AI-Powered COFFIES: Predicting Solar Storms & Sunspots with AI

Space weather forecasting has entered a new predictive era as a joint team of astrophysicists and data scientists within NASA’s COFFIES DRIVE Science Center developed a machine-learning model that spots the emergence of active solar regions up to 12 hours before they break the surface.

How Machine Learning Reads Solar Acoustic Precursors Beneath the Surface

The Sun’s interior churns continuously, driving localized magnetic fields upward to form sunspots, which serve as visible catalysts for severe space weather events like solar flares and coronal mass ejections. Because these magnetic structures remain hidden during their initial ascent, researchers must track indirect acoustic signatures. “We cannot directly see the magnetic structure while it is still rising through the solar interior. Instead, we must look for indirect effects – very small changes in the magnetic field and in the pattern of acoustic waves continually traveling through the Sun,” according to Alexander Kosovichev, a COFFIES co-investigator at NJIT. The newly published technique captures these subtle rhythm shifts in solar sound waves using high-performance computing resources located at NASA Ames Research Center in California’s Silicon Valley.

Comparing AI Prediction Against NOAA and Air Force Operational Forecasts

Traditional operational forecasting relies on monitoring active regions only after they become visible on the solar disk. The National Oceanic and Atmospheric Administration’s Space Weather Prediction Center and the United States Air Force currently track these surface features to evaluate flare probabilities. In contrast, the COFFIES artificial intelligence model employs a sliding-window transformer architecture designed for long data sequences. This system moves a fixed-size viewing window across historical timelines captured by NASA’s Solar Dynamics Observatory, isolating tiny reductions in acoustic activity and magnetic field variations that earlier deep-learning architectures missed. This capability allows forecasters to anticipate sunspot locations ahead of surface emergence rather than reacting to existing spots.

Protecting Artemis Astronauts and Deep-Space Missions

Accurate early warnings carry direct implications for crew safety as NASA advances human exploration through the Artemis missions to the Moon and planned journeys to Mars. Operating alongside NOAA, specialized teams such as NASA’s Moon to Mars Space Weather Analysis Office (M2M SWAO), the Space Radiation Analysis Group, and the Community Coordinated Modeling Center work around the clock to translate these research tools into robust operational frameworks. “The COFFIES AI model is exciting to our team because it could provide us with new capabilities towards predicting potential flaring locations ahead of time,” notes Michelangelo Romano, M2M SWAO deputy director. By providing advanced notice of high-energy radiation and charged particle storms, these collaborative modeling efforts aim to secure critical satellite infrastructure and safeguard astronauts against volatile solar weather.