Deep learning 3D inversion of gravity data at volcanic seamounts: An example of Conception Bank volcano (Canary Islands)

Deep learning (DL) approaches enable rapid, high-resolution imaging of subsurface density structure based on satellite gravity data, effectively capturing both lateral and vertical complexities that are challenging for conventional inversion methods. DL-based inversion can provide a robust preliminary model of inaccessible undersea volcanoes, where no local gravity measurements exist. Although the fit to observed gravity data is expected to be moderate, inherent to DL inversion, it rapidly generates geologically plausible density structures. This study presents a 3D gravity inversion based on DL applied to volcanic seamounts, using Conception Bank (Canary Islands) as a case study. The relative age of the Conception Bank seamount volcanoes and islands suggests that Conception Bank formed prior to the main Canary Islands volcanic edifices. However, it remains a geological enigma because, despite its symmetric and extensive uplift, it never emerged to form a subaerial volcanic island. First, the robustness of the proposed 3D DL gravity inversion is validated in a well-constrained volcanic setting at El Hierro Island, demonstrating its ability to accurately capture subsurface density variations. Subsequently, application to Conception Bank reveals four distinct magma intrusions extending to depths of approximately 8 km, with structural connections between some intrusions at 10–12 km depth along the Canary Island chain. The spatial distribution of these intrusions explains the observed symmetric bathymetric uplift and indicates a volcanic style characterized by moderate melt supply.