Seismic reflection data are often recorded on the Earth surface. Therefore, an equation that describes the vertical slowness as a function of the horizontal one, the dispersion relation, is a key equation for imaging such data. In fact, reflection seismic data explicitly provides horizontal slowness information given by the slope of the reflections.

However, a simple *P*-wave analytical equation that describes the vertical slowness as a function of
the horizontal one does not exist in a practical form for orthorhombic media. Because obtaining such an equation
requires solving for the roots of a cubic polynomial as a function of the squared vertical slowness. On the other
hand, setting
all three shear wave velocities to zero will reduce the cubic equation to a linear one. The influence of the shear-wave velocities on *P*-wave propagation is small. This is a general statement that holds for most anisotropies, but have been proven extensively for transversely isotropic media. The setting of the
shear velocity to zero, as we will see later, will not compromise the accuracy of the equations for kinematic
or dynamic uses.

To obtain the dispersion relation for orthorhombic anisotropy we must first derive the Christoffel equation for such media. A general form for the Christoffel equation in 3-D anisotropic media is given by

with where
For orthorhombic media, the Christoffel equation slightly simplifies,
and in its matrix form (using *A*_{ij} instead of *a*_{ijkl})
is
given by

(14) |

Setting all shear wave velocities (*v*_{s1}, *v*_{s2}, and *v*_{s3}) to zero and using
Tsvankin's 1997 parameter representation, the Christoffel equation reduces to

(15) |

(16) |

Setting equation (16) to zero and solving for *p*_{z} provides the dispersion relation for
orthorhombic media as follow

*p*_{z}^{2}=

(17) |

Note that the vertical velocity appears only once in equation (17) and thus the vertical-time based dispersion relation, like in the case of VTI media, become vertical velocity independent. This feature is shown approximately by Grechka and Tsvankin (1997) using numerical methods, and is shown exactly here.

8/21/1998