SummaryThe standard workflow for velocity model building (VMB) in complex regimes is an interpretive process that requires time-consuming manual intervention, and remains an error-prone process that can produce suboptimal results. We discuss an application of Full Waveform Inversion (FWI) to automate the refinement of legacy velocity models generated by conventional workflows. We demonstrate our solution on a full-azimuth (FAZ) survey acquired in the Gulf of Mexico using dual-sensor streamers and blended sources in the form of simultaneous long-offsets (SLO). The dual-sensor acquisition provides low-frequency data, while the SLO configuration enables the recording of long offsets in excess of 16 km; thereby providing a unique dataset to meet the traditional data dependencies for robust FWI stability. The long offsets and low frequencies were key to using both refractions and reflections to update the deeper parts of the velocity model. Our solution directly inverts the simultaneous data as acquired in the field, and is particularly relevant to the industry growth in blended seismic data acquisition. We also employed an FWI velocity gradient that eliminates the migration isochrones; thereby removing the reflectivity imprint from the model updates. The FWI application to the field survey successfully refined the geometry of the salt bodies, including the base salt and the intra-salt enclosures. It also improved the RTM image; particularly the salt flanks and the subsalt reflectors.