The Regional Differentiation on the Spatial Distribution and Influencing Factors of Potential Landslides Across the Entire Loess Plateau, China, Based on InSAR and Subregion XGBoost-SHAP Model

Abstract

The Loess Plateau (LP), with an area of 6.4 × 105 km2, has experienced numerous landslides triggered by earthquakes, rainfall, and anthropogenic activities for long history. However, it still lacks a comprehensive investigation on both the spatial distribution and inherent influencing factors of potential landslides across the entire LP due to its vast extension, active tectonic movement, diverse geomorphic types, and climate variations. We first apply interferometric synthetic aperture radar technology to identify 2052 potential landslides across the LP, which exhibit three landslide concentrated zones. Then, we adopt XGBoost (eXtreme Gradient Boosting) to model the effects of topographic, geomorphic, geological, hydrological factors on the landslides, and SHAP (Shapley additive explanation) algorithm to boost the interpretability and diaphaneity of modeling process. Considering the spatial differentiation of influencing factors for a wide area, we compare the patterns revealed by whole region modeling and subregion modeling. The results indicate that the subregion modeling is superior to whole region modeling, especially in revealing the distribution patterns of influencing factors with spatial heterogeneity. The influencing factors can be divided into two categories, those with spatial heterogeneity (i.e., elevation, fault, and road) and those without spatial heterogeneity (i.e., rain, river, and NDVI). The subregion modeling results indicate that rainfall has the greatest contribution to landslide development in all zones, and fault, NDVI and elevation are the subdominant factors in three zones, respectively. These findings can provide a reference framework for landslide detection and influencing factors analysis over large area.

Keywords

• Influencing factors
• Loess Plateau (LP)
• potential landslides
• spatial distribution
• spatial heterogeneity