New method for landslide susceptibility mapping supported by spatial logistic regression and GeoDetector: A case study of Duwen Highway Basin, Sichuan Province, China

Landslides are destructive not only to property and infrastructure but also to people living in landslide-prone regions. Landslide susceptibility mapping (LSM) is critical for preventing and mitigating the negative impacts of land-slides. However, many previously proposed LSM modeling techniques included only the attribute information of spatial objects and ignored the spatial structural information of spatial objects, which led to suboptimal LSM. In addition, the selection of condition factors was not objective to such an extent that it may have reduced the reliability of LSM. To address these problems, a new method based on GeoDetector and a spatial logistic regression (SLR) model is proposed. GeoDetector is used to select condition factors based on the spatial distribution of landslides. The SLR model is used to make full use of the structural and attribute information of spatial objects simultaneously in LSM. The GeoDetector-SLR model is validated using data from the Duwen Highway Basin, which indudes the epicenter of the May 12, 2008 Wenchuan earthquake in southwestern China. Prediction accuracy of the GeoDetector-SLR model is found to be 86.1%, which is an 11.9% improvement over the traditional logistic regression model, indicating an improved and reliable solution for evaluating landslide susceptibility. (C) 2018 Published by Elsevier B.V.