Structurally oriented coherent noise filtering
All seismic data, whether 2D, 3D, post-stack or pre-stack, contains noise. Typically, this noise is comprised of both coherent and random components. Coherent noise presents itself as regular patterns in the seismic data. It may appear to be random or coherent depending on the orientation of the slice on which it is being observed. For example, coherent noise associated with acquisition may appear random on vertical slices through the volume, with its coherent nature becoming apparent on horizontal slices through the volume. When seismic data is processed, an effort is made to reduce the coherent noise in the seismic data by applying a variety of signal processing techniques. However, when the seismic volume is delivered to the client, it often includes remnant noise that processing was unable to remove without having a deleterious effect on the amplitudes and bandwidth of the seismic data. Since the processed seismic volume typically contains coherent noise, managing that noise within an interpretation system is critical, as it has a significant negative impact on semi-automatic and automatic interpretation workflows and techniques including autotracking of horizons, imaging of faults and fractures, automatic extraction of faults, and interpretation of stratigraphy and geomorphology. An obvious example is the effect of noise on edge attributes (i.e., ‘coherence’ class attributes). Any noise in the seismic data, whether random or coherent, will appear as edges in the seismic attribute volume and may obscure the geologic features of interest in the edge attribute volume (faults, fractures, stratigraphy, and geomorphology). Finally, many interpreters today must use older 3D seismic volumes, merged seismic volumes of a variety of acquisition designs and processing workflows. Even if the pre-stack data is still available, their company may not be in a position to spend the money required to re-process the data. Yet they still need and expect to be able to use modern advanced interpretation technology on the data. In response to this need a post-stack structurally oriented coherent noise filtering process has been developed and is described, with an application to 3D seismic surveys from the North Sea. The results are evaluated by comparing seismic volumes, edge attribute volumes, and amplitude and phase spectra pre- and post-filtering.