News

NVIDIA Unveils GenAI Model to Enhance Extreme Weather Predictions

• By John K. Waters
• 08/21/2024

Amid increasing severity and frequency of extreme weather events, NVIDIA Research has announced a new generative AI model for emulating high-fidelity atmospheric dynamics called StormCast. Unveiled during the peak of the Atlantic hurricane season, StormCast promises to enhance the accuracy of weather forecasts, particularly at the mesoscale—a critical range for disaster planning that lies between storm and cyclone scales.

Detailed in a paper written in collaboration with the Lawrence Berkeley National Laboratory and the University of Washington, this model enhances the resolution of weather predictions to a three-kilometer scale with hourly updates, a significant improvement over traditional methods. Early tests indicate that StormCast can deliver forecasts that are up to 10% more accurate than those from the National Oceanic and Atmospheric Administration (NOAA).

"It's just one example of how generative AI is supercharging thundering breakthroughs in climate research and actionable extreme weather prediction," said Mike Pritchard, a climate scientist at NVIDIA and co-author of the study, in a blog post, "helping scientists tackle challenges of the highest stakes: saving lives and the world."

StormCast is part of NVIDIA's broader efforts in climate research, exemplified by its Earth-2 platform. Earth-2 combines AI, physical simulations, and computer graphics to simulate and visualize weather and climate predictions on a global scale. In Taiwan, for instance, the National Science and Technology Center for Disaster Reduction is leveraging Earth-2's CorrDiff model to improve typhoon predictions. CorrDiff has the ability to "super-resolve" atmospheric data from 25 kilometers down to 2 kilometers, doing so 1,000 times faster and with 3,000 times less energy than traditional methods.

StormCast also integrates "autoregressive prediction" capabilities, allowing it to predict future weather outcomes based on past data. This makes it an invaluable tool for regional weather prediction, particularly in areas prone to severe weather, the company says.

With StormCast, NVIDIA is not only advancing climate science but also providing practical tools to address the escalating challenges posed by climate change, the company said.

"Given both the outsized impacts of organized thunderstorms and winter precipitation, and the major challenges in forecasting them with confidence, the production of computationally tractable storm-scale ensemble weather forecasts represents one of the grand challenges of numerical weather prediction," said Tom Hamill, head of innovation at The Weather Company, in a statement. "StormCast is a notable model that addresses these challenges, and The Weather Company is excited to collaborate with NVIDIA on developing, evaluating and potentially using these deep learning forecast models."