Robustness in the Estimation of BRDF Model Parameters Using Limited Number of Observations

Bidirectional Reflectance Distribution Function (BRDF) is reported to represent a characteristics of reflectance’s angular dependency, and each land cover has different type of BRDF. BRDF products are operationally produced from Moderate Resolution Imaging Spectroradiometer (MODIS) data, and the estimated BRDF parameters are used to produce MODIS albedo products. Albedo is one of the most important parameters for heat and water budge analysis. However, observed data are often contaminated by errors caused by imperfect cloud mask and atmospheric correction. And, rainy season also limits available observations. Therefore, robust estimation of BRDF model parameters is required. Operationally, albedo is estimated from remotely sensed data through the production of Bidirectional Reflectance Distribution Function (BRDF). However, a limited number of observations is one of the most severe problems for the estimation of the BRDF. In the present paper, experimental results are reported that examine the robustness with respect to the scarcity of the number of observations. In the present research, the LSM was examined in terms of robustness with respect to the scarcity of the number of observations. In the experiments, BRDFs were estimated by the LSM using MODIS reflectance data on paddy fields in Japan in March, June and August, 2002 were used. The results revealed that the degree of the robustness depends on the condition of the land cover, even if the land cover is classified as the same category.