Direct-Estimation Algorithm for Mapping Daily Land-Surface Broadband Albedo From MODIS Data

Abstract

Land surface albedo is a critical parameter in surface-energy budget studies. Over the past several decades, many albedo products are generated from remote-sensing data sets. The Moderate Resolution Imaging Spectroradiometer (MODIS) bidirectional reflectance distribution function (BRDF)/Albedo algorithm is used to routinely produce eight day (16-day composite), 1-km resolution MODIS albedo products. When some natural processes or human activities occur, the land-surface broadband albedo can change rapidly, so it is necessary to enhance the temporal resolution of albedo product. We present a direct-estimation algorithm for mapping daily land-surface broadband albedo from MODIS data. The polarization and directionality of the Earth's reflectance-3/polarization and anisotropy of reflectances for atmospheric sciences coupled with observations from a Lidar BRDF database is employed as a training data set, and the 6S atmospheric radiative transfer code is used to simulate the top-of-atmosphere (TOA) reflectances. Then a relationship between TOA reflectances and land-surface broadband albedos is developed using an angular bin regression method. The robustness of this method for different angular bins, aerosol conditions, and land-cover types is analyzed. Simulation results show that the absolute error of this algorithm is ${\sim}{0.009}$ for vegetation, 0.012 for soil, and 0.030 for snow/ice. Validation of the direct-estimation algorithm against in situ measurement data shows that the proposed method is capable of characterizing the temporal variation of albedo, especially when the land-surface BRDF changes rapidly.