Extending the Velocity Resolution of Waveform Inversion Below the Diving Waves Using AWI

Abstract

Summary The combination of conventional FWI and adaptive waveform inversion is used to invert a broad-band narrow azimuth shallow-water towed-streamer dataset, recovering an accurate velocity model, to 20 Hz, above, within and below high velocity chalk to a total depth of around 5000 m. FWI alone cannot achieve this depth of penetration from this dataset. Conventional full-waveform inversion of surface seismic data is normally able to recover an accurate and well-resolved velocity model down to the depth of penetration of the deepest recorded diving waves. This depth depends upon the acquisition geometry and the background velocity model, but it seldom extends much deeper than about 2 to 3 km below mud line, and can be shallower where there are strong velocity inversions. Below this depth, conventional FWI applied directly to reflection data can usefully update the high-wavenumber velocity model but normally makes little useful contribution to longer wavelengths. Various flavours of reflection FWI, that attempt to separate the migration and tomographic aspects of FWI, are capable of modifying the long-wavelength velocity model successfully below the diving waves, but typically such updates have rather poor vertical resolution and are often only able to make relatively minor adjustments to an already-mature velocity model. Here, we apply adaptive waveform inversion (AWI) ( Warner & Guasch, 2016 ) to a conventional NATS dataset from the Danish sector of the North Sea, recovering the velocity model to 5-km depth including a highly anisotropic low-velocity clastic section underlying thick high-velocity chalk. We used an inversion scheme that interleaves AWI and FWI, and that takes advantage of AWI's enhanced sensitivity to reflection moveout, and lack of edge effects below high-velocity refractors, to build a model which is well resolved over a kilometre below the diving wave zone with the model verified by downhole sonics.