Researchers have unveiled artificial intelligence models capable of forecasting solar active regions nearly nine hours before they emerge on the sun’s surface, using acoustic and magnetic data from NASA’s Solar Dynamics Observatory to offer critical early warnings for space weather events.
While astronomers have long tracked solar activity after eruptions occur, recent developments in machine learning are shifting the paradigm toward predictive forecasting. Researchers are now harnessing long-term observations from space-based solar observatories to detect faint precursor signals before sunspots and flares materialize, providing a potential buffer for vulnerable technology infrastructure on Earth.
EarlyDetect Model and Helioseismic Analysis
Long before dark sunspots appear on the sun’s surface, a new active region begins showing subtle signs of its formation. In a study published on August 14 in the Journal of Geophysical Research: Machine Learning and Computation, a research team led by the New Jersey Institute of Technology introduced an artificial intelligence model called EarlyDetect. The system is designed to identify precursor signals of active region emergence by analyzing the sun’s acoustic activity and magnetic field measurements.
Active regions are magnetically intense areas where sunspots form, typically emerging over several hours while taking days to reach full development. As magnetic fields rise toward the sun’s surface, they leave faint signatures in acoustic waves that scientists detect through helioseismology, the study of solar vibrations. The EarlyDetect model utilizes a Transformer architecture—the same underlying AI technology used in large language models—to learn patterns in solar observations rather than text.
Tirona developed the approach alongside NJIT computer scientists, solar physicists, and collaborators at Princeton University and NASA’s Ames Research Center, drawing on observations from Miragenews. The acoustic maps analyzed by the model are derived from sound-wave observations recorded every 45 seconds by the Helioseismic and Magnetic Imager aboard the spacecraft.
Overcoming Filtering Hurdles in Solar Data
Developing the forecasting tool presented unexpected challenges during the research phase. After joining the project, the science team discovered that a filtering technique initially applied to help the AI model isolate important patterns was actually degrading its predictive performance.
“That surprised us most. We initially expected it to help isolate useful short-timescale patterns. Instead, it averaged away the very faint fluctuations that provided the earliest warning.”
Alexander Kosovichev, distinguished professor of physics at NJIT, via Miragenews
Kosovichev explained that the main difficulty in forecasting lies in the fact that active regions develop beneath the visible surface of the sun, where direct observation of the magnetic structure is impossible. Instead, instruments monitor very small changes in the magnetic field and acoustic wave patterns. Kosovichev described the task as detecting a slight change in rhythm within a very noisy orchestra. Removing the restrictive filters allowed the AI to factor in those critical fluctuations, enabling the best-performing version of the model to forecast active region emergence an average of 9.24 hours ahead.
Broader AI Deployments and Surya Heliophysics Integration
The EarlyDetect project is part of a broader wave of artificial intelligence integration within heliophysics. NASA and IBM developed the Surya Heliophysics Foundational Model, an AI system trained on nine years of continuous observations from the Solar Dynamics Observatory. Surya generates visual predictions of solar flares two hours into the future, surpassing existing benchmarks by 16 percent.
While advanced models show promise, researchers emphasize that they are not yet fully prepared for automated real-time alerts. EarlyDetect can occasionally produce false alarms or late predictions, and an emergence warning does not guarantee that a solar flare or coronal mass ejection will immediately follow.
Mitigating Infrastructure and Spaceflight Vulnerabilities
Improved forecasting models carry significant implications for modern technology infrastructure. Powerful solar storms can disrupt satellite communications or power grid companies, which can potentially lead to damage.
The stakes are particularly high for human spaceflight missions operating beyond Earth’s magnetic field. NASA’s Moon to Mars Space Weather Analysis Office closely monitors solar eruptions ahead of crewed missions like Artemis 2, where precise radiation predictions are essential for astronaut safety.
