AI Tool Offers New Insights Into Climate Change’s Impact on Heat Waves

Scientists from Stanford and Colorado State University have developed a fast and cost-effective method for analyzing the impact of global warming on specific extreme weather events. Published in the journal Science Advances, the study used machine learning to estimate how much recent heat waves in the United States and other countries have been influenced by global warming.

The method demonstrated high accuracy and could revolutionize how scientists study and predict the effects of climate change on various extreme weather events. The findings are also relevant to litigation seeking damages related to climate change and can help guide climate adaptation efforts.

We have seen the impacts that extreme weather events can have on human health, infrastructure, and ecosystems. To design effective solutions, we need to better understand the extent to which global warming drives changes in these extreme events.

Jared Trok, Ph.D. Student and Study Lead Author, Earth System Science, Doerr School of Sustainability, Stanford University

Trok and his coauthors trained AI models to predict daily maximum temperatures based on local weather patterns and average global temperatures. They utilized data from an extensive database of climate model simulations, covering the period from 1850 to 2100, to train the AI algorithms.

To refine and validate the AI models, the researchers used actual weather data from specific real-world heat waves. They then used this data to predict how hot those heat waves would have been under different levels of global warming. By comparing these projections at various levels of global warming, they were able to assess how climate change has influenced the frequency and intensity of past weather events.

Case Studies and Beyond

The researchers first applied their AI technology to analyze the 2023 Texas heat wave, which led to a record number of heat-related deaths in the state. They found that this historic heat wave was 1.18 °C to 1.42 °C (2.12 °F to 2.56 °F) hotter than it would have been without the influence of climate change.

The researchers also discovered that their method accurately predicted the intensity of record-breaking heat waves in other parts of the world, with results consistent with previously published analyses of these events.

Using this information, the team employed AI to forecast the potential severity of future heat waves if similar meteorological patterns were to occur under higher levels of global warming. They found that if global temperatures rise by 2.0 °C above pre-industrial levels, events similar to some of the most extreme heat waves experienced in Europe, Russia, and India over the past 45 years could occur more than once every decade. Currently, global warming is approaching 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 enable us to run more generalized virtual experiments on the Earth system. AI has not solved all the scientific challenges, but this new method is a really exciting advance that I think will get adopted for a lot of different applications.

Noah Diffenbaugh, Study Senior Author and Kara J Foundation Professor, Stanford University

Diffenbaugh, a Professor of Earth System Science at the Stanford Doerr School of Sustainability, highlights that the new AI technique leverages real historical weather data to predict the impact of global warming on extreme events. This approach addresses some limitations of previous methods, including those developed at Stanford, by eliminating the need for costly new climate model simulations since the AI can be trained using existing simulations.

These advancements will enable more accurate and affordable analyses of extreme events across a broader range of regions, which is crucial for developing effective climate adaptation strategies. Moreover, it opens up opportunities for rapid, real-time research on the effects of global warming on extreme weather.

The team plans to enhance the AI models by expanding the range of extreme weather events to which their method can be applied. This effort will include employing new techniques to measure the full spectrum 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. We hope that this study helps promote future research into using AI to improve our understanding of how human emissions influence extreme weather, helping us better prepare for future extreme events.

Jared Trok, Ph.D. Student and Study Lead Author, Earth System Science, Doerr School of Sustainability, Stanford University

Video Credit: Stanford University

Journal Reference:

Trok, J. T., et al. (2024) Machine learning–based extreme event attribution. Science Advances. doi.org/10.1126/sciadv.adl3242.