Design and applications of optimum multichannel linear moveout filters

Abstract

Abstract Practical aspects of the optimum multichannel least-squares (OMLS) linear moveout filter design parameters are investigated by examining them in the frequency-wavenumber domain and applying them to simulated and field reflection seismograms. The results of these studies show that the so-called 'chatter parameter' should be omitted from the input design models of the OMLS moveout filters. Random noise relative weightings of about 0.3 to 0.5 are adequate for stabilizing the autocorrelation matrix of the normal equations. Linear moveout filters with weights higher than 0.5 act as high-frequency filters and reduce the resolution of the data. A serious limitation of the OMLS linear moveout filters for seismic reflection applications is aliasing of the reject moveout window into the pass window which prevents weightings of the reject moveout window from being much higher than for pass-moveout windows when higher weightings are required in practical applications such as grand-roll reduction.Simulated data indicate that multichannel filtering is superior to stacking in improving the ratio of signal to correlated noise, especially when there are large static shifts in the input traces.