sciamachy toa reflectance correction effects on aerosol optical depth retrieval
DESCRIPTION
Atmospheric Science Conference. SCIAMACHY TOA Reflectance Correction Effects on Aerosol Optical Depth Retrieval. W. Di Nicolantonio, A. Cacciari, S. Scarpanti, G. Ballista, E. Morisi, R. Guzzi. ESA ESRIN Frascati, 8-12 May 2006. ESA ESRIN Frascati, 8 - 12 May 2006. Contents. - PowerPoint PPT PresentationTRANSCRIPT
SCIAMACHY TOA Reflectance Correction SCIAMACHY TOA Reflectance Correction Effects on Aerosol Optical Depth RetrievalEffects on Aerosol Optical Depth Retrieval
W. Di Nicolantonio, A. Cacciari, W. Di Nicolantonio, A. Cacciari, S. Scarpanti, G. Ballista, E. Morisi, R. GuzziS. Scarpanti, G. Ballista, E. Morisi, R. Guzzi
ESA ESRIN Frascati, 8-12 May 2006ESA ESRIN Frascati, 8-12 May 2006
Atmospheric Science ConferenceAtmospheric Science Conference
SCIAMACHY data Processing for Aerosol retrievalSCIAMACHY data Processing for Aerosol retrieval
Measured Reflectance Correction effects on AOP Measured Reflectance Correction effects on AOP retrievalretrieval
Comparison with AERONET measurementsComparison with AERONET measurements
ConclusionConclusion
ContentsContents
ESA ESRIN Frascati, 8 - 12 May 2006
A S PA S P
PRE-ProcessingPRE-Processing -PIXELPIXEL SELECTION onSELECTION on> Cloud Coverage Fraction (FRESCO SC-v3) <0.05> SZA < 70 deg> | latitude| < 70 deg > sea OR land pixel ( no coast)
DATA READING DATA READING ( based on BEAT library ) > Earthshine radiance> Solar Irradiance > Geolocation> Observation Geometry> Acquisition Time
WAVELENGTHS SELECTIONWAVELENGTHS SELECTION> CRITERIA: Tg(i) > 99.7 % and
i+1 - i > 20 nm
sea set: 364, 387,429, 683,754,775 nmland set: up to 683 nm
MEASURED SPECTRAL MEASURED SPECTRAL REFLECTANCESREFLECTANCES @ @
selected aerosol wavelengthsselected aerosol wavelengths
AOD @ 500 nmAOD @ 500 nmAEROSOL CLASSAEROSOL CLASS
PROCESSING PROCESSING
AEROSOL DB >Spectral Optical Properties >Aerosol Profile
DEM DB GTOPO30 ( 0.25º res.)
SURFACE ALBEDO DB Spectral monthly MLER from GOME data ( 1º res.)
RTM DOWNSTREAM for simulated reflectances INVERSE METHOD
2 min. by LMFM for simulated and measured spectral reflectances on AOD and aerosol class
Aerosol Level-2 ProductAerosol Level-2 Productandand
selected pixels for selected pixels for AODAOD
Comparison / ValidationComparison / Validation
POST-ProcessingPOST-Processing
QUALITY CHECK Data Filtering on:
>2 values and >Discrimination Index on aerosol class
VALIDATION Spatial Coincidence check on AERONET sites
VISUALIZATION Data Format for:>daily/monthly map visualization
LEVEL 1B DATALEVEL 1B DATA > from KNMI (via SDC link) > from ESA (via ftp or DVD)
NADIR ObservationsNADIR Observations Selection &Calibration
by SciaL1C from EnviView
Data downloading &
calibration
Level 1-C data
SCIAMACHY data Processing for Aerosol retrieval : SCIAMACHY data Processing for Aerosol retrieval : main Phases and Toolsmain Phases and Tools
ESA ESRIN Frascati, 8 - 12 May 2006
SCIAMACHY data Processing for Aerosol retrieval : SCIAMACHY data Processing for Aerosol retrieval : Aerosol types in useAerosol types in use
ESA ESRIN Frascati, 8 - 12 May 2006
Aerosol Class(RH)
Component MixingRatio bynumber
Radius (m)lognormal
Refractiveindex
@ 500 nm
Reference
Maritime - Ma(0%,50%,80%)
Maritime OceanicOriginSmall Rural
0.990.01
3.30.16
0.350.4
1.53 – i 6.6 x 10-3
1.5 - i 1.0 x 10-3Shettle &
Fenn
DesertBackground - Ds
NucleiAccumulationCoarse
0.92740.0724610.9961 x10-4
0.010.02186.24
0.3280.5050.277
1.53 – i 5.5 x 10-3 D’Almeida
Maritime Polluted-MP(0%,50%)
Water SolubleSootOceanic
0.59390.40519.6 x 10-4
0.02850.01180.3
0.350.3010.4
1.53 – i 5 x 10-3
1.75 – i 0.451.5 – i 1 x 10-8
D’Almeida
Clean Continental-CC(0%,50%,80%)
Water SolubleDust Like
0.99991 x10-4
0.02850.471
0.3500.4
1.53 – i 6 x 10-3
1.53 – i 8 x 10-3D’Almeida
Rural -Ru(0%,50%)
Small RuralLarge Rural
