Modellizzazione delle valanghe di roccia e ghiaccio: applicazione del modello fisico RAMMS al ghiacciaio di Solatset, Valpelline, Italia

This research investigates whether, and to what extent inherent uncertainties in the simulation of complex natural processes influence the behaviour of rock–ice avalanches, with a particular focus on runout distances and potentially affected areas. For summer scenario variations in initial water content in the unstable glacier mass are analysed using physically based simulations generated with the RAMMS modelling framework, followed by a statistical sensitivity analysis of model outcomes. For winter scenario the same modelling framework is used to evaluate the effects of different snow cover depths over the glacier and along the flow path. A review of the literature shows that, while statistical approaches have been applied to other mass-movement processes, a dedicated statistical evaluation of rock–ice avalanche dynamics is still lacking. Simulating such hazardous events is increasingly important due to accelerated glacier retreat, which alters thermal and hydrological conditions. In the European Alps, where these events occur in densely populated regions, understanding rock–ice avalanche dynamics is essential for hazard assessment. Compared to seasonal snow avalanches, rock–ice avalanches are characterized by greater volume, density, and harmful consequences. The study focuses on the Solatset Glacier in the Aosta Valley, Italy whereas the Marmolada glacier collapse from 2022 was used as a reference event to calibrate friction parameters and relevant physical properties. Based on this, four release scenarios for the Solatset Glacier were simulated, representing different seasonal conditions and release volumes. A statistical framework was applied to evaluate the influence of key variables, particularly water content and ice temperature for summer scenario and seasonal snow amount and density for winter scenario. Water content and critical saturation were described using probabilistic distributions, while ice temperature was derived from an established relationship with water content following Ryser et al.. Due to uncertainty, no cross-correlation between the water content and critical saturation was assumed. Hazard extent appears to be more strongly controlled by release volume than by seasonal conditions. Comparison with results from r.avaflow indicates that the physically based modelling approach applied in this study produces more realistic flow behavior, particularly in terms of lateral spreading and counter-slope run-up. Overall, simulation results exhibit limited dispersion, with runout distances clustering in similar areas. This behavior may be influenced by input parameter distributions, sample size, or topographic constraints, and should be addressed in future work.