In Japan, the risk assessment of rainfall-induced landslide disasters is based on the risk base line of sediment disasters occurrence (CL: Critical Line) set for each 1 km mesh using two values: the 60-minute accumulated rainfall (RI: Rainfall Intensity), a short-term rainfall index, and the soil precipitation index (SWI: Soil water Index), a long-term rainfall index. However, the current Japanese early warning system is set without considering the ground and topographical characteristics that are the endogenous factors of slope failures, and it assumes debris flows and large-scale landslides that occur intensively and does not target small-scale landslides such as cut slope collapse and embankment collapse. This study aims to develop a wide-area slope disaster risk assessment method for small-scale landslides that occur locally along roads by proposing a method for setting slope disaster risk base line that can be applied to various regions with the machine learning based on the numerical data representing regional characteristics, including topography and geology, in addition to meteorological data such as temperature and precipitation.
Ishikawa et al. (Thu,) studied this question.