IMAGE EXTRACTION OF MAQU WETLANDS AND ACCURACY ANALYSIS
WANG Xun, XU Dan-Dan, LI Wen-Long
College of Pastoral Agriculture Science and Technology, Lanzhou University; Key Laboratory of Grassland Agro-ecosystem,Ministry of Agriculture,Lanzhou 730000,China
This study is focused on the extraction of information concerning regional wetland based on Remote Sensing Multi-spectral Thematic Mapper data in the study area of Maqu Country, Gansu Province. For this purpose, local Digital Elevation Model, Normalized Difference Vegetation Index, and the first principal component extracted from principal components analysis were taken as the main factors of classification. A decision tree model was built based on these factors through spatial, spectral and statistic analysis for extracting wetland information. A comparison between the output of this model and that of unsupervised classification or maximum likelihood classification indicates that the multi-feature decision tree classification can be applied to the extraction of wetland information and that this classification is quite suitable for the study area.