Analysis of different modeling strategies for forecasting snowmelt during rain-on-snow events

Rain-on-snow (RoS) floods pose a threat to human life and infrastructure, and snowmelt can be a significant contributor to these events. Associated with the current climate warming phase, RoS events are increasing in prevalence and are occurring in places traditionally not prone to such conditions. Short- and long-term planning based on model simulations play key roles in establishing community preparedness. This research focuses on short-term snowmelt forecasts during RoS events. Snowmelt rates during these events are very sensitive to turbulent heat fluxes. These fluxes can be quite strong when high winds accompany an event and lead to strong melting of the snowpack. Unfortunately though, even small changes in the ambient conditions can dramatically change the available melt energy. Hence, in operational settings forcing data uncertainty must be accounted for.

Physics-based, energy-balance (EB) snow models - robust to rapidly changing conditions and the weather extremes often present in RoS events - are widely considered the ideal tool for evaluating RoS snowmelt. Tight time constraints in operational settings however, limit the number of EB simulations possible. Probabilistic forecasts capable of accounting for potential weather forecast errors require numerous simulations and may not be possible using the physics-based solutions. Conceptual models, such as degree-factor models, have far lower computational requirements but have demonstrable weaknesses in rain-on-snow events dominated by turbulent heat exchanges. In this research we ask the question whether physics-based turbulent flux calculations from a designed experiment can be aptly parameterized to allow solutions that are fast enough to permit extensive probabilistic analysis for operational applications while maintaining the characteristics of the more complex EB solutions.

In this presentation we compare and contrast the above RoS modeling strategies at four long-term snow observation sites located across a range of elevations and climate zones: Weissfluhjoch (2540 masl) in the Swiss Alps, Col de Porte (1325 masl) in the Chartreuse Mountains of France, and wind-sheltered (2067 masl) and wind-exposed sites (2091 masl) in Reynolds Mountain East, U.S.A. We further demonstrate spatial applications over entire Switzerland.