Minimum entropy deconvolution

Abstract Minimum Entropy Deconvolution (MED—Western Geological Company of America Service Mark) was developed to aid the extraction of detailed reflectivity information from amplitude anomalies in reflection seismic recordings. Interpreting such reflectivity information depends on accurate compensation for the combined effects of the source wavelet and earth filtering. MED approaches this problem by finding a linear operator that maximizes the “spike-like” character of a representative set of traces. Unlike predictive deconvolution, the MED process makes no assumptions about the phase characteristics of the “seismic wavelet”, nor does it assume that the reflection series is white—certainly a poor assumption over short-time windows where “bright spots” dominate. Rather than seeking to whiten data, the MED process seeks the smallest number of large spikes that is consistent with the data. Synthetic data examples show that the method can determine effective operators by exploiting quite small differential moveouts of events across unstacked traces. While maximizing the spikiness of output traces, the derived MED operator selectively suppresses frequencies over which the ratio of coherent signal to random noise is low. This noise suppression characteristic makes unstacked, output traces particularly suited for analysis of offset-dependent variations in reflectivity across bright spots.