Analyses of bedrock groundwater levels as a factor of deep landslides by using functional models

Ken’ichirou KOSUGI, Yoshiki SANDO, Masamitsu FUJIMOTO, Yosuke YAMAKAWA, Naoya MASAOKA, Takahisa MIZUYAMA, Shin’ya HIRAMATSU, Taijirou FUKUYAMA and Takashi JITOUSONO

This study evaluated four functional models developed by the previous study (Kosugi et al., 2013), which calculate bedrock groundwater level (BGL) changes caused by rainwater infiltration by using antecedent precipitation indices(APIs). BGLs observed at 25 boreholes excavated at 5 different locations were used for the evaluation. Results showed that Pw2 model, which assumes power functions between BGLs and two types of APIs, was selected to be the best model for the14boreholes. Pw1 and Li1 models, which assume power and linear functions, respectively, and use a single API, were selected as the best model for10and1boreholes, respectively. Optimized half]life times (HLTs) of the APIs were generally longer than those used for shallow landslide and debris flow predictions. In general, the optimized parameters showed typical values for each location, suggesting that geological characteristics for each location had more dominant effects for BGL changes than characteristics of each point situated in the same geological settings. The proposed models can readily estimate BGLs anteceding storms and earthquakes. Moreover, they can be used for indicating the biggest BGL increases ever. Thus, the proposed models are effective for establishing evacuation systems against deep landslides and detecting vulnerable slopes.