Creating probability maps for avalanche hazardous areas reflecting snowpack uncertainty updates

Snow avalanches in snowy regions can become natural disasters when their flow paths intersect settlement areas or affect infrastructures. Hazard maps indicating avalanche runout areas can assist in decision-making to mitigate avalanche damage. Avalanche simulators can predict runout areas but require uncertain input variables, such as initial flow thickness and basal friction parameters. This study represents the uncertainty of input variables using probability distributions based on meteorological and snowpack measurements, and delineates hazardous regions with exceedance probabilities. By combining these distributions with a simulator, a probability map was created. The exceedance probabilities represent the likelihood of the simulator output (e.g., maximum flow thickness) exceeding a threshold value, and the probability map shows its spatial distribution. As probabilistic evaluations require a large number of numerical simulations, a polynomial emulator based on the polynomial chaos quadrature method was developed and validated for specific cases. We targeted a specific release area and assumed probability distributions based on data from an automated weather station near the release area. The probability maps, obtained using three different input probability distributions assumed for distinct cases, demonstrated that the input distributions influence both the probability values and runout zones. Sensitivity analyses identified influential input variables for each case. The simulated hazardous areas under appropriate input conditions correspond well to observed runout areas of a historical avalanche. Additionally, temporal variations in probability maps, which account for snow conditions, provide dynamic assessments of hazardous areas based on daily snowpack conditions. • Probabilistic avalanche hazard maps created using a polynomial chaos emulator. • Maps reflect snow condition variability with accuracy and low computational cost. • Sensitivity analysis identifies key inputs affecting runout extent. • Seasonal changes in snow conditions reflected on hazard maps.