We propose a method to extract the feature of Bi-directional Reflectance Distribution Functions （BRDF） over Badain Jaran Desert using Landsat-TM/ETM+ and ASTER GDEM data. Badain Jaran Desert is characterized with homogeneous and rugged terrain

which forms a natural Bi-directional Reflectance data sets with hypotheses that the surface structure of each slope element does not vary with the variations of slope and aspect. Therefore

we use nadir view Landsat-TM/ETM+ imagery to reconstruct the BRDF characterization of this experimental site. To begin with we use ASTER GDEM product to retrieve the sun-view geometry for each slope element corresponding to one pixel in remotely sensed imagery. Then the observations from Landsat-TM/ETM+ imagery and the retrieved sun-view geometries are used together to extract the BRDF feature of the experimental site. Finally

we simulate surface reflectance for 25 scenes of different Landsat-TM and ETM+ images and compare the simulations with the actual Landsat-TM and ETM+ imageries. Results show that the difference between the simulation and "true" surface reflectance are 2.80%

1.92%

2.68% and 2.32%

respectively

which is higher than the radiometric calibration error requirement of 5% to 7%. Therefore

the combination of nadir viewing Landsat-TM/ETM+ imagery with ASTER GDEM data can be used to cross-calibrate remotely sensed imagery with large viewing observation.