Landslide susceptibility assessment in the Anfu County, China: comparing different statistical and probabilistic models considering the new topo-hydrological factor (HAND)

摘要: The present study is aimed at producing landslide susceptibility map of a landslide-prone area (Anfu County, China) by using evidential belief function (EBF), frequency ratio (FR) and Mahalanobis distance (MD) models. To this aim, 302 landslides were mapped based on earlier reports aerial photographs, as well as, carrying out several field surveys. inventory was randomly split into training dataset (70%; 212landslides) for the models remaining (30%; 90 landslides) cast off validation purpose. A total sixteen geo-environmental conditioning factors considered inputs to models: slope degree, aspect, plan curvature, profile new topo-hydrological factor termed height above nearest drainage (HAND), average annual rainfall, altitude, from rivers, roads, faults, lithology, normalized difference vegetation index (NDVI), sediment transport (STI), stream power (SPI), soil texture, land use/cover. maps evaluated under receiver operating characteristic curve (AUROC). As results, FR outperformed other with an AUROC 84.98%, followed EBF (78.63%) MD (78.50%) percentage classes each model revealed that managed build compendious focused highly susceptible areas (high very high classes) overall approximately 17%, (22.76%) (31%). premier attested five mostly influenced occurrence in area: NDVI, HAND. Interestingly, HAND could manifest clearer pattern regard compared such SPI, STI, rivers. Lastly, it can be conceived landsliding more subjected complex environmental set rather than anthropological ones (residential roads). This upshot make platform further pragmatic measures regarding hazard-planning actions.