[H-DS08] 地すべりおよび関連現象

[HDS08-P02] Landslide Susceptibility Analysis and Mapping in Northern Kyushu Area Using Slope Hazard History and Geological Information

*川畑 大作¹、宮地 良典¹、阿部 朋弥¹、斎藤 眞¹、井柳 卓也²、小林 央宜²、大石 博之²、内村 公大²、水落 裕樹¹、宮川 歩夢¹ (1.国立研究開発法人　産業技術総合研究所、2.西日本技術開発株式会社)

キーワード：地すべり、すべりやすさマップ、シームレス地質図

The basic factors for landslide occurrence are influenced by multiple conditions such as topography, geology, and vegetation. Spatial information on these factors and past landslide histories can be used to determine which conditions predispose to the occurrence of landslides. In this study, an inventory dataset of topography, geology, and disaster history was compiled mainly for the northern Kyushu region, where heavy rainfall disasters have occurred frequently in recent years. Using 10-m mesh data from the Fundamental Geospatial Data (FGD) by GSI, we developed a topographic inventory of topographic features such as slope and relative relief, and a topographic inventory of watersheds. For the geologic inventory, we used Seamless digital geological map of Japan V2 1:200,000 and classified the geologic divisions into age data and lithologic facies. The same geological features that have been contact metamorphosed by igneous intrusions are harder and less susceptible to collapse. Therefore, the distribution of contact metamorphic zones should be classified in terms of landslide susceptibility and landslide occurrence among geological features of the same type. As data on past landslides to compare with these predispositions, we collected digital archive for landslide distribution map by the National Research Institute for Earth Science and Disaster Resilience (NIED) and information maintained by local governments and organizations and created an inventory data of landslide history.
In this study, using the inventory data, we analyzed the relationship between past landslides and their predisposing factors by dividing them into three categories: rainfall-induced, earthquake-induced, and other factors. In recent years, many studies have also used machine learning and other techniques for hazard mapping. Landslide susceptibility was mapped using multiple methods in this study.