Monitoring methods for wetland changes of Maduo County of China based on Landsat TM data

The remote sense is a useful method to extract the information of wetland. To select an optimal method for extracting wetland information in the Tibetan Plateau, this study tested the accuracy of spectrum, normalized difference water index, modified normalized difference water index and unsupervised classification to extract the wetland information with Landsat TM data in the Maduo County of Qinghai Province. This study showed that the four methods successfully extracted the wetland information at a large scale. The accuracy of MNDWI to extract the wetland information was better than that of other three methods when the wetland distributed at above 4 500 m elevation regions and its accuracy was 80%. At low and mid regions, the accuracy of NDWI was higher than that of MNDWI, but the latter was higher than those of spectrum and unsupervised classification. This study suggested that the MNDWI was better to extract the wetland information due to minimizing the negative effects of the clouds and terrain.