[H-DS09] Human environment and disaster risk

[HDS09-P04] Analysis of the factors influenced the landslides caused by 2018 heavy rain in the Hiroshima area in Japan

*Hiromu Daimaru¹ (1.Forestry and Forest Products Research Institute)

Keywords:July 2018 rain storm, shallow landslides, risk evaluation, hazard map, Hiroshima Prefecture, Random Forest

In eastern Hiroshima Prefecture, heavy rains in July 2018 caused numerous shallow landslides. The author analyzed the factors that influenced the occurrence of the landslides by using airborne LiDAR data conducted by the Forestry Agency after the disaster. Landslide polygons produced by the Forestry Agency were used as the objective variable. An overlay analysis was conducted by using 100-meter mesh system, and the effects of explanatory variables such as rainfall, slope, elevation, geology, NDVI, and vegetation height on the occurrence of landslides were evaluated using random forest analysis. The results showed that the importance of vegetation height, elevation, NDVI, slope, and rainfall was particularly high. The influence of geology could not be fully evaluated because most of the study area is underlain by Cretaceous granitic and rhyolitic rocks. Actually, a concentrated area of collapsed land is recognized that cannot be explained by the results of this analysis. Aerial photographic observation showed that many landslides originating from the embankment of the forest road network occurred in these concentrated areas. In order to improve the accuracy of the landslide risk evaluation model, it is necessary to include the status of artificial landform modifications such as forest road networks in the evaluation.