0.9998751.25 x10-4
0.0270.43
0.350.4
1.53 – i 5.9 x 10-3 Shettle &Fenn
Biomass Burning-BB
BB FineBB Coarse
0.9997952.05 x10-4
0.080130.70549
0.1740.32
1.51 – i 0.021 Dubovik
Maritime Lanai -ML
ML FineML Coarse
0.998821.18 x10-3
0.080150.6744
0.2080.295
1.36 – i 1.5 x 10-3 Dubovik
Urban - Ur Ur FineUr Coarse
0.9995014.99 x10-4
0.082580.4436
0.1860.343
1.4 – i 0.9 x 10-3 Dubovik
&
SCIAMACHY data Processing for Aerosol retrieval : SCIAMACHY data Processing for Aerosol retrieval : Aerosol types in useAerosol types in use
ESA ESRIN Frascati, 8 - 12 May 2006
Aerosol Class(RH)
Component MixingRatio bynumber
Radius (m)lognormal
Refractiveindex
@ 500 nm
Reference
Maritime - Ma(0%,50%,80%)
Maritime OceanicOriginSmall Rural
0.990.01
3.30.16
0.350.4
1.53 – i 6.6 x 10-3
1.5 - i 1.0 x 10-3Shettle &
Fenn
DesertBackground - Ds
NucleiAccumulationCoarse
0.92740.0724610.9961 x10-4
0.010.02186.24
0.3280.5050.277
1.53 – i 5.5 x 10-3 D’Almeida
Maritime Polluted-MP(0%,50%)
Water SolubleSootOceanic
0.59390.40519.6 x 10-4
0.02850.01180.3
0.350.3010.4
1.53 – i 5 x 10-3
1.75 – i 0.451.5 – i 1 x 10-8
D’Almeida
Clean Continental-CC(0%,50%,80%)
Water SolubleDust Like
0.99991 x10-4
0.02850.471
0.3500.4
1.53 – i 6 x 10-3
1.53 – i 8 x 10-3D’Almeida
Rural -Ru(0%,50%)
Small RuralLarge Rural
0.9998751.25 x10-4
0.0270.43
0.350.4
1.53 – i 5.9 x 10-3 Shettle &Fenn
Biomass Burning-BB
BB FineBB Coarse
0.9997952.05 x10-4
0.080130.70549
0.1740.32
1.51 – i 0.021 Dubovik
Maritime Lanai -ML
ML FineML Coarse
0.998821.18 x10-3
0.080150.6744
0.2080.295
1.36 – i 1.5 x 10-3 Dubovik
Urban - Ur Ur FineUr Coarse
0.9995014.99 x10-4
0.082580.4436
0.1860.343
1.4 – i 0.9 x 10-3 Dubovik
Derivative of aerosol TOA reflectance - in SS approx - with respect to AOD @ 500 nm
Reflectance Correction effects : Reflectance Correction effects : linear correction for TOA Refllinear correction for TOA Refl
Tilstra et al. 2005Tilstra et al. 2005 derived spectral derived spectral correction factors in the range 240 - correction factors in the range 240 - 400 nm by comparing SCIAMACHY 400 nm by comparing SCIAMACHY spectral reflectance and spectral reflectance and DAK DAK simulated reflectancesimulated reflectance over different over different surfacessurfaces
Under the hypothesis of a linear relation Under the hypothesis of a linear relation between SCIAMACHY and between SCIAMACHY and MERIS MERIS reflectancesreflectances, , Acarreta & Stammes 2005Acarreta & Stammes 2005 provided a set of correction factors at provided a set of correction factors at 442, 510, 665, 708, and 885 nm taking 442, 510, 665, 708, and 885 nm taking account different underlying surfacesaccount different underlying surfaces
Known SCIAMACHY underestimation forKnown SCIAMACHY underestimation for UV UV and and VIS-NIRVIS-NIR reflectance by about reflectance by about 10 to 25%10 to 25% and and 10 to 20%10 to 20%
Rscia-corra) = Csla) Rscia(a) + Coffa)
ASP a
(nm)Csl
sl Coff
off
364 1.158 0.028 - -387 1.143 0.042 - -429 1.130 0.040 -0.01 0.0100683 1.157 0.020 -0.01 0.0080754 1.187 0.022 -0.01 0.0082775 1.191 0.023 -0.01 0.0084
ESA ESRIN Frascati, 8 - 12 May 2006
Reflectance Correction effects : Reflectance Correction effects : Spectra for clouds free pixelsSpectra for clouds free pixels
FRESCO ccf = 0
ESA ESRIN Frascati, 8 - 12 May 2006
ASP a
(nm)Csl
sl Coff
off
364 1.158 0.028 - -387 1.143 0.042 - -429 1.130 0.040 -0.01 0.0100683 1.157 0.020 -0.01 0.0080754 1.187 0.022 -0.01 0.0082775 1.191 0.023 -0.01 0.0084
IIn then the NIR NIR, the adopted correction doesn’t , the adopted correction doesn’t introduce a significant variation in the TOA introduce a significant variation in the TOA reflectance values, mainly over ocean. reflectance values, mainly over ocean. This can be due to the hypothesis of linearity This can be due to the hypothesis of linearity assumed also in low reflectance domain assumed also in low reflectance domain (< 0.2) (< 0.2) in in SCIAMACHY - MERIS refl. comparisonSCIAMACHY - MERIS refl. comparison
