New edge detection methods for seismic interpretation

Brian H. Russell, Claude Ribordy

In this article, we will discuss a new set of edge detection methods for seismic interpretation that involve derivatives of the seismic volume in the time, inline and crossline directions. The methods discussed in this paper were adapted from techniques that were initially applied to potential field data for edge enhancement. These methods are similar to algorithms that have already been proposed for seismic edge detection, which involve spatial differences between traces. However, the difference between these new approaches and the original methods is the way in which the vertical derivative is introduced into the computation. The edge detection methods discussed here are also very similar to coherency and semblance-based methods for seismic discontinuity detection, which in their initial form were based on cross-correlations between traces rather than derivatives. However, we can show that coherency estimates can also be implemented using temporal and spatial instantaneous attributes, which in turn depend on derivatives of the seismic volume and its Hilbert transform in the time, inline and crossline directions. Since these methods require differentiation and multiplication of two separate seismic volumes the potential for numerical instability is greater. The methods discussed in this article will be illustrated by a structurally complex seismic volume recorded in the North Sea.