1. Introduction
The Turpan oasis is a typical fragile environment that lies in an arid region of eastern Xinjiang and is affected by natural conditions and human activities. The severity of the land degradation and desertification in this area is increasing and it has adverse effects on people living and society economic development; therefore, desertification monitoring and controlling is important for environmental management of the region. Satellite remote sensing has several advantages for land use/cover change monitoring and desertification information extraction. Unlike aerial photography or in situ fieldwork, satellite remote sensing is cost-effective and less time-consuming, especially for large geographic areas [1]. Many kinds of remote sensing images, such as MODIS, TM, ETM+, SPOT and ALOS AVNIR-2 images had been applied to conduct land use/cover classification and thematic information extraction. However, due to the presence of same object with different spectra and different objects with same spectrum, the accuracy of remote sensing classification is limited. In recent years, several ground parameters which represent spectral, textural and spatial characteristics had been increasingly used in the remote sensing classification. In these studies, NDVI, MNDWI, SLOPE, principal components, textural features were always chosen [2], and these studies had achieved satisfactory results in many complicated tasks, such as forest ecosystem mapping, urban shaded area classifying, snow mapping and so on [3], [4]. For example, Jiang et al. used NDVI and elevation, and based on decision rules algorithm in conjunction with maximum likelihood classification, successfully extracted coastal wetland [5]. Fraser et al. used reflectance, temperature, and textural information for detecting large-scale forest cover change [6].