GB1363383A - Multistage detection of velocity estimation from seismograms - Google Patents

Abstract

1363383 Seismic signal processing MOBIL OIL CORP 6 June 1972 [29 July 1971] 26310/72 Heading H4D Velocity data are computed automatically from a set of seismic reflection traces by a twostage method, the first stage comprising searching the traces at instantaneous values of trace time to determine estimates of the reflection times and of acoustic velocity, the second stage comprising searching the traces within time windows at the estimated reflection times to determine velocity within a velocity range centred on the estimated acoustic velocity. In one embodiment an input common depth point (CDP) set of digitized traces is searched along assumed normal moveout (NMO) curves as in Specification 1,304,248 (Division H4) to determine the estimated times T N and velocities V N of successive reflections. This is the first stage and is denoted by steps 20 to 36; it is repeated for each CDP set. Estimated velocity is plotted as a function of trace time T and horizontal distance X along a survey line. In the first stage search the trace amplitude values across a CDP set, at trace times lying on an assumed NMO curve, are noted and summed at 33. Each sum exceeding a threshold α is stored at 37 to derive a respective amplitude sumversus-time trace for each CDP set. At 38 a search is made across these traces, along assumed dip lines, to determine the dip D N of each subsurface reflector. In the second stage a time window of duration corresponding to the expected length of a seismic wavelet is selected at 48 at a time tx on each trace of a CDP set, values of tx being calculated at 47 relative to a selected reflection time T FIRST so as to take into. account set-to-set dip and trace-to-trace moveout (dependent on an assumed velocity V FIRST in a range centred on the estimated velocity V N ) across the set. The signal in each window is detected at 49 by, e.g. a summation of amplitudes or by a zero-lag cross-correlation of window pairs, the correlation products being summed to give at 50 a measure of signal power. By iteration, the signal power for other assumed velocities within the range is determined and the assumed velocity giving the maximum power is stored at 53 as being the best estimate V NEW of the actual velocity. Steps 47 to 53 are iterated for other estimated reflection times T N .