INTERPRETING NONORTHOGONAL SPLIT SHEAR-WAVES FOR SEISMIC ANISOTROPY IN MULTICOMPONENT VSPS

There are two main sources of non-orthogonality in multicomponent shea r-wave seismics: inherent non-orthogonal split shear waves arising fro m substantial ray deviation in off-symmetry planes due to strong aniso tropy or complex overburden, and apparent non-orthogonal split shear w aves in the horizontal plane due to variation of the angle of incidenc e even if the two shear waves along the raypath are orthogonal. Many t echniques for processing shear-wave splitting in VSP data ignore these kinds of non-orthogonality of the split shear waves. Assuming inheren t non-orthogonality in zero-offset VSPs, and apparent nonorthogonality in offset VSPs, we derive equations for the four-component data matri x. These can be solved by extending the linear-transform technique (LT T) to determine the shear-wave polarizations in zero-offset and offset VSPs. Both full-wave synthetic and field data are used to evaluate th e technique and to examine the effects of non-orthogonal polarized spl it shear waves. If orthogonality is incorrectly assumed, errors in pol arization measurements increase with the degree of non-orthogonality, which introduces a consistent decreasing trend in the polarization mea surements. However, the effect of non-orthogonality on the estimation of geophone orientation and time delays of the two split shear waves i s small and negligible in most realistic cases. Furthermore, for most cases of weak anisotropy (less than 5% shear-wave anisotropy) apparent non-orthogonality is more significant than inherent non-orthogonality . Nevertheless, for strong anisotropy (more than 10% shear-wave anisot ropy) with complicated structure (tilted or inclined symmetry axis), i nherent non-orthogonality may no longer be negligible. Applications to both synthetic and real data show that the extended linear-transform techniques permit accurate recovery of polarization measurements in th e presence of both significant inherent and apparent non-orthogonality where orthogonal techniques often fail.