中文摘要：

以位于水蚀风蚀复合区的陕西省横山县为例,通过分析土地覆被类型在TM影像上的光谱特征,选择不同的特征参数,逐层提取有关信息：首先用TM5/Tm^3提取水体,然后用NDVI分离植被和非植被;在非植被区,用NDBI和NDVI相结合提取城镇用地,再用TM5将沙地和荒草地分离;在植被区,首先根据NDVI分为三大类：草地、旱地和灌木林、水浇地和有林地,然后利用TM4区分旱地和灌木林,用TM5-TM4的差值把水浇地和有林地分离。精度评价结果表明,分类总精度为86.8%,Kappa系数为0.8506,达到分类要求标准。

英文摘要：

Taking water-wind erosion region of Hengshan County in Shaanxi Province as a study area,certain information was extracted layer by layer through analyzing spectral characteristics of different land cover type by TM image and choosing different characteristic parameter and classification method： first extracting water area by TM5/Tm^3,and then dividing the other region into vegetation region and non-vegetation region by NDVI.In non-vegetation region,town land was extracted through NDBI and NDVI,and then sandy land and waste grassland were separated by TM5.While the vegetation region was divided into three classes by NDVI： grassland,dryland and shrub forest,irrigated land and forest land.Then dryland and shrub forest were separated in class 2 by TM4,irrigated land and shrub forest in class 3 by the difference of TM5-TM4.The result of accuracy evaluation showed that the overall classification accuracy is 86.8%,the Kappa coefficient is 0.8506 and they satisfied the classification accuracy standard.