Abstract: A technique includes spatially filtering a signal that is derived from a seismic acquisition. The filtering is associated with a filter length, and the filtering includes varying the filter length with frequency. The filtering may be used in connection with adaptive noise attenuation, which is applied to decomposed subbands. Furthermore, the filtering may be applied during the reconstruction of the signal from the subbands.

Abstract: A technique includes spatially filtering a signal that is derived from a seismic acquisition. The filtering is associated with a filter length, and the filtering includes varying the filter length with frequency. The filtering may be used in connection with adaptive noise attenuation, which is applied to decomposed subbands. Furthermore, the filtering may be applied during the reconstruction of the signal from the subbands.

Abstract: A method of attenuating noise in three dimensional seismic data using a projection algorithm is disclosed. A frequency—space—space (“f-xy”) projection algorithm is used which is a generalization of the f-x projection algorithm. The predictability of the seismic signals in the f-xy domain constitutes the basis of the algorithm. Specifically it is demonstrated that if the seismic events are planar in the t-xy domain, then in the f-xy domain they consist of predictable signals in the xy-plane for each frequency f. A crucial step of the 2-D spectral factorization is achieved through the helical coordinate transformation. In addition to the disclosed general algorithm for arbitrary coherent noise, a specialized algorithm for random noise is disclosed. It has been found that the disclosed projection algorithm is effective even in extreme cases of poor signal to noise ratio. The algorithm is also signal preserving when the predictability assumptions hold.