Interpretation of velocity spectra

One major effort in processing seismic data is that of correcting for the geometric effects or normal moveout introduced by multichannel data gathering procedures. Usually, this involves detecting coherent reflected signals, determining a time (t) versus moveout (Δt) relationship for these signals, time correcting the individual receiver traces, and combining selected traces into a stacked or composite section. Most of these operations are automated. Several velocity analysis techniques are available for detecting reflections automatically, and individual trace correcting and combining procedures are trivial. However, determination of a moveout function which will discriminate among valid reflections, multiple reflections, and noise is still largely a human operation. We describe a machine method devised to determine automatically these moveout, or t‐Δt, functions, as they are commonly called. The method makes extensive use of the observed phenomenon that the t‐Δt function is well approximated by a segmented straight‐line curve when plotted on a logarithmic moveout (Δt) versus linear time (t) graph. Criteria for determining these straight‐line functions, which include a constraint on the line segment slopes, are discussed; several examples resulting from their use are shown. In the examples, seismic sections corrected by our machine‐derived moveout functions are compared with those corrected by more conventionally determined functions.