A method for reducing uncertainty in estimates of fish-school frequency response using data from multifrequency and multibeam echosounders

Type Article
Date 2009-07
Language English
Author(s) Berger Laurent1, Poncelet CyrilleORCID1, Trenkel VerenaORCID2
Affiliation(s) 1 : IFREMER, Ctr Brest, Dept NSE, F-29280 Plouzane, France.
2 : IFREMER, Dept EMH, F-44311 Nantes 03, France.
Source ICES Journal of Marine Science (1054-3139) (Oxford university press), 2009-07 , Vol. 66 , N. 6 , P. 1155-1161
DOI 10.1093/icesjms/fsp113
WOS© Times Cited 10
Keyword(s) Species identification, Scomber scombrus, Sardina pilchardus, Pelagic fish, Acoustics
Abstract Fish schools can be insonified simultaneously with multifrequency echosounders (e.g. Simrad EK60s) and a multibeam echosounder (e.g. Simrad ME70). This paper presents a method for combining these data to improve estimates of the relative frequency response r(f) of fish schools. Values of r(f) are now commonly used to classify echoes in fishery surveys. The data from the roll- and pitch-stabilized, high-resolution ME70 are used to correct beam-width effects in the multifrequency EK60 data. First, knowing the exact position and orientation of the transducers and the position of the vessel, the echoes are placed into a common geographic coordinate system. Then, the EK60 data are rejected if they do not include a significant percentage of the fish school imaged with the multibeam echosounder. Echoes that exceed the overlap threshold are used to estimate the r(f). The proposed method is applied to simulated and actual data for sardine and mackerel schools in the Bay of Biscay to estimate their r(f) values. The results for different overlap thresholds are compared with the results of a different method, one that uses adaptive thresholds on volume-backscattering strength Sv. The proposed method reduces uncertainty in estimates of r(f) for schools with an overlap of greater than 80%, and it outperforms the Sv-thresholding technique.
Full Text
File Pages Size Access
publication-6590.pdf 10 339 KB Open access
Top of the page