LEAST-SQUARES REVERSE TIME MIGRATION OF BLENDED DATA WITH LOW-RANK CONSTRAINT ALONG STRUCTURAL DIRECTION

摘要

The simultaneous-source acquisition has been developed to tremendously improve the marine acquisition efficiency. The economic benefit is compromised by the complexity of recorded seismic data. We propose an effective direct imaging algorithm for migrating the blended simultaneous-source data without preprocessing, i.e., source separation. While the least-squares reverse time migration can help attenuate the crosstalk noise significantly, there are still a huge amount of artifacts existing in the final image. We propose to apply a low-rank constraint to regularize the model during the least-squares inversion. The low-rank constraint is applied along the geological structure of the reflectivity image in order to fully take advantage of the structural patterns of seismic images. Two numerical examples with different complexity validate the effectiveness of the proposed method.