Optimal estimation of the signal of multi-channel seismic record in case of complicated model of regular and irregular noise
© Yu.K. Tyapkin, Yu.V. Roganov, I.A. Nekrasov
A method is described for estimating a signal with a complicated, more realistic as compared with the conventional one, mathematical model of the multichannel seismic record containing regular and irregular noise. It is based on the Wiener optimality criterion and the related multichannel filter. The model supposes that, besides a difference in their arrival times, the signal and regular noises are independent random processes and have amplitudes that vary across the channels in any manner. Irregular noise is characterised by an arbitrary trace-dependent variance. The spectral (correlative) properties of all the above constituents of the record are trace-independent. Under certain conditions, the method may be reduced to successively subtracting all the regular noises and estimating the signal on the residual record. Both processed use optimum weighted stacking, with reference to the related arrival times. Algorithms are suggested for estimating the parameters needed to introduce the method. The efficiensy of the method in eliminating severe noise is demonstrated with field data sets.