Forecast:  There are signs pointing to raised European windstorm activity this coming winter, relative to the past 15-20 years. Uncertainty in loss outcomes is large.

Why is there a raised risk, relative to past 15-20 years?

Storm activity is modulated by interannual drivers, and their status in November 2022 points to more damage than the average over the past 15 years.

This winter will have an ongoing La Niña, a westerly QBO, moderate sea-ice (relative to past 15 years) and quite average solar activity. Labitzke and Kunz (2009) showed the joint effects on the late-winter (February) polar vortex due to three of these drivers in their Figure 4. The westerly QBO and moderate solar flux (about 120 solar flux units) point to a deeper polar vortex in late winter, and the La Niña is linked to stronger polar vortices in this part of QBO-solar flux phase space. The tendency for a deeper polar vortex favours a stormier Europe at the same time, in late winter. Some research focuses on European climate anomalies, such as Bronnimann (2007) who analysed the joint impact of westerly QBO and ENSO, and his Figure 13 suggests this winter would be expected to produce a signal which is around 25% of the total variability.

Sea-ice anomalies this autumn are quite average for the past 15 years, hence unlikely to drive a storm anomaly relative to the same period. Though latest research on sea-ice is worth mentioning. Smith et al. (2022) reported results from an experiment conducted by many modelling groups, in which sea-ice changes were tested (no other change), and model results were improved by novel post-processing. They found a negative NAO response which was 30% of the observed interannual deviation, due to a sea-ice reduction expected from almost 1C of global-mean warming. This represents quite a small impact on the NAO, however, European storm damage is not the same as time-mean circulation over the North Atlantic (for example, the current decadal mean of the winter NAO is approaching values last seen 30 years ago, but losses are much lower today than in the 1980s and '90s). I hope to discuss the relation between NAO and European storm damage in a future blog, and bridge this gap between insurance and research.

Other parts of the climate system can drive year-to-year changes in storminess. Notably, there has been no highly explosive tropical volcanoes injecting huge amounts of sulphur into the stratosphere, hence this driver (probably the strongest one) is inactive. The northern Atlantic has had slightly stronger north-south temperature gradients over the past decade which, at the very least, ought to slightly strengthen existing storms through air-sea heat exchange.

Why use a shorter climate context of past 15-20 years?

The forecasted windstorm risk is placed in the context of the past couple of decades, because (1) there is an observed shift from the stormy 1980s/90s to the quieter 21^st century climate, and (2) this shift is understood to be caused by rapid Arctic warming which has reduced low-level north-south temperature gradients, and leads to less windstorm damage in modern times.

The first point is underpinned by a wide variety of observations, including insurance loss experience. The second point is explained by the analysis in Wang et al. (2017), with more information in Chang (2018). Further, I’ve been reviewing the storm index that was presented in the two-part blog series earlier this summer, and found the current storm activity was lower than indicated in those blogs, and the most plausible cause is reduced low-level baroclinicity. For these reasons, the forecast for this winter is placed in the context of recent storm levels. 20^th century storm levels were from a different climate.

Why is uncertainty so large?

It’s reasonable to assume around half of all interannual variations is random variability, or at least unexplained at present. In contrast, the signal this winter (westerly QBO and La Niña) was given above as around one quarter of interannual deviations, hence the unpredictable internal variability is dominant. This means that while there is a higher chance of it being in the top tercile, the loss outcome could also finish in the bottom tercile of the modern storm climate.

We are quite sure to have a La Niña winter, though it has two unusual features. First, it is the third consecutive winter with La Niña conditions, and it is unclear whether its longevity may have conditioned the climate system to be more or less stormy in Europe. Second, the La Niña has an unusual spatial pattern with distinctly larger anomalies south of the equator. As a result, the Southern Oscillation Index is positive and suggests more flood and TC risk for Australia. However, the impacts on windstorms from a La Niña with diminished anomalies north of the equator are unknown.

Summary

The signs point to slightly higher European windstorm activity, relative to a 21st century norm, due to the combination of a La Niña and westerly QBO this winter.

The storm activity is expressed relative to climate over the past 15-20 years, because the rapid Arctic warming has moved the climate system to a new, lower state of storm activity this century, so far.

While the interannual drivers shift probabilities of loss outcomes this winter, the natural, random variability is larger than the signal. Further, the current La Niña is unusual in two respects (its longevity and north-south asymmetry) which injects greater uncertainty into the forecast.