SEMBLANCE AND OTHER COHERENCY MEASURES FOR MULTICHANNEL DATA

The concept of semblance is introduced, along with a descriptive review of several of the more common likeness or coherence measures. Measures are considered from three points of view: the domain in which they are applied, the philosophy of their design, and the manner in which they are used.Crosscorrelation, the most familiar of the likeness criteria, is examined in detail. Differences of design philosophy are noted as expressing themselves by a change in the normalization. Semblance is shown to be related to an energy-normalized crosscorrelation and to share certain features of the summation method or stack which has been used recently as a coherence measure.Several coherence measures, including semblance, are considered in a problem environment--the determination of stacking velocities from multiple ground coverage seismic data. A noise-free synthetic example is studied in order to compare discrimination thresholds of the various methods. Semblance, when properly interpreted, proves to have the greatest power of discrimination among the candidates examined for the particular application.

Перевод пока недоступен