Using Multi-spectral Remote Sensing Data to Extract and Analyze the Vegetation Information in Desert Areas

The coverage and spatial distribution of vegetation in desert areas are fundamental indexes to estimate the desertification severity, and acquiring vegetation information is very beneficial to carry out desertification monitoring and evaluation. Based on Landsat TM image of 2007, vegetation information of the western Gurbantunggut Desert was extracted with vegetation indexes and spectral mixture analysis. The results show that: the U-min method (one of the SMA methods) is the best one and this method was used for further analysis. Moreover, a significant linear relationship is found between vegetation coverage and vegetation fraction extracted from SMA, with a correlation coefficient of 0.858. This indicate that vegetation coverage in desert areas can be extracted through remote sensing images indirectly.