In this article, we introduce a noise-reduction method, edge-preserving smoothing (EPS), that we have successfully applied as a preprocessing step before running algorithms to detect sharp edges in seismic data. These edges often correspond to important geologic features (e.g., faults, fractures, and channels).

Suppressing random noise is important before applying edge-detection (or coherence-cube) algorithms because reflection data near faults/fractures are usually more complex and noisier than in other areas and most edge-detection algorithms, which attempt to highlight local rapid changes in seismic data, are sensitive to noise.

Usually, prediction-error filtering (PEF) or f-x deconvolution are used to precondition data before...

You do not currently have access to this article.