Human impact index in landslide susceptibility mapping

Human activities which are represented by human impact index in landslide susceptibility mapping have great impact on the landslide occurrences in urban areas, especially in those fast developing areas. It is important to study how to determine a reasonable human impact index based on available factors, such as buildings, roads, and land prices, to ensure that landslide susceptibility is accurately determined. In this paper, twelve representation methods for human impact index based on available source factors are evaluated through a case study of Shenzhen in which general additive model and k-fold validation are applied. Various representation methods for human impact index lead to different landslide susceptibility maps in terms of AUC, SD, and probability distribution. It can be concluded that the human impact index should be carefully determined according to the factual situation of the study area to improve the landslide susceptibility model. This study may provide a guide for future studies on landslide susceptibility mapping.