First experiences with a seamless weather forecast for severe weather products at MeteoSwiss

At MeteoSwiss, a suite of Numerical Weather Prediction (NWP) models is regularly run to provide the data basis for generating weather forecasts and severe weather warnings for the public. These models produce forecasts that differ in spatial resolution and extent as well as initialization frequency and lead time length. We combine those sources into a single probabilistic, gridded weather forecast that is seamless in space and time, without any postprocessing applied. The user can retrieve the seamless forecast at the spatial and temporal resolution, temporal aggregation and spatial extent they need. 

The resulting seamless forecast serves as input data for the Extreme Weather Identifier (EWI), a tool that summarizes the forecasting information focusing on occurrence probabilities of warning levels for different temporal aggregations for severe rain and wind events. 

Since February of this year, the EWI has been delivering three products to the forecasters every time a new seamless forecast becomes available: forecast percentiles (10,50,90), occurrence probabilities of warning levels at grid point level, and for warning regions. Our first systematic feedback collection among forecasters has confirmed that the forecasters appreciate the support by EWI products when deciding whether to issue warnings, with different products being most helpful in different weather conditions.  

We further evaluate the merit of these forecasts quantitatively by analyzing reforecasts of the seamless forecast and the contributing individual models. This happens for a set of past severe weather events over Switzerland. Reforecasts are compared with surface wind and rain observations as well as with a gridded precipitation product that combines radar and rain gauge information.  

In future developments, the EWI will be expanded to produce actual warning proposals for forecasters from this seamless forecast. Furthermore, seamless forecasts are not just desirable in the context of severe weather warnings but also for e.g., drought prediction or hydrological runoff modelling and can facilitate uncertainty communication. Hence, our seamless weather forecast is planned to be disseminated to customers from these sectors as well in the future.