Abstract: Hazard mapping is a prevailing part of spatial analysis of landslides. Previous researches use runout distances to map the hazard ranges. In this paper, we present an improved methodology by using the dataset that contains all runout distances of landslide locations in Heifangtai area. According to the runout distances, the landslide locations are categorized into six groups. For each group, the kernel density estimation and Value-at-Risk (VaR) measurement are conducted for statistical modeling. Statistical results indicate a kernel density with MISE=68.7238 fit the probability distributions of runout distances best. Furthermore, for each group, hazards are mapped according to the runout distances at different levels of risks (D_R). According to the experimental results, a preventive construction measure is proposed in the location computed as D_R for Groups I, II, III and VI. Meanwhile, the correlation between moisture content and runout distance in Group IV and V is derived by further numerical analysis.

Key words: hazard mapping, Kernel density estimation, Value-at-Risk, runout distance of landslide, Value-at-Risk measurements of runout distances