Paul A. Baxley
Ocean and Atmospheric Sci. Div., Code 541, NRAD, NCCOSC, San Diego, CA 92152-5000
Robert Bruce Williams
NRAD, NCCOSC, San Diego, CA 92152-5000
William H. Hodgkiss
Scripps Inst. of Oceanogr., San Diego, CA 92152-6400
Matched-field processing (MFP) is well-suited for tracking applications because of its inherent source localization capability, as demonstrated by Wilmut et al. [J. Acoust. Soc. Am. 93, 2374 (A) (1993)]. If exact track determination is not an objective, improved detection at a low signal-to-noise ratio may be obtained using a simple track-before-detect technique that averages range-shifted MFP ambiguity-surface snapshots, with the range shift being determined by an assumed range rate. After searching through all candidate range rates, the signal function is approximately reconstructed in the average surface corresponding to the correct range rate. If peak-to-sidelobe relationships in the signal function vary little between snapshots, both the sidelobes and the main peak are maximized in the average surface for the correct range rate; otherwise, only the main peak is maximized. In either case, this maximization provides a detection clue along with an estimate of range rate. This detection enhancement at low signal-to-noise ratios is demonstrated via simulations for a 48-element full-water-column vertical array in a thickly sedimented shallow-water environment (500-m water depth) typical of the southeastern Mediterranean sea. A 50-Hz source moving at constant range rate and source depth was simulated using a normal mode program, while noise samples were obtained using a PE-based shipping noise model and the HITS II shipping distribution database.