Remote Sensing Dynamic Analyse Of Desertification Evolution In Qinghai Hu

Land desertification is one of the environmental disasters in the world. Investigating the distribution and dynamic change trend of desertification has important significance for Desertification Prevention. Remote sensing technology has been applied to investigation on land desertification. There are two classification methods to investigate desertification by applying Remote sensing technology: computer automatic classification and interpretation on screen. This paper made use of Landsat data investigating desertification situation of three periods in Qinghai Hu using Remote sensing technology. Studying on computer automatic classification on desertification by selecting test area.The paper including:Many methods have been used to enhance the multi-spectral bands of TM data, such as linear contrast enhancement and color transforms and the principal components transformation, analyzing the effect of above methods on enhancing desertification. As a result, contrast enhancement and color transforms have been good for enhancing desertification but the principal components transformation in the test area.Applying Landsat data(MSS,TM,ETM) investigated three periods distribution of desertification in Qinghai Hu and acquired three desertification classification maps. The paper analyzed the dynamic change trend in Qinghai Hu in the past thirty years, and found that the area of slight desertification had been reduced but the area of medium and severe desertification had been increased year after year. The cause of desertification in Qinghai Hu is the effect of nature and human being.Based on investigation of previous efforts on desertification index and source of Remote sensing data(TM), this paper proposed a desertification index system using Remote sensing technique. Three desertification indices (MSAVI, FVC, Albedo) were retrieved from Landsat(TM) satellite data in the test area. Supported by the desertification index system, the desertification status in the test area was classified by decision tree classifier. The classification accuracy was low when using the desertification index system. So we divided the test area into two parts: firstly, using the desertification index system extracted the part the classification accuracy of which was high, secondly, the two parts in the test area were classified by different decision tree, and the precision of classification was enhanced.Virtual reality of Qinghai Hu was realized by applying VirtualGIS model of ERDAS image, it made the observer interpret the studying region more accurately and corrected the result of interpretation on desertification.