Estimation of Heihe region surface albedo based on a priori knowledge by using HJ1-a satellite images

Prior knowledge can significantly improve the retrieval of surface spectral albedo from satellite observations. This paper compares two methods that derive HJ-1 surface albedo in Heihe region by using prior knowledge based on kernel-driven BRDF model, with that derived by assuming Lambertian surface. The first algorithm (algorithm I) uses the backup algorithm of operational MODIS BRDF/Albedo product; the second algorithm (algorithm II) is developed by Li et al. (2001) that bases on the Bayesian inference theory to use prior knowledge from sets of field measurements. Our results show that both algorithms reduce the relative error by up to 10%~12% in the red and near-infrared band. Further analysis shows that the albedo would be retrieved with higher accuracy if view zenith angles provided by satellite sensor are larger than that of HJ-1.