Principles of Remote Sensing Using MultiAngular Information

1
Principles of Remote Sensing Using Multi-Angular
Information
Richard Kleidman SSAI, NASA Goddard Space Flight
Center David J. Diner Jet Propulsion Laboratory,
California Institute of Technology Eugene E.
Clothiaux Department of Meteorology, The
Pennsylvania State University François-Marie
Bréon Laboratoire des Sciences du Climat et de
lEnvironnement
2
Over time remote sensing instruments of
increasingly greater complexity have been
developed to allow greater exploration of the
Earth, Sun, space and other planetary
bodiesand objects.

• This presentation is one of several that examines
  different remote sensing techniques that are used
  in Earth Science remote sensing.

• Lets start by taking a brief look at the
  capabilities of a few remote sensing instruments.

3
Meteosat weather satellite first launched in 1977

The next few slides illustrate the
capabilities that some sensors have for the
detection of atmospheric aerosols.
One channel One direction
METEOSAT
Reflectance
Scattering Angle
4
AVHRR first launched in 1978 Weather and Earth
Science research applications
AVHRR
Reflectance
Two channels One direction
5
MODIS-Terra launched in 2001
Overall MODIS capabilities36 spectral channels
in one view direction.Data is used to create
over 44 different geophysical products.
How spectral information is used to create
geophysical products is discussed in
the presentation Principles of Earth Remote
Sensing Spectra
The seven primary channels used for aerosol
retrieval are illustrated here.
How spectral information is used to create
geophysical products is discussed in
the presentation Principles of Earth Remote
Sensing Spectra
Reflectance
Reflectance
Many channels One direction
Scattering Angle
Many channels One direction
Scattering Angle
6
MISR launched in 2001 Views in 9 different
directions. Each view has 4 spectral channels.
MISR

Multi-directional measurements
Reflectance
Scattering Angle
7
Principles of Remote Sensing Using Multi-Angular
Information

• Remote Sensing Approaches with Satellite Examples
• Spatial
• High resolution land studies (e.g., Landsat,
  SPOT, ASTER)
• Moderate resolution for aerosol/cloud and ocean
  studies
• Coarse resolution for IR (HIRS) and microwave
• atmospheric sounding (SSM/I, AMSR-E)
• Multispectral (with appropriate spatial
  resolution)
• Filter radiometers AVHRR, MODIS, ATSR, TOMS, OMI
  
• Hyperspectral Hyperion, SCHIAMACHY, TES, AIRS,
  IASI, CrIS
• Multi-angular (with appropriate spectral and
  spatial resolution)
• MISR, POLDER
• Polarimetry
• POLDER (CCD w/filter wheel) on ADEOS-1/2 and
  PARASOL, APS (also multidirectional, to be
  launched on GLORY)
• Active Remote Sensing
• Radar TRMM PR, CloudSat
• Lidar LITE (STS-64), GLAS (ICESat), CALIOP
  (CALIPSO)

There are several types of measurements used in
remote sensing to help us learn about the Earths
environment. This presentation attempts to
explain some of the things we can learn from
remote sensing instruments that use the technique
of multi-angular measurements. Many examples
from the MISR instrument will be used to
illustrate the types of information we can obtain
from multi-angular measurements.
passive remote sensing
8
MISR characteristics
Four spectral bands at each angle 446 nm 21 nm
558 nm 15 nm 672 nm 11 nm 866 nm 20 nm
MISR has 9 cameras pointed at different angles
relative to its flight direction. 4 Forward
pointing cameras 70.5º. 60.0º, 45.6º,
26.1º Nadir camera 4 Backward pointing
cameras 26.1º, 45.6º, 60.0º, 70.5º
Flies onboard the Terra spacecraft
Although MISR also has multispectral capabilities
we will concentrate on what we can learn from the
multi-angle observations
MISR has the potential to capture 9 images of the
same feature from different angles within a short
time span.
9
Why multi-angle?
1. Change in reflectance with angle distinguishes
different types of aerosols, and surface structure
2. Oblique slant paths through the atmosphere
enhance sensitivity to aerosols and thin cirrus
3. Stereo imaging provides geometric heights of
clouds and aerosol plumes
4. Time lapse from forward to backward views
makes it possible to use clouds as tracers of
winds aloft
5. Different angles of view enable sunglint
avoidance or accentuation
6. Integration over angle is required to estimate
hemispherical reflectance (albedo) accurately
10
Changes in scene brightness with angle
Oblique view looking at forward scattered light
MISR flight direction
11
Changes in scene brightness with angle
Less oblique view looking at backward scattered
light
MISR flight direction
12
Visualizing surface texture
Hudson and James Bays 24 February 2000
multi-spectral compositing
nadir blue band
nadir green band
nadir red band
Using multi-spectral information from a single
angle can give us some information about
surface texture.
13
Visualizing surface texture
Hudson and James Bays 24 February 2000
multi-angle compositing
stratocumulus cloud
70º forward red band
nadir red band
70º backward red band
Using multi-angle information from a single
wavelength can reveal much more information
about features in the image.
pack ice (rough)
fast ice (smooth)
14
Cloud and ice bidirectional reflectance
The large differences in the directional
reflectance of clouds, pack ice and smooth ice as
we go from forward to backward looking cameras is
what makes this distinction possible.
15
Why multi-angle?
1. Change in reflectance with angle distinguishes
different types of aerosols, and surface structure
2. Oblique slant paths through the atmosphere
enhance sensitivity to aerosols and thin cirrus
3. Stereo imaging provides geometric heights of
clouds and aerosol plumes
4. Time lapse from forward to backward views
makes it possible to use clouds as tracers of
winds aloft
5. Different angles of view enable sunglint
avoidance or accentuation
6. Integration over angle is required to estimate
hemispherical reflectance (albedo) accurately
16
Changes in geometric perspective with angle
Forward-viewing camera
MISR flight direction
cloud-top height
apparent cloud position
17
Changes in geometric perspective with angle
Backward-viewing camera
MISR flight direction
cloud-top height
parallax
18
MISR Stereo Cloud Heights from Parallax

• Cycling through the nine views of this scene of
  clouds over Florida, notice the displacement of
  the clouds
• This is due to a geometric effect called
  parallax, and not true motion

19
Multi-angle flyover Florida and Cuba
6 March 2000
The degree of movement of the cloud layers
relative to the ground and each other with the
changing camera angle reveals the relative
heights of the cloud layers
20

• Additional details and examples about cloud
  height retrieval using multi-angular observations.

