Desertification mapping of Horqin sandy land, inner Mongolia, by means of remote sensing

Abstract: The extent of today's land degradation and desertification in the arid, semi-arid. and sub humid regions of China is a serious threat to the social and economic development in these areas. A national programme for monitoring and classification of drylands has been implemented by the Chinese Academy of Science. Areas are classified into degrees of desertification on the basis of vegetation cover, biomass change, deflation, and deposition. This study evaluates the use of Landsat MSS derived vegetation indices for desertification mapping as an alternative to the labour intensive field measurements used in the currently used classification system. Ratio-, difference-, normalized-, and soil adjusted vegetation indices images, together with MSS band 6, the first principal component, and maximum likelihood classified images are used and compared. Accuracy is evaluated using a thematic desertification map of Horqin Sandy Land, Inner Mongolia (1991), compiled by the Lanzhou Institute of Desert Research, Chinese Academy of Science, for validation. The green vegetation cover estimates images all showed comparable results, and no one stood out as superior to the others. Total accuracy is approximately 48% for all indices, while mean accuracy varies between 26.8% and 64.1% for individual classes. Annual plants are theorised to be the cause of the class confusion, hence the low accuracy.