Seasonal forecasts of European winter storms - skill, dynamics and constraints

Degenhardt, Lisa ORCID: 0000-0002-5291-204X (2023). Seasonal forecasts of European winter storms - skill, dynamics and constraints. University of Birmingham. Ph.D.

Preview

Degenhardt2023PhD.pdf
Text - Accepted Version
Available under License Creative Commons Attribution.
Download (45MB) | Preview

Abstract

Severe winter windstorms are one of the most damaging extreme events for the extra-tropics, especially in Europe. Companies like the SwissRe have estimated losses of around USD 190 million for European windstorms in alone 2022 (SwissRe, 2023). These rare events are a seasonal phenomenon. Hence, they occur especially on a seasonal scale, and their amount and strength vary each season. Besides the scientific purpose, politics, insurances and the general public are interested in understanding forecasts of these extreme events. They want to know how reliable those seasonally averaged forecasts are and how they can be improved. Hence, forecasts with a few months’ lead times are crucial for such events on a seasonal scale. The investigation covers multiple aspects of seasonally predicted winter windstorms over Europe. Firstly, the general forecast skill of seasonally averaged storm characteristics (frequency and intensity). Secondly, the link of these forecasts not only to known large-scale patterns like the North Atlantic Oscillation (NAO), Scandinavian Pattern (SCA) and East-Atlantic Pattern (EA) but also to more physics-based dynamical factors like the mean sea-level pressure (MSLP) gradient or measure for the instability of the atmosphere (Eady Growth Rate or equivalent potential temperature). The third part studies the constraints of the seasonal forecast model regarding windstorms and focuses on the signal-to-noise paradox.

This study investigates the seasonal forecast model of the UK Met Office, Global Seasonal forecasting system version 5, GloSea5, and compares it with different statistical measures to an observational data set, ERA5 from the European Centre for Medium-RangeWeather Forecasts (ECWMF). Windstorms are tracked over the core winter season, December-January-February, with an impact-based algorithm that focuses on the 2% strongest wind events over Europe and the North Atlantic.

The results of this study can be separated by the mentioned study aims. Firstly, a positive forecast skill, as correlation, for storm frequency and for the first time for storm intensity measures have been found over coherent regions over the British Isles and southern Scandinavia. Secondly, large-scale patterns like the NAO, SCA and EA have been linked to windstorms and explain up to 80% variability of windstorm frequency and 60% of intensity when statistically combined. But small-scale physical factors are investigated as well and it can be concluded that the seasonal forecast model is showing windstorm forecast skill out of the right physical connections between windstorms and atmospheric factors. The third part is about model constraints and therefore, the signal-to-noise paradox has been assessed within seasonal windstorm prediction. The paradox can limit the predictability of the model. For windstorm forecasts, this paradox exists strongly in the same regions as the significant forecast skill. A hypothesis that the disturbances of windstorms are causing the paradox has been investigated. Indications of a stronger signal-to-noise paradox within stormy seasons have been found in the GloSea5 model.

This all concludes that GloSea5 as a seasonal forecast model is capable of predicting the seasonal average of rare and extreme events like winter windstorms. This prediction is robust as it results from the right physical connections but is still reduced because of the existence of the signal-to-noise paradox.

Type of Work:

Thesis (Doctorates > Ph.D.)

Award Type:

Doctorates > Ph.D.

Supervisor(s):