In order to explore the evaluation model suitable for the semi-mountainous areas,the landslide disaster in was selected as the research object,and 9 evaluation factors including landform,slope,slope direction,curvature,relief,rainfall,distance from road,slope structure and elevation were selected to test their independence based on correlation analysis and collinearity diagnosis.The optimal Maxent model and weighted information model were used to calculate the probability of landslide occurrence. The natural discontinuity method based on GIS evaluates the result grade,and compares the rationality and accuracy of the two models through rationality test,sensitivity test and field test.The results show that:Both two models can pass the rationality test,and the AUC values are 0.902 and 0.769,respectively.The optimized Maxent model has higher accuracy,and the main impact factors of typical disaster points are consistent with the optimized Maxent model.Its stable,low,medium and high areas are 1585.74 km2 (26.70％),2138.42 km2 (36.90％),1594.48 km2 (26.85％)and 567.37 km2 (9.55％),respectively.
关键词
半山区/滑坡/优化Maxent模型/加权信息量模型
Key words
Semi-mountainous areas/landslip/optimization maxent model/weighted information value model
万方数据