In a previous study, we inverted shear‐wave birefringence observations from an azimuthal vertical seismic profile (VSP) experiment conducted at the Conoco Borehole Test Facility, Oklahoma (Horne and MacBeth, 1994; Horne, 1995). Our results indicate that the observations can be interpreted in terms of two distinctly different transversely isotropic (TI) models (Figure 1). The first model predicts the symmetry axis to be at N165°E and dipping 10° to the northwest. This orientation coincides with geological information relating to the fracture system that strikes between N50°E and N75°E (Queen and Rizer, 1990). Thus, this first model is consistent with a priori information, so that a possible source of the anisotropy can be identified. However, the second model derived from the inversion results suggests the symmetry axis to be at N200°E and dipping 30° to the southwest. If we interpret this result in terms of an equivalent medium resulting from aligned cracks or fractures, then this inferred crack‐fracture strike would lie in a direction conflicting with the a priori measurements. The bimodal nature of this solution can be readily understood if we examine the shear‐wave behavior for the different models shown in Figure 2. In this plot, the symmetry axis is chosen to be the [Formula: see text]. If we consider the near‐vertical propagation directions that are typically measured in VSP experiments, it can be seen that the qS1 polarizations lie either perpendicular (model 1) or parallel (model 2) to the symmetry axis. Since these polarizations are usually interpreted in terms of aligned crack‐fracture systems, the inferred strike would lie in the [Formula: see text] plane for model 1 and the [Formula: see text] plane for model 2. This interpretation is completely incorrect for model 2, since this inferred alignment is actually orthogonal to the alignment implied by the symmetry of the TI system. This situation represents a worrying aspect to the interpretation of shear‐wave surveys used to characterize crack‐fracture systems. The question that we address is whether anisotropic materials that possess properties similar to those of model 2 can be constructed from equivalent media resulting from cracks or fractures. We also consider other sources of anisotropy that may lead to this behavior.
Bayesian recognition of a moving object information-measuring system state: a priori information weight correction