GENEVA, SWITZERLAND – In a significant leap for volcanology and risk management, a team of researchers from the University of Geneva (UNIGE) and the National Institute of Geophysics and Volcanology (INGV) in Italy has successfully created a highly-detailed 3D model of the internal structure of an active volcano. Using a novel combination of advanced seismic networks and artificial intelligence, the team was able to “x-ray” the Vulcano volcano in northern Sicily, providing an unprecedented look into its depths.
The groundbreaking research, published in Nature Communications, offers a new way to understand the complex processes that unfold miles beneath the surface, a critical challenge for scientists monitoring the world’s over 1,500 active volcanoes.
A Picture of Unprecedented Detail
The team’s success hinged on two key innovations. First, they deployed a nodal seismic network of approximately 200 portable sensors across the island of Vulcano for a month. These state-of-the-art seismometers recorded the subtle, natural ground vibrations known as seismic ambient noise.
Second, the researchers used neural networks (a form of artificial intelligence) to process the massive volume of data collected. This AI-assisted approach allowed them to create a high-resolution, three-dimensional image of the volcano’s internal structure. As lead author Douglas Stumpp noted, this breakthrough is comparable to the transition from ultrasound to MRI in medicine, offering a far more detailed and precise view than previous methods.
The modeling specifically reveals the distribution of magmatic fluids in the volcano’s upper regions. This is particularly relevant given that Vulcano entered a period of unrest in late 2021, characterized by seismic events that indicate the movement of magma and gas.
The Future of Volcanic Risk Management
While the current findings do not allow for real-time eruption prediction, they represent a significant step toward that goal. The study demonstrates that integrating machine learning and deep learning with seismic data processing could eventually lead to the real-time analysis of a volcanic system’s behavior. This capability could, in the future, enable dynamic and adaptable evacuation plans as a volcanic crisis unfolds.
“The ultra-fast processing of such massive volumes of data remains a major technical challenge, but the integration of machine learning…shows that this prospect is now becoming feasible,” said study leader Matteo Lupi. The research offers a promising path forward for better anticipating volcanic activity and protecting the more than 800 million people who live in proximity to these geological giants.