Comparison of landslide susceptibility assessment based on multiple hybrid models at county level: A case study for Puge County, Sichuan Province

Abstract

In order to effectively predict the spatial probability of landslide occurrence at county scale, the quantitative evaluation and comparative study of landslide susceptibility were carried out based on different statistical coupling models in Puge County, Southwest Sichuan. Six evaluation factors including slope, slope direction, elevation, engineering geological rock group, distance from fault and slope structure are selected to construct the evaluation index system. Information model (I), certainty factor model (CF), weight of evidence model (WF) and frequency ratio model (FR) are coupled with logistic regression model (LR) respectively to conduct landslide susceptibility evaluation. The results show that the evaluation results of each coupled model and the zoning of susceptibility are reasonable. The extremely high susceptibility areas with 129.04 −183.43 km2 (accounting for 6.77% −9.62%) are mainly distributed in the slope zones on both sides of Zemu River and Heishui River Valley. The evaluation accuracy decreases from WF-LR model (AUC=0.869), I-LR model (AUC=0.868), CF-LR model (AUC=0.866), to NFR-LR model (AUC=0.858). The research results can provide an important reference for the quantitative evaluation of county landslide susceptibility in mountainous areas of Southwest Sichuan.

Keywords

• landslides susceptibility
• information model
• certainty factor model
• weights of evidence model
• frequency ratio model
• logistic regression model
• puge county