Abstract ID: 3.10591 | Accepted as Talk | Talk/Oral | TBA | TBA

Numerical weather and climate predictions have improved significantly in recent years. This is mainly due to increased computational power and hence the ability to process more observations, include better physical parameterizations or increasing the horizontal as well as vertical resolution of used numerical weather and climate models. In general, one has to distinguish between short term weather prediction of a couple of days and long-term climate prediction for decades and even centuries. Similar to the numerical weather prediction the assessment of winter natural hazards requires an accurate nowcasting of the state of the snow cover in terms of stability, e.g., for avalanche warning or for snow loads on infrastructures. This is typically done by observers doing manual observations depending on the target, i.e. snow profiles and stability tests or measurements of the snow water equivalent (SWE). However, only by adding the information of a weather forecast to the current state a warning for the next day or days becomes possible. For climate predictions knowing the current state is of less importance and due to increased computational costs and a high demand on data storage future scenarios for snow are typically limited to simple question such as the future evolution of the snow extent. However, for short-term and mid-term planning of resources for warning purposes or even winter tourism local authorities would benefit from a prediction into the near future, i.e., a season or a couple of weeks or month. Seasonal forecast, i.e., forecast of a couple of months could bridge the gap between weather and climate timescales. In this study we explore the potential of coupling snow cover models with seasonal winter forecast for avalanche warning purposes as well as prediction of critical snow load events by using an ensemble of seasonal hindcasts of the last 20 winters. Note that by forcing snow cover models with data form atmospheric models there is no direct coupling and therefore no feedback between both models. Therefore, this pilot study shows also the benefits of fully coupled sophisticated snow cover models for a seamless prediction from weather to climate timescales.