PARAXIAL RAY TRACING IN 3D INHOMOGENEOUS, ANISOTROPIC MEDIA

Many investigations of the propagation elastic waves within the earth require a technique for producing synthetic seismograms which is capable of modelling 3D propagation effects. Ray methods are an excellent option for these problems, because they can be made fully 3D and allow a relatively quick and flexible computation of synthetic seismograms. However, the two point problem of finding the ray which connects exactly a specific source and receiver, may still be difficult and time consuming. Therefore, application of the paraxial method, which allows extrapolation of the information on a given ray to nearby receiver locations, is very valuable. With this approach, great savings in computation time and significant simplification of computer codes are possible. We investigate the application of the paraxial ray method to two problems in which the effects of 3D seismic wave propagation are important. The first is a model of a reef structure. In this case, we consider synthetic seismograms for a VSP experimental configuration. When the SV source and well are located along the axis of the reef, only 2D propagation effects are observed. If the source-receiver plane is located to the side of the reef, however, the 3D shape of the reef causes significant amplitudes to be predicted for shear-wave arrivals on the transverse component of the synthetic seismograms. The second example is a 1D, layered earth model, but it includes two layers which are azimuthally anisotropic due to the presence of aligned, vertical fractures. This anisotropy leads to 3D raypaths. Synthetic seismographs are presented for a cross-hole geometry both for an equivalent isotropic model and for the direction parallel to the fractures and at an angle of 45-degrees to the fractures in the anisotropic case. These synthetics show that the differences between the isotropic case and the case for source and receivers aligned with the plane of the fractures are small and subtle. On the other hand, the predictions for the direction at 45-degrees to the cracks show shear-wave splitting and significant transverse component signal. These results have important implications for both modelling and for applications such as tomography. It is clear that for some of the cases considered, a 2D algorithm will lead to errors in interpretation of data. In addition, ray-based tomographic techniques will have great difficulty in obtaining a well-defined 2D planar image when the signals are propagating in regions outside the image plane.