To retrieve albedo from air-borne WIDAS based on a prior BRDF database

A method to derive land surface albedo based on a prior archetypal BRDF (Bidirectional Reflectance Distribution Function) database is presented. The algorithm was based on kernel driven BRDF models, the 69 sets of field observations were classified into four classes according to AFX (Anisotropic Flat Index) which can indicate basic dome-bowl anisotropic reflectance patterns of terrestrial surface, and then the archetypal BRDF shapes database was created. In the inversion of surface albedo, we fit the observations using the four archetypal BRDF shapes respectively to select the shape that has least fitting error as the underlying surface anisotropy prior knowledge. The archetypal BRDF shapes do not depend on land cover. An albedo datasets for air-borne WIDAS is produced with this scheme. At last, we obtained the shortwave spectral albedo of WIDAS in the Yingke station in WATER Campaign. Comparison of the albedo with field observations shows that the absolute error is less than 0.05. This study will provide a possible method for space-borne albedo retrieval which lacks sufficient multi-angular observations.