Volume 72

Taxonomical classification of reef fish based on a swimbladder BEM, broadband echosounder modeling; and Bayesian, SVM, and KNN estimators

Roa, C; K. Boswell; G. Pedersen; C. Taylor; M. Bollinger; S. Labua
Download PDF Open PDF in Browser

Other Information

Date: November, 2019

Pages: 368

Event: Proceedings of the Seventy-Two Annual Gulf and Caribbean Fisheries Institute

City: Punta Cana

Country: Dominican Republic


The recent development and commercial availability of broadband echosounders have the potential to classify acoustic targets based on their scattering responses, which are expected to be a function of their species-specific morphological and physiological properties. This is particularly important in complex environments with biologically diverse fish assemblages. Using theoretical acoustic scattering models, we examined the potential to taxonomically classify dominant reef fish based on the fine-scale gas-bearing swimbladder morphology quantified from three-dimensional computed-tomography models. Echoes of the swim bladder for an incident broadband sound source (30 – 200 kHz) and orientation angles with respect to the fish between +/- 45o from normal incidence were acoustically simulated using the boundary element method (BEM). They presented characteristics that were consistent within species and distinguishable among them. We used a Bayesian, Support Vector Machine and K- Nearest Neighbor estimators to classify the broadband echoes and compare them to a multi-frequency case. The classifiers had accuracies between 80% and 90%, performing better in the broadband case. The modeling and classification approach presented here indicates that a taxonomic distinction based on morphologically-dependent scattering responses is possible. Furthermore, it represents an important step toward improving marine ecosystem acoustics for managing and assessing reef fish communities

PDF Preview