SUMMARY A g*-Hamiltonian method for tracing real rays was developed that can handle cusps and triplication of quasi-shear wave in a general viscoelastic anisotropic medium. We demonstrate that the g*-Hamiltonian method can produce homogeneous ray-velocity vectors (with parallel real and imaginary parts) and the slowness vectors of reflected and transmitted waves at the interface based on the real Snell's law (RSL), which constrains only the continuity of the real parts of the slowness vectors, or the real slowness direction (RSD) method, which ignores the inhomogeneous component of the slowness vector. These methods are based on the characteristic lines with different Hamiltonians. Our research indicates that these methods are limited to pre-critical incidence ranges. Moreover, we derived a complex energy velocity vector (energy flux velocity) expression, which is always homogenous. We found that directions of corresponding energy velocity calculated with complex Snell's law (CSL) at a contact of two viscoelastic anisotropic materials well match the solutions of the RSL and RSD methods for all R/T waves except post-critical incidence in which the RSL and RSD methods fail to obtain homogenous ray velocities. The RSL and RSD methods result in discrepancies in the ray quality factor, R/T coefficients, and energy ratios, especially for post-critical incidence. However, the exact critical angle determined by the RSD method approximates the ‘critical’ angle for anelastic/inhomogeneous waves, which was a previous challenge. Our calculations suggest that the energy velocity and energy quality factor obtained with the CSL method can be used for real ray tracing at the interface of viscoelastic anisotropic media, and the complex energy flux velocity vector is always exactly homogeneous. For the post-critical incidence, the RSL and RSD methods fail because the ray quality factor drastically changes from the infinite down to near 2, which contradicts homogeneous ray velocity even in elastic anisotropic materials for RSD method. However, the energy flux quality factor for the elastic-anisotropic material is all infinite, even for post-critical incidence, which is correct. We also show that the CSL method has the same efficiency as the RSD method.
Read full abstract