A Research on Extracting Information of the Arid Regions´ Vegetation Coverage Using Improved Model of Spectral Mixture Analysis

As the low vegetation coverage in arid region, the pixels with full vegetation coverage could not be found for medium and low resolution images and it is influenced by the soil background when vegetation coverage information is extracted using vegetation index. Thus, the accuracy of extracting vegetation coverage in arid areas is not high. To solve this problem, this paper presented the model, which was fit for the extraction of the arid zones´ vegetation coverage. In this model, TM image was decomposed by linear spectral mixture mode and then the vegetation component of TM image was amended by the high-resolution satellite image. Compared with traditional and improved dimidiate pixel models respectively, this model is more suitable for quantifying vegetation coverage of arid and semi-arid regions, and for large-scale macro-monitoring by medium-resolution satellite images. Based on what have been mentioned above, the model has a very high practical value.