Integrated analytical hierarchy process and numerical groundwater flow modelling approach for mapping landslide susceptibility in Kegalle District, Sri Lanka

Abstract

Landslides are becoming more frequent and destructive as the variability in climate and changes in the natural environment intensified. Numerous studies have considered geological, geomorphological, and human factors in landslide susceptibility analysis. However, studies considering subsurface hydrological processes in basin scale for landslide susceptibility analysis are relatively rare. This study investigated the importance of groundwater gradient as a triggering factor along with six other influencing factors belonging to four groups: topography, geological conditions, hydrological environment, and human engineering activities for landslide susceptibility in the Kegalle District area in Sri Lanka. The groundwater flow model was developed using the USGS modular finite-difference flow model (MODFLOW), and the spatial distribution of hydraulic gradient was estimated over a 1,523 km2 area. The estimated hydraulic gradient was then combined with slope, elevation, distance from the river, rainfall, land use, and soil texture for landslide susceptibility analysis using the analytical hierarchy process. The study area is divided into five groups, from low to high risk of susceptibilities. The results show a 60%-75% match between landslides that occurred in the past and the categories with extreme hydraulic gradients. The final landslide susceptibility map generated incorporating all other influencing factors demonstrates an F1-score of 0.814, highlighting the potential of the proposed methodology as a valuable tool for disaster risk assessment and management.