Development of shallow landslide susceptibility maps incorporating relative spacing index for forest management

This study aims to obtain valuable insights into forest management that can reduce the risk of shallow landslides. We developed shallow landslide susceptibility maps (SLSMs) incorporating tree height and the relative spacing index as indicators of forest management for Cryptomeria japonica and Chamaecyparis obtusa plantations, focusing on the shallow landslides that occurred during the July 2012 Northern Kyushu heavy rain in the Aso region of Kumamoto Prefecture. We created SLSMs by developing a Random Forest model. The explanatory variables included tree height and relative spacing index, topography, geology, rainfall, and other factors totaling eleven factors. The area under the curve for the model was 0.94, indicating very high accuracy. The importance of the relative spacing index was comparatively low. The impact of forest management on shallow landslides was quantitatively assessed via a partial dependence plot for the relative spacing index. The results indicated that the relative spacing index that minimized the landslide probability was 23.7, below the cut-off value that distinguishes between landslide and non-landslide. In addition, the SLSMs indicated that implementing forest management with a uniform relative spacing index of 23.7 could reduce the ‘Very High' susceptibility area by 77.76 ha compared to the case without forest management. The threshold of the relative spacing index obtained in this study are crucial for future plantation forest management practices aimed at reducing the risk of shallow landslides. • In the Aso region of Japan, shallow landslide susceptibility maps were proposed using indicators for forest management. • It was quantitatively demonstrated that forest management can reduce the risk of shallow landslides. • The shallow landslide susceptibility maps will contribute to facilitating consensus-building among forestry workers, local residents, and government.