摘要
Abstract
In order to scientifically and accurately analyze the susceptibility of landslides,this paper took the surrounding areas of Baoshan Basin as the study area and selected nine key factors contributing to and inducing landslides,namely,elevation,slope,aspect,Normal-ized Difference Vegetation Index(NDVI),engineering geological rock formations,distance to roads,distance to faults,distance to water systems and disaster point density,as evalua-tion indicators.Models based on the Certainty Factor(CF)method and a coupling of the CF method with Random Forest(CF-RF)were established to assess the landslide susceptibili-ty around the basin.The results revealed that the frequency ratios of landslide susceptibility levels significantly increased from low to extremely high susceptibility zones in both the CF and CF-RF models,effectively evaluating the susceptibility of landslides around the Baos-han Basin.Furthermore,after the CF model was refined with the introduction of the Ran-dom Forest algorithm,the CF-RF model achieved improved evaluation accuracy,with an increase in the AUC value by 0.034.Additionally,the importance of each evaluation factor was calculated and evaluated using two methods:mean decrease accuracy and mean decrease Gini coefficient.The results indicated that the disaster point density factor had a significant impact on the susceptibility of landslides around the Baoshan Basin.This study provides technical references for the monitoring,early warning,and prevention of landslides in the surrounding areas of the Baoshan Basin.关键词
确定性系数法/随机森林模型/滑坡易发性/耦合模型/保山盆地周边Key words
deterministic coefficient method/random forests/landslide susceptibility/coupled model/Baoshan basin perimeter分类
天文与地球科学
