Multi-component seismic event correlation in coherent noise

Abstract

A new correlation and picking algorithm has been developed which uses multi-component seismic data from two stations. A conventional single-station covariance matrix analysis is well known to provide polarization estimates of the particle motion at a given point in space. This single-station technique is successful if any noise present on the data is purely random. However, in the presence of coherent noise, the single-station analysis breaks down. A two-station (binocular) covariance matrix provides similar information at two points in space, as well as indicating the correlation between the two stations. The major advantage of the procedure is that it can successfully identify overlapping arrivals (in the time domain) where a single-station analysis fails. The binocular analysis may be sequentially applied to a multi-station seismic gather using the same reference signal to identify a coherent event. The technique was applied to synthetic data and then successfully used to pick overlapping events on data acquired in a physical scale-model experiment.