Regional Land Degradation Mapping Using MODIS Data and Decision Tree (DT) Classification in a Transition Zone between Grassland and Cropland of Northeast China

In Northeast China, Grassland in Inner Mongolia borders the Songnen Plain has a semi-arid climate that is highly vulnerable to degradation, salinization, even desertification. Over the past decades, it has undergone significant environmental degradation, which implies not only the climate-related changes, but also the human-induced alterations, especially the inappropriate land use of overgrazing, agricultural intensification. The paper provides an effective method for MODIS data classification by decision tree (DT) classification methods. The MODIS data are thus classified by the decision tree methods. Meanwhile, the classification was also conducted by the maximum likelihood classification. The results of two classification methods were further compared to evaluate the DT method. The Kappa values and overall accuracies from the error matrices are 0.3476 and 48.9346% for ML classification method, and they are 0.4542 and 64.4610% respectively for DT classification method. The DT classification approach outperformed the ML classification approach for regional land degradation mapping. Thus, the cropland, grassland, water and degraded soil mapping at 500 m are generated every day for the study area by DT classification method.