2025-11-20 ユニバーシティ・カレッジ・ロンドン（UCL）
＜関連情報＞
- https://www.ucl.ac.uk/news/2025/nov/cause-santorini-earthquake-swarm-uncovered
- https://www.science.org/doi/abs/10.1126/science.adz8538
2025年のサントリーニ島の活動が明らかに：マグマの岩脈の反動による地震活動の誘発 The 2025 Santorini unrest unveiled: Rebounding magmatic dike intrusion with triggered seismicity
Anthony Lomax, Vasilis Anagnostou, Vasileios Karakostas, Stephen P. Hicks, and Eleftheria Papadimitriou
Science Published:20 Nov 2025
DOI:https://doi.org/10.1126/science.adz8538
Editor’s summary
Early in 2025, an intense seismic swarm near the volcanic island of Santorini forced thousands of people to evacuate. To understand the cause of the unrest, Lomax et al. used data from seismic stations on Santorini and surrounding islands to locate about 25,000 earthquakes over an 8-week period (see the Perspective by Pinel). Their machine learning–derived maps provide an unusually detailed picture of the seismicity, including episodic tremor bursts just a few hours long. Changes in failure stress indicators suggest that the unrest was caused by pump-like intrusions of magma into newly opened dikes 12 kilometers below the seafloor. —Angela Hessler
Structured Abstract
INTRODUCTION
In early 2025, between the islands of Santorini and Amorgos in the Aegean Sea, intense swarm seismicity alarmed the local population, resulting in school closures and disruptions to tourism. Santorini poses a substantial hazard because it is an active stratovolcano and hosted one of the largest known historical eruptions, the massive and devastating Minoan eruption of ~1620 BCE. The swarm seismicity also occurred just southwest of the destructive moment magnitude (Mw) 7.7 Amorgos earthquake rupture of 1956. Whether the 2025 unrest was caused by a magmatic dike intrusion or tectonic fault slip remains controversial. Imaging and explaining this unrest are therefore essential for basic scientific understanding, public information, hazard assessment, and eruption forecasting.
RATIONALE
Magmatic intrusion in Earth can lead to hazardous volcanic eruptions, but the kinematic and dynamic processes involved remain largely hidden from direct observation. Space-time imaging and modeling of the geometry, emplacement, and internal magma flow of dikes, which are essential to understanding volcanic systems and forecasting eruptions and associated earthquakes, rely on seismological and geodetic measures at Earth’s surface, resulting in loss of resolution and constraint of dike activity at depth. To overcome these difficulties, we investigated the cause of the 2025 Santorini-Amorgos unrest using precise, machine learning–derived seismicity (~25,000 earthquakes) as virtual, at-depth measures of Coulomb stress change surrounding possible intrusive and fault slip sources.
RESULTS
Our relocated seismicity reveals intense and complex activity extending up to 50 km northeast of Santorini and more than 15 km in depth. Before mid-January 2025, seismicity was sparse and mainly confined to Santorini’s caldera, simultaneous with surface uplift and inferred magmatic inflation under the caldera since mid-2024. From 26 January, dense, swarm-like seismicity developed ~20 km northeast of Santorini, followed by rapid expansion of activity to the northeast between 3 and 6 February. Intense, migrating swarm seismicity between 6 and 19 February extended more than 30 km further to the northeast, widening in a fan-shaped cloud. Using this seismicity as a virtual measure for Coulomb stress change imaging, we demonstrated that the source of the unrest was horizontal magmatic dike propagation and not tectonic fault slip. We imaged, in detail, the space-time evolution of the dike along a ~30-km swath as multiscale rebounding waves of dike opening, magma pressure, and breaking of barriers, together with triggered seismicity.
CONCLUSION
Our relocated seismicity and stress imaging resolve a more complex, rebounding feedback mechanism for dike emplacement than previously recognized, advancing understanding of dike physics and volcanic-tectonic-seismogenic feedback mechanisms. The exemplary time and space resolution of Coulomb-seismicity-stress imaging offers a foundation for advancing the physics-based modeling of dikes and for driving machine learning–based, data-driven procedures for tracking intrusions and forecasting eruptions in near real time.
Seismicity and imaged dike.
Machine learning–relocated seismicity for 1 January to 28 February 2025, with symbol size proportional to magnitude and color-coded according to occurrence time (OTime). The yellow patch shows the magmatic dike imaged using the seismicity as virtual measures of stress change due to dike opening. The large gray disk shows the epicentral area of the Mw 7.7 Amorgos earthquake of 1956. The inset shows the location of the study area (white filled rectangle) in the Aegean Sea.CREDIT: TOPOGRAPHIC AND BATHYMETRIC BASEMAPS WERE MADE WITH GEOMAPAPP (WWW.GEOMAPAPP.ORG) CC BY FROM W. B. F. RYAN ET AL. GEOCHEM. GEOPHYS. GEOSYST.10, 2008GC002332 (2009)
Abstract
Magmatic intrusion in Earth’s crust can lead to hazardous volcanic eruptions, but the physical processes involved remain largely hidden from direct observation. We used machine learning–derived seismicity as virtual stress meters at depth to study the disruptive 2025 seismogeodetic unrest in Greece between the Santorini volcano and the epicenter of the devastating moment magnitude 7.7 Amorgos earthquake that occurred in 1956. We show that the cause of unrest was magmatic dike propagation, which we imaged with ~25,000 relocated earthquakes occurring over 2 months. The dike propagated horizontally ~30 kilometers as multiscale rebounding waves of dike opening, magma pressure, and breaking of barriers while triggering intense surrounding seismicity. Our results establish magmatic intrusion as a more complex feedback process than previously recognized and can facilitate physics-based and data-driven modeling and eruption forecasting.
