Verification of a Two-Stage Slope Condition Estimation Method Using Real-Time Monitoring Records of a Rainfall-Induced Landslide

Shallow landslides caused by heavy rainfall pose significant hazards in mountainous Japan, necessitating improved early warning methodologies. Despite technological advances, the quantitative understanding of failure mechanisms based on comprehensive field observations remains limited. This study presents a multi-parameter monitoring system to detect precursory phenomena and proposes a two-stage estimation methodology based on a shallow landslide triggered by Typhoon Hagibis in 2019. The system combined rainfall gauges, inclinometers, soil moisture sensors, and tensiometers with 10-minute recording intervals. Analysis revealed relationships between groundwater level rise and displacement development, with initial movement occurring at critical groundwater thresholds and a two-phase deformation process corresponding to soil stratification. The proposed approach enables continuous monitoring during normal conditions and quantitative evaluation before failure. The first stage analyzes historical data to estimate instability progression based on hydraulic threshold monitoring, while the second stage applies the inverse velocity method to estimate failure time with excellent statistical reliability ( R 2 >0.9) and ±15-minute accuracy using post-acceleration data. This methodology provides a framework for real-time monitoring, demonstrating that shallow landslides develop through temporally staged deformation processes triggered by groundwater fluctuations.