Conclusions

The diffraction separation performance of the traditional PWD filter is limited in complex geological structure, thereby causing an inaccurate slope estimation and the inadequacy of the plane-wave assumption of the PWD filter. We have proposed a new effective diffraction separation method based on the localized rank-reduction method. The rank-reduction method assumes the diffractions to be high-rank and reflections to be low-rank locally, and thus the diffraction energy can be easily separated via a simple rank-reduction filter. The localized rank-reduction method shows great advantages over the global rank-reduction method because the rank to be chosen in a localized method has a narrower range than in the global method and thus is easier to choose. More importantly, the localized rank-reduction is free of artifacts that commonly exist in global methods. The ranks of the localized rank-reduction method can be either constant or better be adaptively chosen. The adaptive rank selection strategy is simple and convenient to use. The separated diffractions and reflections based on the localized rank-reduction method are accurate both kinematically and dynamically. Synthetic and field examples demonstrate the effectiveness of the proposed method in diffraction separation and further in imaging of the separated diffractions.