Air Pollution May Have Supercharged Europe’s Worst Winter Storms, Study Finds

For decades, insurers and meteorologists have puzzled over a striking pattern in European windstorm damage: losses soared to extraordinary heights in the 1980s and 1990s, then collapsed to levels comparable with the quieter 1960s and 1970s. Now a new study suggests that a surprising culprit may have helped drive that multidecadal swing — the very air pollution that industrialized nations pumped into the atmosphere during the twentieth century. According to research published in the journal Natural Hazards and Earth System Sciences, anthropogenic aerosols may have amplified European windstorm losses by an average of 45 percent in the late twentieth century compared with preindustrial conditions, with the effect ranging from nothing at all to a doubling of losses depending on which climate model is examined.

The research, conducted by Stephen Cusack of Stormwise Ltd in the United Kingdom, draws on a recently developed set of historical storm reconstructions that were extensively validated against insurance loss data. Those reconstructions revealed that insured property damages across Europe, indexed to 2022 values, rose from roughly EUR 2.8 billion per year in 1960–1979 to EUR 6.7 billion in 1980–1999, before falling back to EUR 2.5 billion in 2000–2019. That threefold variation over just a few decades dwarfs anything that can be explained by gradual global warming alone, and understanding its causes could fundamentally change how this costly risk is managed by governments, reinsurers, and coastal communities.

To isolate the role of aerosols, the study turned to a specialized suite of experiments from the Detection and Attribution Model Intercomparison Project, or DAMIP, part of the sixth phase of the Coupled Model Intercomparison Project (CMIP6). In these experiments, climate models simulate the industrial period from 1850 to 2014 with one type of forcing set to its historical values while all others are held fixed at preindustrial levels. By comparing simulations driven only by anthropogenic aerosol forcing against control runs with no external forcing at all, researchers can extract the fingerprint of aerosols alone — a powerful technique for separating human influences from the background noise of natural climate variability.

The physical mechanism at work is subtle but increasingly well documented. Sulphate aerosols, produced mainly by burning coal and oil in Europe, Russia, and North America, reflect sunlight back to space and cool the surface beneath them. Because these emissions were concentrated in the northern mid-latitudes, they altered the latitudinal gradient of net radiation at the top of the atmosphere, which in turn modified the poleward transport of energy by both the atmosphere and the ocean. Previous studies have shown that this perturbation strengthened the North Atlantic storm track in the second half of the twentieth century, steepening meridional pressure gradients over Europe and intensifying the winter winds that sweep in from the Atlantic.

The scale of the historical aerosol burden is remarkable. Sulphates generated the largest multidecadal anomalies in shortwave radiative forcing of the entire industrial era, growing from a negligible influence at the start of the twentieth century to peak negative forcing values in the 1980s and 1990s — precisely the period when European windstorm losses hit their documented maximum. Since then, aggressive air quality legislation on both sides of the Atlantic has slashed sulphur emissions, and the aerosol burden has fallen dramatically. Crucially, research indicates that the northern hemisphere storm track is particularly sensitive to emissions from the mid-latitude source regions where most of that sulphur originated, making the timing of the rise and fall of aerosol pollution a plausible match for the rise and fall of storm damages.

Converting raw model winds into monetary losses required a well-established damage function first published by Klawa and Ulbrich in 2003 and since validated against insurance loss timeseries and widely adopted in the field. The method extracts daily maximum near-surface winds across a domain of sixteen northern and central European countries that experience the vast majority of insured wind losses, identifies the 98th percentile wind threshold for each grid cell, and cubes the excess of winds above that threshold — a nonlinear relationship that reflects how damage grows disproportionately with wind speed. Storm events of up to three days are identified, and losses are weighted by population density so that the resulting index tracks where people and property actually lie. By holding population fixed at a common year, the analysis ensures that all modelled loss anomalies reflect changes in storm climate alone, isolated from socio-economic factors.

