A coherent approach to broadband matched-field processing: application in the Hudson Canyon

In matched-field processing, sophisticated models of the ocean channel are used to provide accurate steering vectors. These steering vectors are then used with standard beamformers (conventional, MVDR, etc.). The Hudson Canyon data set is one of a few publicly available data sets in a form directly suitable for matched-field processing. Using these data we show successful tracking out to a range of 5 km. Exploitation of broadband information is critical for that processing. However, we show that the MVDR processor is only successful when the broadband information is exploited coherently.