NVIDIA StormCast AI Model Delivers More Accurate Extreme Weather Forecasts
NVIDIA introduces StormCast, a generative AI model that improves extreme weather predictions by providing high-resolution, hourly updates. The new system outperforms traditional forecasting methods to support better disaster planning.
NVIDIA Research unveils StormCast, a new generative AI model designed to emulate high-fidelity atmospheric dynamics and improve extreme weather predictions. Developed in collaboration with the Lawrence Berkeley National Laboratory and the University of Washington, this model focuses on the mesoscale, which is a critical range for disaster planning that sits between storm and cyclone scales.
StormCast enhances weather prediction resolution to a three-kilometer scale with hourly updates, achieving up to 10% better accuracy than traditional methods from the National Oceanic and Atmospheric Administration (NOAA). The model uses autoregressive prediction capabilities to forecast future weather outcomes based on historical data, making it an invaluable tool for regional forecasting.
This new model is part of NVIDIA's broader Earth-2 platform, which combines AI, physical simulations, and computer graphics to visualize weather on a global scale. Organizations like Taiwan's National Science and Technology Center for Disaster Reduction already use Earth-2's CorrDiff model to predict typhoons with significantly faster speeds and lower energy consumption than traditional approaches.