Elastic reverse time migration based on the similarity between PP and PS images at each instant

Elastic reverse time migration is one of the most accurate migration methods of the multi-component data at present. The multi-component data are used as the boundary conditions for wavefield extrapolation, generating non-physical reflection waves in the back-propagated wavefields due to the conversion between P- and S-waves at the boundaries. Consequently, lots of migration noises are generated in PP and PS images and significantly reduce the quality of PP and PS images. To overcome this problem, a new elastic reverse time migration method based on the similarity between the PP and PS images at each instant is proposed. The similarity coefficients between the PP and PS images at each instant and imaging point are calculated and used to construct a new vector-wave-based imaging condition. The forward-extrapolated P-wave, backward-extrapolated P- and S-waves meet at the reflection point at the same time while the non-physical converted P- and S-waves do not meet due to the velocity differences. Thus, the corresponding PP and PS images of the former are with high similarity and the latter with low similarity. Utilizing the similarity coefficients between the PP and PS images at each instant and imaging point, the artifacts are suppressed and high-quality PP and PS images are obtained. Numerical tests demonstrate the correctness and effectiveness of the proposed method.