ICES Journal of Marine Science: Journal du Conseil Advance Access originally published online on May 9, 2009
ICES Journal of Marine Science: Journal du Conseil 2009 66(6):1238-1244; doi:10.1093/icesjms/fsp122
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
This article appears in the following ICES Journal of Marine Science issue: The Ecosystem Approach with Fisheries Acoustics and Complementary Technologies [View the issue table of contents]
Measurement and visual verification of fish target strength using an acoustic-optical system attached to a trawlnet
1 CSIRO Marine and Atmospheric Research, PO Box 1538, Hobart 7001, Australia
2 National Institute of Water and Atmospheric Research Ltd, PO Box 14-901, Kilbirnie, Wellington, New Zealand
Correspondence to T. E. Ryan: tel: +61 3 62325291; fax: +61 3 62325000; e-mail: tim.ryan{at}csiro.au.
Ryan, T. E., Kloser, R. J., and Macaulay, G. J. 2009. Measurement and visual verification of fish target strength using an acoustic-optical system attached to a trawlnet. – ICES Journal of Marine Science, 66: 1238–1244.It is difficult to make acoustic target-strength (TS) measurements of fish behaving naturally in deep-water habitats. The fish may avoid the acoustic instrumentation, and, if measured, there is uncertainty about their species and their orientation relative to the incident sound. To address these issues, a novel acoustic-optical system (AOS) has been developed, which combines a battery-powered, dual-frequency, split-beam acoustic system with a low-light video camera. The AOS attaches to the headline of a commercial deep-water demersal trawlnet that herds fish past the AOS and to the codend. This paper describes initial trials of the AOS to measure calibrated TS of New Zealand orange roughy, validated with video images. The fish species were visually identified, and their behaviour and orientation were approximated. The trawl catch provided associated samples for species identification and measurements of their length and other biological metrics. The combination of acoustics and optics in a net-mountable system constitutes a powerful sampling tool with broader applications in fishery research and ecosystem investigations.
Keywords: acoustic-optical system, acoustics, deep-water ecosystem, optics, orange roughy, target strength
Received 7 August 2008; accepted 14 December 2008; advance access publication 9 May 2009.
 |
Introduction |
|---|
 Top Introduction MethodsResultsDiscussionReferences |
|---|
The combination of acoustics and optics in underwater research is an emerging field, and both technologies are evolving rapidly. Each has its advantages and limitations. Acoustics provide non-invasive measurements over long ranges, but cannot identify the species directly from which the reflections originated (Horne, 2000). Conversely, optical methods may allow visual identification of species and quantification of their behaviour and orientation, but they have short observational ranges because of rapid propagation losses and decreases in resolution with distance. The complementary advantages of both methods can be exploited by making coincident measurements with the two techniques.
Accurate estimates of fish target strength (TS) are required to calculate biomass using echo-integration methods (Simmonds and MacLennan, 2005). TS measurements of in situ fish are generally preferred over those obtained from ex situ fish (Ehrenberg, 1983; Kloser et al., 1997). However, reliable TS measurements of in situ fish can be difficult to obtain because measurements can be biased by multiple targets; information on the species, and their size and orientation, is generally uncertain; and fish may not behave naturally as a consequence of the presence of the acoustic measuring instrument (Koslow et al., 1995). Consequently, the uncertainty in TS for the focus species of this paper, orange roughy (Hoplostethus atlanticus), remains a major component of uncertainty for acoustic-based biomass estimates. These issues are seemingly intractable with existing methods (e.g. vertically lowered acoustic platforms) and have led to novel approaches (Anon., 2005).
One of these was a battery-powered, dual-frequency, internal-logging, acoustic system, which had been developed to make TS measurements and investigate the species composition of deep-water ecosystems (Kloser et al., 2007). This concept was extended by designing a new platform that added a low-light video camera and a high-efficiency LED lighting system to create an acoustic-optical system (AOS). The additional optics allow the subjects of acoustic TS measurements to be verified visually (Jaffe et al., 1998; Warren et al., 2001). The AOS is attached to the headline of a commercial fishing-net system. This novel approach allows the fish in target regions to be sampled throughout the deployments.
The AOS project has four main objectives: (i) restrict fish avoidance by herding fish past the AOS; (ii) collect TS measurements of individual fish at two frequencies; (iii) simultaneously record information on species and estimate their orientation and behaviour; and (iv) retain samples of the fish.
