SHALLOW LANDSLIDE DETECTION BASED ON HYDROLOGICAL-GEOTECHNICAL MODELING AND REMOTE SENSING DATA: A CASE STUDY IN SOUTH CHINA

Among the world’s most destructive natural hazards, rainfall-induced shallow landslides are becoming even more frequent and widespread. Due to the combined impact of monsoon climate and hilly topography, South China suffers frequent severe shallow landslides during the rainy season, usually resulting in heavy casualties and huge property losses. To illustrate the risk map of shallow landslide in certain areas, we integrated the empirical analysis method and physically based modeling to estimate the possibility and to detect the occurrence of shallow landslides. In this study, first, a weighted combination method was developed to illustrate the susceptibility of landslide in South China based on the geospatial remote sensing data of soil, slope, elevation, land cover and drainage density. Then, the rainfall intensity and amount were used to further determine the likelihood of landslides. For regions with high possibility, a physically based model should be implemented to estimate the timing and location of landslides. In this study, we chose Jianxi River (a tributary of Min River) basin in Fujian Province, South China as the testbed to setup a coupled hydrological-geotechnical model called Coupled Routing and Excess Storage and SLope-Infiltration-Distributed Equilibrium (CRESLIDE). Specifically, the 3B42V7 rainfall product of Tropical Rainfall Measuring Mission (TRMM) based Multi-satellite Precipitation Analysis (TMPA) at 3-hour temporal resolution was compared with the ground observation. Using the 3B42V7 rainfall as the model input, the spatial and temporal (3h/90m) distribution of the factor of safety showed a similar pattern with that derived from ground observed rainfall during the period of June to October, 2010. When further validated with the available landslide events from wire-service news reports, results indicated that most events could be detected by CRESLIDE model, even using satellite rainfall. This study provides a useful prototype in mapping characteristics and detecting the occurrence of shallow landslides in South China. It can be extended to further applications if more data (e.g., forecasting precipitation, streamflow observation, historical shallow landslide events) is available.