External authors:

• Alberto Ardid ( University of Canterbury )
• David Dempsey ( University of Canterbury )
• Corentin Caudron ( Universite Libre de Bruxelles , WEL Res Inst )
• Shane Cronin ( University of Auckland )
• Ben Kennedy ( University of Canterbury )
• Tarsilo Girona ( United States Geological Survey )
• Diana Roman ( Carnegie Institution for Science )
• Craig Miller ( GNS Science - New Zealand )
• Sally Potter ( GNS Science - New Zealand )
• Oliver D. Lamb ( GNS Science - New Zealand )
• Anto Martanto ( Ctr Volcanol & Geol Hazard Mitigat )
• Yesim Cubuk-Sabuncu ( Iceland Met Off )
• S. Ruiz ( Universidad de Chile )
• Rodrigo Contreras ( Universidad Catolica de Temuco )
• Javier Pacheco ( Universidad Nacional Costa Rica )
• Mauricio M. Mora ( Universidad Costa Rica )
• Silvio De Angelis ( University of Liverpool , Istituto Nazionale Geofisica e Vulcanologia (INGV) )

Abstract:

Seismic data recorded before volcanic eruptions provides important clues for forecasting. However, limited monitoring histories and infrequent eruptions restrict the data available for training forecasting models. We propose a transfer machine learning approach that identifies eruption precursors-signals that consistently change before eruptions-across multiple volcanoes. Using seismic data from 41 eruptions at 24 volcanoes over 73 years, our approach forecasts eruptions at unobserved (out-of-sample) volcanoes. Tested without data from the target volcano, the model demonstrated accuracy comparable to direct training on the target and exceeded benchmarks based on seismic amplitude. These results indicate that eruption precursors exhibit ergodicity, sharing common patterns that allow observations from one group of volcanoes to approximate the behavior of others. This approach addresses data limitations at individual sites and provides a useful tool to support monitoring efforts at volcano observatories, improving the ability to forecast eruptions and mitigate volcanic risks.