21
Stereo height retrieval

• Uses 275-m resolution red-band (672 nm) data
• Operational product uses 0º and 26º cameras
• Heights are generated on 1.1-km grid by pattern
  matching of image patches
• Purely geometric approach
• No reliance on temperature profile, cloud
  emissivity, or radiometric calibration
• Operational matcher designed to be
  computationally fast and autonomous
• Disparities calculated in integer pixels
• Height precision 500 m

22
Hurricane Jeanne, 24 Sept. 2004
23
Typhoon Sinlaku, 5 Sept. 2002
24
Cloud height parallax retrieval using additional
information from cloud reflection in the water.
Less oblique MISR camera
MISR flight direction
apparent cloud position
reflection position
25
Cloud reflection in water
Very oblique MISR camera
MISR flight direction
apparent cloud position
reflection position
26
Cloud reflection in the water and cloud height
parallax
Georgian Bay, Ontario, 6 March 2000
Nadir (An)
70º forward (Df)
27
Cloud reflection in the water and cloud height
parallax
Nadir (An)
60º forward (Cf)
28
Cloud reflection in the water and cloud height
parallax
Nadir (An)
46º forward (Bf)
29
Cloud reflection in the water and cloud height
parallax
Nadir (An)
26º forward (Af)
30
Why multi-angle?
1. Change in reflectance with angle distinguishes
different types of aerosols, and surface structure
2. Oblique slant paths through the atmosphere
enhance sensitivity to aerosols and thin cirrus
3. Stereo imaging provides geometric heights of
clouds and aerosol plumes
4. Time lapse from forward to backward views
makes it possible to use clouds as tracers of
winds aloft
5. Different angles of view enable sunglint
avoidance or accentuation
6. Integration over angle is required to estimate
hemispherical reflectance (albedo) accurately
31
Capturing the time dimension
This is not only important for determining winds
aloft but also to determine stereo cloud height
Time step 1
apparent cloud position
32
Capturing the time dimension
Time step 2
apparent cloud position
33
Stereo wind retrieval - how the method works
2. Horizontal displacement of each feature is
determined displacements result from two factors
stereo parallax due to height and (wind)
advection during the time interval between
views 3. The displacements are converted to
heights and wind vectors at least 3 look angles
are needed to separate parallax and advection
effects observation from satellite altitude is
required, using Earth curvature to overcome
equation degeneracy 4. Multiple features are
analyzed to find predominant wind on 70.4-km grid
flight direction
34
MISR extra-tropical cyclone heights and winds
S. hemisphere
N. hemisphere
35
Tropical Storms and Hurricanes
Katrina, 30 August 2005
Daniel, 25 July 2006
height
36
Climatological wind fields vs. height
Fall 2005 Winds 0.5 - 1.0 km Winds 11 - 13
km
37
Indian coast Godavari River Delta Approx. 16.4ºN,
81.8ºE 26 December 2004
86 km
49 km
MISR 60º fwd - 70º aft 0511 - 0517 UTC cloud
motion is due to parallax resulting from their
height above the surface tsunami waves are at
sea level and show actual motion
10 km
38
Von Karman vortex street near Jan Mayen Island
39
Why multi-angle?
1. Change in reflectance with angle distinguishes
different types of aerosols, and surface structure
Lets go back and examine how multi-angle
observations allow us to determine something
about aerosol types.
40
Aerosol particles scatter incoming solar
radiation in characteristic ways based on their
size, shape and chemical properties
If we plot the amount of light scattering in each
direction we create a Scattering Phase Function
41
Below are shown some sample scattering phase
functions for different types of aerosols.
Instrument which can only make observations at a
single angle (such as MODIS) cannot determine the
scattering phase function. MISR has enough
observational angles to determine aerosol phase
functions.
42
MISR sensitivity to aerosol particle properties
O. Kalashnikova et al. (2005), JGR
43
AERONET The ground based satellite
Aeronet is a worldwide network of
ground instruments that measure aerosols and
has been indispensible in the study of aerosols
and in satellite measurement validation.
44
Aerosol Robotic Network http//aeronet.gsfc.nas.go
v

• PI Brent Holben, NASA GSFC

45
Aeronet instruments make two types of
measurements
Diffuse sky The diffuse sky measurements record
information from many angles as the instrument
sweeps around the sky at discrete azimuth angle
along the suns principle plane (constant
?0) Uses RT (LUT) to invert spectral/angular
radiance to retrieve aerosol size, shape and
composition.
Diffuse Sky

• Direct sun
• Derives spectral aerosol optical depth (AOD)

46

• A significant number of slides in this
  presentation are from
• MISR overview and observational principles
• which can be found at
• http//eosweb.larc.nasa.gov/PRODOCS/misr/table_mis
  r.html
• By David J. Diner
• Jet Propulsion Laboratory, California Institute
  of Technology
• Eugene E. Clothiaux
• Department of Meteorology, The Pennsylvania State
  University

Also contributed by François-Marie
Bréon Laboratoire des Sciences du Climat et de
lEnvironnement