William M. F. Grey,1
Peter R. J. North,1
and Sietse O. Los1
1The authors are with the Climate and Land Surface Systems Interaction Centre, School of Environment and Society, Swansea University, Singleton Park, Swansea, SA2 8PP, United Kingdom (e-mail for W. Grey, w.m.f.grey@swan.ac.uk).
William M. F. Grey, Peter R. J. North, and Sietse O. Los, "Computationally efficient method for retrieving aerosol optical depth from ATSR-2 and AATSR data," Appl. Opt. 45, 2786-2795 (2006)
We present a robust and computationally efficient method for retrieving aerosol optical depth (AOD) from top-of-atmosphere ATSR-2 (Along-Track Scanning Radiometer) and AATSR (Advanced ATSR) reflectance data that is formulated to allow retrieval of the AOD from the 11 year archive of (A)ATSR data on the global scale. The approach uses a physical model of light scattering that requires no a priori information on the land surface. Computational efficiency is achieved by using precalculated lookup tables (LUTs) for the numerical inversion of a radiative-transfer model of the atmosphere. Estimates of AOD retrieved by the LUT approach are tested on AATSR data for a range of global land surfaces and are shown to be highly correlated with sunphotometer measurements of the AOD at
. (Pearson's correlation coefficient
is 0.71.)
Lieven Clarisse, Daniel Hurtmans, Alfred J. Prata, Federico Karagulian, Cathy Clerbaux, Martine De Mazière, and Pierre-François Coheur Appl. Opt. 49(19) 3713-3722 (2010)
Jacqueline Lenoble, Timothy Martin, Mario Blumthaler, Rolf Philipona, Astrid Albold, Thierry Cabot, Alain de La Casinière, Julian Gröbner, Dominique Masserot, Martin Müller, Thomas Pichler, Günther Seckmeyer, Daniel Schmucki, Mamadou Lamine Touré, and Alexis Yvon Appl. Opt. 41(9) 1629-1639 (2002)
You do not have subscription access to this journal. Cited by links are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.
You do not have subscription access to this journal. Figure files are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.
You do not have subscription access to this journal. Article tables are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.
You do not have subscription access to this journal. Equations are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.
Perturbed Values of Atmospheric Ozone and Water Vapor Concentrations and Surface Elevation Used in the Sensitivity Study
Parameter
Forward Modeled
Minimum Perturbation
Maximum Perturbation
Ozone (cm atm)
0.35
0.10
0.50
Water vapor (g∕cm2)
2.5
0.0
6.0
Elevation (km)
0.5
0.0
2.0
Note:The values used in forward runs of the model for creating the top-of-atmosphere simulated data are also given.
Table 2
Conditional σ of the Retrieved Estimates of AOD by Using the Full Inversions for Perturbed Values of Atmospheric Ozone and Water Vapor Concentrations and Surface Elevation
Land Surface (and Aerosol Model)
Simulated AOD
Retrieved AOD for Ideal Case
Curvature of Brent Parabola (10−4)
Water Vapor
Ozone
Elevation
Total
Relative Azimuth Angle
Solar Zenith Angle
Dense forest (smoke)
0.7
0.682
0.31
0.002
0.011
0.065
0.068
0
30
0.4
0.395
4.56
0.000
0.001
0.011
0.011
180
30
0.1
0.076
6.17
0.000
0.003
0.014
0.015
90
60
Sparse forest (smoke)
0.7
0.159
0.01
