Shear anisotropy analysis from direct and converted wave VSP data using various algorithms

Abstract

Shear-wave exploration cannot be performed rigorously without considering the effects of azimuthal anisotropy. Exploration for fractures can directly utilize measurements of anisotropy to infer fracture existence and orientation, while exploration using shear waves for imaging or AVO studies must take into account the distortion that anisotropy will add to the data. Therefore it is necessary to be able to measure the magnitude, orientation and variation with depth of shear-wave anisotropy in an area of interest from the surface to the zone of interest. Three different algorithms were applied to a synthetic and a real dataset. Both datasets contained direct surface-generated shear waves with orthogonal polarization as well as converted shear waves from a compressional surface source. The algorithms are: four-component rotation including layer stripping, which was applied to the orthogonal shear-source data only; cross-correlation prediction modeling, and parametric inversion, which were applied to both direct shear waves and to shear waves generated from mode conversions in the subsurface. Both of the latter algorithms can be applied to data from a source with a single polarization. All three algorithms give similar correct results when applied to the synthetic data. The results on real data are broadly similar for the three algorithims, but they displayed different sensitivities to the characteristics of the data. The analysis of the real converted-wave data was encouraging because these data can be expected to have a much broader spectrum than the surfacegenerated shear data. The compressional source offset was quite smalll, 300 m in a -2,100-m well, so converted shear-wave travel paths were close to vertical, preventing contamination of the anisotropic response by the transversely isotropic medium. The analysis of the real data found some evidence of birefringence in the bottom part of the well, possibly as much as 5%, though the anisotropic layers were interspersed with isotropic ones and the natural axes of the anisotropic layers were not collinear. There are some effects visible in-the converted-wave data near the bottom of the well that cannot be easily understood in terms of azimuthal anisotrop polarization of the wavefields. y, though they affect the apparent