Slip Slidin’ Away – Some Practical Implications of Seismic Velocity Anisotropy on Depth Imaging

We can no longer ignore seismic velocity anisotropy in seismic data processing. Laboratory and field studies have provided compelling evidence that shales exhibit intrinsic transverse isotropy (TI), in which the seismic velocity parallel to the laminations is greater than that perpendicular to the layering, with the difference being as high as 30%. In flat-layered clastic sequences such as in the Plains of Alberta, the TI axis of symmetry is essentially vertical. Isotropic depth migration of reflection seismic data tends to overestimate the true depths of reflectors since the imaging velocity is generally greater than the true vertical velocity. In structured areas, such as the Rocky Mountain fold and thrust belt, strata are often rotated in folds and in the hanging walls of thrust faults. These strata form coherent panels with high angles of dip, so that the TI axis of symmetry is no longer vertical. In these situations, significant depth and position errors of targets occur if seismic velocities are assumed to be isotropic during data processing.

The practical impact of velocity anisotropy on seismic imaging is being investigated by researchers in the Foothills Research Project at the University of Calgary, in conjunction with the Project sponsors, using numerical and scaled physical seismic modelling studies and analysis of field seismic data. Models of thrust faults incorporating anisotropic layers demonstrate time-structural anomalies and discontinuities on events from horizontal reflectors below the thrusts. These effects are induced by velocity anisotropy in the overlying, dipping layers. Another model consists of a simple step function underlying a layer of anisotropic material that has its axis of symmetry, or bedding plane normal, dipping at 45 degrees. Isotropic depth migration of seismic data collected over this model yielded errors in depth and lateral position of approximately 5% and 20% of the thickness of the anisotropic layer, respectively. Anisotropic depth migration correctly positioned the target, but required a more sophisticated velocity model which includes the known magnitude and dip of the slow velocity vector as well as the Thomsen anisotropy parameters ε and δ. For processing field data, however, these anisotropy parameters have to be determined independently, and assigned as initial values during interpreter-driven depth imaging. We have established a range of values for these parameters through multiazimuth refraction seismic surveys in areas where steeply dipping strata outcrop at surface. Examples of seismic data from the western Canadian Foothills show improved quality and positional accuracy of images when anisotropy is incorporated into the velocity model building process.