AI trained on Mars crater data discovers 73 unknown submarine volcanoes

An artificial intelligence algorithm originally designed to detect impact craters on Mars has identified 73 previously unknown submarine volcanic calderas beneath the world's oceans. The finding, published in Nature Communications Earth & Environment and led by volcanologist Andrea Verolino of Université Paris-Saclay, more than doubles the known global inventory of these structures Source: elconfidencial; Source: infobae.

The discovery

The team adapted a machine-learning algorithm built for Martian crater detection to analyse bathymetric maps of the ocean floor. From an initial pool of 87,435 possible formations, manual verification narrowed the list to 78 candidates, of which five were already documented—confirming the method's reliability Source: elconfidencial. The 73 new calderas are distributed across oceanic ridges (8), volcanic arcs (9), and intraplate regions such as seamount chains (61) Source: elconfidencial. The study focuses on calderas from the Pleistocene-Holocene period (last 2.5 million years) and provides a reproducible framework for future surveys Source: infobae.

Why it matters

Submarine calderas can produce explosive eruptions, tsunamis, and shockwaves, as demonstrated by the 2022 Hunga Tonga–Hunga Haʻapai eruption Source: infobae. The researchers stress that their catalogue is conservative and open to refinement as higher-resolution bathymetry becomes available. Seven of the newly identified structures are flagged as priority targets for detailed investigation Source: elconfidencial. The work fills a critical observational gap: over 70% of the ocean floor lacks high-resolution bathymetric data Source: infobae.