How well do seasonal forecasts reproduce European winter storm clustering and its relationship to large-scale climate drivers?

The inter-annual variability of the European windstorm season is dependent on a number of large-scale climate drivers and conditions, for example the North Atlantic Oscillation. For seasonal forecasts to provide valuable information to decision makers about the potential severity of the winter windstorm season, they must capture this relationship between large-scale climate drivers and seasonal windstorm frequency in advance. Here, we examine the performance of the latest state of the art ECMWF seasonal forecast product (SEAS5) in capturing this climate response. We apply a statistical model previously shown to well reproduce the explained behaviour of European windstorms from large-scale climate drivers (Walz et al. 2018) to SEAS5, and examine the choice of statistically significant drivers. The model applied is a stepwise Poisson regression approach to account for serial clustering within inter-annual variability of windstorms, the resultant of which categorizes each windstorm season as either active, neutral or inactive. In particular, we focus on the European region where the explained variance of the statistical model in observations is highest (Walz et al. 2018), the British Isles. In addition to comparing the performance of the model in SEAS5 and in observations, we examine which relationships are not recreated in the seasonal forecast successfully from a dynamical perspective, to provide further insight into the current ability of seasonal forecasts to represent European windstorm inter-annual variability.

 

Reference:

Walz, M. A., Befort, D. J., Kirchner‐Bossi, N. O., Ulbrich, U., & Leckebusch, G. C. (2018). Modelling serial clustering and inter‐annual variability of European winter windstorms based on large‐scale drivers. International Journal of Climatology, 38(7), 3044-3057.