Determination of the Shortwave Anisotropic Function for Clear-Sky Desert Scenes from Meteosat Data

Abstract

Determination of planetary albedo on the basis of satellite observations of reflected shortwave (SW) radiances requires taking into account the anisotropic (non-Lambertian) reflectance properties of the earth?atmosphere system, depending both on the cloud cover and the nature of the underlying, surface. One approach frequently used has been to represent these properties by a limited set of normalized bidirectional reflectance functions (BDRF) for different scene types. The construction of the normalized BDRFs used to process the Earth Radiation Budget Experiment (ERBE) measurements was based mostly on data from the sun-synchronous Nimbus-7 mission, observing close to local noon. Consequently, because desert zones are fairly restricted in latitude, only a small range of solar zenith angles was sampled. Here the authors consider, for clear-sky desert areas, the improvements that can be made using data from the geostationary satellite Meteosat, which samples all solar zenith angles that occur. The authors define BDRF ratios (between two instants on the same day for the same area) that depend on viewing geometry (five angles for a geostationary satellite) and that together with infrared window radiance measurements allow to distinguish clear and cloudy desert scenes. Using three to five Meteosat images per day over the year 1985, and considering 42 areas in desert zones, the authors compute roughly 12 500 clear-sky BDRF ratios (representing 4.5 million B2 pixels), and sort these into bins in five-dimensional angular space. Values of the BDRF ratio are well defined and stable in each of these bins. Application of the Helmholtz reciprocity principle yields data for angular bins not directly observed. After spectral corrections and normalizations, the authors obtain a completely defined SW angular model (i.e., normalized anisotropic function and directional albedo) for clear-sky desert scenes. This model is quite different from that used in the ERBE analyses. The authors discuss the improvements obtained.