0.004
0.070
0.055
0.104
0
30
0.4
0.393
2.31
0.000
0.002
0.008
0.008
180
30
0.1
0.068
0.90
0.000
0.006
0.011
0.013
90
60
Homogenous canopy bright soil green vegetation (continental)
0.7
0.703
0.33
0.000
0.014
0.075
0.076
0
30
0.4
0.377
2.83
0.000
0.005
0.005
0.007
180
30
0.1
0.084
2.28
0.000
0.006
0.018
0.019
90
60
Homogenous canopy bright soil
0.7
0.678
0.08
0.007
0.034
0.062
0.071
0
30
sensecent vegetation
0.4
0.358
0.83
0.001
0.015
0.005
0.016
180
30
(continental)
0.1
0.068
0.63
0.001
0.022
0.021
0.030
90
60
Homogenous canopy dark soil green vegetation (smoke)
0.7
0.721
0.37
0.001
0.011
0.061
0.062
0
30
0.4
0.392
3.68
0.000
0.003
0.004
0.005
180
30
0.1
0.089
2.49
0.000
0.005
0.017
0.018
90
60
Homogenous canopy dark soil sensecent vegetation (smoke)
0.7
0.750
0.08
0.003
0.040
0.068
0.079
0
30
0.4
0.386
1.09
0.001
0.013
0.004
0.013
180
30
0.1
0.087
0.80
0.001
0.018
0.019
0.026
90
60
Bare soil (desert dust)
0.7
0.739
0.16
0.001
0.009
0.095
0.096
0
30
0.4
0.404
4.93
0.000
0.000
0.014
0.014
180
30
0.1
0.113
2.02
0.000
0.001
0.019
0.020
90
60
Bare soil (continental)
0.7
0.694
0.15
0.001
0.010
0.081
0.082
0
30
0.4
0.403
8.80
0.000
0.000
0.018
0.018
180
30
0.1
0.100
3.89
0.000
0.001
0.024
0.024
90
60
Note:Total σ for all parameters is also presented. The inversions are performed on the simulated data set over a range of land surfaces, aerosol models, and sun-sensor geometries. The continental aerosol model is predominantly composed of desert dust. The retrieved aerosol for AOD is calculated, assuming an ideal characterization of the atmospheric profile.
Table 3
Summary of Statistics of Intercomparison between the AATSR Derived Estimates of AOD at 550 nm Calculated by Using the Full Model Inversions and the LUT Inversions for All Sites Combined at Different Sampling Densities
AERONET
Full Inversion
VZA Interval
RAZ Interval
SZA Interval
AOD at 550 nm Interval
Number of Samples in LUT
r2
RMSE
Mean Error
r2
RMSE
Mean Error
5.0
20.0
5.0
0.05
212000
0.71
0.18
0.008
1.00
0.02
0.003
5.0
20.0
5.0
0.10
109200
0.74
0.17
0.037
0.99
0.05
0.032
5.0
20.0
5.0
0.20
57200
0.74
0.17
0.041
0.99
0.05
0.036
5.0
60.0
5.0
0.05
85280
0.73
0.17
0.044
0.97
0.08
0.040
5.0
20.0
10.0
0.05
114800
0.73
0.17
0.039
0.99
0.05
0.034
10.0
20.0
5.0
0.05
68880
0.74
0.17
0.055
0.99
0.07
0.051
Note:The ozone and surface elevation intervals are 0.4 cm atm and 2 km, respectively and are the same for all table sizes, whereas the view zenith angle (VZA), relative azimuth angle (RAZ), and solar zenith angle (SZA) intervals are changed.
Table 4
Summary of Statistics of Intercomparisons between AATSR Estimates of AOD Derived using the Largest LUT, Sunphotometer Observations, and Estimates of AOD Retrieved from the Full Inversions at Individual Sites
Italicized r2 values are significant to the 95% confidence level.
Tables (4)
Table 1
Perturbed Values of Atmospheric Ozone and Water Vapor Concentrations and Surface Elevation Used in the Sensitivity Study
Parameter
Forward Modeled
Minimum Perturbation
Maximum Perturbation
Ozone (cm atm)
0.35
0.10
0.50
Water vapor (g∕cm2)
2.5
0.0
6.0
Elevation (km)
0.5
0.0
2.0
Note:The values used in forward runs of the model for creating the top-of-atmosphere simulated data are also given.
