Application of matching pursuit algorithm for dispersive acoustic signals

Abstract

Dispersive acoustic signals contain significant amounts of information concerning the shallow water environment structure. It is possible to unravel the modal propagation dispersion characteristics using a proper time-frequency analysis. Recent experiments such as ShelfBreak Primer and ASIAEX collected long range propagation data in shallow water on vertical line arrays from SUS charge explosive sources. Signals collected by the vertical arrays were processed using Morlet wavelets to observe the modal arrival structure due to group speed dispersion. Although the wavelet transform can extract the time-frequency information, it is not sufficient to get high resolution for high frequencies. The matching pursuit algorithm is an alternative approach that provides excellent spectral localization. We present an improved processing algorithm based upon the matching pursuit algorithm to analyze the data from ASIAEX-2001. This analysis yields the dispersion characteristics of the acoustic signals and clearly shows the modal arrivals, especially for higher modes. These modal dispersion diagrams from the results form the basis for long range sediment tomography. [Work supported by ONR.]