Hydroacoustic Signal Classification of Fish Schools by Species

Features extracted from hydroacoustic backscatter from fish schools enabled classification by species. Target species were cod (Gadus morhua), capelin (Mallotus villosus), and mackerel (Scomber scombrus), observed in the northern Gulf of St. Lawrence during 1985–86. Two features of internal school d...

Full description

Bibliographic Details
Published in:Canadian Journal of Fisheries and Aquatic Sciences
Main Authors: Rose, G. A., Leggett, W. C.
Format: Article in Journal/Newspaper
Language:English
Published: Canadian Science Publishing 1988
Subjects:
Online Access:http://dx.doi.org/10.1139/f88-073
http://www.nrcresearchpress.com/doi/pdf/10.1139/f88-073
id crcansciencepubl:10.1139/f88-073
record_format openpolar
spelling crcansciencepubl:10.1139/f88-073 2023-12-17T10:30:23+01:00 Hydroacoustic Signal Classification of Fish Schools by Species Rose, G. A. Leggett, W. C. 1988 http://dx.doi.org/10.1139/f88-073 http://www.nrcresearchpress.com/doi/pdf/10.1139/f88-073 en eng Canadian Science Publishing http://www.nrcresearchpress.com/page/about/CorporateTextAndDataMining Canadian Journal of Fisheries and Aquatic Sciences volume 45, issue 4, page 597-604 ISSN 0706-652X 1205-7533 Aquatic Science Ecology, Evolution, Behavior and Systematics journal-article 1988 crcansciencepubl https://doi.org/10.1139/f88-073 2023-11-19T13:38:56Z Features extracted from hydroacoustic backscatter from fish schools enabled classification by species. Target species were cod (Gadus morhua), capelin (Mallotus villosus), and mackerel (Scomber scombrus), observed in the northern Gulf of St. Lawrence during 1985–86. Two features of internal school density (20 log R amplification) were the best discriminators. These were mean standardized peak to trough distance (SPT) and mean distance between voltage peaks (PP). Quadratic discriminant functions based on the variables SPT, PP, an inverse coefficient of variation, school depth, and off-bottom distance correctly classified 93% of schools (1986). These functions also correctly classified 93% of cod and capelin schools acoustically sampled independently during 1985. The target strength of individual fish was a less successful discriminator of species. For cod and capelin of known length, average target strength (TS) was a linear function of length: TS (decibels) = −65 + 20 log 10 length (centimetres). Mackerel had target strengths that were 8–12 dB less than those of cod of equivalent length. Quadratic discriminant functions based on target strength, school depth, and off-bottom distance correctly classified 77% of schools by species. Our methods are generalized to any schooling species or environment. Article in Journal/Newspaper Gadus morhua Canadian Science Publishing (via Crossref) Canadian Journal of Fisheries and Aquatic Sciences 45 4 597 604
institution Open Polar
collection Canadian Science Publishing (via Crossref)
op_collection_id crcansciencepubl
language English
topic Aquatic Science
Ecology, Evolution, Behavior and Systematics
spellingShingle Aquatic Science
Ecology, Evolution, Behavior and Systematics
Rose, G. A.
Leggett, W. C.
Hydroacoustic Signal Classification of Fish Schools by Species
topic_facet Aquatic Science
Ecology, Evolution, Behavior and Systematics
description Features extracted from hydroacoustic backscatter from fish schools enabled classification by species. Target species were cod (Gadus morhua), capelin (Mallotus villosus), and mackerel (Scomber scombrus), observed in the northern Gulf of St. Lawrence during 1985–86. Two features of internal school density (20 log R amplification) were the best discriminators. These were mean standardized peak to trough distance (SPT) and mean distance between voltage peaks (PP). Quadratic discriminant functions based on the variables SPT, PP, an inverse coefficient of variation, school depth, and off-bottom distance correctly classified 93% of schools (1986). These functions also correctly classified 93% of cod and capelin schools acoustically sampled independently during 1985. The target strength of individual fish was a less successful discriminator of species. For cod and capelin of known length, average target strength (TS) was a linear function of length: TS (decibels) = −65 + 20 log 10 length (centimetres). Mackerel had target strengths that were 8–12 dB less than those of cod of equivalent length. Quadratic discriminant functions based on target strength, school depth, and off-bottom distance correctly classified 77% of schools by species. Our methods are generalized to any schooling species or environment.
format Article in Journal/Newspaper
author Rose, G. A.
Leggett, W. C.
author_facet Rose, G. A.
Leggett, W. C.
author_sort Rose, G. A.
title Hydroacoustic Signal Classification of Fish Schools by Species
title_short Hydroacoustic Signal Classification of Fish Schools by Species
title_full Hydroacoustic Signal Classification of Fish Schools by Species
title_fullStr Hydroacoustic Signal Classification of Fish Schools by Species
title_full_unstemmed Hydroacoustic Signal Classification of Fish Schools by Species
title_sort hydroacoustic signal classification of fish schools by species
publisher Canadian Science Publishing
publishDate 1988
url http://dx.doi.org/10.1139/f88-073
http://www.nrcresearchpress.com/doi/pdf/10.1139/f88-073
genre Gadus morhua
genre_facet Gadus morhua
op_source Canadian Journal of Fisheries and Aquatic Sciences
volume 45, issue 4, page 597-604
ISSN 0706-652X 1205-7533
op_rights http://www.nrcresearchpress.com/page/about/CorporateTextAndDataMining
op_doi https://doi.org/10.1139/f88-073
container_title Canadian Journal of Fisheries and Aquatic Sciences
container_volume 45
container_issue 4
container_start_page 597
op_container_end_page 604
_version_ 1785583348563836928