Using AI to link heat waves to global warming

Researchers from Stanford and Colorado State University have developed a fast, inexpensive approach to studying how individual extreme weather events have been affected by global warming. Their method, detailed on August 21 in Science Advancesused machine learning to determine how much global warming has contributed to heat waves in the US and elsewhere in recent years.

The approach proved highly accurate and could change how scientists study and predict the impact of climate change on a variety of extreme weather events. The results can also help guide climate adaptation strategies and are relevant to lawsuits trying to collect compensation for damages caused by climate change.

“We have seen the impacts that extreme weather events can have on human health, infrastructure and ecosystems,” said study lead author Jared Trok, a Ph.D. student in Earth system science at the Stanford Doerr School of Sustainability. “To design effective solutions, we need to better understand the extent to which global warming is driving changes in these extreme events.”

Trok and his co-authors trained AI models to predict daily maximum temperatures based on the regional weather conditions and the global average temperature. To train the AI ​​models, they used data from a large database of climate model simulations spanning from 1850 to 2100.

But once the AI ​​models were trained and verified, the researchers used the actual weather conditions of specific real-world heat waves to predict how hot the heat waves would have been if the exact same weather conditions had occurred but at different levels of global warming.

They then compared these predictions at different levels of global warming to estimate how climate change affected the frequency and severity of historic weather events.

Case studies and beyond

The researchers first put their AI method to work analyzing the 2023 Texas heat wave, which contributed to a record number of heat-related deaths in the state that year. The team found that global warming made the historic heat wave 1.18 to 1.42 degrees Celsius (2.12 to 2.56 F) warmer than it would have been without climate change.

The researchers also found that their new technique accurately predicted the magnitude of record heat waves in other parts of the world, and that the results were consistent with previously published studies of those events.

Based on this, the researchers used the AI ​​to predict how severe heat waves could be if the same weather patterns that caused previous record-breaking heat waves instead of under higher levels of global warming.

They found that events equivalent to some of the worst heat waves in Europe, Russia and India in the past 45 years could happen multiple times per decade if global temperatures reach 2.0 C above pre-industrial levels. Global warming is currently 1.3°C above pre-industrial levels.

“Machine learning creates a powerful new bridge between the actual meteorological conditions that cause a specific extreme weather event and the climate models that allow us to run more generalized virtual experiments on the Earth system,” said senior author Noah Diffenbaugh. , the Kara J Foundation. Professor and Professor of Earth System Science at the Stanford Doerr School of Sustainability.

“AI has not solved all scientific challenges, but this new method is a really exciting advance that I think will be adopted for many different applications.”

The new AI method addresses some limitations of existing approaches – including those previously developed at Stanford – by using actual historical weather data in predicting the effect of global warming on extreme events. It does not require expensive new climate model simulations, as the AI ​​can be trained using existing simulations.

Together, these innovations will enable accurate, low-cost analyzes of extreme events in more parts of the world, which is crucial for developing effective climate adaptation strategies. It also opens up new possibilities for rapid, real-time analysis of the contribution of global warming to extreme weather.

The team plans to adapt their method to a wider range of extreme weather events and refine the AI ​​networks to improve their predictions, including using new approaches to quantify the full range of uncertainty in the AI ​​predictions .

“We have shown that machine learning is a powerful and efficient new tool for studying the impact of global warming on historical weather events,” said Trok.

“We hope this study will help advance future research into using AI to improve our understanding of how human emissions affect extreme weather, and help us better prepare for future extreme events.”