Finite-difference modelling of wavefield constituents

The finite-difference method is among the most popular methods for modelling seismic wave propagation.Although the method has enjoyed huge success for its ability to produce full wavefield seismograms in complex models, it has one major limitation which is of critical importance for many modelling applications; to naturally output up-and downgoing and P-and S-wave constituents of synthesized seismograms.In this paper, we show how such wavefield constituents can be isolated in finite-difference-computed synthetics in complex models with high numerical precision by means of a simple algorithm.The description focuses on up-and downgoing and P-and S-wave separation of data generated using an isotropic elastic finitedifference modelling method.However, the same principles can also be applied to acoustic, electromagnetic and other wave equations.