Integrated Landscape Pattern and Classification Tree for Land Use Classification in Inland River Basin of China Arid Region

Inland river basins are a significant geomorphologic unit in the arid region of northwestern China, which are composed of mountains, oasis, desert, river, lake, and wetland. In spite of global climate change influenced regional environment, human activities and water utilization are the major reasons for land degradation and desertification in the inland river basin. Land use/land-cover (LUC) mapping is always an important step in study the relationship between anthropogenic causes and natural environment factors. Decision trees represent nonparametric methods, handle nonlinear relationships, and allow for missing values. In this paper, we integrated ecoregions to map LUC in the heterogeneous inland river basin. Based on confusion matrix analysis we found that the method is significant and effective in LUC classification in inland river basin in China arid regions, with the overall accuracy 72.1% and Kappa statistic 0.69.