Validation and Refinement of Modis Aerosol Optical Depth Product over Coastal Urban Areas
Post on 05-Jan-2016
27 Views
Preview:
DESCRIPTION
Transcript
NOAA/NESDIS Cooperative Research Program Second Annual Science Symposium
SATELLITE CALIBRATION & VALIDATION July 13-14 2004
Barry Gross (CCNY)Brian Cairns (NASA-GISS)Bill Lawrence (Bowie State)
Michael Hagigeorgiou (Ugrad CCNY)Min Min Oo (Grad CCNY)
Istvan Laszlo (NOAA-NESDIS)Stephan Ungar (NASA-GSFC)Thomas Brakke (NASA-GSFC)
Validation and Refinement of Modis Aerosol Optical Depth
Product over Coastal Urban Areas
NOAA/NESDIS Cooperative Research Program Second Annual Science Symposium
SATELLITE CALIBRATION & VALIDATION July 13-14 2004
Motivation
• Air Pollution forcasting (Ozone, Aerosols, etc) has become a major NOAA responsibility in support of EPA
• MODIS AOT algorithms are global in nature.• Application to regional areas may be difficult due to local
ground albedo anomolies.• Colocated matchups show that MODIS optical depth
retrievals often overestimate optical depth measurements on the North East coast in urban areas
• Separate aerosol and land contributions using high spatial resolution data to help– Examine urban ground reflectance to tune MODIS – Retrieve aerosols in urban areas
NOAA/NESDIS Cooperative Research Program Second Annual Science Symposium
SATELLITE CALIBRATION & VALIDATION July 13-14 2004
Colocated matchup procedure
• Intercompare MODIS Optical Depth with CIMEL optical depth. (Need to use only spatially homogeneous datasets)
• CIMEL Optical Depth taken between NYC and Brookhaven (5 hour mean to agree within 10%)
• MODIS 10km products. 3 x 3 cells to have std < 20% mean
• Few Points satisfy these conditions
brookhaven
CCNY
NOAA/NESDIS Cooperative Research Program Second Annual Science Symposium
SATELLITE CALIBRATION & VALIDATION July 13-14 2004
Intercomparision betweenCIMEL Sky radiometer and Satellite
CART Site CCNY Site
CIMEL
MODIS
CIMEL
MODIS
NOAA/NESDIS Cooperative Research Program Second Annual Science Symposium
SATELLITE CALIBRATION & VALIDATION July 13-14 2004
MODIS Aerosol algorithms over land
25.660
25.470
2160,
660,
2160,
470,
c
c
g
g
g
g
Surface correlations
NOAA/NESDIS Cooperative Research Program Second Annual Science Symposium
SATELLITE CALIBRATION & VALIDATION July 13-14 2004
MODIS-MISR BRDF Productexamining VIS-MIR correlation
1)470(4~)470(/)2160( cgg Red pixels Consistent with MODISassumptions
As we go toYellow, MODISUnderestimatesVIS ground albedo
NOAA/NESDIS Cooperative Research Program Second Annual Science Symposium
SATELLITE CALIBRATION & VALIDATION July 13-14 2004
Path Radiance (Regression) Approach
• Many Surfaces have covarying spectral responses between the VIS and MIR
• If the spectral responses covary in a similar way (one component model), a strong ‘linear’ correlation between the VIS and MIR bands occurs
,,,,
,
Atmvis
TOAMIRud
TOAvis
SurfVIS
SurfMIR
RRTTMIRVIScR
RMIRVIScR
slopeY intercept
Single scattering+small Lanbertian surface albedo
NOAA/NESDIS Cooperative Research Program Second Annual Science Symposium
SATELLITE CALIBRATION & VALIDATION July 13-14 2004
Application and validation of ground reflectance correlations
(MODIS over vegetation)
Vis
ible
/ N
IR R
efle
ctan
ceAtmvis
TOAMIR
TOAvis m
Y intercept gives atmospheric reflection while slope (m) is proportional to the ground correlation
1) Correlation betweenground reflection for different channels will result in correlations at the TOA
2) This can be used to separate ground andatmosphere components byplotting the MIR and VIS reflectances anddetermining their regression coefficients
MIR Reflectance TOAMIR (2160nm)
47.~)660(
18.~)470(
C
C
NOAA/NESDIS Cooperative Research Program Second Annual Science Symposium
SATELLITE CALIBRATION & VALIDATION July 13-14 2004
Hyperion as model of Special Events Hyperspectral Imager
30 meter pixel resolution
S/N between 60 and 150 (oh well)from blue to red.
