next up previous print clean
Next: Numerical examples Up: Shan: Implicit migration for Previous: Table-driven implicit finite-difference migration

Phase correction filter

In the 3D case, as in the isotropic migration, the dispersion relation is split into x and y components as follows:
\begin{displaymath}
\frac{\partial}{\partial z}P=i\frac{\omega}{v_p}\left [ \fra...
 ...frac{v_p^2}{\omega^2}\frac{\partial^2}{\partial y^2}}\right ]P.\end{displaymath} (13)
This two-way splitting causes numerical anisotropy, which can be remedied by a phase-correction filter Li (1991) in the Fourier domain as follows:
\begin{displaymath}
P=Pe^{i\Delta zk_L},\end{displaymath} (14)
where
\begin{displaymath}
k_L=\sqrt{\frac{1-(1+2\varepsilon_r)\frac{k_r^2}{(\omega/v_p...
 ...{v_p^r}k_y)^2}{1-\beta_1^r(\frac{\omega}{v_p^r}k_y)^2}\right ],\end{displaymath} (15)
where vpr is the reference vertical velocity, $\varepsilon_r$ and $\delta_r $ are the reference anisotropy parameters, and $\alpha_1^r$and $\beta_1^r$ are the optimized finite-difference coefficients corresponding to the anisotropy parameters $\varepsilon_r$ and $\delta_r $.
next up previous print clean
Next: Numerical examples Up: Shan: Implicit migration for Previous: Table-driven implicit finite-difference migration
Stanford Exploration Project
4/5/2006