Matched‐field source localization using data‐derived modes

Acoustic modes can be derived from the cross‐spectral density matrix of a vertical line array in an ocean wave guide, when spatially distributed ambient noise or a moving source causes the modes at the array to be uncorrelated. This typically produces modes that are noisy and incomplete (only some modes are recovered, at some frequencies, and only at depths spanned by the array). Furthermore, without wave numbers, only incoherent localization algorithms are possible, providing less gain than coherent MFP using wave numbers. It will be shown how to overcome these problems, assuming the sound speed profile has been measured in the water column but assuming no a priori knowledge of bottom properties. A data‐dependent shooting method is presented which extrapolates the data‐derived modes to the entire water column and estimates both the wave numbers and bottom properties. Having bottom parameters enables modes to be calculated at all frequencies, rather than being limited to frequencies of sources of opportun...