An approach to improve hot spot effect for the MODIS BRDF/Albedo algorithm

Abstract

The RossThick-LiSparse-Reciprocal (RTLSR) Bidirectional Reflectance Distribution Function (BRDF) model has been developped to derive the operational Moderate Resolution Imaging Spectroradiometer (MODIS) BRDF/Albedo product due to its simplicity and the underlying physics; however, early research showed that this model deficiency mainly comes from its underestimation of the hotspot directional signatures near the Sun's illumination direction. In this paper, we developed an approach to improve the hot spot effect for the RTLSR model by improving the volumetric scattering kernel with an exponential approximation of the hot spot kernel (Chen and Cihlar, 1997). Compared with the RTLSR model and the further-developed model that modify the hotspot directional signatures of RTLSR model based on the calculation of an overlay function of the intersection of viewed and sunlit leaf areas (Jupp and Strahler, 1991, thereafter named RTJLSR), this newly-corrected model that modifies the hot spot effect of RTLSR model based on the theory of calculation of a canopy gap size distribution function (Chen and Leblanc, 1997, thereafter named RTCLSR) preserves the linear form of kernel-driven model, but flexibly adjust hotspot magnitude and width through two additional parameters C1 and C2. Initial validation result with airborne cloud absorption radiometer (CAR) data shows that the RTCLSR model can significantly improve the model-observation fits in hotspot region. In near future, we will focus on determining C1/C2 values from spaceborne POLDER BRDF database provided by the POSTEL Service Centre. With C1/C2 are predetermined, the newly-correctly RTCLSR model is promissing for global application with a minor update from the RTLSR model.