Clearly, the AOS allows TS measurements to be made on fish that have been disturbed by the act of fishing and that are not technically in situ or in their ideal natural state. Nevertheless, the possibility of identifying the measured species concurrently and of quantifying their behaviour constitutes a major advance in fisheries acoustic research (Anon., 2005). This paper describes the AOS, and its application to New Zealand’s orange roughy in a deep-water environment, as an example of its use.
|
Methods |
|---|
|
TopIntroductionMethodsResultsDiscussionReferences |
|---|
Sampling area
Experiments with the AOS were conducted at the Chatham Rise (42°50'S 177°12'W; 600–1000 m depth) in July 2007, where New Zealand’s largest orange-roughy population aggregates to spawn each winter. The sampling platform was RV "Tangaroa", a 70-m research vessel with deep-water trawling capability, owned and operated by New Zealand’s National Institute of Water and Atmospheric Research (NIWA).
Description of AOS and net system
The AOS is a self-contained system (i.e. battery powered with data logged locally and no data transferred to the surface (Figure 1 and Table 1), with 38 and 120 kHz, split-beam transceivers and transducers. Two 38-kHz transducers [Simrad ES38DD (7° beam width) and EDO 6989 (14° beam width), with pulse duration of 256 and 512 µs, respectively], were tested for optimal system performance. At each frequency, measurements of compensated TS and target position were estimated using a split-beam algorithm (Myriax, 2007).
|
|
The video system includes a PAL digital video camera, video-capture card, and two high-efficiency LED lights. For synchronization with the acoustic data, date, time, and platform depth were overlaid on the video images. Two red scaling lasers (400 mm separation) allowed the capture of size metrics for fish and the seabed in some instances.
The AOS was attached to the headline of a NIWA, rough-bottom, orange-roughy trawl, which was deployed via the stern ramp in its usual manner (Figure 1). The forepart of the net was 300 mm mesh, with a robust groundgear of steel and rubber bobbins. The codend was 100 mm mesh. To minimize interfering reflections from the surrounding net structure, 80 mm mesh was attached to the top panel, and the AOS mounting frame was attached to the centre of the headline. The sled-style AOS was bolted to the mounting frame and ten 12-inch trawlnet floats were attached to make the system neutrally buoyant.
Door spread and headline height were measured (SCANMAR 4000 and Furuno CN24/CN22 system, respectively) and displayed on the bridge for close monitoring of trawl sets. The wingspread was 26 m, and the headline height was 4–6 m when the footrope touched the seabed.
The acoustic and video data, and measurements of platform motion (collected at 10 Hz), were time-referenced with the system clock. This synchronization allows the effects of platform motion on TS measurements to be explored. For example, if a fish has a narrow scattering directivity, the angle between it and the AOS will significantly modulate the measured TS.
Acoustic calibration
To calibrate the on-axis sensitivities of the acoustic systems, the AOS was removed from the net system, attached to a trawl warp, and lowered to 800 m, with a 38.1-mm diameter, tungsten-carbide sphere suspended 13 m below the transducers (Foote et al., 1987).
Concurrent video and acoustic measurements and analysis
Acoustic and video measurements were made simultaneously as fish passed beneath the AOS (Figure 2). When defining shapes were resolved, fish species were identified from the video images. The single video camera did not provide direct estimates of fish orientation. Instead, fish-orientation angles were coarsely estimated from their appearance and behaviour in the video record. Although these estimates are insufficient to characterize TS accurately vs. incidence angle, they describe whether the fish were orientated horizontally or at a steep angle, and this is of great value. Behaviours were classified as vigorously swimming downwards (
30–60° relative to horizontal) or vigorously swimming or drifting approximately horizontally. Angles between the fish and the AOS were estimated from the alignment of fish images relative to a compass rose, with 10° resolution, overlaid on the video images.
|
Acoustic data were processed with Echoview V4.30 (Myriax), resulting in single-target detections, including a time reference, the three-dimensional position of the target, the range, angular positions in two orthogonal planes, and the TS. In cases where the target species were identified visually, the temporally coincident TS measurements at 38 and 120 kHz were exported for further analysis. TS data without visual confirmation of species were also exported for fish tracked using the "alpha–beta tracking algorithm" (Myriax, 2007) and regions with large concentrations of orange roughy in the video record.
