Numerical modelling of ice avalanches from the Planpincieux glacier (Italy)

Forward simulations of ice avalanches are important to inform risk management. However, the reliability of such simulations often suffers from the dependency of model parameters on the process magnitude, hampering the simulation of unprecedented events in a given area. We suggest a reliable, straightforward and practically applicable work flow for the forward simulation of ice avalanches for the purpose of risk management with regard to the Planpincieux glacier, located on the Italian side of Mont Blanc massif.

Since 2013, the Planpincieux glacier, has been studied to analyse the dynamics of ice collapses in a temperate glacier. Several documented ice avalanches and glacial floods (1929, 1952, 1982, 2005, 2017), which, in some cases, threatened the village of Planpincieux and damaged the municipal road, have been linked to this glacier. Starting from the summer of 2019, a fast moving ice volume, partially separated by the rest of the glacier tongue by a large crevasse, has drastically increased the occurrence of a new collapse with possible implications for the valley floor. Considering the potential risk, a glacier constant monitoring (GbInSAR) and an avalanche early warning system (avalanche Doppler radar) were deployed, and numerical modelling of ice avalanches from this glacier was made.

Thereby, we couple an empirical-statistical model with a physically-based mass flow model: (I) the rules of Alean (1985) for the angle of reach are fed into the software r.randomwalk in order to estimate worst-case reference travel distances for various scenarios of starting volumes, (II) the basal friction angle used in the physically-based tool r.avaflow is optimized against those reference travel distances for each volume scenario, (III) the travel distances and travel times are checked for plausibility against well-documented events, (IV) flow pressures and flow kinetic energies are computed with r.avaflow for each volume scenario.

The model results are well supported by empirical data for smaller events, whereas direct reference data for the larger scenarios are not available. Interpretation of the results further has to take into account that (A) for some scenarios, the empirical relationships had to be extended beyond their known range of validity, introducing additional uncertainty, and (B) the relationships do not work for snow-covered trajectories, that, for example, would decrease the friction and lead to longer travel distances. As a result, the outcomes can be, with some care, considered as worst-case assumptions for ice avalanches in summer, but are not valid for ice avalanches during the other seasons.