Construction of a Bayesian inversion method for determining seafloor sediment properties from simulated transmission loss data

Abstract

Geoacoustic inversion of transmission loss measurements for determining sediment properties greatly reduces the time and cost needed to characterize the seafloor compared to cores or other in situ measurements. This study sought to determine the efficacy of a Bayesian inversion scheme to determine sediment parameters from transmission loss using simulated data, created using a hybrid finite element/ray acoustic model. For simplicity, the sediment was described as a fluid. A reduced order inversion parameter set was chosen to limit computational time and parameter intercorrelation, as well as to limit the search space to what is believed to be the most contributive parameters in the high frequency regime. This set includes sound speed, density, attenuation, and the von Karman spectral strength and exponent to describe interface roughness. The inversion scheme yielded marginal posterior probability distributions for each of the parameters, as well as information about parameter resolvability and covariance. [Work supported by the Office of Naval Research, Task Force Ocean.]

Influence of noise on model selection in geoacoustic parameter inversion

“…Under the assumption of Gaussian data error, an error function is established to measure the matching degree between the theoretical vector acoustic field and the measured vector acoustic field based on the likelihood function [5,6]. The Bartlett mismatch device, K represents the number of hydrophones, F represents the number of frequency points used, and p represents the acoustic pressure field.…”