NASA says its COFFIES machine-learning model can flag solar active regions up to 12 hours before sunspots become visible on the Sun’s surface.
COFFIES makes those forecasts from indirect signals—surface magnetic-field measurements and acoustic waves—rather than direct observations of the Sun’s deeper magnetic flows.
The system is designed to spot patterns linked to sunspot formation, giving heliophysicists an earlier read on solar activity that can feed into their own analysis.
That extra lead time could improve warnings ahead of severe geomagnetic storms, including rare Carrington Event-class disruptions that threaten satellites, power grids and communications.