An optimum data editing technique for the reduction of noise in multi-channel geophysical data

Abstract

We develop a data editing method, for the optimum interpolation of multichannel time series containing time-coincident data gaps, in one, several, or all channels based upon the autoregressive data model. The method is applied to a set of very long-period seismic data recorded during the 19 August 1977 Indonesian earthquake, which shows several unassociated bursts of noise. Spectral analysis following editing and interpolation of the record indicates existence of systematic signals with periods higher than 1 hr and perhaps as long as 2 hr. The individual spectral peaks in this subseismic band have not been identified.