Source tracking in the Hudson Canyon experiment

In the Hudson Canyon experiment, a broadband source, transmitting simultaneously in four frequencies, moved in range at a constant depth and bearing. Using broadband matched-field processing we demonstrate that the source can be localized and tracked. Incoherent broadband approaches for matched-field processing, based on the averaging of ambiguity surfaces obtained with the narrowband Bartlett and Minimum Variance processors, are compared to a new coherent variant. Localization is very successful when either the incoherent or coherent Bartlett estimator is used. The Minimum Variance processor performance in determining the source location is very poor in the incoherent case but improves significantly with the introduction of the coherent scheme.