Abstract: Three Gorges reservoir area is one of the ecologically sensitive areas in China. With the segment from Fengjie to Badong area as a case study, the main influencing factors of rock desertification are collected by remote sensing(RS) and GIS technology. The distribution map of rock desertification is extracted through the improved rock desertification enhance index. Moreover, the decision tree C5.0 algorithm is employed to obtain the importance of variables and rule sets. The relations between rock desertification and various factors are found out. Results show that rock desertification is caused by multi-factors instead of isolated factor. Vegetation coverage is the most intuitive external expression of rock desertification. High content of carbonate in rock near residential area is more likely to lead to rock desertification. When vegetation coverage is between 0.4-0.6, with slope of 35°-90°far from residential area, rock desertification is less likely to happen. And when vegetation coverage is low, with slope of 15°-35°and high content of carbonate in the rock near the residential area, rock desertification is highly possible. This research lays foundation for the quantitative analysis of rock desertification.

Key words:  rock desertification, desertification enhance index, remote sensing, decision tree, data mining