Inversion of geoacoustic model parameters using ship radiated noise

Estimation of geoacoustic model parameters is an underlying research issue in understanding acoustic propagation in shallow water environments, where the propagation is generally bottom limited. Inversion methods based on matched field processing have become widely used in applications with experimental data. Traditionally, the experiments have been carried out with controlled source geometries and either vertical or horizontal arrays. This paper presents an experimental approach that makes use of the random noise radiated by passing ships as the sound source for the inversion. The inversion uses a recently developed cross-relation matched field processor that is insensitive to the waveform of the random noise signal radiated by the ship. Ship noise data were measured on a 16element vertical line array in shallow water off the west coast of Vancouver Island. The data were filtered into low (73-113 Hz) and high (190-270 Hz) frequency bands, and processed in an inversion algorithm based on back propagation of the spectral components of the noise signal. The geoacoustic model that generated the most accurate focus at the source location was taken as the best estimate. The band limited data allowed estimation separately of geoacoustic model parameters of the sea floor with the high frequencies, and then for the deeper layers using the low frequencies. The estimated model parameters compared well with ground truth data from a seismic survey and from sediment samples at the site.