Applications of soil moisture for three-dimensional landslide thresholds

In analyzing the trigger of landslides, numerous studies have paid attention to the importance of hydrological variables. Above all, precipitation is the main factor triggering landslides and debris flows. Since pore water pressure rise influenced by rainfall can lead to the reduction of slope stability, many studies tried to determine the rainfall-driven threshold to figure out the conditions of landslide initiation. Though rainfall-driven threshold (e.g., Intensity-duration curve) is simple and straightforward, universal use has been constrained due to the site-specific features, such as hydraulic parameters, soil texture, and anthropogenic activities. Recently, soil moisture is widely applied to enhance the detecting capability of thresholds. Since soil moisture reflects the condition of the ground directly, it can be used more effectively to identify fluctuations in pore pressure. Therefore, this study attempted to use both rainfall and soil moisture for determining the landslide thresholds. Daily precipitation from Global Precipitation Measurement (GPM) IMERG Final run and 3-hourly surface soil moisture from Global Land Data Assimilation System (GLDAS) L4 V2.1 were used to produce hydrological characteristics (i.e., Antecedent Precipitation Index (API) 24-hr accumulated precipitation, antecedent soil moisture, daily soil moisture, and soil moisture increment). Very firstly, two-dimensional relationships were shown to analyze the corresponding reactivity of each factor in the event of landslides. Based on these results, a three-dimensional critical plane was determined. In order to reflect the site-specific characteristics depending on the region, the thresholding process was divided into 2 steps. After obtaining the national scale threshold based on the probability distribution, regional-scale thresholds were optimized for each area. The capability was verified through validation. Results showed compared to the two-dimensional threshold, the three-dimensional critical plane showed similar accuracy rates but lower False Alarm Rates (FAR). In other words, soil moisture increase can detect landslides effectively and the three-dimensional critical plane can help understand the process of landslide occurrence. Furthermore, it seems possible to quantify the landslide vulnerability depending on the critical plane section.