Matlab software for the analysis of seismic waves recorded by three-element arrays

Abstract

We develop and implement an algorithm for inverting three-element array data on a Matlab platform. The algorithm allows reliable estimation of back-azimuth and apparent velocity from seismic records under low signal-to-noise conditions. We start with a cubic-spline interpolation of the waveforms and determine the differences between arrival times at pairs of array elements. The time differences are directly computed from cross-correlation functions. The advantages of this technique are: (a) manual picking of the onset of each arrival is not necessary at each array element; (b) interpolation makes it possible to estimate time differences at a higher resolution than the sampling rate of the digital waveforms; (c) consistency among three independent determinations provides a reliability check; and (d) the value of apparent velocity indicates the nature of the recorded wavelet and physically checks the results. The algorithm was tested on data collected by a tri-partite array (with an aperture of ~250 m) deployed in 1998 by the National Data Center of Israel, during a field experiment in southern Israel, 20 km southwest of the Dead Sea. The data include shallow explosions and natural earthquakes under both high and low signal-to-noise conditions. The procedure developed in this study is considered suitable for searching of small aftershocks subsequent to an underground explosion, in the context of On-Site Inspections according to the Comprehensive Nuclear-Test-Ban Treaty (CTBT).