Abstract

Random seismic noise, present in all 3D seismic data sets, hampers manual interpretation by geoscientists and automatic analysis by a computer program. As a result, many noise-suppression techniques have been developed to enhance image quality. Accurately suppressing seismic noise without damaging image details is crucial in preserving small-scale geologic features for channel detection. The automatic detection of channel patterns theoretically should be easy because of their unique spatial signatures and scales, which differentiate them from other common 3D geobodies. For example, one notable channel characteristic has high local linearity: Spatial coherency is much greater in one direction than in other directions. A variety of techniques, such as spatial filters, can be used to enhance this “slender” channel feature in areas of high signal-to-noise ratio (S/N). Unfortunately, these spatial filters may also reduce the edge detectability in areas of low S/N. In this paper, I compared the effectiveness of three noise reduction filters: (1) running average, (2) redundant wavelet transform (RWT), and (3) polynomial fitting. I demonstrated the usefulness of these filters prior to edge detection to enhance channel patterns in seismic data collected from Saudi Arabia. The data examples demonstrated that RWT and polynomial fitting can successfully preserve, enhance, and delineate channel edges that were not visible in conventional seismic amplitude displays, whereas the running average filter further smeared the detectability of channel edges.

You do not currently have access to this article.