The National Aeronautics and Space Administration (NASA) and IBM have unveiled Surya, the most advanced open-source foundation model to date for interpreting high-resolution solar data and predicting space weather.
Trained on nine years of imagery from NASA’s Solar Dynamics Observatory, Surya is designed to forecast solar flares and other space weather events that can disrupt satellites, navigation, power grids, and astronaut safety.
“By developing a foundation model trained on NASA’s heliophysics data, we’re making it easier to analyze the Sun’s complexities with unprecedented speed and precision. This empowers broader understanding of how solar activity impacts critical systems we all rely on,” Kevin Murphy, NASA’s Chief Science Data Officer, said.
Traditional space weather forecasts have been limited by incomplete solar observations. Surya overcomes these challenges with unprecedented resolution, improving solar flare classification accuracy by 16 percent and, for the first time, visually predicting where flares may occur up to two hours in advance.
Released on Hugging Face, Surya will be openly available for researchers worldwide, extending NASA’s mission to advance data-driven science and safeguard technologies essential to modern life.