Surface albedos and angle-corrected NDVI from AVHRR observations of South America

This study investigated the use of a semiempirical land surface bidirectional reflectance distribution function (BRDF) model to derive surface spectral albedo and correct for angular effects on remotely sensed images. The Ambrals RossThick-LiSparse BRDF model has been applied to a 31-day multiangular data of daily global area coverage (GAC) images over South America in July 1987. A BRDF-based angle-corrected normalized difference vegetation index (NDVI) image was generated along with corresponding images of land surface albedo in the visible and near-infrared bands, computed from the BRDF model parameters retrieved. For comparison, the monthly maximum value NDVI image and the land surface visible and near-infrared albedo images were created based on the corresponding single reflectance observations. Visual evaluation and histogram analysis of these images shore evident differences between the surface spectral albedo images produced by the Ambrals BRDF nzodel and those by the Lambertian conventional approach, and between the angle-corrected NDVI image and the monthly composite NDVI image. These differences were further explored fur the area of the Large Scale Biosphere-Atmosphere Experiment (LBA) in Amazonia. Results indicate that 1) the angle-corrected NDVI is an improvement over the monthly maximum value NDVI with more consistency with surface properties, such as land cover type and vegetation height, and 2) the surface spectral albedo values calculated by the Ambrals BRDF model are In better agreement with these surface characterizations than the albedo values derived from a maximum-value composited reflectance. Due to limitations inherent in the AVHRR data, especially with respect to atl,atmospheric correction, the results obtained are considered primarily as indicative of what will be possible with the global data sets to be obtained by NASA's MODIS and MISR sensors in the near future, illustrating the potential of rising a surface BRDF model for advanced earth system studies. (C) Elsevier Science Inc., 2000.