On February 3, 2022, SpaceX lost 38 of 49 freshly launched Starlink satellites. The culprit was not a malfunction or a collision. It was a moderate geomagnetic storm that altered the density of Earth's upper atmosphere enough to drag the satellites out of orbit before they could reach their operational altitude. The incident set back SpaceX's constellation expansion and cost tens of millions of dollars.
That event is now one of the central motivations behind a new European Space Agency project called Swarm-AWARE, short for Swarm Investigation of Space Weather and Natural Hazards Effects. Researchers presented findings from the project at the European Geosciences Union General Assembly in late April 2026.
The core challenge the project addresses is a detection problem. When solar storms interact with Earth's magnetosphere and ionosphere, they generate electromagnetic signals that can look nearly identical to signals produced by earthquakes, volcanic eruptions, and other natural hazards. Distinguishing one from the other in real time is difficult, and getting it wrong has consequences for early-warning systems, power grids, rail networks, and satellite operations.
Georgios Balasis of the National Observatory of Athens leads the research team. His group is pulling together data from ESA's three Swarm satellites, which measure Earth's magnetic field, plasma densities, plasma temperatures, and electric fields, alongside ground-based sensors and atmospheric observations from the Copernicus Sentinel-5P satellite. The goal is to build a dataset rich enough to train machine learning models that can reliably separate the two types of signals.
The 2022 eruption of Hunga Tonga in the South Pacific gives researchers a well-documented test case. The eruption sent shockwaves into the upper atmosphere that disturbed ionospheric density over a wide area. "The waves triggered electric fields that traveled along magnetic field lines, causing instantaneous changes on the opposite side of the Pacific Ocean," Balasis said. Swarm magnetometers recorded all of it. Researchers can now use those measurements as a reference to see how a volcanic signal looks compared to a geomagnetic one, and begin building the classification tools needed to tell them apart automatically.
The practical stakes are significant. Infrastructure operators, satellite controllers, and disaster-response agencies increasingly rely on electromagnetic monitoring systems. If a power company's sensors pick up a sharp spike in magnetic field activity, knowing within minutes whether that spike came from an approaching solar storm or from a nearby fault rupturing is the difference between a precautionary shutdown and a missed earthquake warning.
The Swarm-AWARE team plans to apply advanced time-series analysis alongside the machine learning work, using the full 36-year archive of satellite and ground observations available to them. The project is explicitly designed to feed into real-time decision-making. Balasis and his colleagues say the system they are building could eventually allow organizations to act on space weather and hazard data with enough confidence to protect critical infrastructure rather than simply observe the disruption after the fact.