Table 2
Conditional σ of the Retrieved Estimates of AOD by Using the Full Inversions for Perturbed Values of Atmospheric Ozone and Water Vapor Concentrations and Surface Elevation
Land Surface (and Aerosol Model)
Simulated AOD
Retrieved AOD for Ideal Case
Curvature of Brent Parabola (10−4)
Water Vapor
Ozone
Elevation
Total
Relative Azimuth Angle
Solar Zenith Angle
Dense forest (smoke)
0.7
0.682
0.31
0.002
0.011
0.065
0.068
0
30
0.4
0.395
4.56
0.000
0.001
0.011
0.011
180
30
0.1
0.076
6.17
0.000
0.003
0.014
0.015
90
60
Sparse forest (smoke)
0.7
0.159
0.01
0.004
0.070
0.055
0.104
0
30
0.4
0.393
2.31
0.000
0.002
0.008
0.008
180
30
0.1
0.068
0.90
0.000
0.006
0.011
0.013
90
60
Homogenous canopy bright soil green vegetation (continental)
0.7
0.703
0.33
0.000
0.014
0.075
0.076
0
30
0.4
0.377
2.83
0.000
0.005
0.005
0.007
180
30
0.1
0.084
2.28
0.000
0.006
0.018
0.019
90
60
Homogenous canopy bright soil
0.7
0.678
0.08
0.007
0.034
0.062
0.071
0
30
sensecent vegetation
0.4
0.358
0.83
0.001
0.015
0.005
0.016
180
30
(continental)
0.1
0.068
0.63
0.001
0.022
0.021
0.030
90
60
Homogenous canopy dark soil green vegetation (smoke)
0.7
0.721
0.37
0.001
0.011
0.061
0.062
0
30
0.4
0.392
3.68
0.000
0.003
0.004
0.005
180
30
0.1
0.089
2.49
0.000
0.005
0.017
0.018
90
60
Homogenous canopy dark soil sensecent vegetation (smoke)
0.7
0.750
0.08
0.003
0.040
0.068
0.079
0
30
0.4
0.386
1.09
0.001
0.013
0.004
0.013
180
30
0.1
0.087
0.80
0.001
0.018
0.019
0.026
90
60
Bare soil (desert dust)
0.7
0.739
0.16
0.001
0.009
0.095
0.096
0
30
0.4
0.404
4.93
0.000
0.000
0.014
0.014
180
30
0.1
0.113
2.02
0.000
0.001
0.019
0.020
90
60
Bare soil (continental)
0.7
0.694
0.15
0.001
0.010
0.081
0.082
0
30
0.4
0.403
8.80
0.000
0.000
0.018
0.018
180
30
0.1
0.100
3.89
0.000
0.001
0.024
0.024
90
60
Note:Total σ for all parameters is also presented. The inversions are performed on the simulated data set over a range of land surfaces, aerosol models, and sun-sensor geometries. The continental aerosol model is predominantly composed of desert dust. The retrieved aerosol for AOD is calculated, assuming an ideal characterization of the atmospheric profile.
Table 3
Summary of Statistics of Intercomparison between the AATSR Derived Estimates of AOD at 550 nm Calculated by Using the Full Model Inversions and the LUT Inversions for All Sites Combined at Different Sampling Densities
AERONET
Full Inversion
VZA Interval
RAZ Interval
SZA Interval
AOD at 550 nm Interval
Number of Samples in LUT
r2
RMSE
Mean Error
r2
RMSE
Mean Error
5.0
20.0
5.0
0.05
212000
0.71
0.18
0.008
1.00
0.02
0.003
5.0
20.0
5.0
0.10
109200
0.74
0.17
0.037
0.99
0.05
0.032
5.0
20.0
5.0
0.20
57200
0.74
0.17
0.041
0.99
0.05
0.036
5.0
60.0
5.0
0.05
85280
0.73
0.17
0.044
0.97
0.08
0.040
5.0
20.0
10.0
0.05
114800
0.73
0.17
0.039
0.99
0.05
0.034
10.0
20.0
5.0
0.05
68880
0.74
0.17
0.055
0.99
0.07
0.051
Note:The ozone and surface elevation intervals are 0.4 cm atm and 2 km, respectively and are the same for all table sizes, whereas the view zenith angle (VZA), relative azimuth angle (RAZ), and solar zenith angle (SZA) intervals are changed.
Table 4
Summary of Statistics of Intercomparisons between AATSR Estimates of AOD Derived using the Largest LUT, Sunphotometer Observations, and Estimates of AOD Retrieved from the Full Inversions at Individual Sites