Particle filtering for geoacoustic characterization of the soft sediment using ship noise

In coastal areas, the properties of the soft sediment are crucial for sound propagation prediction and relevant sonar applications. The noise of intense ship traffic is a convenient acoustic source in these areas for geoacoustic characterization. This paper introduces a sequential particle filter (PF) technique that can take advantage of the continuous ship sound field to predict the soft sediment properties along ship track. The PF is based a point mass (or “particle”) representation of the probability density function associated to different parameters. The particle set evolves recursively according to newly input data, which allows updating the estimation in real time and suitable for dynamical system monitoring if used properly. The approach is tested on synthesized vertical waveguide impedance data to characterize the sediment properties for the obtained bottom geoacoustic model offshore the Amazon River mouth. The vertical waveguide impedance is proven to be source spectrum independent and valuable for passive geoacoustic inversion. Simulation results demonstrate that the PF provides a statistical estimation of the geoacoustic parameters close to true values versus range, in addition, the intrinsic features of the PF filtering technique in successively updating the estimate suggest our approach is capable of characterizing more complex environments with varying geometries.