| A multi-disciplinary reservoir characterization project was undertaken in the Glenn Pool Field, Oklahoma. The geophysical objective was to aid in reservoir description by providing subsurface images using crosswell seismic tomography. Tomography is a method of estimating the subsurface velocity distribution based on observed and modeled traveltime differences. It requires accurate ray path and traveltime calculations.; Seismic anisotropy is the angular variation of velocity with propagation angle, at one particular point. Conventional geophysical forward and inverse methods often assume isotropic subsurface conditions, ignoring anisotropy. Due to its severity, in the Glenn Pool Field the effects of seismic anisotropy on traveltime calculations and tomography cannot be ignored. Isotropic tomography fails to yield a reasonable subsurface image.; This research follows a five-step approach to address the anisotropy problem: (1) Show the evidence of anisotropy, (2) Characterize anisotropy, (3) Develop an anisotropic ray tracing code, (4) Develop a velocity updating scheme for anisotropic tomography, and (5) Demonstrate the theoretical developments on synthetic and field data.; This methodology was successfully applied to the Glenn Pool data. The tomographic result is an acceptable solution which yields modeled traveltimes consistent with the observed traveltime data. The velocities obtained from tomography are in good agreement with sonic derived velocities. Nevertheless, crosswell seismic imaging bears the uncertainties and variability common to all inverse problems. The results vary, depending on how the processing constraints are imposed or how well the physics of energy propagation is characterized. A variety of processing algorithms applied to the same input data resulted in significantly different tomograms, demonstrating non-uniqueness of the problem. The result is in good agreement with the geological work carried out independently from tomography.; The effects of anisotropy, if not taken into account properly, will be detrimental to image quality. The methods introduced in this dissertation are useful in dealing with anisotropy, without resorting to weak anisotropy assumptions. |