Biological sampling
Biological samples collected during each AOS trawl were sorted by species and analysed for length, weight, sex, and gonad stage. Additionally, to capture small fish, a fine-mesh, multiple-opening pelagic net (MIDOC) was deployed (Kloser et al., 2002).
|
Results |
|---|
|
TopIntroductionMethodsResultsDiscussionReferences |
|---|
Field operations
During the 14-day voyage, the AOS was deployed on 22 trawls and the system was fully operational for 17 trawls. Protected by the sled-style design, and its position on the headline, the AOS was not damaged. Seven tows were made close to the seabed, but these caught insufficient fish, or the observed fish were generally too close to the seabed to be resolved acoustically.
Six trawls were made 150 m above the seabed at two seamounts (Cameron’s 43°08'S 174°17'W and Smith’s 42°57'S 174°25'W), to characterize the species composition using net catch, measure TS and volume-backscatter strength (Sv) of fish at two frequencies, and concurrently video-image the targets. Four trawls towed well above the bottom (headline 
20 m above bottom; footrope not touching the seabed) targeted a large plume of spawning orange roughy, and this proved to be the most successful strategy for obtaining TS measurements. Care was required, however, to avoid large and potentially net-damaging catches [catches of 50 t are common from 1- to 2-min tows on orange-roughy spawning plumes (Clark, 1995)]. For this reason, the codend was closed for only one of these tows.
The AOS was stable once the net was at depth and the headline fully tensioned. Its stability is attributed to the high tensioning forces from the heavy trawl doors and water flow through the net system. The average pitch and roll of the AOS was generally within ±2° of horizontal for each deployment. A typical peak-to-peak variation was 3° at 0.2 Hz for pitch and <1° at 1 Hz for roll.
Acoustic calibration
The AOS echosounders were calibrated with the 38-kHz EDO and Simrad ES38DD and the 120-kHz Simrad ES120–7 D transducers on one, two, and three occasions respectively (Figure 3). Variations in on-axis sensitivities over multiple deployments were estimated from the minima and maxima of profiles from previous voyages (Figure 3). At a nominal working depth of 800 m, the ranges between minima and maxima were ±0.25 and ±0.7 dB for the Simrad ES38DD and ES120–7D transducers, respectively. Midpoint profiles were used to estimate gains for the system configured with each transducer and operating at 800 m (Table 2). This was the first deployment with the wide-beam configuration of the EDO transducer; therefore, no previous calibration results were available for comparison.
|
|
Optimal settings for the AOS acoustic system were established during initial deployments. When configured with the 38-kHz EDO transducer and transmitting 256 µs pulses, the Sv echogram at 20–30 m range with the display minimum set to –72 dB included noise generated by contact of the net system with the seabed. However, using a pulse duration of 512 µs, the noise spikes were suppressed. Similarly, noise was reduced using the Simrad ES38DD and ES120–7D transducers with pulse durations of 512 and 256 µs, respectively. The results of the acoustic-calibration experiments were applied to the TS data during post-processing.
Optical performance
Meeting a principal objective, the AOS allowed visual identification of orange roughy coincident with the TS measurements (Table 3). In regions of high orange-roughy densities, 30–80 fish could be identified per video frame, and thousands were observed during the trawl. Other large fish species less frequently observed included Johnston’s cod (Halargyreus johnsonii), basketwork eels (Diastobranchus capensis), and deep-water sharks, and their respective TS measurements were obtained in some instances.
|
Visual detection and identification of fish species is probably dependent on their colour. Light-coloured fish, such as orange roughy and Johnston’s cod, were observed routinely, whereas probably present, dark-coloured fish, such as large myctophids and oreos, were not identified in the video. Light backscattered because the turbidity was negligible at 800 m depth and did not limit the optical performance of the AOS.
The MIDOC catches included diverse small species, including myctophids, deep-water prawns, and unidentified gelatinous animals. Visual identification of such small and/or fast-moving biota was usually impossible because the images were often blurred or nebulous. Furthermore, there were many instances where biota with reasonably high TS (e.g. –50 dB at 38 kHz, –62 dB at 120 kHz) were acoustically tracked within 4–8 m of the video system, but could not be detected on the corresponding video images. These animals were presumably small myctophids with gas bladders (Kloser et al., 2002).
Visually verified TS measurements
Thousands of video observations of orange roughy and two-frequency TS measurements were made as the AOS was trawled through the spawning plume. A fish-tracking algorithm was applied to these TS measurements, reducing the number of accepted targets to those contained within the hundreds of tracked individuals. Of those tracked fish, only 47 could be unequivocally identified in the video as orange roughy and confidently paired with corresponding TS measurements. This is a very small subset of the 20 000–50 000 fish estimated, based on trawl catches and the number of acoustically tracked fish, to have passed through the net during the 17 AOS deployments. The TS measurements of visually verified individual orange roughy ranged from –65 to –45 dB. An estimation of the mean TS from these data should consider the effects of the transducer nearfield, single-target detection criteria, and fish length and orientation. This is beyond the scope of this paper.
