Author(s): Jie Liu; Yuji Sugihara
Linked Author(s): Yuji Sugihara
Keywords: Landslide deep learning heavy rainfall slope unit risk assessment
Abstract: Kitakyushu City, located in the northern part of Kyushu Island, Japan, has frequently experienced natural disasters caused by heavy rainfalls. In particular, landslides by short-time heavy rainfalls have caused enormous economic losses and human suffering. In recent years, many studies have focused on evaluating landslide susceptibility with AI technologies. In this study, we investigated potential relationships among influencing factors regarding landslides by applying a data augmentation technique based on the slope units to landslide disasters. Moreover, through a transfer learning approach, a pixel-point-based Deep Learning Neural Network (DLNN) model was constructed to make the spatial distribution of the landslide susceptibility in the Kitakyushu area.
Year: 2025