Abstract: Disaster susceptibility mapping is the first step to improve the ability of control and prediction. In view of the problem that Jinsha River located in Yunnan province due to steep terrain and fragile ecological environment, combined with the increase of human activities in recent year, its basin has become the most frequent incident area of geological disaster in china. Taking Jinsha River of Deqin-Huapingas an example, this paper uses two methods(maxent and random forest model) for attribution and prediction of spatial distribution of landslides, and compares their differences. The result show that the prediction accuracy of random forest is higher than Maxent, and the AUC is 0.72.

Key words: landslide forecast, random forest, Maxent model, Jinsha River, susceptibility mapping