Abstract

Accurate velocity-depth modeling in a complex geologic setting especially in a deep-water environment, using towed streamer marine seismic data, is very challenging. Interval velocities derived using normal move-out (NMO) analysis of the reflected seismic signals for shallow reflectors (<1 km below the seafloor) is significantly compromised by the combination of a long wave path in the water column and the complex ray paths due to topography, leading to small move-out differences between reflectors. Additionally, refraction arrivals associated with the low sediment velocities and deep water only appear at far offsets, containing information about deeper structures. In this study, we present an innovative method where a 12 km long towed streamer seismic data are downward continued to the seafloor leading to the collapse of the seafloor reflection and the emergence of refraction events as first arrivals close to zero offset, which are used to determine a high-resolution near surface velocity-depth model using an efficient tomographic method These velocities are then used to perform pre-stack depth migration using a coincident 5.5 km long, shallow streamer data. Our findings infer that the velocity-depth model derived from tomography of downward continued towed streamer data provides a far superior pre-stack depth migrated image than those produced from velocity-depth models derived from conventional velocity estimation techniques.