Imaging of the Deep Galicia margin using ocean bottom seismic data

The Galicia margin, west of Iberia, is one of the most studied magma-poor rifted margins to understand the rifting process leading to continental breakup. Seismic imaging has been instrumental in understanding rifting in the Galicia margin. In this work, I derived a high-resolution P-wave velocity model of the Deep Galicia margin (DGM) where the final breakup of the continental crust happened. The velocity model is derived employing a 3D acoustic full waveform inversion(FWI) technique in the time domain using sparsely acquired wide-angle ocean bottom seismometer(OBS) data. The model is validated by tracking phase changes of the first arrivals during the inversion, and by comparing the predicted waveforms with the observed for all the instruments. In addition, the anomalies introduced by FWI were validated by performing synthetic inversion runs by recovering the anomalies using a synthetic dataset predicted using the final FWI model as observed dataset. The final model shows an improved alignment with the structures observed on 3D prestack time migrated multichannel seismic images compared to the starting model. Comparison of the 3D FWI model result with 2D result derived along a profile through the 3Dseismic volume highlighted the differences between the imaging methods in a real world setting .The 3D FWI result constraints better the complex faulting within the pre- and syn-rift sediments, crystalline crust and below a detachment fault, known as the S-reflector, compared to the 2D result. Below the S-reflector, 3D FWI has enhanced the pattern of serpentinisation compared to the starting model with local low velocity zones occurring around fault intersections. Based on my comparison, I conclude that the use of 3D data can partially mitigate the problem of receiver sparsity in FWI. Using the high-resolution 3D model, I attempted to understand the nature of the crystalline crust by comparing the velocity range of the crystalline crust in the DGM with other similar tectonic settings. The velocity limits of the crystalline crust in the DGM include velocities of both the upper and lower crust observed in other similar settings, indicating that it is comprised of both the upper and lower crust. Unlike in many other settings, there is no clear evidence in the P-wave velocity profiles for a separate upper and lower crust within the crystalline crust. The high-resolution model also shows evidence for exhumation of the lower crust under the footwall of the fault blocks to accommodate the extension. I generated a serpentinisation map of the DGM at a depth of 100 ms below the S-reflector and compared the map with a map generated by training a machine learning algorithm using velocities from 2D FWI. A mean serpentinisation of ~33 % is estimated below the S-reflector using the 3D FWI model. Seismic images of the DGM are developed in time and depth domains using the first-order multiples from the OBS data and a technique called mirror imaging. In this technique, the seafloor along with the OBS is mirror imaged with respect to the sea-surface and placed at a depth of twice the water column depth. Such an adjustment allows incorporation of the multiples in to migration algorithms just like primary reflections. I observe that mirror imaged sections show a good match within the sediment column with the multichannel images, but the quality of the images deteriorates below the top of the basement partially due to weak signal strength of the multiples. Primary reflections from the OBS illuminate very narrow sections of the subsurface, hence the quality of the image using them is poor compared to the mirror images. Mirror imaging can become a standard processing step in studies where no multichannel data are available.