Sequential vs batch processing of ship noise data in inversion for geoacoustic properties

Abstract

As part of ECOAO 13 acoustic fisheries survey off Senegalese coast, geoacoustic measurements were performed by deploying a newly-designed vector sensor array to record the noise field generated by N/O Antea survey ship. This paper describes the application of a particle filter (PF) to estimate sequentially bottom geoacoustic parameters during the ship passage. The PF estimates the parameters by constructing their posterior probability density functions (PDFs) using a set of weighted particles. The PDF embodies all available statistical information and provide a complete solution to the estimation problem, from which the optimal solution and its associated confidence can be derived. In this paper, the acoustical observables are the range-varying amplitudes of vertical impedance at multiple frequencies, as measured from different pairs of array elements and sequentially processed by the PF. The obtained PDFs demonstrate that the PF approach provides performance similar to that of global optimization, suggesting that the bottom can be well described acoustically as a half space of sand sediment. This is in good agreement with a Log-Probit interpretation of sieving data from surface grab samples taken at the same location as the acoustic data. The combination of mobile compact array and particle filtering appears to be well suited for environmental mapping of coastal areas with space-or time-varying bathymetry.