NVIDIA Earth-2 makes weather predictions easier and faster for everyone, from scientists to businesses and governments. It offers a complete set of open tools and AI models for forecasting weather and climate, available for use worldwide.
At the American Meteorological Society's Annual Meeting, NVIDIA introduced Earth-2, a set of open weather AI tools that can handle everything from processing weather data to creating global 15-day forecasts or local storm predictions.
These AI tools save time and money compared to traditional supercomputers that use physics-based models. Now, organizations can run their own weather prediction systems as well as customize them for their needs.
NVIDIA Earth-2 includes different types of AI models:
- Earth-2 Medium Range: This model offers accurate forecasts up to 15 days in advance, covering over 70 weather variables like temperature, wind, as well as humidity.
- Earth-2 Nowcasting: A short-term prediction model that uses AI to forecast storms and hazardous weather in just minutes, offering predictions for the next few hours.
- Earth-2 Global Data Assimilation: This model creates accurate weather snapshots worldwide in seconds, speeding up forecasts by hours compared to traditional methods.
These models operate together to improve forecasting speed and accuracy, helping industries like energy, agriculture, and finance make better decisions based on weather data.
Executive Statement
According to Julian Green, cofounder and CEO of Brightband, the revolution of new AI weather tools for forecasting is very exciting and continues to gather speed with new models like NVIDIA Earth-2 Medium Range. Brightband is among the first to run Earth-2 Medium Range operationally, and the model being open source speeds up innovation, allowing easier comparison and improvements by other members of the weather enterprise.