NASA’s COFFIES model aims to flag solar active regions up to 12 hours early
NASA reported in August 2026 that its COFFIES team developed a machine-learning approach to predict emerging solar active regions from subsurface acoustic-wave data. The result is a research forecast method, not a guarantee of a particular storm.
Full Disclosure Administrator · Published August 6, 2025 · Updated October 17, 2026NASA reported August 14, 2026 that researchers with the COFFIES science center developed a machine-learning model to predict the emergence of solar active regions up to 12 hours before they appear on the surface. The team analyzed observations from NASA’s Solar Dynamics Observatory and used supercomputing resources to study acoustic-wave fluctuations associated with magnetic regions rising beneath the Sun’s visible surface.
Sunspots and active regions can produce solar flares and coronal mass ejections, so earlier identification could give forecasters another signal to evaluate. The model is a research result published in a scientific journal; it should not be read as a guaranteed prediction of when a storm will occur or how severe its effects will be. Operational forecasting depends on validation, false-alarm rates, and combining multiple observations.
Machine learning can find patterns in complex data, but good evaluation requires comparison against held-out observations and transparent error measures. The useful question is not whether a model sounds intelligent; it is how often it succeeds, where it fails, and whether it improves decisions over existing methods.