Maddison et al. (2026) Using seasonal forecasts to enhance our understanding of extreme wind and precipitation impacts from extratropical cyclones

Identification

• Journal: Natural hazards and earth system sciences
• Year: 2026
• Date: 2026-02-19
• Authors: Jacob Maddison, Jennifer L. Catto, Sandra Hansen, Ching Ho Justin Ng, Stefan Siegert
• DOI: 10.5194/nhess-26-827-2026

Research Groups

• Department of Mathematics and Statistics, University of Exeter, Exeter, UK
• Guy Carpenter & Company Limited, London, UK

Short Summary

This study utilizes nearly 700 years of seasonal forecast model output to quantify the likelihood of unprecedented wind and precipitation impacts from European extratropical cyclones (ETCs). It finds that the probability of an ETC having an impact more extreme than any observed is generally between 0.5 % and 1.6 % for wind and 0.2 % and 0.7 % for precipitation, with the North Atlantic Oscillation strongly influencing wind impact likelihood.

Objective

• To produce a dataset of extratropical cyclone tracks and their associated wind and precipitation footprints, spanning several hundred years, for studying extreme storms and aiding catastrophe model development and evaluation.
• To use this dataset to gain insight into the potential impacts of higher return period storms, specifically to estimate the probability of unprecedented ETC impacts (more extreme than any observed in reanalysis) from both wind and precipitation.

Study Configuration

• Spatial Scale: Europe, focusing on 14 countries: Austria, Belgium, Switzerland, Czechia, Germany, Denmark, France, Finland, Great Britain, Ireland, Norway, Netherlands, Poland, and Sweden.
• Temporal Scale: 672 extended winter seasons (October–March), derived from seasonal hindcasts covering 1993–2016, with each hindcast running for 215 days.

Methodology and Data

• Models used:
  • Global Seasonal Forecast System version 6 (GloSea6-GC3.2 system 601) from the Met Office.
  • Objective feature tracking algorithm TRACK (Hodges, 1994, 1995, 1999) for cyclone identification and tracking.
  • Storm Severity Index (SSI) (Klawa and Ulbrich, 2003) for wind impact estimation.
  • Quantile mapping (Thrasher et al., 2012) for converting wind speed SSI to wind gust SSI.
  • Generalized Pareto Distribution (GPD) fit using the Peaks Over Threshold (POT) method for return period estimation.
• Data sources:
  • Primary: GloSea6 seasonal hindcasts (1993–2016), 7 ensemble members, 4 initiation dates (1, 9, 17, 25 September). Data retrieved from Copernicus Climate Data Store (CDS, 2023). Fields include mean sea level pressure and 10 m horizontal wind components (6-hourly), and daily precipitation totals. Spatial resolution: F128 Gaussian grid (approximately 0.7° × 0.7° latitude–longitude).
  • Verification: ERA5 reanalysis (Hersbach et al., 2020) for 1993–2016, at F128 spatial and 6-hourly/daily temporal resolutions, for 10 m wind components, maximum 10 m wind gust, mean sea level pressure, and daily precipitation totals.

Main Results

• GloSea6 accurately represents North Atlantic storm track characteristics (track density, mean intensity) and surface weather (10 m wind speed, ETC-associated precipitation) with generally small biases compared to ERA5.
• Unprecedented wind impacts are found in GloSea6 for most countries, with the most impactful storms being approximately 1.5 times stronger than the maximum in ERA5. The likelihood of an ETC having an unprecedented wind impact is generally between 0.5 % and 1.6 % across the considered European countries.
• Unprecedented precipitation impacts are also identified in GloSea6, with maximum impacts around 1.5 times larger than in ERA5. The likelihood of an ETC having an unprecedented precipitation impact is lower, typically between 0.2 % and 0.7 %.
• The North Atlantic Oscillation (NAO) is strongly related to European ETC wind impact: strongly positive NAO values approximately double the likelihood of an unprecedented wind impact, while strongly negative values halve it.
• The NAO has a weaker relationship with ETC-related precipitation impacts, with seasonal precipitation totals only weakly correlated to the seasonal mean NAO index.
• For high return periods, GloSea6 SSI curves generally align with ERA5 extrapolations, though for Great Britain and the Netherlands, GloSea6 suggests a flattening of the SSI curve for the most extreme events compared to ERA5 extrapolations.

Contributions

• First application of the UNprecedented Simulated Extreme ENsemble (UNSEEN) methodology to quantify both wind and precipitation impacts from extratropical cyclones.
• Generation of a unique 672-year dataset of European ETC tracks and their associated wind and precipitation footprints, providing a significantly larger sample size for studying high-return-period events than observational records.
• Quantification of the probabilities of unprecedented ETC wind and precipitation impacts across multiple European countries, offering critical insights for risk assessment in the insurance sector.
• Demonstration of the influence of the North Atlantic Oscillation on the likelihood of unprecedented wind impacts, suggesting a potential avenue for extended seasonal predictability of extreme wind events.
• Development and validation of a quantile-mapping approach to convert wind-speed-based Storm Severity Indices (SSIs) to wind-gust-based SSIs, making model output directly comparable to industry-standard catastrophe models.

Funding

• Guy Carpenter & Company Limited

Citation

@article{Maddison2026Using,
  author = {Maddison, Jacob and Catto, Jennifer L. and Hansen, Sandra and Ng, Ching Ho Justin and Siegert, Stefan},
  title = {Using seasonal forecasts to enhance our understanding of extreme wind and precipitation impacts from extratropical cyclones},
  journal = {Natural hazards and earth system sciences},
  year = {2026},
  doi = {10.5194/nhess-26-827-2026},
  url = {https://doi.org/10.5194/nhess-26-827-2026}
}