Remote sensing of surface reflective properties: Role of regularization and a priori knowledge

Abstract

Linear kernel-driven bidirectional reflectance distribution function (BRDF) models have been used for mapping albedo with single field-of-view satellite measurements such as Moderate Resolution Imaging Spectroradiometer (MODIS). Due to limited samplings and poor angular configurations available from these satellite remotely sensed data, BRDF models inversion is often plagued by numerical instability. In order to overcome the ill-posedness of the BRDF model inversion and robustly estimate terrestrial surface albedo, a regularization technique is employed for the cases where the number of observations is insufficient, or the angular distribution is poor. Emphasis is also placed on the combination of a priori knowledge with the regularized inversion. Numerical performances and case study results with ground measurements and MODIS observations suggest that the method is sound and robust for ill-posed BRDF inverse problems. The method presented in this study is promising for land surface reflective parameters retrieval even for regions where only sparse observations are available.