Extracting desertification from landsat imagery using a feature space composed of vegetation index and albedo——a case study of Zhangye oasis and its adjacent areas

In this study,the Landsat TM image of Zhangye Oasis and its adjacent areas were used to retrieve the vegetation index and land surface albedo.A spectral feature space composed of normalized difference vegetation index and land surface albedo was constructed to evaluate present status and degree of desertification quantificationally.The classification error was evaluated by the data acquired from field survey.It indicated that the method was helpful to the extraction of desertification and the overall accuracy of about 82% was satisfying.Compared with the classifying method of only using spectrum information,the method using NDVI-albedo feature space produces a better effect on the classification of the present status and degree of desertification,and the result was tested.