Extraction of Land Cover Information in East China based on MODIS data

Extraction of Land Cover Information is important to make a study of global change. Taking East China as a study area, after MODIS data of this area in the four seasons are preprocessed, spectrum analysis of typical surface features are carried out. On these basis, by using decision tree classification, selecting spectral characteristics, NDVI and classification results of the maximum likelihood method as test variables, using proper thresholds for setting discriminating rules, a decision tree model is built. An assessment is given to the result by random samples. It is proved that the decision tree classification can get better accuracy result on the basis of the traditional supervised classification method. According to the method of Decision Tree, the image was classified in the consideration of multi-temporal and multi-spectral information. The research discusses a kind of classification scheme on extracting land cover information by the moderate resolution and high temporal resolution MODIS images, which can provide the technical support for rapid extraction of land cover information and contribute to real-time dynimic monitoring.