MISR/MODIS intercomparisons and data fusion
MODIS Science Team Meeting23 March 2005
David J. Diner, MISR Principal Investigator
Jet Propulsion Laboratory,California Institute of Technology
and the MISR team
MISR characteristics
9 view angles at Earth surface:70.5º forward to 70.5º aftward
Four spectral bands at each angle:446 nm ± 21 nm 558 nm ± 15 nm672 nm ± 11 nm866 nm ± 20 nm
275 m sampling in all nadir bands and red band of off-nadir cameras1.1 km for the other channels
400-km swath: Complete zonal coverage 9 days at equator2 days at poles
7 minutes to observe each sceneat all 9 angles
Digitization: 14 bits
Standard products are processed andarchived at the NASA LangleyAtmospheric Sciences Data Center eosweb.larc.nasa.gov
Radiometry
MISR On-Board Calibrator
AirMISR
MISR lunarimages
Railroad Valley
LunarLake Vicarious calibrations and validations
over desert playas and dark water sites
MISR calibration
A comprehensive review of thesedata sources in 2003-2004 led todownward revision of the MISRradiances by 3% in the red and 2% in the NIR
MISR ocean aerosol optical depths prior to 2004 calibration revision
MISR-AERONET difference was ~0.05 at 558 nm
Similar differences were seen with airborne sunphotometer (AATS-14) data and with MODIS
MO
DIS
AO
DM
OD
IS A
OD
MISR AODR. Kahn, W-H. Li, W. Abdou, L. Remer
Effect of recent product revisions on retrieved MISR AOD over water
AOD, old calibration
AO
D,
ne
w -
old
cal
ibra
tio
n
-0.25
-0.20
-0.15
-0.10
-0.05
0.00
0.05
0.10
0.15
0.20
0.25
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
AOD, 24 mixturesAOD, old dust models
AO
D,
ne
w -
old
du
st m
od
els
Calibration revisions New dust models
R. Kahn, O. Kalashnikova, D. Diner
Initial MISR-AERONET-AATS 14 comparisons suggest that the MISR ocean AOD bias has been eliminated
Medium-sized weakly-absorbing (1% hematite) grain-shaped dust models provide the best fit to MISR radiances over dust plumes
Comparison of revised radiometric scale with5 years of MISR vicarious calibration history
C. Bruegge, W. Abdou
An camera vicarious calibration results
-5.0
-4.0
-3.0
-2.0
-1.0
0.0
1.0
2.0
3.0
4.0
5.0
Blue Green Red NIR
06-Jun-2000: LL
30-Jun-2001: RRV
10-Jun-2002: Ivan
07-Jul-2003: BRD
22-Jul-2003: RRV
22-Jun-2004: RRV
10-Jun-2002: RRV
13-Jun-2003: RRV
29-Jun-2004: RRV
Mean
center of field of viewedge of field of view
MISR revised calibration compared to MODIS and other sensors
C. Bruegge, W. Abdou
Railroad Valley22-July-2003
85
95
105
115
125
135
145
155
165
400 500 600 700 800 900
Wavelength (nm)
Radiance (W m-2 sr-1 um-1)
MISR
AirMISR
MERIS
MODIS
Landsat
VC MISR
MISR/MODIS radiance intercomparison in the Arctic
Arctic Ocean/Greenland/Baffin Bay/Baffin IslandTerra path 26—15 orbits from 2002
Red band
E. Clothiaux
MODIS image
Details of MISR/MODIS radiance intercomparison
MISR-MODIS difference (%)as function of radiance
MISR-MODIS difference (%)as function of location
MISR: 672 nmMODIS: 645 nmNot adjusted for spectral shift, ozone
E. Clothiaux
Terminator:long solar slant path
Dark, openwater
3% difference
---
Aerosols and cirrus
MISR/MODIS standard product complementarity:MISR retrievals over brightsurfaces
R. Kahn, J. Martonchik, W. Abdou, B. Gaitley, D. Diner
Scattering angle (deg)
Ph
ase
fu
nct
ion
O. Kalashnikova, R. Kahn, D. Diner
Synergistic sensitivities to aerosol particle properties
MODIS: broad spectral coverageenhances sensitivity to size, particularly coarse mode
MISR: multiangle dataprovides sensitivity to shape
MISR/MODIS cirrus case study at ARM Oklahoma site
MISR nadir view image
Millimeter wave radar reflectivity
Ice scattering properties from Yang et al. (2000)
S. McFarlane, R. Marchand, T. Ackerman
MISR is sensitive to particle shapeMODIS is sensitive to particle size
Ice scattering properties from Yang et al. (2000)
Cloud structure, heights, and detection
Morphological diversity of marine stratocumulus clouds
R. Davies
A. Horvath, R. Davies
Multiangle data provides a physical consistency check on MODIS 1-D cloud retrieval assumption
Cloud morphology, along with cloud microphysics, plays a major role in determining TOA bidirectional reflectance
Multiangle tests of cloud homogeneity
1-D theory fits MISR observations
1-D theory does not fit MISR observations
A. Horvath, R. Davies
Application to retrievals of cloud liquid water: MODIS/MISR/TMI fusion
Reasonable consistency (±25%) is obtained between MODIS and TRMM Microwave Imager (TMI) LWP for shallow, boundary layer clouds.
MISR is used as a check on the applicability of 1-D retrieval theory.
MODIS/TMI consistency is much poorer for deep clouds.