The headline result emerged clearly from the multimodel mean of 62 simulations spanning six climate models. European windstorm losses in the aerosol-forced runs peaked at 45 percent above preindustrial control levels at the end of the twentieth century, a signal that remained more than four standard errors above zero throughout the final two decades of the century. The timing of this peak coincides closely with the maximum in aerosol forcing from northern mid-latitudes, strengthening the physical case for causation. Yet the spread between individual models was enormous: four of the six models produced statistically significant increases in storm losses, with two showing near-certain signals, while two models simulated no significant aerosol effect at all. The overall mean was driven largely by two particularly responsive models, CanESM5 and HadGEM3-GC31-LL.

The study did not stop at the raw model output. Recognizing the wide spread, Cusack assessed the credibility of the modelled storm responses using independent evidence from earlier research, focusing on two key diagnostics: the strength of the aerosol radiative forcing itself, and the response of the Atlantic Meridional Overturning Circulation (AMOC) to that forcing. The logic follows from the fact that aerosol-driven changes in poleward energy transport manifest as anomalies in both the AMOC and the storm track, so a model that gets the forcing and the ocean response right should also get the storm response right. The assessment found that the models’ global aerosol effective radiative forcing averaged about 13 percent weaker in magnitude than the best observational estimate, and that the models tended to simulate AMOC responses larger than reconstructions of historical ocean behaviour — both biases that would push modelled storm responses downward rather than upward.

That validation carries an intriguing implication: if the models systematically underestimate the aerosol influence, the true contribution of air pollution to late twentieth century storm losses may sit at the higher end of the zero-to-100 percent range. The circulation evidence supports this reading. In the multimodel mean, aerosol forcing produced meridional gradients of 500 hPa geopotential height anomalies over the North Atlantic of about 12 meters for the 1970–1999 period relative to 1940–1969 — roughly one third of the magnitude seen in the ERA5 reanalysis, mirroring the ratio found for the loss anomalies themselves. Models with steeper circulation gradients, such as CanESM5, also produced the largest loss anomalies, confirming that the modelled damage changes are physically connected to large-scale shifts in the atmospheric circulation rather than being statistical artifacts.

Significant uncertainties remain, and the study is careful to acknowledge them. Reliable observational records of the European storm climate span only a few decades, direct measurements of the AMOC began only in 2004 with the RAPID array, and pre-instrument estimates of ocean circulation trends vary so widely that even the sign of twentieth century changes is debated. The validation framework also operates at a zonal-mean scale that cannot resolve processes local to the Atlantic sector, and it relies partly on the behaviour of a single climate model. Nevertheless, the evidence points to a provocative conclusion: the cleanup of air pollution that has delivered enormous public health benefits since the 1990s may also have removed an unintended amplifier of European storminess. Because future aerosol emissions from northern mid-latitudes are relatively predictable compared with stochastic forcings such as major volcanic eruptions, accounting for this aerosol-storm link could sharpen forecasts of windstorm risk in the decades ahead — a finding with direct consequences for the insurance industry, infrastructure planners, and everyone who lives in the path of Europe’s winter gales.

Subject of Research: The influence of anthropogenic aerosol forcing on multidecadal variability of European windstorm losses in CMIP6 DAMIP climate model experiments

Article Title: Anthropogenic aerosol forcing of European windstorms in CMIP6 climate models

Article References: Cusack, S. (2026). Anthropogenic aerosol forcing of European windstorms in CMIP6 climate models. Natural Hazards and Earth System Sciences, 26(10), 4663-4673. https://doi.org/10.5194/nhess-26-4663-2026

Image Credits: AI Generated

DOI: 10.5194/nhess-26-4663-2026

Keywords: European windstorms, anthropogenic aerosols, CMIP6, DAMIP, storm losses, climate models, North Atlantic storm track, AMOC, sulphate emissions, radiative forcing, insurance losses, multidecadal variability