Broadband and statistical characterization of echoes from random scatterers : application to acoustic scattering by marine organisms

Abstract

The interpretation of echoes collected by active remote-sensing systems, such as sonar and radar, is often ambiguous due to the complexities in the scattering processes involving the scatterers, the environment, and the sensing system. This thesis addresses this challenge using a combination of laboratory and field experiments, theoretical modeling, and numerical simulations in the context of acoustic scattering by marine organisms. The unifying themes of the thesis are 1) quantitative characterization of the spectral, temporal, and statistical features derived from echoes collected using both broadband and narrowband signals, and 2) the interpretation of echoes by establishing explicit links between echo features and the sources of scattering through physics principles. This physics-based approach is distinct from the subjective descriptions and empirical methods employed in most conventional fisheries acoustic studies. The first part focuses on understanding the dominant backscattering mechanisms of live squid as a function of orientation. The study provides the first broadband backscattering laboratory data set from live squid at all angles of orientation, and conclusively confirms the fluid-like, weakly-scattering material properties of squid through a series of detailed comparisons between data and predictions given by models derived based on the distorted-wave Born approximation. In the second part, an exact analytical narrowband model and a numerical broadband model are developed based on physics principles to describe the probability density function of the amplitudes of echo envelopes (echo pdf) of arbitrary aggregations of scatterers. The narrowband echo pdf model significantly outperforms the conventional mixture models in analyzing simulated mixed assemblages. When applied to analyze fish echoes collected in the ocean, the numerical density of fish estimated using the broadband echo pdf model is comparable to the density estimated using echo integration methods. These results demonstrate the power of the physics-based approach and give a first-order assessment of the performance of echo statistics methods in echo interpretation. The new data, models, and approaches provided here are important for advancing the field of active acoustic observation of the ocean.