From storm-induced failure to runout simulations: A Monte Carlo-based probabilistic assessment of potential landslide scenarios in the Granitztal (Austria)

Runout analyses are widely used to simulate the propagation of landslides and debris flows in order to predict deposition and flow heights for hazard assessment and risk management. Among the available approaches and methods, physically based numerical models require the definition of multiple input parameters and boundary conditions, including rheological properties and potentially unstable volumes. Especially in a predictive context (forward analysis), a key challenge relates to the assumptions adopted during the model parametrization to realistically simulate the material behavior, often resulting in a strong user dependence of the modelling outcomes. Particularly, this uncertainty can mask the predictive accuracy of the simulations, affecting both the spatial distribution of deposits and the assessment of potentially affected areas. Such limitations are significantly evident where site-specific soil properties or documented past events are not available, further increasing subjectivity and underlining the need for approaches that explore a wide range of scenarios.  
In this study, a probabilistic framework based on a Monte Carlo approach is presented to evaluate runout simulations implemented in the software RAMMS::Debrisflow (Rapid Mass Movement Simulation:: Debrisflow). Instead of defining “best-fit” parameter sets, the Monte Carlo approach allows the analyses of a large number of simulations, each performed using an independent set of input parameters randomly sampled from defined statistical distributions. The framework is applied to assess potential mobilizations of a complex earth-slide system in the Granitztal (Carinthia, S Austria), which initially occurred in August 2023 following an extreme rainfall associated to the “Zacharias” storm that triggered multiple earth-slides and mudflows across the region. As the slope has not fully failed, it represents an ongoing hazard for residents and threatens the buildings located in the lower part of the slope. 
Field investigations and multi-temporal monitoring were conducted using Electrical Resistivity Tomography (ERT) and UAV-derived data to provide spatially distributed information on the subsurface structure of the slope and surface morphologies, identifying features of progressive deformation and potentially mobilizable volumes. These datasets are used to constrain release volumes in the RAMMS simulations, allowing data-driven runout patterns within the explored scenarios. The resulting large set of simulations is analyzed statistically to derive runout metrics and to evaluate the spatial variability of predicted deposition heights. By including a broader range of scenarios, this study demonstrates the value of data-driven, probabilistic runout modelling in reducing user dependence and improving the robustness of predictive hazard assessment.