Localization and Attenuation of Diffracted Seismic Noise in Shallow Water

In this paper, we propose an automatic approach for diffracted seismic noise (DSN) localization. In the proposed method, the apexes of the potential DSNs in each shot are first recognized to calculate their locations, respectively. After that, all these potential DSNs recognized in one seismic line are used together to improve the location estimations further, according to their locations distribution. By supposing that the location of one DSN is fixed or slowly changed during the seismic data acquisition, for a true DSN, its location estimates obtained from different shot gathers should focus together. Therefore, a clustering algorithm is applied to obtaining the location estimations of the final DSNs and removing the false recognized DSNs at the same time. After the DSNs are localized, they are flattened along their trajectories and extracted by multi-channel filters, as do the other existed methods. A real 3D marine data example demonstrates that the proposed method obtains some promising results.