Dynamical models for calculating snow avalanche motion have gained growing importance in recent years for avalanche hazard assessment. Nevertheless, inherent uncertainties in their input-data specification, although well acknowledged, are usually not explicitly incorporated into the analysis and considered in the mapping results. In particular, the estimate of avalanche release conditions is affected by strong uncertainties when associated to a return period. These sources of error are normally addressed through sensitivity analysis or conservative parameters estimate. However, each of these approaches has limitations in assessing the statistical implications of uncertainties. In the present paper the problem of release scenarios randomness is looked at following a Monte Carlo procedure. This statistical sampling-analysis method allows the evaluation of the probability distributions of relevant variables for avalanche hazard assessment - such as runout distance and impact pressure - once the release variables - essentially release depth and release length - are expressed in terms of probability distributions, accounting explicitly for inherent uncertainties in their definition. Both the theoretical framework of this procedure and its application to a real study case are presented. As initial step of this research in the present work the attention is mainly focused on flowing avalanches descending on open slopes. Therefore, the one-dimensional version of VARA dynamic models is used for avalanche simulations.
