Land surface albedo from Meteosat Second Generation (MSG) observations

In the framework of the EUMETSAT Satellite Application Facility for land surface analysis (Land-SAF) we develop a surface albedo product which is generated from observations provided by the SEVIRI instrument on board of the geostationary MSG satellite. Cloud-cleared top-of-atmosphere reflectances in the 0.6 /spl mu/m, 0.8 /spl mu/m, and 1.6 /spl mu/m channels are atmospherically corrected with a simplified and fast radiative transfer code (SMAC). The algorithm for the determination of albedo relies on the inversion of a linear kernel-based BRDF model. Spectral directional-hemispherical and bi-hemispherical albedos as well as associated error estimates are derived. Broadband albedo values are derived from the spectral quantities by using suitable linear regression coefficients. In the case of poor angular sampling the information content provided by the MSG observations needs to be augmented by a priori information, e.g. by using a BRDF data base derived from measurements of instruments with multiple view angle capabilities such as POLDER. A method based on Bayes' Theorem has been developed and integrated for that purpose. In addition, this approach makes it possible to add the information coming from an additional sensor such as the EPS/AVHRR instrument which is planned to be included in the Land-SAF activities. First applications of the prototype to simulated MSG images as well as to real SPOT/VEGETATION data are presented.