A groundbreaking AI driven tool developed by a University of Canterbury led research team significantly enhance prediction of volcanic eruptions worldwide. The tool based on machine learning models potential to save lives and protect infrastructure by improving early warning systems.
Dr. Alberto Ardid and Associate Professor David Dempsey from UC’s Civil and Natural Resources Engineering department analyzed seismic data from 41 past eruptions across 24 volcanoes, including three in New Zealand. Their findings suggest that eruption warning signals follow repeatable patterns, which can be applied to under-monitored volcanoes.
Dr. Ardid explained “This could be a breakthrough in eruption forecasting, allowing us to transfer knowledge from well-studied volcanoes to improve risk mitigation at sites with limited data”.
With 29 million people living within 10 km of active volcanoes, timely forecasting is crucial. The AI tool offers a scalable and cost effective method particularly benefiting developing regions in Southeast Asia and Central America, where monitoring resources are scarce.
Dr. Ardid recently won the New Zealand Geophysics Prize for this research emphasized the tool impact stating that collaboration with international volcano observatories ensures its findings are actionable. The models will be shared as open access software to support global disaster management efforts.
UC volcanologist Professor Ben Kennedy said “This research challenges the long held belief that eruption precursors are unique to each volcano, we have a model that uses data from multiple volcanoes to improve forecasts at sites with little recorded history.”
The study published in Nature Communications a collaborative effort involving the University of Auckland and 18 researchers from nine countries supported by the New Zealand government Endeavour Fund and international partners.