Paper submitted to ADIPEC 2026, by Øystein Korsmo, John Brittan and Mokhtar Raafat (TGS)
Abstract
This study presents an elastic Full Waveform Inversion (FWI) workflow applied to an Ocean Bottom Node (OBN) dataset, with the aim of generating reliable velocity models for accurate structural depth imaging and geological interpretation. The important elements of the geology in the study area a carbonate-basement system with shallow gas accumulations and fractured basement structures. A secondary objective is to evaluate elastic multi-parameter FWI (eMP-FWI) as an extension of the workflow, targeting the recovery of additional subsurface properties: P-wave impedance, absolute density, and Vp/Vs ratio. The derivation of these properties is a key element of using eMP-FWI for seismic attribute analysis and integrated interpretation.
The elastic FWI workflow updates P-wave velocity (Vp) using an elastic modelling engine, with initial shear wave velocity (Vs) and density models derived from well log data and Gardner's equation. A phase-driven dynamic matching objective function, combined with a quasi-Newton L-BFGS optimizer, drives stable convergence while reducing sensitivity to amplitude mismatches and cycle skipping. A data reconstruction approach is applied to suppress unmodelled wave phenomena (e.g., Scholte waves, attenuation effects) as this improves the utilization of the full dataset in the inversion. The eFWI workflow used multiple frequency stages up to 25 Hz. The eMP FWI stage then simultaneously inverted Vp, Vs, and P-wave impedance, using an inverse scattering imaging condition to decouple kinematic and dynamic sensitivities.