The Comparative Analysis of Landslide Susceptibility Assessment of Dabie Mountain Area,Anhui Province Based on Different Models
Dabie Mountain area in Anhui Province is one of the areas in China with serious landslide disasters.Con-ducting a susceptibility assessment of landslides provides a scientific basis for determining the spatial distribution and causes of landslide-prone areas.In this study,extreme gradient boosting algorithm,K-nearest neighbor,logis-tic regression,support vector machine,and Stacking model fusion method were used,and Bayesian algorithm was used to optimize the model.The rainfall,vegetation cover,topography,geology,hydrology and other data in Dabie Mountain area from 1959 to 2020 were selected as inputs.The results are as follows:(1)The AUC of the XGBoost model on the validation set is 92.06%,and the Precision,Accuracy,Recall,and F1-score are high,indicating good generalization ability and suitability as a prediction model for the research area.The extremely high and high susceptibility areas determined by the model account for 23%and 16.2%of the total area,respectively,mainly distributed in Jinzhai County,Huoshan County,the southern part of Shucheng County,the northern part of Qianshan County,and the eastern part of Taihu County.(2)The feature importance ranking of the XGBoost model shows that lithology,slope,and rainfall in August are the most important influencing factors,while curvature and TWI are the least important influencing factors.
landslidemachine learningDabie Mountain area in Anhui Province
NETL
NSTL
万方数据