NOAA/NESDIS Cooperative Research Program Second Annual Science Symposium
SATELLITE CALIBRATION & VALIDATION July 13-14 2004
Probing aerosol retrieval and ground correlations over urban scales
Regions of interest include vegetation(central park – green), Urban areas (red-black) the river (orange), andlower manhattan
Scan Column
Sca
n Li
ne
New York observed through Hyperion (30 meter resolution)
NOAA/NESDIS Cooperative Research Program Second Annual Science Symposium
SATELLITE CALIBRATION & VALIDATION July 13-14 2004
Correlation Coefficient vs wavelength
Urban environmentshave many wavelength independent reflection(geometric) mechanisms that improve the correlations between the VIS and MIRchannels
Note a sharp differencefor lower Manhattandue to shadowing /urban effect
500 600 700 800 900 1000 1100 1200-0.2
0
0.2
0.4
0.6
0.8
1
1.2light urban vegatation middle urbannear wtc
Shadow/Urban effect
TOAMIR
TOAVISCorr ,
)(nm
NOAA/NESDIS Cooperative Research Program Second Annual Science Symposium
SATELLITE CALIBRATION & VALIDATION July 13-14 2004
Hyperion Retrieval of Aerosol Reflection over heavy urban zone
Larger Deviationsabove 700nm dueperhaps to partial Vegetated scenes but Still much smaller than for vegetated scenesthemselves
400 500 600 700 800 900 1000 1100 12000
2
4
6
8
10
12
14
Aerosol Reflection based on Aeronet estimates of AOT and Phase Function
Total Reflection Single Scattering
Rayleigh Scattering]94[.a
Hyperion Retrieval
)(nm
TO
A R
efle
ctan
ce %
NOAA/NESDIS Cooperative Research Program Second Annual Science Symposium
SATELLITE CALIBRATION & VALIDATION July 13-14 2004
430nm (Quite noisy)
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
5 10 15 20 25 30
5
10
15
20
25
30 9
10
11
12
13
14
5 10 15 20 25 30
5
10
15
20
25
30
50 100 150 200 250
100
200
300
400
500
600
700
800
900
1000
Hyperion Scene
Correlation between VIS and MIROnly keep 80.
Aerosol Reflection (bad pixelsMasked with navy blue) Increased loading to WTC observed
WTC
NOAA/NESDIS Cooperative Research Program Second Annual Science Symposium
SATELLITE CALIBRATION & VALIDATION July 13-14 2004
Ground reflection correlation frequency histogram
No angle dependence (Lambertian) assumption
Correlation larger than MODISassumption of 0.5
This has been Observed elsewhereas seen below
2160,660c
Fre
quen
cy
* Source MODIS ATBD Document
Light Urban
Central Park
Heavy Urban
NOAA/NESDIS Cooperative Research Program Second Annual Science Symposium
SATELLITE CALIBRATION & VALIDATION July 13-14 2004
Gnd ref 0.02
Gnd ref 0.05
Gnd ref 0.1
Gnd ref 0.2
Pe
rce
nta
ge
Of
De
lata
Ta
u
Aerosol optical thickness (Tau)
Percentage of Optical depthOverestimation
AOT overestimatesseem to be fairly consistent with ground albedo ~0.05
MIR albedo Urban
NOAA/NESDIS Cooperative Research Program Second Annual Science Symposium
SATELLITE CALIBRATION & VALIDATION July 13-14 2004
Conclusions• Spatial Regression over urban areas using high spatial
resolution sensors can isolate aerosol Path Radiance directly and help decouple ground albedo from atmosphere.
• MODIS correlation coefficients are too low for urban scenes and leads to an overestimate of optical depth from MODIS.
• Preliminary Radiative Transfer calculations show that the 30% overestimate of VIS ground reflectance can help explain observed AOD overestimates.
• More urban data including high aerosol optical depth events needed.
NOAA/NESDIS Cooperative Research Program Second Annual Science Symposium
SATELLITE CALIBRATION & VALIDATION July 13-14 2004
Additional Slides
NOAA/NESDIS Cooperative Research Program Second Annual Science Symposium
SATELLITE CALIBRATION & VALIDATION July 13-14 2004
Aeronet Optical Depth
Hyperion Fly By
From GISS112th St
NOAA/NESDIS Cooperative Research Program Second Annual Science Symposium
SATELLITE CALIBRATION & VALIDATION July 13-14 2004
Water leaving radiance using AVIRIS (subtracting ground decoupled atmosphere signal from total water signal)
400 500 600 700 800 900 1000-0.01
-0.005
0
0.005
0.01
0.015
0.02
Obtained using high shadowingBetter water leaving retrieval
Weak shadowingPoor water leaving retrieval
Wat
er le
avin
g R
adia
nce
(Ref
lect
ion
Uni
ts)
)(nm
100 200 300 400 500 600 700
50
100
150
200
250
300
350
400
450
500
shadowing
Littleshadowing
water
top related