Europe faces an increasing threat from devastating heatwaves, underscoring the urgent need for reliable seasonal forecasts to protect lives and livelihoods.
The CMCC Foundation – Euro-Mediterranean Center on Climate Change recently published a study in Nature Communications Earth & Environment detailing a revolutionary machine learning (ML) system that can predict these dangerous summer events with unprecedented accuracy and efficiency.
The research highlights the potential of integrating cutting-edge artificial intelligence with climate science to address one of the continent’s most pressing challenges.
The new data-driven system matches and, in some cases, outperforms traditional forecasting models. Additionally, it delivers improvements in previously challenging areas, such as Scandinavia and northern-central Europe. The system achieves its greatest predictive skill by analyzing atmospheric, oceanic, and land variables, a crucial 4–7 weeks before summer (mid-March).
AI Offers Europe Heatwave Warning Months in Advance
Explore Tomorrow's World From Your Inbox
Get the latest science, technology, and sustainability content delivered to your inbox.
I understand that by providing my email address, I agree to receive emails from Tomorrow's World Today. I understand that I may opt out of receiving such communications at any time.
Early warning capabilities are vital for enabling public health services and the agricultural industry to implement effective mitigation strategies. CMCC researcher Ronan McAdam said, “Early warning of extremely hot summers could help society prepare to mitigate against economic losses and reduce risk to life.”
The model has already demonstrated its success. Researchers say it accurately forecasted real-world heatwaves from 1993 to 2016, including extreme events such as the deadly summers of 2003 and 2015.
Additionally, the system uses an optimization-based feature selection framework. This allows the system to efficiently sift through approximately 2,000 potential predictors to pinpoint the most critical ones, such as European soil moisture and distant signals from the tropical Pacific. As a result, it provides valuable insights into the physical mechanisms that drive extreme heat.
Moreover, the system is also cost-effective compared to traditional models that demand massive supercomputing resources. According to researchers, their new technique requires only “a tiny fraction of the computational resources of traditional approaches.” Importantly, this makes seasonal forecasting more accessible.
Researchers also used the ML system on paleoclimate simulations (years 0–1850) to overcome the scarcity of modern observational data. Despite learning from a simulated past, the system successfully transferred its knowledge to predict real-world events.
McAdam noted that this demonstrated success occurred because “There is not yet enough real-world data to train the forecast sufficiently, so the ML models actually learned about heatwave drivers in a model world but successfully applied the training to the real world.”