From NVIDIA: 2025-02-24 09:00:00

Weather agencies and scientists worldwide are utilizing NVIDIA CorrDiff, a generative AI weather model, to provide kilometer-scale forecasts of wind, temperature, and precipitation. This model is part of the NVIDIA Earth-2 platform and is being used to enhance forecasts for extreme weather events, renewable energy management, and agricultural planning.

CorrDiff uses generative AI to improve the resolution of coarse-resolution weather models, predicting atmospheric data at a 2-kilometer scale. It can also forecast variables not present in input data, such as radar reflectivity. The model was trained on the Weather Research and Forecasting model to generate weather patterns at 12x higher resolution.

CorrDiff has been optimized on Taiwan weather data and scaled to cover the continental United States, making it 500x faster and 10,000x more energy-efficient than traditional high-resolution numerical weather prediction using CPUs. The model continues to be enhanced to better resolve small-scale details in different environments and capture rare or extreme weather events.

Meteorological agencies and companies globally are using CorrDiff for regional forecasting, renewable energy, and disaster management. Taiwan’s National Science and Technology Center for Disaster Reduction has deployed CorrDiff to support disaster alerts, leading to significant energy savings and improved typhoon preparedness. Attendees can learn more about Earth-2 at NVIDIA’s GTC conference.



Read more at NVIDIA: NVIDIA Earth-2 Features First Gen AI to Power Weather Super-Resolution for Continental US