On the use of high-frequency broadband sonar to classify biological scattering layers from a cabled observatory in Saanich Inlet, British Columbia

Abstract

This paper explores the use of broadband acoustics to differentiate between biological scattering layers using observatory-based acoustic observations with minimal supporting biological observations. Targets and layer assemblages were classified based on 85–155 kHz acoustic data collected on the VENUS observatory in Saanich Inlet, B.C. between March 2008 and February 2010 using a clustering algorithm and different broadband acoustic data descriptors. First, a 6-h segment of data, for which there were coincident depth-resolved net-tow data, was analyzed. Clustering based on the calibrated spectrum of volume scattering strength for each target resulted in clusters that were distributed just as those resulting from clustering based on 120 kHz narrowband data because the clustering was dominated by the scattering level, rather than the spectral shape. When the target spectra were normalized, the clustering results were consistent with the different taxa found in the net samples, but often could not distinguish taxonomic groups. However, layers with distinct species assemblages had different distributions of target classifications, suggesting the assemblages could be distinguished using frequency-dependent scattering information. Ensemble-averaging the scattering observations and converting the spectral data to a 3-descriptor acoustic color representation prior to clustering was (1) more effective at distinguishing the dominant scattering layers based on their assemblages and (2) much more efficient in terms of computational cost. Clustering two years of acoustic-color data identified 4 main groups (diel migrating euphausiids and chaetognaths, fish, and a mix of pteropods and bottom-to-oxycline migrating amphipods) that were consistent with contemporaneous and historical observations of zooplankton in the inlet. A wider frequency band might be effective in better distinguishing individual zooplankton targets.