REMOTE SENSING OF NATURAL RESOURCES

1
Preprocessing RS Data And Ground Truth
2
Preprocessing

• Many corrections that need to be made before RS
  can be analyzed.
• Preprocessing RS data is a function of the
  application and sensor type
• Satellite-Airborne-Handheld
• Imaging Non Imaging
• Some satellite data is delivered ready to go.

3
Preprocessing

• Preprocessing can be extensive
• Atmospheric Mixed Pixels
• Radiometric Georegistration
• Elevation Off-Nadir

4
The Digital Value

• So what shape is the raw data in?
• Essentially the raw data represents the amount
  of energy received by the sensor in any one
  band. This is only radiant energy and referred
  to as a digital value (DV).

An electron will move into the conduction band
with applied energy. The amount of energy in the
conduction band is recorded as a DV.
e-
e-
e-
e-
e-
5
The Atmosphere

• The raw DV is often converted to energy or
  percent reflectance.
• It is an important because atmospheric
  conditions can significantly affect the amount
  of energy received at the sensor.
• Corrections are done in many ways
• Atmospheric Models Image Based Algorithms
• Reflectance Calculations

6
The Atmosphere

• ATMOSPHERIC MODELING
• Attempts to model atmospheric scattering and
  converts DV into energy values
• Often used on satellite and highly calibrated
  airborne systems.
• Requires
• Altitude Elevation Aerosol Distribution
• Visibility Water Vapor Sensor
  Configurations

7
The Atmosphere
REFLECTANCE CALCULATIONS Reflectance
Radiance/Irradiance
8
The Atmosphere
REFLECTANCE CALCULATIONS In airborne systems,
reflectance may be estimated using ground control
points of known reflectance Painted
Tarps Asphalt Concrete
9
The Atmosphere
REFLECTANCE CALCULATIONS Handheld Units are
slightly different
10
The Atmosphere
Image Based Algorithms Designed to calculate the
average reflectance in an image. The average
reflectance is assumed to approximate a flat
field in each band. Each pixel in the image is
divided by the average reflectance.
11
The Atmosphere
Image Based Algorithms Just like the vegetation
ratios, the ratio between pixel reflectance and
average scene reflectance will reduce atmospheric
effects. Caution Use only over large areas
with representative amounts of water, urban,
vegetation.
12
Off- Nadir

• Nadir is the center most point directly beneath
  the sensor.
• Because a camera lens has a fish-eye effect,
  the farther from nadir you go the pixels become
  distorted.
• Off-Nadir affects are particularly important
  with airborne images.

13
Pixels at Nadir (pt.1) 1 m2 Pt. 2 1.02
m2 Pt. 2 1.05 m2 Pt. 2 1.22 m2
1524 m
X1
X2
X3
1
2
3
4
14
Radiometric Corrections

• Sometimes satellite systems are radiometrically
  calibrated.
• This means each DV can be converted to an
  energy value.

15
Radiometric Corrections

• These are NOT atmospherically corrected! The
  energy values calculate energy at the sensor
  only. They are not calibrated for atmospheric
  attenuation/contribution.

16
Georegistration

• For many applications it will be important to
  locate the image on the surface of the earth.
• Georegistration is simply assigned coordinates
  to an image.

17
Georegistration
Registered DOQQ
Unregistered RS Image
18
Georegistration
Registered DOQQ
Unregistered RS Image
19
Georegistration
The computer resamples the image. Coordinates
are assigned coordinates based on selected points
in each reference image. GIGO Applies!
Geo registered RS Image
20
Elevation Corrections

• Ground elevation can cause distortions in an
  image.
• Digital Elevation Models can be used to account
  for this distortion
• The resulting image is referred to as a digital
  ortho quadrangle.

21
Elevation Corrections

• Digital Elevation Models (DEM) can also be
  applied to imagery to show terrain elevation.

22
Elevation Corrections
23
Mixed Pixels

• A mixed pixel is a pixel composed of more than
  one attribute.
• It is highly dependent on spatial resolution.

24
Mixed Pixels

• How do you handle mixed pixels?

• Vegetation Indices

• Ground Truth

25
Ground Truthing

• How do we know whats on the ground?
• How accurate are our interpretations of the RS
  data?
• How are other site-specific factors affecting
  those interpretations?

26
Ground Truthing

• Examples
• Ground Control Points for the
  Georegistration Windshield Surveys
• Yield
• Soil Samples

27
Ground Truthing

• This is a critical component in designing an RS
  acquisition. You must consider what kinds of
  data will be necessary beforehand.
• Ground Truthing will be a function of
• Total Coverage Area
• Frequency
• The Question Asked

28
Ground Truthing

• Example
• Fire ants are a nuisance and a pest. The GA EPD
  wants to know where fire ants are likely to be
  and target pest control efforts accordingly.
  They would like to use satellite imagery as a
  tool and have asked for your help. What kinds of
  ground truth would be useful?
• Hint We cannot actually detect the mound
  itself.

29
Ground Truthing

• Mounds will not always be evident, but are
  usually found in open areas such as lawns,
  pastures, along roadsides and unused cropland.
  Mounds are rarely found in frequently cultivated
  areas. This species could easily be confused with
  the common coastal brown ant and as well as
  some local native ants.

30
Ground Truthing

• Using modelling and satellite imagery to assist
  fire ant eradication
• The Queensland Government is now employing
  satellite imagery to find areas of land
  unsuitable for fire ants in a bid to eradicate
  fire ants in the State's south-east and protect
  the rest of Australia from this dangerous exotic
  pest.
• The Fire Ant Control Centre has been working in
  collaboration with a postgraduate researcher from
  the University of Queensland, Robert George, to
  develop a model to predict preferred habitats and
  likely spread-patterns of fire ants. Using
  satellite imagery, this research has been able to
  identify the land types not suitable to fire ants
  in what is called the Fire Ant Habitat
  Identification System.