This paper contains a brief review of the operations that are involved in digital processing of array seismic recordings by the methods of velocity filtering, summation, cross-multiplication and integration, and by combinations of these methods (the UK method and multiple correlation).
Analyses by the several techniques were made on array recordings that were obtained by the U. S. Geological Survey of seismic waves from chemical and nuclear explosions in the western United States.
Seismograms were synthesized, using recordings of seismic noise and a Pn signal, such that the signal-to-noise ratio, onset time and apparent velocity of the signal were predetermined for the synthetic records. These records were then analyzed by summation, cross-multiplication and the UK method, and the results were compared.
From the examples presented, it appears that for recordings with initial peak-signal-to-rms-noise ratio less than about 0.8, the best improvement can be obtained by the process of velocity filtering and summation of all the traces. For a six-element array improvement can be obtained by this method on records which have an original signal-to-noise ratio of 0.4 or greater. When the initial signal-to-noise ratio is greater than 0.8, multiplicative compounding yields about the same improvement as a process involving additive compounding and subsequent raising of the summation trace to large powers.