Landslide susceptibility assessment using triangular fuzzy number-analytic hierarchy processing (TFN-AHP), contributing weight (CW) and random forest weighted frequency ratio (RF weighted FR) at the Pengyang county, Northwest China

Landslides in the loess area of China pose a great threat to the lives of local residents. Therefore, it is necessary to study the landslide susceptibility in this region. This paper takes Pengyang County in northwest China as the research area, selects 12 influencing factors [including elevation, slope, slope aspect, plan curvature, profile curvature, distance to rivers, groundwater types, land use, normalized difference vegetation index (NDVI), rainfall, lithology and peak ground acceleration(PGA)], and uses triangular fuzzy number-analytic hierarchy process (TFN-AHP), contributing weight (CW) and random forest weighted frequency ratio (RF weighted FR) to carry out the landslide susceptibility. The weight of landslide influencing factors and classified influencing factors calculated by TFN-AHP, CW and RF weighted FR are ranked. The AUC values of the landslide susceptibility assessment results of TFN-AHP, CW and RF weighted FR are all above 0.70, and the prediction effect is good. In addition, the new landslides found in the three field investigations is consistent with the superposition of high union very high landslide susceptibility areas, indicating that the landslide susceptibility assessment results made in this paper are reliable. The research results of this paper are helpful to explore the relationship between landslide and influencing factors in loess area, which can be used as a reference for the susceptibility assessment in loess area of China, and can provide a basis for disaster prevention and reduction for government departments in Southern Ningxia of China.