Dynamic Landslide Susceptibility Analysis that Combines Rainfall Period, Rainfall, and Geospatial Information

Landslides occur due to irregularities in natural phenomena and are often difficult to predict. Typically, landslide disaster management relies on the anticipation of events using rainfall and spatial information characteristics using historical data. Also, in this study, a probability-based collapse model was developed considering 174 cases of occurrence in a time period unit. However, this study has three strengths. First, the landslide case survey was conducted based on the time of occurrence, as opposed to date. Second, drone photography was performed at recent event sites to confirm the exact locations of landslides, whereas only the exact time and place were selected for the older cases. Finally, unlike existing methods that apply rainfall and spatial information characteristics in separate dataset, the susceptibility of landslide was determined by combining the rainfall and spatial information. Rainfall and spatial information linkage models were developed using the PowerSim software. Two types of rainfall models were considered: a general cumulative model that determines the amount of rainfall generated during the previous antecedent period based on the point of occurrence and a rainfall inter-event time definition model that considers the drying time of the land. The results indicated the possibility of predicting the point of occurrence based on three variables: geospatial information, rainfall period, and rainfall. When the collapse possibility model is set to a period from the time of collapse, susceptibility can be analysed based on the amount of rainfall accumulated for a predetermined period for each geospatial piece of information. Excel data of susceptibility were extracted from the developed model and subsequently can link to the Geographical Information System to identify the possibility of collapse in real time according to changes in rainfall, thereby supporting forecast landslide occurrences. We expect that this research will help to reduce the landslide damages by contributing the landslide warning standards and the rainfall thresholds research field.