NASA AI Model Predicts Solar Active Regions 12 Hours Before Emergence
A new machine-learning tool identifies precursor signals in acoustic waves to forecast potential space weather storms.
Researchers from NASA's COFFIES DRIVE Science Center have developed an AI model capable of predicting the emergence of solar active regions up to 12 hours before they become visible on the Sun's surface. This breakthrough allows scientists to identify potential storm-causing regions before they manifest as sunspots.
The model utilizes a "sliding-window transformer architecture" to analyze long sequences of data, specifically detecting minute reductions in magnetic fields and acoustic activity. By processing data captured by NASA's Solar Dynamics Observatory (SDO) using supercomputing resources at NASA Ames Research Center, the AI identifies precursors that were previously difficult to isolate. Alexander Kosovichev, a COFFIES co-investigator at the New Jersey Institute of Technology (NJIT), described the technique as identifying a "slight change in rhythm within a very noisy orchestra.
The Science of Solar Storms
Sunspots are the visible markers of active regions, which are intense concentrations of magnetic fields. These regions are the primary drivers of solar flares and coronal mass ejections—violent eruptions that create space weather storms. Such events have the potential to disable satellites, disrupt global radio communications, and pose significant radiation risks to astronauts. Currently, operational forecasting conducted by the U.S. Air Force and NOAA typically relies on monitoring active regions that have already appeared on the solar disk.
Implications for Space Exploration
This predictive window provides a critical early warning system for space weather forecasters. By identifying flaring locations before they emerge, NASA can better safeguard critical infrastructure and human life during high-stakes missions. Michelangelo Romano, M2M SWAO deputy director, noted that the model is exciting because it provides new capabilities for predicting potential flaring locations ahead of time. This is particularly vital for deep-space initiatives, such as the Artemis lunar missions and future crewed explorations of Mars, where astronauts lack the protective shield of Earth's atmosphere.
Research and Development
The project was a multi-institutional collaboration involving Princeton University, the New Jersey Institute of Technology, and NASA's Ames Research Center. The findings were published in the Journal of Geophysical Research: Machine Learning and Computation. Moving forward, the team aims to refine these predictive capabilities to further extend the warning time and accuracy of solar activity forecasts.