Sep 26, 2004 Sep 26, 2004
SCIAMACHY AOD @ 500 nm
Reflectance Correction effects on AOP : Reflectance Correction effects on AOP : Example of daily global retrievalExample of daily global retrieval September 15, 2004September 15, 2004
ESA ESRIN Frascati, 8 - 12 May 2006
SCIAMACHY AOD @ 500 nm
Reflectance Correction effects on AOP : Reflectance Correction effects on AOP : Example of daily global retrievalExample of daily global retrieval September 15, 2004September 15, 2004
ESA ESRIN Frascati, 8 - 12 May 2006
Reflectance Correction effects on AOD : Reflectance Correction effects on AOD : Example of monthly global retrievalExample of monthly global retrieval September 2004September 2004
ESA ESRIN Frascati, 8 - 12 May 2006
Reflectance Correction effects on AOD : Reflectance Correction effects on AOD : Example of seasonal global retrievalExample of seasonal global retrieval Autumn 2004Autumn 2004
ESA ESRIN Frascati, 8 - 12 May 2006
ASP tendency to AOD understimationASP tendency to AOD understimation AOD = 0.03 0.06 AOD
Comparison with AERONET measurements: Comparison with AERONET measurements: 2003 and 2004 corresponding data2003 and 2004 corresponding data
AERONET AOD Level 2 data
9 AERONET sites
Land : y=0.85 x + 0.07; r= 0.93
Ocean : y=0.24 x + 0.13; r= 0.38 HP: non-linearity effects for small TOA reflectance not considered in the correction factors derivation, introducing a systematic error, resulting in a bias mainly in the retrieval of AOD lower than 0.4
ESA ESRIN Frascati, 8 - 12 May 2006
Reflectance Correction effects on AOP retrieval: retrieval uncertainties Reflectance Correction effects on AOP retrieval: retrieval uncertainties
CCsl-maxsl-max = C = Cslslsl sl ; ; CCoff-maxoff-max = C = Coffoff + + offoffCCsl-minsl-min = C = Cslslslsl; ; CCoff-minoff-min = C = Coffoff - - off off
ESA ESRIN Frascati, 8 - 12 May 2006
AOD13.008.0
AOD20.003.0AOD
AOD > 0.1 in 50% and 20% aerosol type changed
AOD > 0.2 in 10% and 5% aerosol type changed
Estimates of retrieval Estimates of retrieval uncertainties associated uncertainties associated to reflectance correction to reflectance correction factors factors
ConclusionConclusion
TOA reflectance correctionTOA reflectance correction procedure was applied to SCIAMACHY nadir measurements to explore the possibility of obtaining reliable information on tropospheric aerosol retrieval :
• Examples of global mapglobal map of AOD at 500 nm, relative to September, October, and November 2004, display the relatively low aerosol loading over oceanic regions, with AOD < 0.2, and the hot spots in AOD over land, presenting values 0.5, can be clearly identified
• SCIAMACHY versus AERONETSCIAMACHY versus AERONET AOD (2003-2004) result in a good agreement (r = 0.93) in the case of significant aerosol loading over land and a very minor agreement (r = 0.38) for the retrieval over ocean. 1 uncertainty evaluated as AOD = 0.03 0.06 AOD
On the other handOn the other hand
• Uncertainties in the reflectance correction factors lead to a significant significant propagated variability in the retrieved AODpropagated variability in the retrieved AOD of the order of 20% for AOD 1 and up to 90% for AOD 0.1
ThenThen
Applied correction encourages the adoption of SCIAMACHY sensor for aerosol retrieval and at the same time strengthens the need to use anthe need to use an official re-calibrated version of official re-calibrated version of Level 1BLevel 1B data to improve the radiometric accuracy of low reflectance measurements and to increase the accuracy degree of the aerosol retrieval.
ESA ESRIN Frascati, 8 - 12 May 2006
ESA ESRIN Frascati, 8-12 May 2006ESA ESRIN Frascati, 8-12 May 2006
Atmospheric Science ConferenceAtmospheric Science Conference