Study on Effective Influencing Factors of Common Landslides Susceptibility Methods

Earthquake-induced landslides are the most common hazard with serious consequences worldwide. Understanding the main controlling factors of the spatial distribution of coseismic landslides is of great significance for future disaster prevention. Mw 6.1 Ludian earthquake happened on 3 August 2014, Yunnan province, China. This study evaluated and compared landslide susceptibility maps (LSM) obtained from common landslides susceptibility methods, including frequency ratio (FR), information value (IV), and weights of evidence (WoE) respectively, and find out the most effective influencing factors of coseismic landslides Through remote sensing image interpretations and field surveys, landslide inventory including 1414 coseismic landslides was founded. In the inventory, 1000 landslides (70%) were randomly selected as training landslide dataset for the modeling using FR, IV, and WoE methods, and the remaining 414 landslides (30%) were used for the testing landslide dataset. The factors considered in the landslide susceptibility analysis are slope degree, aspect, elevation, lithology, distance to coseismic fault (BXF), distance to river, distance to road and intensity. Four different combinations of eight factors were taken into consideration in three methods. The results show that all three methods show good predictive capability according to AUCs. However, there is not much improvement in the accuracy of the results in each method. This paper further reveals that slope gradient and distance to coseismic fault are the most important factors controlling the spatial distribution of coseismic landslides. In addition, the model, which considers only these two factors, has sufficient accuracy.