University of Western Ontario - Electronic Thesis and Dissertation Repository

Degree

Doctor of Philosophy

Program

Geophysics

Supervisor

Dr. R. Gerhard Pratt

Delay of Publication

1

Abstract

Waveform Tomography, combines starting models from traveltime tomography with tomographic waveform inversion. This approach to inversion recovers a “best fit” velocity model by iteratively minimizing the misfit between the observed, frequency-domain waveform data and modelled, synthetic waveforms. The method is both computationally efficient and an improvement over ray-based tomographic methods. Synthetic benchmark studies and experiments with long-offset, 2-D, exploration seismic field data, demonstrate that Waveform Tomography can yield accurate, high resolution velocity models, vital in seismic exploration for hydrocarbons.

I first present an application of Waveform Tomography to long-offset, 2-D, synthetic marine seismic data using the 2004 BP velocity benchmark. Due to the importance of refracted energy, the limited offset of marine seismic streamers limits the depths tomographic techniques can accurately image. Using data simulated with a multi-ship acquisition, providing maximum offsets up to 32 km, I show that Waveform Tomography of long-offset data can, potentially, accurately delineate top-salt, image shallow, low-velocity anomalies, and determine sub-salt velocity structures in marine seismic exploration.

In thrust-fold belt tectonic settings, such as the Canadian Foothills, conventional exploration seismic processing is inadequate, given the effects of steep dips, rough topography, and near-surface weathering. Depth imaging is often necessary, requiring an accurate estimation of the macro-velocity model. Long-offset data, acquired in these areas to "undershoot" steeply dipping faults and a severely weathered near-surface, may provide the necessary input for successful Waveform Tomography.

To investigate this, synthetic data were generated in a geologically realistic model from an area of the Foothills in northeastern B.C. Using this dataset, I demonstrate the ability of Waveform Tomography to accurately reconstruct the velocity model and the complex geological structures contained within. If the data do not contain frequencies below 3 Hz, the results are severely degraded.

Finally, using seismic field data recorded with an appropriately designed long-offset survey, I develop pre-conditioning and inversion strategies which effectively increase the low-frequency content of the data while mitigating the effects of ground roll. I apply these strategies to 2-D field data from southern Alberta, demonstrating that Waveform Tomography, using 4 Hz data, can produce complex velocity models containing geologically interpretable structures, correlatable with results from pre-stack time migration.