Prolonged heatwaves, cold spells, and other extreme temperature events, can have major impacts on healthcare, food production, infrastructure, and ecosystems. Being able to reliably anticipate these events in advance can help governments and businesses to plan ahead, reduce risks, and improve climate resilience. Climate centres around the world routinely produce forecasts for the coming season or long-term outlooks for the decade, but it is less well understood how well we can predict extreme temperature events one to two years ahead. 

This study assesses how well a leading multi‑year climate prediction system can globally predict, up to two years in advance, changes in the number of days per calendar season that are associated with extreme warm and cold events.

Key findings 

• Changes in the number of  extreme temperature events can be predicted up to two years ahead in many regions, with the greatest predictability in the tropics. 
• Central and northern South America and Southeast Asia emerge as the most promising regions for extended-seasonal forecasts of extreme temperature events. 
• At higher latitudes, predictability is more limited, though significant skill is found in parts of North America during winter and spring, extending beyond the first forecast year. Europe shows limited predictability of warm extreme events during winter, but there is higher predictability during other seasons, particularly in southeastern Europe. 
• Most of the predictability is driven by outside influences such as volcanic eruptions and greenhouse gases. However, the study also found that the El Niño Southern Oscillation (ENSO), a naturally occurring pattern of variability in Pacific sea surface temperatures with strong influences on global weather, also plays an important role in determining forecast skill for warm and cold extreme events.

What’s next?

Further research should test other forecasting systems and look for factors that influence how predictable these extreme events are. This would help improve forecast accuracy in regions where the current model does not perform as well. Advancing research in this area will improve understanding of what drives temperature extremes to make forecasts more useful for decision-making.  

Read the full study in Environmental Research Letters.  

Reference: Tsartsali, E.E., Yeager, S.G., Athanasiadis, P.J., Tibaldi, S. and Gualdi, S., (2025). Predictability of temperature extremes in multi-annual forecasts. Environmental Research Letters, 20(10), p.104004. 

Keywords: temperature extremes, multiannual forecasts, interannual predictability, prediction skill, extreme event forecasting