geostationary surface albedo retrieval error estimation
DESCRIPTION
Geostationary surface albedo retrieval error estimation. Y. Govaerts (1) and A. Lattanzio (2). (1) EUMETSAT, Germany (2) Makalumedia, Germany. 2nd CEOS/WGCV/Land Product Validation (LPV) Workshop on Albedo Products April 27-28, 2005. METEOSAT MISSIONS STATUS. 6 bits data. 8 bits data. - PowerPoint PPT PresentationTRANSCRIPT
Geostationary surface albedo retrieval error estimation
Y. Govaerts(1) and A. Lattanzio(2)
(1) EUMETSAT, Germany(2) Makalumedia, Germany
2nd CEOS/WGCV/Land Product Validation (LPV) Workshop on Albedo Products
April 27-28, 2005
METEOSAT MISSIONS STATUS
Pre-operationalVIS 6 bits
OperationalVIS 8 bits
24 years of archive
Need to be consistently calibrated and navigated
6 bits data
8 bits data
The Meteosat mission has been conceived in the early seventies. The primary objective of VIS band data was the near real-time qualitative observation of meteorological systems (i.e., to take picture).
METEOSAT MISSIONS STATUSMajors drawbacks concerning the quantitative use of the VIS band concern:•The unreliable SSR characterization
Development of a SSR model to characterize the error
Govaerts, Y.M., Clerici, M., and Clerbaux, N. (2004) Operational Calibration of the Meteosat Radiometer VIS Band, IEEE TGARS, 42, 1900-1914.
METEOSAT MISSIONS STATUSMajors drawbacks concerning the quantitative use of the VIS band concern:•The unreliable SSR characterization•The width of the VIS band (surface changes, WV absorption, …) Surface reflectance
Aerosol transmittance (τ=0.4)Gaseous transmittance
METEOSAT MISSIONS STATUSMajors drawbacks concerning the quantitative use of the VIS band concern:•The unreliable SSR characterization•The width of the VIS band (surface changes, WV absorption, …)• Navigation problem in the eighties.
The proposed approach relies on a reliable estimation of the retrieved surface albedo retrieval error that explicitly accounts for the observation system uncertainties.
The Meteosat system has not been design to fulfil any climate monitoring requirements
METEOSAT-7 VIS Band Calibration
Estimated calibration errorTarget characterisation error : 4.1%SSR error contribution : 3.8%Random error : 1.6%Total calibration error : 6% The SSR error should increase in time
The loss of transmittance depends on the wavelength
Desert
Sea
Govaerts, Y.M. Clerici, M. Clerbaux, N., 2004, Operational calibration of the Meteosat radiometer VIS band, IEEE TGARS, 42, 1900- 1914
Surface Albedo from Geostationary Obs.The Reciprocity Principle is applicable at a
spatial scale of a few km
Absorbing atmosphere
Scattering atmosphere
Assumptions•Atmosphere is composed of one absorbing gas layer and one scattering layer
•US76 atmospheric profile•Continental aerosol type•Atmospheric and surface scattering properties are constant along the day
•Surface scattering properties can be represented by the RPV BRF model
Model :•ozone (TOMS)•Total column water vapour (ECMWF)
State variables
Retrieved :•aerosol optical thickness (1)•surface reflectance level (1)•surface anisotropy (2)
Pinty, B., Roveda, F., Verstraete, M.M., Gobron, N., Govaerts, Y., Martonchik, J.V., Diner, D.J., and Kahn, R.A. (2000) Surface albedo retrieval from Meteosat: Part 1: Theory, Journal of Geophysical Research, 105, 18099-18112.
INVERSION
GEOSTATIONARY SURFACE ALBEDO
Absorbing atmosphere
Scattering atmosphere
Inversion of the forward model
2
2 )ˆ(
y
m xyy
INVERSION : Nile delta (winter)
GEOSTATIONARY SURFACE ALBEDO
τ = 0.1, DHR = 0.17
INVERSION : Nile delta (summer)
GEOSTATIONARY SURFACE ALBEDO
τ = 0.4, DHR = 0.21
Meteosat -2/7 Albedo Comparison
Meteosat-7Launch date : 1997Sub-satellite point : 0o Repeat cycle (archive) : 30 minDigitalisation : 8 bitsCalibration accuracy : 6%Daily Rad. Noise : 7-9%
Meteosat-2Launch date : 1981Sub-satellite point : 0o Repeat cycle (archive) : 30/60 minDigitalisation : 6 bitsCalibration accuracy : 15%Daily Rad. Noise : 12%
5.915.3TOTAL
1.61.2Random
3.814.2SSR. Error
4.14.1Rad. Transfer.
72Meteosat
5.915.3TOTAL
1.61.2Random
3.814.2SSR. Error
4.14.1Rad. Transfer.
72Meteosat
Calibration error budget
Meteosat -2/7 Albedo Comparison
Met-2: 1-10 May 1984 Met-7: 1-10 May 2004
SURFACE ALBEDO COMPARISON OVER STABE DESERT
Min prob.: 90%Max albedo rel. error: 10%
6.8%
5.0%
MET-2 has more high values
DETECTION OF SIGNIFICANT CHANGES
1. Remove the pixels with a QI (probability of the goodness of the fit) lower than 90%
SOLUTION PROBABILITY
Met-2 Met-7
DETECTION OF SIGNIFICANT CHANGES
1. Remove the pixels with a QI (probability of the goodness of the fit) lower than 90%
2. Analyse the albedo difference with respect to the retrieval error
RADIOMETRIC RELATIVE ERROR
Met-2 Met-7
SURFACE ALBEDO RELATIVE ERROR
Met-2 Met-7
DETECTION OF SIGNIFICANT CHANGES
1. Remove the pixels with a QI (probability of the goodness of the fit) lower than 90%
2. Analyse the albedo difference with respect to the retrieval error
3. Keep only differences larger than the respective error (including calibration error):
22
2727 || AAAA
DETECTION OF SIGNIFICANT CHANGES
-75%< <+75%
(A2-A7)/A7
A2 : 1-10 May 1984A7 : 1-10 May 2004
CONCLUSION•Geostationary observations offer a decisive advantage
to retrieve surface albedo climate data set thanks to the frequent daily sampling used to estimate the surface anisotropy and to the duration of the archive (+20 years).
•Met-2 and Met-7 agrees within 6% over desert area. It should be possible to detect significant (15%) temporal surface albedo changes from the Meteosat archive.
•Uncertainty in the SSR characterization is the major limiting factor.
GLOBAL SURFACE ALBEDO
Broadband (0.3 – 3.0µm) surface albedo derived from GOES-8/10, MET-5/7 and GMS-5 in 1-10 May 2001