Spatial Non-Stationarity-Based Landslide Susceptibility Assessment Using PCAMGWR Model

Landslide Susceptibility Assessment (LSA) is a fundamental component of landslide risk management and substantial area geospatial research. Previous researchers have considered the spatial non-stationarity relationship between occurrences Conditioning Factors (LCFs) as fixed effects. The effects consider scale different LCFs an average value, which represented by single bandwidth in Geographically Weighted Regression (GWR) model. present study analyzes effect landslides explains influence factor correlation on LSA. A Principal-Component-Analysis-based Multiscale GWR (PCAMGWR) model proposed for susceptibility mapping, hexagonal neighborhoods express proximity extract input. under receiver operating characteristic curve other statistical indicators are used to compare PCAMGWR with GWR-based models global regression models, has best prediction effect. Different scales obtained improve accuracy compared scale.