AI model detects solar storm warning signs hours in advance

Scientists have developed an artificial intelligence system that can detect solar storm warning signs nearly nine hours before the underlying activity becomes visible on the Sun’s surface, a breakthrough that could eventually give satellite operators and power grid companies precious extra time to prepare for incoming space weather. The model, called EarlyDetect, comes from a research team led by the New Jersey Institute of Technology (NJIT) and described in a study published Aug. 14 in the Journal of Geophysical Research: Machine Learning and Computation.

[...]

EarlyDetect analyzes hourly maps of that acoustic activity together with magnetic field measurements gathered by NASA’s Solar Dynamics Observatory (SDO), pulling its acoustic data specifically from sound wave readings the Helioseismic and Magnetic Imager captures aboard that spacecraft every 45 seconds.

[...]

Tirona still cautions that EarlyDetect cannot yet function as a real-time forecasting system, since it can still generate false alarms and detecting an emerging active region never guarantees that a solar flare or coronal mass ejection will actually follow, meaning years of additional testing across far more solar events still lie ahead before this approach reaches genuine operational use.