Considering antecedent rainfall to improve susceptibility assessment of rainfall-earthquake-triggered landslides on unsaturated slopes

Rainfall and earthquakes are two major triggering factors for landslides. Previous studies have often used critical rainfall for assessment of occurrence of landslides induced by simultaneous rainfall and earthquakes; there has been minimal exploration of antecedent rainfall, which directly affects water content and saturation degree of soil before landslides occur due to both rainfall and earthquakes. Here, this study developed a model to assess the landslide probability of an unsaturated slope considering earthquake effect and under different critical and antecedent rainfall conditions. The proposed model includes two steps. In step 1, antecedent and critical rainfall are used as initial and boundary conditions, respectively, to simulate rainfall infiltration into an infinite unsaturated soil slope; this determines the depth of the wet zone. In step 2, factor of safety is calculated with consideration of seismic effects in the infinite slope model, using the wet zone depth determined in step 1. The spatial distribution of these factors of safety is used to classify susceptibility levels of landslide occurrence. The proposed model was assessed using data from a landslide event in Atsuma, Japan that had been caused by rainfall and earthquakes. The results showed that the area under the receiver operating characteristic curve was 83.4%, while confusion matrix-based accuracy was 63.7%. The proposed model was applied to Mt. Umyeon, Korea to evaluate the likelihood of landslide occurrence due to rainfall and earthquakes, according to four susceptibility levels: very low, low, moderate, and high. The results indicated that antecedent rainfall strongly affects the spatial probability of landslides induced by simultaneous rainfall and earthquakes. The proposed method may provide reasonable predictions for landslide susceptibility assessments.