Retrieving BRDF of desert using time series of MODIS imagery

Desert plays a very import role on earth radiation budget and calibration research. In this paper, we propose a new algorithm for retrieving BRDF of desert using time series of MODIS imagery. The central idea of this algorithm is to detect the "clearest" observation during a temporal window for each pixel. For desert, the temporal window can be one year since its surface is highly stable. The clear observations are then used to fit the desert BRDF. Finally, the fitted BRDF is used to simulate the MODIS images under "real" conditions and the simulated MODIS images are compared with MODIS surface reflectance product (MOD09 and MYD09), which shows that the R2 and RMSE of the simulated surface reflectance is much better than those of MOD09 product. Therefore, the derived BRDF from this new algorithm much more accurately describe the directional characterization of the desert site.