Calderas Beneath the Waves: AI-Powered Exploration of Subaqueous Volcanism
Andrea Verolino 1
Affiliations: 1 Earth Observatory of Singapore, Nanyang Technological University, 50 Nanyang Ave, Singapore, 639798, Singapore
Presentation type: Talk [Invited]
Presentation time: Friday 12:55 - 13:10, Room R380
Programme No: 8.1.8
Abstract
Volcanic calderas, vast depressions generally formed by large explosive eruptions, have shaped Earth's landscapes and even influenced global climate. While subaerial calderas are extensively studied, their submarine counterparts remain largely unexplored, despite accounting for a significant portion of Earth's volcanic activity. Their detection is essential, as underwater eruptions --such as Hunga Tonga-Hunga Ha'apai (2022) -- can generate devastating tsunamis and inject climate-altering gases into the atmosphere. To bridge this gap, we applied a machine learning-based crater detection algorithm (CDA), originally developed for Martian impact structures, to systematically identify submarine calderas using GEBCO bathymetry data. The algorithm autonomously detects sub-circular depressions ≥4 km in diameter, followed by manual classification, supported by principal component analysis. This research not only advances caldera detection methodologies underwater, but also lays the foundation for future deep-sea investigations, offering critical insights into the dynamics of subaqueous explosive volcanism and directly contributing to hazard assessment and disaster