A multifrequency method to classify and evaluate fisheries acoustics data

Abstract

Acoustic surveys have been conducted on Georges Bank from 1998 to present to estimate Atlantic herring (Clupea harengus) population abundance. Acoustic data were collected with a 12 or 18, 38, and 120 kHz Simrad EK500 scientific echo sounder. A pelagic trawl and underwater video images were used to collect biological information and to verify the species composition of acoustic backscatter. A multifrequency classification method was developed to improve the efficiency and accuracy of classifying species from acoustic echograms. In this method, a volume backscatter (Sv) threshold was applied equivalently to all echograms, and then a composite echogram was created based on which frequencies had Sv greater than or less than the Sv threshold. The results of this method were compared with the standard method of visually scrutinizing regions, and metrics were developed to evaluate the accuracy of classification algorithms relative to current methods, as well as to assess the effects of classification methods on population abundance estimates. In general, this method matched visually scrutinized Atlantic herring regions, but with consistent biases in classifying 38 kHz backscatter. The metrics highlighted spatial and temporal changes in the acoustic landscape, which may be indicative of intra- and inter-annual biological changes.