IBM and NASA Unveil Surya: Open-Source AI for Early Solar Forecasts

IBM and NASA unveiled Surya, an open-source AI model designed to forecast solar weather with unprecedented lead times, including the ability to predict solar flares two hours ahead and to visualize where on the Sun’s surface those flares are likely to originate. Trained on nine years of high-resolution imagery from NASA’s Solar Dynamics Observatory, Surya uses a 366 million-parameter foundation to interpret solar images and generate forward-looking predictions. The model, now available to researchers on Hugging Face, represents a significant step toward closing gaps identified in earlier readiness exercises that tested how well scientists can anticipate space weather and its cascading effects on satellites, power grids, and communications networks.

Solar activity poses persistent risks to modern infrastructure. When energetic particles from the Sun interact with Earth’s magnetic field, geomagnetic storms can disrupt satellites, weather power grids, and interfere with radio and navigation systems. Insurance and risk analyses have warned that severe solar events—sometimes likened to the historical Carrington-class storms—could cause enormous economic damage if unanticipated....