Bayesian inversion of seabed reverberation and scattering data.

Abstract

This paper describes a Bayesian approach to the inversion of ocean acoustic reverberation data for scattering and geoacoustic parameters of the seabed. The seabed is modeled as a sediment layer over a semi‐infinite basement. Interface scattering occurs at the (rough) upper and lower boundaries of the sediment layer, and volume scattering occurs within the layer (the scattering mechanisms are considered to be independent and are modeled using perturbation theory and the Born approximation). Unknown parameters include geoacoustic properties (sediment‐layer thickness and sound speeds, densities, and attenuations for sediments and basement) and scattering properties (roughnesses and scattering strengths for upper and lower sediment boundaries, and volume scattering strength). One dimensional (1‐D) and two‐dimensional (2‐D) marginal probability distributions are computed from the multidimensional posterior probability density (PPD) using Metropolis–Hastings sampling applied in a principal‐component parameter space to provide efficient sampling of correlated parameters. Results indicate that reverberation inversion is a strongly nonlinear inverse problem, with highly multimodal marginal distributions and strong interparameter correlations. Addressing this nonlinearity is of key importance to understanding the information content of reverberation data. [Work funded by the ONR.]