agroecosystem
◾Use thermal images to infer plant water stress
◾Theory: Net Radiation=Latent Heat + Sensible Heat
▪ Latent Heat =λET, λ: late heat of vaporization of water (44kJ/ mol)
▪ ET: Evapotranspiration
▪ Sensible Heat = f (canopy temperature, air temperature)
Precision management in
agroecosystem

Variable rate nitrogen technology

Crop field across the world can be very
heterogeneous
If you apply N uniformly
Under-fertilizing: short of N for crop growth NDVI
Over-fertilizing: high risks of leaching into
the groundwater

Variable Rate Nitrogen (VRN) is to apply

N fertilizer at different rates in a field to
meet site-specific demand 60
Variable rate nitrogen application
• Maize with insufficient N
will have dwarfed
seedlings and yellow
leaves, leading to low
carbon assimilation and
yield.

• Over-fertilization causes
high risks of water
contamination and nitrous
oxide emissions.

• Variable Rate Nitrogen

(VRN) is to apply N
fertilizer at different rates
in a field to meet site-
specific demand
Variable rate nitrogen application

SSURGO Land Unit RapidEye zone

Nightlight Monitoring

https://www.youtube.com/watch?v=8dc58ZrOuck
Nightlight Monitoring and
Economic applications

(188) Using the Right Night Lights Data in

Applied Economics - YouTube
Limitations
Short data spans of satellite data
Time length of available observations

Time length (year) Terrestrial ECV

0~9 Biomass, Glacier and ice
caps
10~19 Land cover, Albedo, fAPAR,
Fire disturbance

20~29 Lakes, LAI

30~39 Soil moisture

40~49 Snow cover

Limitations
Biases associated with instrument
▪ Inadequate spatial resolution and
temporal frequency
▪ Poor calibrations
▪ Merging data from different systems

Uncertainties in retrieval algorithms

▪ Radiative transfer models
▪ Uncertainties in common inputs