Use this URL to cite or link to this record in EThOS:
Title: Full waveform inversion of narrow-azimuth towed-streamer seismic data
Author: Kalinicheva, Tatiana
ISNI:       0000 0004 9350 9307
Awarding Body: Imperial College London
Current Institution: Imperial College London
Date of Award: 2019
Availability of Full Text:
Access from EThOS:
Access from Institution:
Full waveform inversion (FWI) is a computational scheme that produces high-fidelity, high-resolution models of the Earth's subsurface from surface seismic data. FWI has become a standard tool in velocity-model building and performs well on full-azimuth long-offset ocean-bottom seismic datasets. However, the majority of marine seismic datasets use narrow-azimuth towed streamers (NATS) which often lack long-offset refracted energy. Here I explore the capability of conventional FWI when it is applied to marine deep-water reflection-dominated NATS field data. I applied FWI to three datasets: the first used a deep-towed 10-km cable and was specifically acquired for 2D FWI; the other two datasets were both 3D reflection-dominated surveys to which FWI had been previously applied with limited success - these datasets were from Gabon and Brazil, and were chosen specifically because FWI had been previously tried and had failed. Applying FWI to these datasets, I reached the following conclusions: 1) When the input data have adequate turning energy and adequate low-frequency energy, acoustic anisotropic FWI can generate accurate high-resolution velocity models of increasing complexity and resolution up to about 40 Hz. 2) Extending FWI to the full bandwidth of the field data produces minimal further change in the macro-velocity model, but nonetheless continues to improve resolution up to and perhaps beyond that which can be recovered by conventional Kirchhoff-based pre-stack depth migration. 3) Applying 2D and full-3D FWI to a single 2D sail line produces similar outcomes. 4) The Gabon dataset proved almost entirely resistant to FWI; the available evidence suggests that the nominally-raw field data were corrupt in some unknown way. 5) The Brazilian dataset was inverted using a third-party FWI code that assumed constant density; this assumption is detrimental to FWI.
Supervisor: Warner, Michael Robert Sponsor: FULLWAVE Research Consortium
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral