Least-Squares Multiple Imaging for 3D Surface-Related Multiple Elimination on Land Data

Abstract

Abstract Shallow reflectors which generate surface related multiples, can deteriorate image quality and hamper amplitude analysis. Typically, in land seismic data, the severe noise level and near-surface complexity make surface multiples difficult to identify and remove. In this paper, we present a least-squares multiple imaging (LSMI) driven de-multiple method which targets short and medium period surface multiples. The process involves inversion for a shallow multiple generating reflectivity, which is then used to drive the multiple modeling. The method allows true amplitude modeling, so minimal adaption is required at the subtraction stage. We demonstrate this method on a high-density land dataset acquired in Algeria. The results show that the multiple generator image gives better near-surface illumination and continuity compared to the conventional primary imaging approach. The strong multiple energy present in the near angle is largely suppressed, leading to less ringing and a more interpretable seismic image. Compared with surface-consistent deconvolution, the proposed de-multiple approach extends the amount of reverberation being attenuated, this is particularly effective on low-frequency multiples.