| This thesis demonstrates that knowledge of velocity gradients and anisotropy is critical to obtain correct subsurface images in areas of complex deformation. For correct depth processing, a four parameter model, defined in this thesis to be the 'Vepsilondeltatheta' model, was developed, where V is the velocity in the symmetry axis direction (V0), epsilon and delta are the Thomsen parameters of anisotropy and theta is the orientation of the axis of symmetry. The depth processing 'Vepsilondeltatheta' model is closely related to the physical properties of the earth, specifically velocity, density, bulk and shear moduli, and should, therefore, be geologically viable, i.e. the parameters should be within the range of values known for rocks, and the structural component should be interpretable using geologic principles.; At several locations within the Western Canada Sedimentary Basin, VSP first-arrival traveltime inversion yielded global anisotropy values ranging from 0.0 < epsilon < 0.20 and -0.2 < delta < 0.30. Pseudo-epsilon logs, based on a proposed empirical relationship between gamma-ray count and epsilon, yielded anisotropy values that provided a good match between the calculated and observed first-arrival traveltimes. VSP source offsets greater than the depth of interest are necessary to constrain epsilon, and near-surface receivers are required to capture the first-arrivals from turning rays to constrain epsilon and delta in the presence of a velocity gradient. When a velocity gradient is present, epsilon can be estimated from the envelope to vertical and horizontal slowness plots. A procedure to design optimum VSP data for anisotropy analysis was also devised, and examination of shales showed that anisotropy was due to layering typically greater than 1 cm in thickness. If isotropic depth migration is undertaken in areas with sediments that exhibit velocity anisotropy, residual pull-up under thrust faults may be mistaken for real structure. Optimum processing for prestack depth migration in the Llanos Basin fold-thrust belt of Colombia was found to include model-based moveout depth statics and a V0 model with both a lithological dependence and a lithostatic velocity gradient based on borehole logs. |
