Karst Collapse Risk Zonation and Evaluation in Wuhan, China Based on Analytic Hierarchy Process, Logistic Regression, and InSAR Angular Distortion Approaches

The current study presents a detailed assessment of risk zones related to karst collapse in Wuhan by analytical hierarchy process (AHP) and logistic regression (LR) models. results showed that the LR model was more accurate with an area under receiver operating characteristic (ROC) curve 0.911 compared 0.812 derived from AHP model. Both models performed well identifying high-risk only 3% discrepancy area. However, for medium- low-risk classes, although spatial distribution zoning were similar between two approaches, extent areas varied final reliability both methods reduced significantly excluding InSAR-based ground subsidence map analysis, presence falling into zone being approximately 14%, absence increased 6.5% on training samples. To evaluate practicality using maps zonation, are weighted angular distortion (WAD) method Wuhan. We find relatively large horizontal gradient values within belts generally spatially consistent class identified AHP- LR-based approaches. WAD-based approach cannot be used alone as ideal it does not include geological natural factors zonation.