Researchers have utilized machine learning to assess the impact of global warming on extreme weather occurrences in the U.S. and other countries in recent years. This innovative approach might transform how scientists analyze and predict the implications of climate change on severe weather.
A team from Stanford and Colorado State University has crafted a swift and cost-effective technique for examining how specific extreme weather events have been influenced by global warming. Their findings, published in an August 21 study in Science Advances, leverage machine learning to evaluate the extent to which global warming has intensified heat waves both in the U.S. and globally. This method has shown a high degree of accuracy and could redefine the ways scientists investigate and forecast the effects of climate change on various extreme weather scenarios. Additionally, the findings may assist in developing climate adaptation strategies and are pertinent for legal cases seeking compensation for climate-related damages.
“The consequences of extreme weather on human health, infrastructure, and ecosystems are evident,” stated Jared Trok, the leading author of the study and a PhD candidate in Earth system science at the Stanford Doerr School of Sustainability. “To create effective solutions, we need a clearer picture of how global warming drives changes in these extreme conditions.”
Trok and his colleagues instructed AI models to forecast daily maximum temperatures using regional weather data and global average temperatures. They trained these models using extensive data from a climate simulation database, covering the years from 1850 to 2100. Once trained and validated, the AI models were applied to real-world data from specific recent heat waves, allowing the researchers to assess how hot these waves would have been at differing levels of global warming. By comparing these projections against various warming scenarios, they estimated the influence of climate change on both the frequency and intensity of past weather events.
Case studies and beyond
The researchers initially applied their AI techniques to analyze the 2023 heat wave in Texas, which led to an unprecedented number of heat-related casualties in the state. Their analysis indicated that global warming increased the temperature of this heat wave by 1.18 to 1.42 degrees Celsius (2.12 to 2.56 degrees Fahrenheit) compared to what it would have been without climate change. The team’s new method also effectively predicted the severity of record-breaking heat waves in various global regions, aligning with previously reported studies on these events.
Building on these results, the researchers employed the AI to project the potential intensity of heat waves if the same weather patterns that triggered past record-breaking heat waves occurred under heightened levels of global warming. Their findings suggested that heat wave events, comparable to some of the most severe occurrences in Europe, Russia, and India over the past 45 years, could happen multiple times each decade if global averages rise by 2.0 degrees Celsius above pre-industrial levels. Currently, global warming is nearing 1.3 degrees Celsius above pre-industrial levels.
“Machine learning serves as a powerful new link between the actual meteorological conditions behind specific extreme weather events and the climate models that allow us to conduct broader virtual experiments on Earth’s systems,” remarked Noah Diffenbaugh, the study’s senior author and a professor of Earth system science at the Stanford Doerr School of Sustainability. “While AI doesn’t resolve all scientific issues, this innovative method is a thrilling advancement that I believe will be widely adopted for various applications.”
The new AI approach addresses certain weaknesses of current methodologies — including those previously created at Stanford — by utilizing historical weather data in its predictions regarding global warming’s effects on extreme events. It eliminates the need for costly climate model simulations since the AI can be trained with existing ones. These developments will facilitate precise, low-cost assessments of extreme weather in diverse regions, which is essential for crafting effective strategies for climate adaptation. Moreover, it paves the way for quick, real-time evaluations of global warming’s contribution to severe weather.
The research team intends to broaden the application of their method to encompass a wider array of extreme weather conditions and enhance their AI networks for better predictive capabilities, employing new strategies to quantify the complete range of uncertainty within the AI predictions.
“Our findings demonstrate that machine learning is a potent and efficient tool for exploring the impact of global warming on historical weather patterns,” said Trok. “We aspire for this study to foster further research in leveraging AI to deepen our understanding of how human-driven emissions affect extreme weather, allowing us to better prepare for future severe events.”
