Event definition

In the XAIDA case study the event was defined using daily and seasonal mean DJF temperature anomalies over Germany and Central Europe, and different attribution methods were applied.

Objectives

XAIDA aims at combining several methods to conclude whether or not cold winters such as in 1962-1963 (1963 hereafter) are still possible, despite climate change. Methods used are:

1. Dynamical adjustment of temperature trends
2. Circulation analogues for translating 1963 circulation to present-day climate
3. CESM2 amplification of tail events to translate the 1963 winter to the present day
4. Unconditional approach to translate 1963 to the present day
5. Empirical importance sampling to evaluate whether a worst-case cold winter such as 1963 is still possible
6. Climate model boosting to evaluate whether a worst-case cold winter such as 1963 is still possible

Study results

Sippel el at. (2024) analysed several aspects of winter 1963 over central Europe. Winter temperatures in central Europe, such as in Germany, have risen by about 2.5°C since the mid-20th century. Most of the year-to-year variability in winter temperatures is driven by changes in atmospheric circulation, which is closely tied to inter-annual temperature variability. However, there is also a residual warming trend not explained by circulation, likely caused by thermodynamic effects like global warming and feedbacks such as snow–albedo effects and the weakening of cold air advection in the mid-latitudes due to Arctic amplification. Changes in atmospheric circulation, with fewer cold spells and more zonal flows, have thus contributed to warmer winters, though models show little evidence for forced changes in circulation patterns. Therefore, the future of these circulation changes remains uncertain, and if driven by natural variability, conditions similar to the winter of 1963 could still occur, posing significant risks.

To explore how a winter like 1963 would play out in today’s climate, we used a circulation analogue method to simulate its effects. This approach focuses on how thermodynamic factors, like global warming, would change temperatures while keeping the atmospheric circulation patterns of 1963 fixed. The results show that such a winter would be about 1.4°C warmer now, with the strongest warming in areas where cold air originates and where the thermodynamical warming is the greatest. 

We compared our findings to simulations from the CESM2 climate model, which accurately reflects historical climate trends in Europe. The model shows that while extreme cold winters like 1963 are becoming less frequent, they can still happen, with cold events amplifying by about 1.6°C for every degree of global warming. This matches closely with the 1.4°C warming estimated using the circulation analogue method. Cold extremes are expected to warm faster than average winter temperatures due to stronger Arctic warming and changes in temperature gradients.

A statistical analysis of historical data from 1900 to 2023 suggests that the 1963 winter event, which was very rare at the time with a return period of 119 years, would be about 1.6°C warmer today due to global warming. Although cold winters like 1963 are becoming less likely, they are still possible, and even with the warming, a similar event could still be the second-coldest winter in Germany in the last 75 years. This highlights the risks of adapting too much to the mild winters of the past decade.

We compare the statistical method results with climate model boosting, which uses simulations to predict extreme events, like the Pacific Northwest heatwave of 2021. Our analysis shows that both atmospheric circulation and snow cover (which affects surface energy) are important drivers of cold winter temperatures, and the CESM2 climate model realistically simulates these factors, explaining the cold winters observed in both 1963 and simulations. Model boosting simulations of a hypothetical cold in 2020 simulations show two hypothetical « worst-case » cold winters, with temperature anomalies around −9 to −10°C, colder than the 1963 winter. Despite these extreme temperatures, the patterns of circulation and albedo are very similar to those in 1963, but more intense. While empirical importance sampling (e.g. Cadiou and Yiou, 2025) uses past data and is limited to the recent period (1972–2021), it still produces winters with similar temperature and circulation patterns, and some simulated winters are even colder than the 1963 event.

Conclusion 

If a 1963-like winter circulation pattern occurred today, it would cause a temperature anomaly of about -4.7 to -4.9°C relative to 1981–2010 averages, around 1.4 to 1.6°C warmer than the 1963 winter in Germany. This would still rank as the second-coldest winter in the last 75 years. Since 1950, changes in atmospheric circulation, with more zonal flow and less blocking, have driven recent warming. However, it is unclear if this change is natural or forced, and models predict little future change. If historical circulation changes were unforced, a reversal in circulation trends could bring back 1963-like patterns and very cold winters.

Using two advanced rare event sampling methods, we showed that temperatures as cold as or colder than those in 1963 are still physically possible in today’s and the near-future climate. While such extreme cold events are unlikely and represent worst-case scenarios, cold extremes will still occur despite global warming.

This highlights the importance of avoiding premature adaptation to the unusually mild winters experienced in central Europe over the past decade.

References

Cadiou, C. and Yiou, P.: Simulating record-shattering cold winters of the beginning of the 21st century in France, Weather Clim. Dynam., 6, 1–15, https://doi.org/10.5194/wcd-6-1-2025, 2025.

Sippel S., Barnes C., Cadiou C., Fischer E., Kew S., Kretschmer M., Philip S., Shepherd T. G., Singh J., Vautard R., and Yiou P.: “Could an extremely cold central European winter  such as 1963 happen again despite climate change?”, Weather Clim. Dynam., 5, 943–957 (2024)  https://doi.org/10.5194/wcd-5-943-2024