MISR 60º image of deep cloud
-90 -60 -30 0 30 60 90
40
60
80
100
120
140
view zenith angle (degree)
boundary layer clouds
MISR allTMI all TMI all +/- 15%
MISR overcast TMI overcast TMI overcast +/- 15%
TMI LWP (g/m2)
MO
DIS
LW
P (
g/m
2 )
boundary layerclouds
nadir only nadir+60°
cloud vertical extent no information 10.5 ± 0.8 km
extinction coefficient no information8 - 22 km-1
(higher at base)
cloud optical depth > 60 150 ± 30
Oblique views make it possible to scan down the sides of deep clouds:
MISR stereo retrievals of cloud top heights (CTH)
Trade wind cumuli: high sensitivity
Typhoon Sinlaku
MISR cloud heights --independent of radiometric calibration, atmospheric temperature profiles, cloud emissivity
0 2.5 5 km
ARM BarrowARM NauruARM SGP
Comparison of MISR stereo heights with radar/lidar
Clear-sky or opaque cloud topBroken boundary layer cloudsSemi-opaque cirrusThin cloud over other clouds (MISR determines the height of the lower cloud tops)Thin cloud over surface
MISR stereo heightRad
ar/l
idar
max
imu
m h
eig
ht
MISR stereo heightMISR stereo height
R. Marchand, T. Ackerman
MISR/MODIS CTH comparisons
MISR and MODIS in good agreement for single-level mid-high clouds (accuracy ~500 m)
MISR performs well for low clouds, MODIS 11 m brightness temperatures problematic
Multi-layer situations problematic for MISR operational (near-nadir) stereo, altitude detected depends on high cloud optical depth
Typically MODIS CTH > MISR CTH (difference ~600 m)C. Naud, J-P. Muller, E. Clothiaux
ChilboltonARM SGP
Discriminating polar clouds from snow and ice with MISR
L. Di Girolamo, M. Wilson, C.Moroney
-0.5 0.5 1.5 2.5BDAS
0.0 0.2 0.4 0.6 0.8Red BRF (nadir)
Red band bidirectional reflectance factor
Band-differencedangular signature
Antarctica, 2 April 2001
Angular reflectance signature
Stereoscopically-derivedcloud-top height
Bidirectional reflectance factor
Arctic, near Queen Elizabeth Islands, 28 August 2000
Stereo
Strength of MISR/MODIS fusion for polar clouds
Overall classification accuracies relative to “expert” labels
MODIS: 86% [detects the highest cloud, whether optically thick or thin]
MISR stereo only: 75% [locates the first (from space) optically thicksurface; thin clouds problematic for near-nadir stereo]
MISR stereo + angular signatures: 92%
MISR + MODIS: 95-97%
Terra CERES-MISR-MODIS fusion Capitalizes on excellent separation of clouds from ice for understanding polar radiation budget
CALIPSO and CloudSat synergy can be used to quantify cloud optical depths giving rise to MODIS high cloud/MISR low cloud detections
T. Shi, B. Yu, A. Braverman, D. Groff, E. Clothiaux
Arctic Ocean/Greenland/Baffin Bay/Baffin IslandTerra path 26—15 orbits from daylight season of 2002
Fires and smoke
TOMS Aerosol Index (AI)Direct height retrieval ofUT/LS smoke from theChisholm fire, 29 May 2001
MISR
QuickTime™ and aTIFF (LZW) decompressor
are needed to see this picture.
MODIS
M. Fromm, D. Diner, C. Moroney, C. Averill
near-nadirstereo
obliquestereo
MODIS fire/MISR stereo fusion for wildfire studies
Southern California, 26 October 2003
MODIS:Thermal channelspinpoints fire locations
MISR:Oblique views enhanceplume sensitivity
Stereo retrieves plume heights
Provides smoke injectiondata for CTMs
Vegetation
Example of MISR surface directional reflectance retrieval
J. Martonchik
MISR/MODIS land surfaceproduct intercomparisons
Land surface albedo(B. Pinty, S. Liang, H. Fang)
0 30 60 90 120 150 180 210 240 270 300 330 3600
1
2
3
LAI
Days (Year 2000)MODIS MISR days 1 2 3 4 5 6 7 8
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
SALSA San Pedro (Shrubland/Woodland)
0 30 60 90 120 150 180 210 240 270 300 330 3600.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
FPAR
Days (Year 2000)MODIS MISR days 1 2 3 4 5 6 7 8
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
SALSA San Pedro (Shrubland/Woodland)
LAI/FPAR over sparsely vegetated land (shrubs account for 25% of global vegetation) (R. Myneni, Y. Knyazikhin, J. Hu)
Bell-shaped BRF:Tree crowns ofmedium-high densityagainst bright background
Bowl-shaped BRF:Sparse vegetationand dense,closed canopies
B. Pinty, N. Gobron, J-L. Widlowski, M. Verstraete
MISR/MODIS synergy/fusion over land surfaces
MISR along-track multiangle data provide independent structural information
MISR/MODIS fusion studies:B. Braswell et al.: Amazon deforestation and regrowthM. Chopping et al.: Woody biomass incursion into arid grasslands
Map: Australian Surveying and Land Information Group
Conclusions
Simultaneous broad spectral coverage from MODIS and wide angular coverage from MISR makes a uniquely valuable combination:
Having independent retrievals of related parameters from each sensor using different methodologies is a key element of a robust observing system.
Data fusion capitalizes on complementary sensitivities to aerosol, cloud, and surface properties.
MISR/MODIS data fusion is currently being done in research mode
Products using joint retrievals can be considered—AerosolsPolar cloudsFiresLand surface
Some of these were described in the Terra extension proposal (Senior Review).
Joint data analysis tool development would also be a great benefit to the scientific community.