Statistically optimal stacking of seismic data

Abstract A theory for weighting seismic records in the stacking process has been developed from a statistical seismic model. The model applies to common-depth-point seismic records which have been statically and dynamically corrected; the same model applies to an ordinary stacking procedure. The model stipulates for the signal and noise components, respectively, of a seismic record that (1) the signal is coincident with and similarly shaped to the signal on other records, and (2) the noise is statistically independent of that on any other record and of the signal and has zero mean value.In accord with the model, a seismic record is completely described for the purpose of weighting by its signal scale and its signal-to-noise energy ratio. Several statistical procedures for evaluating these parameters for seismic field data are presented. The most favorable procedure is demonstrated with both synthetic and field seismic records.