Developing a matched‐filter kernel for localization of complex sources in a dispersive ocean waveguide with array invariance.

Abstract

The array invariant method was previously developed by Lee and Makris [J. Acoust. Soc. Am. (2006)] for instantaneously localizing both transient and continuous broadband acoustic sources after conventional plane‐wave beamforming and matched filtering array measurements in an ocean waveguide. The approach has been demonstrated with experimental data for a source transmitting linear frequency modulated pulses and theoretically for multiple uncorrelated broadband noise sources. Spectral analyzes of field‐recorded data acquired during the ocean acoustic waveguide remote sensing Gulf of Maine 2006 Experiment show that acoustic signatures from many random sources, such as marine mammal vocalizations, are complex and exhibit irregular patterns that vary over time. Furthermore, the signals embedded in noise may also contain interfering sources correlated with the signal over some time duration. Here, we extend the array invariant method for range and bearing estimation of such complex sources. A robust matched filter kernel based on the acoustic properties of field‐recorded whale vocalizations is developed to maximize the signal‐to‐noise ratio. Examples that simulate the beamformed and matched filtered array measurements of complex whale vocalizations embedded in noise in a random ocean waveguide are provided to demonstrate that range and bearing estimation using array invariance shows good agreement with true whale location.