Projects in the Eastern Caribbean, Iceland and subduction zones in Chile, Alaska and the Cascades are deploying dense sensors, fibre-optic cables, machine learning and laboratory experiments.
Researchers say some volcanoes, including Etna, Stromboli, Kilauea and Iceland's Reykjanes Peninsula, can already be forecast hours ahead, but many still give only short or inconsistent warning signals.
About 800 million people live within 100km of an active volcano, and scientists say much broader, decades-long monitoring is needed to uncover shared magma physics and extend warnings to days or weeks.
Can AI outsmart volcanoes, or will the most catastrophic eruptions remain fundamentally unpredictable?
By drilling into magma for the first time in 2027, are scientists mastering volcanoes or just tempting fate?
Integrating Seismic Jerk Signals, Satellites, and AI for Reliable Volcano Eruption Warnings
Overview
The Jerk method revolutionizes volcanic eruption forecasting by detecting subtle ground movements caused by magma intrusions fracturing rock deep beneath volcanoes. This low-cost technique, requiring only a single seismometer, was successfully automated at Piton de la Fournaise, providing real-time warnings for 92% of eruptions with lead times up to 8.5 hours. Integrating this seismic data with satellite observations and AI enhances early warning systems, as demonstrated during Mauna Loa's 2022 eruption. Despite challenges like sensor durability and false alarms, emerging technologies such as Distributed Acoustic Sensing and advanced AI promise more reliable, global monitoring, ultimately improving community preparedness and resilience against volcanic and other geohazards.