U.S. West research partners have harnessed the capabilities of machine learning to deduce how and when heat waves occur amid changing climate conditions.
Their low-cost new tool, detailed in Science Advances on Wednesday, serves to help clarify connections between global warming and individual extreme weather events and shift the way scientists predict future such phenomena.
In developing this approach, the authors expressed hope that the results could both steer climate adaptation strategies and provide evidence for plaintiffs seeking legal compensation for climate-related injuries.
“We’ve seen the impacts that extreme weather events can have on human health, infrastructure, and ecosystems,” said lead author Jared Trok, a doctoral candidate in Earth system science at the Stanford Doerr School of Sustainability, in a statement.
“To design effective solutions, we need to better understand the extent to which global warming drives changes in these extreme events,” he added.
Trok and his colleagues — from Stanford and Colorado State University — trained AI-powered models to predict daily maximum temperatures based on both regional weather conditions and global mean temperature.
Following the training, which relied on climate simulations from 1850 to 2100, the authors shifted focus to real-world heat waves: predicting how hot those events would have been under the same weather conditions but different levels of warming. They also factored in how climate change has influenced the frequency and severity of historical weather events.
In their first real-world analysis, which looked at the fatal 2023 Texas heat wave, the team determined that global warming caused the event to be 2.12 to 2.56 degrees Fahrenheit hotter than it would have been without the effects of climate change.
Events on par with record-breaking heat waves in Europe, Russia and India, they found, could occur multiple times per decade if global temperatures continue to surge, according to the study.
While the authors viewed the initial utility of their approach with optimism, they acknowledged that further analysis is necessary before the tool “can be used for high-stakes applications such as improving adaptation decisions, attributing climate damages and informing climate litigation.”
Yet just as artificial intelligence has become crucial in divulging complex connections within climate datasets, so too could it open a window “into the historical and future influence of climate change on extreme events,” the scientists stated.
“AI hasn’t solved all the scientific challenges,” senior author Noah Diffenbaugh, a professor of Earth system science at the Stanford Doerr School, said in a statement.
“But this new method is a really exciting advance that I think will get adopted for a lot of different applications,” he added.