The quiet Aegean façade concealed a brutal underground drama. A magma surge moved through a hidden conduit beneath Santorini, Amorgos, and Anafi for months, triggering a swarm of quakes that rattled the region. The unfolding activity, felt as many tremors over magnitude 5.0 and lasting from January 2025, sent tourists fleeing and raised fears about the nearby Kolumbo underwater volcano.
Scientists combined physics with artificial intelligence to turn tremors into a data-rich orchestra. By treating each quake as a sensor and building a 3D map of the surrounding crust, researchers tracked evolving stress and movement. The resulting model pinpointed a 30 km underground channel linking Santorini and the Kolumbo region, through which magma traveled horizontally at depths of >8 km.
Lead author Anthony Lomax described the approach: “The tremors act as if we had instruments deep in the Earth, and they’re telling us something. When we analyse the pattern those earthquakes make in our 3D model, it matches very well what we expect for magma moving horizontally.” The team calculated that the volume of magma moving through the crust could fill about 200,000 Olympic-sized pools, as it smashed through rock layers and sparked thousands of tremors.
The researchers published their findings in Science and together with lead scientist Dr. Stephen Hicks emphasized a hopeful takeaway: AI paired with physics could transform volcano monitoring and, crucially, improve forecasting of eruptions. For now, the magma remained deep—>
more than 8 km below the surface—and has since cooled and stalled, lessening immediate eruption risk. Yet scientists caution that unrest in volcanic regions can last years and single events can precede rapid surface activity, as history has shown in the same seismic corridor that includes the 1956 magnitude 7.7 quake.
Implications for future monitoring are substantial. This study demonstrates a scalable framework: treat earthquakes as a distributed sensing network, feed them into physics-based crust models, and apply AI to extract meaningful patterns that hint at magma movement and potential eruption timelines. If replicated globally, such approaches could sharpen early warnings for seismically active populations around the world.