Rocky Desertification Feature Extraction and Spatio-temporal Evolution Analysis based on Decision Tree

Karst rocky desertification is a kind of land degradation phenomenon which is harmful and difficult to control. Rocky desertification monitoring based on remote sensing technology is an effective method. In this paper, Chongzuo and Qianxinan Prefecture in Guangxi Province are used as experimental areas, Landsat8 TM / OLI remote sensing images and ASTER GDEM elevation data are used as data sources, and three factors of vegetation coverage, bedrock exposure and slope are used as characteristic variables. The decision tree is constructed to extract the information of rocky desertification in the study area, and the rocky desertification grade is divided according to the experimental results. Finally, the temporal and spatial evolution characteristics are analyzed and discussed. The results show that: 1. The area of rocky desertification extracted in this paper is 94.5 % consistent with the land survey data. 2.the area without rocky desertification in Chongzuo increased by 11.25 % from 2010 to 2020; the area without rocky desertification in southwestern Guizhou increased by 8.36 % and the area of severe and extremely severe rocky desertification decreased by 10.31 % and 5.58 %, respectively. 3. The methods and conclusions of this paper can provide methods and data support for the evaluation of ecological security and rocky desertification control.