High-resolution exploration of the submarine crust in the North Sea

This study presents a 3D high-resolution characterization of the upper crust using FWI applied to dense seismic data. The subsurface is parameterized by P-wave velocity, density, quality factor, and VTI anisotropy. Starting from an initial velocity model obtained by reflection tomography and a density derived from Gardner’s law, the inversion uses the reciprocity principle and random source subsampling to reduce computational costs. The computations were carried out on the ADASTRA HPC platform, operated by CINES within the GENCIframework

The resulting models accurately reproduce P-wave velocity and density down to 2–3 km depth, validated by well log measurements, and reveal fine geological structures such as paleo-channels, low-velocity hydrocarbon-rich zones, and networks of fluid-filled faults. Although the study currently uses a visco-acoustic approximation, future extensions toward fully visco-elastic FWI with multicomponent data will allow better characterization of S-waves and fluid discrimination. These results demonstrate the potential of multiparameter FWI for detailed quantitative crustal characterization, bridging the gap between seismic imaging and reservoir analysis.