Acoustic backscatter from zooplankton and fish explored through an optimized model framework

Abstract

The purpose of this work has been to test complementary methods in order to classify marine organisms, with particularly attention to zooplankton and fish. Algorithms to separate fish and zooplankton have been developed and implemented at IMR and at IRD. A novel optimised model framework based on known scattering models are used to classify zooplankton and to separate these from fish. Acoustic data from up to 6 frequencies were collected to test the scattering model framework, while concurrent biological samples from multi-net oblique or horizontal MOCNESS tows, WP2 vertical net hauls and pelagic trawl were also obtained and analysed. Great attention are given on one side to the inter calibration and the comparability of all the frequencies, and to the space and time coherence between the samples collected and the acoustical data which are processed. All algorithms involve zooplankton scattering models, the high-pass ones from Stanton, or the more complex ones like the truncated fluid sphere from Holliday or the DWBA set of models from Chu and Stanton. A set of reliable acoustical and biological data has been chosen in order to proceed to comparisons between the results of the acoustic data processing through the classification algorithms and the results of the biological processing.