Snow avalanches are a major natural hazard, endangering human life and infrastructure in mountainous areas throughout the world. In many countries with seasonally snow‐covered mountains, avalanche‐forecasting services reliably warn the public by issuing occurrence probabilities for a certain region. However, at present, a single avalanche event cannot be predicted in time and space. Much about the release process remains unknown, mainly because of the highly variable, layered character of the snowpack, a highly porous material that exists close to its melting point. The complex interaction between terrain, snowpack, and meteorological conditions leading to avalanche release is commonly described as avalanche formation. It is relevant to hazard mapping and essential to short‐term forecasting, which involves weighting many contributory factors. Alternatively, the release process can be studied and modeled. This approach relies heavily on snow mechanics and snow properties, including texture. While the effect of meteorological conditions or changes on the deformational behavior of snow is known in qualitative or semiquantitative manner, the knowledge of the quantitative relation between snow texture and mechanical properties is limited, but promising developments are under way. Fracture mechanical models have been applied to explain the fracture propagation, and micromechanical models including the two competing processes (damage and sintering) have been applied to explain snow failure. There are knowledge gaps between the sequence of processes that lead to the release of the snow slab: snow deformation and failure, damage accumulation, fracture initiation, and fracture propagation. Simultaneously, the spatial variability that affects damage, fracture initiation, and fracture propagation has to be considered. This review focuses on dry snow slab avalanches and shows that dealing with a highly porous media close to its melting point and processes covering several orders of scale, from the size of a bond between snow grains to the size of a mountain slope, will continue to be very challenging.

The failure of a weak snow layer buried below cohesive slab layers is a necessary, but insufficient, condition for the release of a dry-snow slab avalanche. The size of the crack in the weak layer must also exceed a critical length to propagate across a slope. In contrast to pioneering shear-based approaches, recent developments account for weak layer collapse and allow for better explaining typical observations of remote triggering from low-angle terrain. However, these new models predict a critical length for crack propagation that is almost independent of slope angle, a rather surprising and counterintuitive result. Based on discrete element simulations we propose a new analytical expression for the critical crack length. This new model reconciles past approaches by considering for the first time the complex interplay between slab elasticity and the mechanical behavior of the weak layer including its structural collapse. The crack begins to propagate when the stress induced by slab loading and deformation at the crack tip exceeds the limit given by the failure envelope of the weak layer. The model can reproduce crack propagation on low-angle terrain and the decrease in critical length with increasing slope angle as modeled in numerical experiments. The good agreement of our new model with extensive field data and the ease of implementation in the snow cover model SNOWPACK opens a promising prospect for improving avalanche forecasting.

Even today, the assessment of avalanche danger is by and large a subjective yet data-based decision-making process. Human experts analyse heterogeneous data volumes, diverse in scale, and conclude on the avalanche scenario based on their experience. Nowadays, modern machine learning methods and the rise in computing power in combination with physical snow cover modelling open up new possibilities for developing decision support tools for operational avalanche forecasting. Therefore, we developed a fully data-driven approach to assess the regional avalanche danger level, the key component in public avalanche forecasts, for dry-snow conditions in the Swiss Alps. Using a large data set of more than 20 years of meteorological data measured by a network of automated weather stations, which are located at the elevation of potential avalanche starting zones, and snow cover simulations driven with these input weather data, we trained two random forest (RF) classifiers. The first classifier (RF 1) was trained relying on the forecast danger levels published in the official Swiss avalanche bulletin. To reduce the uncertainty resulting from using the forecast danger level as target variable, we trained a second classifier (RF 2) that relies on a quality-controlled subset of danger level labels. We optimized the RF classifiers by selecting the best set of input features combining meteorological variables and features extracted from the simulated profiles. The accuracy of the models, i.e. the percentage of correct danger level predictions, ranged between 74 % and 76 % for RF 1 and between 72 % and 78 % for RF 2. We assessed the accuracy of forecasts with nowcast assessments of avalanche danger by well-trained observers. The performance of both models was similar to the agreement rate between forecast and nowcast assessments of the current experience-based Swiss avalanche forecasts (which is estimated to be 76 %). The models performed consistently well throughout the Swiss Alps, thus in different climatic regions, albeit with some regional differences. Our results suggest that the models may well have potential to become a valuable supplementary decision support tool for avalanche forecasters when assessing avalanche hazard.