The dataset contains a unique longterm time series of surface albedo derived from first generation Meteosat satellites. The original dataset is generated by EUMETSAT (www.eumetsat.int). The uniqueness of the data set is the fact that the high temporal sampling of Meteosat (every 30 minutes) significantly increases the probability to obtain cloud free observations for a particular area. By collecting cloud free data from different sun zenith angles, the retrieval algorithm enables the quantitative derivation of surface albedo. For details the reader is referred to the references given in Loew and Govaerts (2010). Meteosat surface albedo is probably currently the best surface albedo information one can get prior to the MODIS era, which started in 2000.
A further unique feature of the Meteosat surface albedo data set is that the original data set comes together with quantitative uncertainty estimates which is used for quality screening of the data.
The original data provided by EUMETSAT was carefully analyzed to generate a homogeneous longterm data record (Loew and Govaerts, 2010). The quality flags in the original data contained information about the error in the retrievals which were used to strictly filter out unreliable data. This filtering removed basically grid points which were significantly affected by aerosols. Monthly means were then calculated from the 10 day original data set using the climate data operators (cdo's).
Up to now the dataset is limited to the geographic domain covered by the Meteosat disc. Currently, EUMETSAT aims to generate a global land surface albedo data set from geostationary data in the frame of the scope-CM project.
The algorithm to retrieve surface albedo is based on the assumption of an isotropic response of the surface. Problems have been observed in very bright regions like deserts with anisotropic structures (sand dunes). Here the surface albedo did exceed a value of 0.6. For the generation of the T63 data set, all grid cells with BHR>0.6 were therefore set to a value of 0.6.