Remote Sensing Dynamic Monitoring And Research Of Sandy Degradation At County Level

Sandy Desertification is one of the most serious environmental and social-economic problems in the world. It directly influences regional economic development and social stability. The entironment is great threatened with sandy desertification in agriculture and husbandry interlace zone of northwest China. Now, it is inefficient to use manual visual interpretation method and assisted by the computer automatic classification method on many monitoring project for degradation. Using remotely sensing automatic-extract techniques for Desertification monitoring at County Level, it can understand the distribution and development trend of sandy desertification with great efficiency and speed. It also can provide effective technical support for the decision and sandy land management project.In the paper, Hengshan County in the province of Shanxi China, is chosen to be the research Area, which is local on the south edge of the Mu Us Sandy Land. Based on the analysis of the Landsat TM(1990) and Landsat ETM(2002) data, it do the monitoring and assessment of sandy desertification on Hengshan County.1. the monitoring of sandy landBased on the analysis of Multi-temporal satellite image and the field investigation data, three methods are contrasted and applied to evaluate the sandy Desertification. There are sandy feature Index (SFI), Line Spectral Mixture Model (LSMM) and Temperature/Vegetation Dryness Index(TVDI).2. the assessment of sandy landThere are two methods to assessment of sandy land. One is Comprehensive assessment method. The assessment facters include five facters as follow: soiltype(ST), land use&land cover (LULC), sandy area ratio in a pixel (SRP) , Modified Soil-Adjusted Vegetation Index(MASVI) and soil wet index(SWI). The images of five factors are recoded, and their weights are decided by analytical hierarchy process. Based on calculating the respective coefficient of impact extent with assessment results of each factor, the sandy land grading is evaluated Comprehensively by GIS. The other is backpropagation Neural Network method. Firstly, the images (SRP, MASVI and SWI )are masked by the extracted image of sandy; Then, the masked images(MASVI and SWI) are inverse extended , the masked SRP image is also extended, these three images as three layer were stacked to a image Secordly, the supervised Classification is done on the stacked image for sandy land level by means of Neural Network method.3. the analysis of the trend of sandy desertificationBased on the change information of sandy land between two period (1990 and 2002) and the social-economic statistic data, the trend and cause ofsandy desertification in hengshan are analysised. it also can advise to mamage to sandy land.