A team of scientists led by the Euro-Mediterranean Center on Climate Change (Centro Euro-Mediterraneo sui Cambiamenti Climatici – CMCC) has developed a machine-learning (ML) prediction system capable of outperforming traditional forecasting systems, helping to protect vulnerable areas of Europe from heatwaves – one of the continent’s most pressing climate challenges.
Trained on data from centuries of climate analysis up to recent years, the system has demonstrated an increase in forecasting efficiency by significantly reducing the computational resources required. This makes it more cost-effective and accessible for a broader number of researchers and institutions.
The new model is said to be able to deliver forecasting insights 4-7 weeks before summer, in mid-March, providing a strong lead time for preparation measures. The researchers say that they were able to accurately forecast real-world heatwaves from 1993-2016, including extreme events like 2003 and 2015.
The model’s early warning capability could be particularly useful for climate services across agriculture, public health, energy and emergency planning, and help mitigate the effects of heatwaves that can often pose a risk to life and to the economy.
The study, Feature selection for data-driven seasonal forecasts of European heatwaves, published in Nature Communications Earth & Environment, is the product of the CMCC’s efforts to integrate the latest artificial intelligence technology with climate science.
Ronan McAdam, one of the lead scientists on the project, said that machine learning is set to become a “fundamental part” of how climate variability is studied. “This study has demonstrated the usefulness of ML in extreme event prediction, but it is only a first step in defining how we do that to receive interpretable and physically meaningful results,” he said.
Methodology
The systems employs an optimization-based feature selection framework that identifies the optimal combination of atmospheric, oceanic and land variables to predict heatwave likelihood across Europe. Using ML techniques, the approach analyzes roughly 2,000 potential predictors to select the most predictive combinations for each geographic location.
While it is capable of matching or even out-performing traditional forecasting systems, it also provides indicators on which predictors were used in the process. This is a valuable scientific resource, as being able to pinpoint which predictors contributed most to forecast skill can inform future research into the physical mechanisms behind heatwaves.
A persistent challenge in seasonal forecasting has been poor performance over Scandinavia and northern-central Europe. In contrast, the new data-driven approach developed in the paper improves skill in these previously problematic areas, the CMCC says. While traditional dynamical systems require enormous supercomputing resources to run, this approach focuses specifically on heatwave prediction with minimal computational overhead.
“Our research has successfully extended data-driven ML-based forecasting to the seasonal timescale using a tiny fraction of the computational resources of traditional approaches,” McAdam commented.
The framework is also said to have the potential to be adapted for other extreme weather events, start dates and target seasons.
In related news, MIT-developed lightning prediction tool targets safety of future aircraft
