Research
Hurricane Fiona, Puerto Rico (11/2022)
The combined effect of rainfall and earthquake can generate hundreds to thousands of landslides across mountainous regions. Yet, the coupled influence of rainfall and seismic activity on slope stability remains insufficiently understood, particularly at the regional scale.
To address this gap, Mirna Kassem and Dimitrios Zekkos developed CRISIS, an open-source, Python-based model for simulating shallow pseudo-3D rainfall- and earthquake-triggered landslides at regional scales.
Understanding how landslide failure mechanisms relate to topography, geomorphology, antecedent hydrological conditions, and subsurface properties, including shear strength and hydraulic characteristics, is essential for identifying the factors that control landslide occurrence across a landscape.
The CRISIS model predicts regional rainfall-induced landslides and distinguishes between two primary failure mechanisms: (1) saturation-induced failure, initiated from the bottom up by a rising groundwater table, and (2) wetting-induced failure, initiated from the top down by a downward-propagating wetting front or the development of a perched water table above less permeable subsurface layers.
An application to a watershed in Utuado, Puerto Rico, during Hurricane Maria (2017) demonstrated that these failure mechanisms are strongly influenced by several interacting factors, including hillslope geomorphic position, initial groundwater conditions, and surface cohesion.
Predicting rainfall-induced landslides across multiple spatial scales requires approaches that balance regional coverage with the physical detail needed to characterize slope failure. Combining global probabilistic models, such as LHASA, with high-resolution, physics-based simulations connects regional-scale hazard screening with detailed watershed-scale assessments, enabling broad identification of landslide-prone areas and closer examination of specific locations.
Applied to Hurricane Maria (2017) in Puerto Rico, this multiscale approach demonstrated how regional predictions across the island can guide high-resolution modeling within critical watersheds. Together, the models provide complementary insights, from identifying regional hotspots to predicting landslide locations, sizes, and timing at finer spatial scales.
Post-event reconnaissance provides a valuable opportunity to collect rich field data on landslide occurrence, geomorphology, geological conditions, and hydrological processes following extreme events. These observations help identify the factors controlling slope failures, improve our understanding of landslide mechanisms, and support the development and validation of predictive models.
We conducted a one-week reconnaissance in Puerto Rico in November 2022 as part of a GEER (Geotechnical Extreme Event Reconnaissance) deployment to investigate landslides triggered by Hurricane Fiona.