[H-DS05] Landslides and related phenomena

[HDS05-P15] Characteristics and susceptibility assessment of the earthquake-triggered landslides in moderate-minor earthquake prone areas at southern margin of Sichuan Basin, China

*Hong Wen^1,2, Xiyong Wu², Sixiang Ling², Chunwei Sun², Guiyu Zhou³, Qiang Liu⁴ (1.Xihua University, 2.Southwest Jiaotong University, 3.Yibin University, 4.Sichuan Institute of Geological Engineering Investigation Group Co. Ltd)

Keywords:Landslides triggered by moderate-minor earthquake, Landslide susceptibility, Weighted information value, Weighted certainty factor, Support vector machine

This study aims to characterize the earthquake-triggered landslides in moderate-minor earthquake prone areas at the southern margin of Sichuan Basin, China, and to assess the susceptibility of these landslides. 284 EQ (Earthquake)-triggered landslides were inventoried from field-based investigation and dominated by rockfall, soil slide, and rock avalanche. Statistical analyses show that the landslide mostly occurred at the elevations of 300–700 m and slope angle of 5–30°. The volume of EQ-triggered landslides ranged from 20 to 1 510 000 m³ with concentrations in the range of 20–10 000 m³, corresponding to 90% of the total number of landslides. Among these landslides triggered by moderate-minor EQs, 38.4% occurred in the areas seriously affected by the mining and 89.8% occurred in the carbonate rock distribution areas. The geological conditions with widely developed karst and goaf amplified the influence of moderate-minor earthquakes. 14 key conditioning factors are filtrated through the colinearity diagnosis of Pearson coefficient and selected to construct landslide susceptibility map by Weighted Information Value Method (E-IV), Weighted Certainty Factor Method (E-CF), and Support Vector Machine (SVM) methods. The results show that the AUC values of E-IV, E-CF, and SVM are 0.867, 0.857, and 0.884, respectively, indicating three methods present good prediction accuracy, among which the SVM has the highest accuracy. Landslides mostly clustered in the coal-mining and carbonate rock area, which are widely affected by karstification (dissolution void or fissures) and goaf. Meanwhile, the very high landslide susceptibility areas are mainly distributed in the Gongchang Anticline tectonic area with 652 km² from the best SVM method. This study can be used for decision-makers in China to prepare and develop intervention strategies for preventing landslides in frequent moderate-minor EQ-impacted regions.