Space Averaged Dynamic Correlation Analysis of Seismic Signals

摘要

1,190,792. Processing seismic records. TEXAS INSTRUMENTS Inc. 9 May, 1967 [16 May, 1966], No. 21411/67. Heading H4D. Seismic data relating to a reflecting horizon is averaged by cross-correlating the commonlycentred time gates of a pair of traces which include reflection from the horizon and were derived using two different energy travel path geometries, cross-correlating commonly-centred time gates of a second pair of traces which include reflection from the horizon and were derived using two different path geometries, and summing the two correlation results. The processing may be applied to at least three traces obtained by common depth point or continuous profiling methods and reproduced from digital magnetic tape records. In the Fig. 1 example for giving six-fold subsurface coverage of depth points a series of 24-trace records is made by shooting at locations A, B, C &c. in turn into the end of a linear array of 24 equispaced geophones. Denoting the respective traces derived using shot points A, B &c. and geophones 1, 2 . . . 24 as A1 . . . A24, B1 . . . B24 &c., then traces F1, E5, D9, C13, B17, A21 (i.e. traces having a " space lag " of four geophone spacings) all involve one common depth point, traces G1, F5, E9 &c. involve a second common depth point and so on. After correction for normal movement error the said time gates of traces F1 and E5 are cross-correlated and similar time gates of traces G1 and F5 are cross-correlated. These two primary " 1,5 " correlations are then summed to give an output correlation signal. Further output correlation signals are similarly derived from " 2,6," " 3,7," " 4,8," &c. primary correlations and the output correlation signals are displayed against a correlation delay time time-base. In practice each output correlation signal may be derived by summing more than two primary correlation signals (e.g. twenty to thirty or more may be summed). The processing is repeated for different time gates along the trace length: the time gate centres may be separated by equal increments of record time and the gates may be of different length and in spaced, contiguous, or overlapping relation. Where the time gates used in the processing are taken at late record time, the primary and multiple events in the display can be distinguished more readily if there is a large residual moveout between traces and to this end the traces selected for the primary cross-correlations are then those with a " space lag " of more than four geophone spacings, e.g. eight, twelve &c. spacings. To determine the static error corrections (which may be applied to the primary correlations before they are summed), the primary correlations are cross-correlated in pairs to give secondary correlations. The time shift of the peak of a secondary correlation, relative to the correlation delay time origin, is the time shift which must be applied as the static error correction between the primary correlations yielding that particular secondary correlation. Additional processing may comprise velocity filtering of the output correlation signals. The application of the invention to processing traces obtained by continuous profiling methods differs in the selection of traces for the primary cross-correlation; details are given.