The Influence of Storm Time Series Characteristics on Landslide Triggering within a Watershed in Utuado, Puerto Rico

Mechanistic models for predicting regional rainfall-induced landslides still face challenges, particularly due to limited input data and high spatial variability at the basin scale. These constraints often necessitate simplifying assumptions, impacting the reliability of results. For example, accurately measuring rainfall, especially in mountainous terrain and during extreme events, remains challenging, making it essential to understand how variations in storm time series influence slope stability predictions. This study presents a novel regional modeling framework that couples a pseudo-three-dimensional slope stability model with the Parallel Integrated Hydrological model (ParFlow) to predict failures. The framework captures two distinct failure mechanisms: (1) bottom-up via a rising groundwater table and (2) top-down via either a downward propagating wetting front or a perched water table. The framework was applied to a 3.25 km² watershed in Utuado, Puerto Rico, using six storms with the same total rainfall but varying in duration, peak intensity, and temporal distribution. Results show that short, intense storms triggered fewer failures than long, less intense events, despite having the same total rainfall. Under high antecedent soil moisture conditions, short, intense storms, where peak rainfall intensity exceeded the subsurface’s saturated hydraulic conductivity, predominantly triggered deeper bottom-up failures due to the rise of the already shallow groundwater table, while longer, less intense storms triggered both failure mechanisms. Conversely, under drier initial conditions with a deeper groundwater table, short, intense storms predominantly triggered top-down failures, while long, less intense ones led mostly to bottom-up failures. Uniform storms caused slightly more failures than Gaussian-shaped events.