Multi-frequency estimates of fish abundance: constraints of rather high frequencies

ISSN： 1054-3139

Source： ICES Journal of Marine Science, Vol.56, Iss.2, 1999-04, pp. : 184-199

Disclaimer: Any content in publications that violate the sovereignty, the constitution or regulations of the PRC is not accepted or approved by CNPIEC.

Previous Menu Next

Abstract

Traditional scientific echosounders operate at discrete frequencies ranging from 38 to 420 kHz. We investigated the applicability and accuracy of length-based population estimates using commercially available acoustic frequencies and the inverse approach under ideal conditions. The inverse approach combines modelled and measured backscatter values to estimate the abundance of organisms in specified length classes. Reference backscatter values of individual fish were calculated using a Kirchhoff-ray mode backscatter model. Single and multi-cohort fish populations were simulated based on length-frequency samples from purse seine catches of threadfin shad (Dorosoma petenense) and used to calculate the total frequency-dependent volume backscatter of each population. A non-negative least squares (NNLS) algorithm was used to estimate total abundance and numbers of fish in each length class. Total abundance estimates were within <1–38% of population numbers. Within length-class estimates were inconsistent among frequency combinations and across length-class criteria. Increasing the number of frequencies does not guarantee improved accuracy of within length-class abundance estimates. Predictability of inverse simulations is non-linear when rather high frequencies are combined with non-monotonic scattering models. Accuracy of length-based abundance estimates is optimized by maximizing the amplitude range of reference backscatter measures and the number of features identified by reference scattering points.