Accuracy and precision of fish-count data from a “dual-frequency identification sonar” (DIDSON) imaging system

Abstract

The reliability of sockeye-salmon (Oncorhynchus nerka) count data collected by a dual-frequency, identification sonar (DIDSON) system is evaluated on the basis of comparisons with visual counts of unconstrained migrating salmon and visual counts of salmon constrained to passing through an enumeration fence. Regressions fitted to the DIDSON count data and the visual count data from the enumeration fence were statistically indistinguishable from a line with slope = 1.0 passing through the origin, which we interpret as agreement in both counts. In contrast, the regressions fitted to the DIDSON count data and the unconstrained visual count data had slopes that were significantly <1.0 (p < 0.001) and are consistent with an interpretation of systematic bias in these data. When counts of both unconstrained and constrained fish from the DIDSON system were ≥50 fish event−1, repeated counts of the DIDSON files were observed to produce the same counts 98–99% of the time, respectively, and based on the coefficient of variation, counts of individual passage events varied <3% on average. Therefore, the DIDSON count data exhibit high precision among different observers. As an enumeration fence provides a complete census of all fish passing through it, we conclude that fish-count data produced by the DIDSON imaging system are as accurate as visual counts of fish passing through an enumeration fence when counts range up to 932 fish event−1, the maximum count recorded during our study, regardless of the observer conducting the count. These conclusions should be applicable to typical riverine applications of the DIDSON system in which the bottom and surface boundaries are suitable for acoustic imaging, the migrating fish are adult salmon, and the transducer is carefully aimed so that the beams ensonify the area through which the salmon are migrating.

Understanding Smolt Survival Trends in Sockeye Salmon

“…Scales and otoliths were collected so that estimates of spawner abundance could be partitioned into age-classes. Since 2006, imaging sonar (DIDSON) has been used to estimate spawner abundance Holmes et al 2006). Numbers estimated by DIDSON and by mark-recapture were similar and, since 2009, DIDSON estimates have replaced mark-recapture (Benner, and T. Cone, DFO Annacis Island, personal communication [s]).…”

Habitat Manipulations Confound the Interpretation of Sockeye Salmon Recruitment Patterns at Chilko Lake, British Columbia

“…The number of Chilko sockeye escaping fi sheries and returning to Chilko Lake to spawn were estimated prior to 2009 by mark recapture (e.g., Schubert and Fanos 1997), except for 1967, when they were estimated by expanding visual counts at Henry's Bridge, 12 km below the lake (Grant et al 2011). Since 2006, imaging sonar (DIDSON) has been used to estimate spawner abundance Holmes et al 2006). DIDSON and mark recapture estimates have been similar enough to rely on DIDSON estimates alone beginning in 2009 (K. Benner, Keri.Benner@ dfo-mpo.gc.ca, and T. Cone, Tracy.Cone@dfo-mpo.gc.ca, pers.…”

Semiautomated Analysis of Data from an Imaging Sonar for Fish Counting, Sizing, and Tracking in a Post-Processing Application

“…Fish counting has been applied not only to salmon but also to jellyfish, steelhead trout, and small-sized fish (Holmes et al, 2006;Burwen et al, 2007;Han and Uye, 2009;Pipal et al, 2010). Salmon size in particular has been an application strongly demanded by scientists and anglers.…”