How much more informative are broadband compared to narrowband echoes for biological interpretation?

Abstract

The recent availability of commercial broadband echosounders has elicited wide interests in their potentials for enhancing the effectiveness, efficiency, and accuracy of acoustic sensing capability for monitoring mid-trophic level marine organisms. However, despite the significantly improved temporal and spatial resolutions, it remains unclear how the additional spectral information provided by broadband echosounders contribute to achieving these goals. In this study, we use a Bayesian inversion framework to compare the estimation uncertainty between broadband and narrowband echo data for biological model parameters, such as organism length, tilt angle, numerical density and aggregation composition. We employ the Markov Chain Monte-Carlo (MCMC) sampling technique to construct the posterior probability density (PPD) of biological parameters given simulated zooplankton and fish echo data in the form of calibrated volume backscattering strength (Sv). The data are simulated for frequency ranges commonly employed in marine ecological and fisheries surveys. We investigate the changes in PPD in response to variations in echo spectral information, with specific emphasis on the correlation structure among model parameters and whether and how broadband information reduces the uncertainty in inferring biological information from acoustic quantities available from field surveys. [Work supported by NMFS Office of Science and Technology Advanced Sampling Technology Working Group.]