Present day computer systems allow computation of an approximate Wiener operator for the time-domain filtering and smoothing of signals of finite (and short) length. However, the memory capacity of a particular computer storage introduces a constraint on the length of the operator and signal. To produce the least mean-square-error, it was helpful to know the extent to which the length of the operator whould be compromised by including additional information (sample points) of the signal. If N is the number of samples of the signal to be equalized and T is the number of points in the optimum digital operator, then the system imposes the constraint that  
where M is the free capacity of the computer memory and K is the number of multiples of signal length required for the correlations.

To find out whether an optimum relation between T and N exists, a program to determine the digital operator and to compute the mean-square-error between the filtered result and the desired signal was written for an IBM 7090 computer. The computed errors for some typical seismic signals are presented as a function of operator length, signal length, signal-to-noise ratio, and the number of records used to determine averages. The results suggest that equalization occurs both as a process of signal shaping and of noise-reduction, but not necessarily simultaneously.

This content is PDF only. Please click on the PDF icon to access.

First Page Preview

First page PDF preview
You do not currently have access to this article.