Jerk, a promising tool for early warning of volcanic eruptions

Forecasting volcanic eruptions remains a great civil defense challenge in many parts of the world. Although instrumental monitoring networks provide a variety of signals interpretable as eruption precursors, volcano observatories still lack a systemic warning tool that can issue unambiguous real-time alerts to stakeholders. Here, we propose such a tool called "Jerk" that permits real-time detection of subtle ground movements associated with magma injections. The Jerk signals consist of very low frequency transients in both horizontal ground motion and tilt which, we show, are generated by fracture openings prior to an eruption. These signals can be detected using one single broadband seismometer. Implementation of fully automated Jerk detection and analysis at Piton de la Fournaise (La Réunion island) has permitted real-time early warnings for 92% of 24 eruptions that occurred in 2014-2023, with alarm times ranging from a few minutes to 8.5 h before magma reaches the surface. The Jerk tool thus promises to be a successful early warning method for forecasting volcanic eruptions and provides a potential alternative for poorly monitored volcanoes.