IBM and NASA have released a new open source AI model for weather and climate that is now available from HuggingFace.

The foundation model, called Prithvi WxC, was trained on 40 years of historical weather data from NASA’s MERRA-2 dataset and can be customized for a variety of weather and climate-related applications. And, it can run on a desktop computer.

The 2.3 billion parameter foundation model can “be quickly tuned for different use cases and served from a desktop computer in seconds,” the announcement states. As described in this paper, Prithvi WxC uses “an encoder-decoder-based architecture, incorporating concepts from various recent transformer models to effectively capture both regional and global dependencies in the input data.”

A featured application involves downscaling, which provides the ability to zoom in on low-resolution data. “By localizing weather and climate projections, downscaling can provide early warning that an extreme flooding event or hurricane force winds are on their way,” the announcement says.

Learn more at IBM.