The small overlapping sampling areas of the acoustic and optical systems greatly limited the number of visually validated TS measurements compared with the number of fish observed on either the acoustic or optical system or caught in the net (Figure 4). The effective cross-sectional sampling areas for the net, video (59° field of view lens), acoustics (7° beam width transducers), and concurrent acoustic/video were estimated as 156, 81, 14, and 3 m2, respectively. High fish densities often precluded acoustic detections of individual targets. Moreover, when high densities of fish were observed in the video, it was not always possible to link TS measurements accurately with individual fish. Consequently, in one deployment, 5200 orange roughy were caught, but visually validated TS measurements were made of only six fish.
|
Observations of fish orientation and behaviour
From the visual inspection of fish forms in the video, the orientation of orange roughy was classified as: (i) approximately horizontal and either motionless or swimming slowly forwards or (ii) swimming vigorously downwards at an angle between
30 and 60°, the latter a less common avoidance behaviour. When diving, the detected tracks revealed increased depth as the fish moved through the acoustic beam. |
Discussion |
|---|
|
TopIntroductionMethodsResultsDiscussionReferences |
|---|
Field operations
These AOS experiments represent three significant advances. For the first time, TS measurements at two frequencies were made for a number of deep-water species that can be confidently linked to species via visual validation. Second, by herding the fish past the measuring system, the formerly intractable problem of avoidance was overcome. Finally, it was demonstrated that a large and heavy instrument could be attached and deployed with high stability on a net, without adversely affecting its trawl characteristics. This approach provides many new opportunities for sampling of the oceans.
Acoustic calibration
Transducer performance was expected to be sensitive to pressure and temperature, and possibly to exhibit hysteresis (Kloser, 1996; Dalen et al., 2003; Demer and Renfree, 2008). The usual method of calibrating the systems by suspending a sphere beneath the transducers throughout the measurements of fish TS (Kloser et al., 2002) was not possible when the AOS was attached to a moving trawl. Instead, results from multiple calibration experiments were compared, summarized, and applied to the AOS data collected during trawls.
Optical performance
The AOS provided visually validated TS measurements of light-coloured, 30–40 cm fish at ranges of 4–8 m, fulfilling the primary objective of this study. However, smaller, darker, or more distant fish, of all sizes and colours, were not detected, so limiting the potential of the system to investigate a broader range of species.
Optimally, such TS measurements should be accompanied by accurate visual measurements of fish orientation and length. However, in this investigation, only pitch angles were estimated from the video images, to approximately ±10° when near-horizontal and ±20° when diving. Nonetheless, these coarse measurements allowed observations of two distinct behaviours, and this capability constitutes a significant advance in TS measurements.
The pixel resolution of the video system (14 mm at 6 m range) must be improved by at least an order of magnitude to identify micronekton, for example, myctophids and deep-water prawns. A planned addition of stereo, digital, still cameras with strobe lights, triggered by TS detections, should greatly improve the optical resolution. Stereo-photogrammetric techniques should allow measurements of target length and orientation. Simulations using a pair of 10 megapixel, digital, SLR cameras, with 50 mm lenses and a separation of 1.1 m, suggest a best-case, point-measurement precision of 0.5 mm horizontally and vertically and 6 mm in range at a distance of 7.0 m from the cameras. The precision of such length measurements could be better than 2 mm at this range. (J. Seager, 2008, pers. comm.).
Visually verified TS measurements
The large mismatch in sampling areas between the net and the concurrent acoustic and video measurements can result in a paucity of visually verified TS measurements, if fish densities are low, or an unmanageably large catch, if densities are high. At the densities of orange roughy encountered in the spawning plume, more single targets were detected with the 7° beam width transducer than with the 14° beam width transducer, because the latter was more likely to have multiple targets within its larger sampling volume and thus fail the criteria for single-target detection. Therefore, to optimize the number of visually verified TS measurements, the acoustic and optical sampling volumes must be optimized for the expected fish densities (Jaffe et al., 1998).
The non-coincident video and acoustic measurements also contain valuable information. For example, video images of fish within and outside the acoustic beam can support the interpretation of the larger set of non-concurrent TS measurements. This approach can provide a greater number of TS measurements to improve statistical robustness and evaluate differences between various datasets (e.g. visually verified vs. unverified TS measurements).
An estimate of the mean TS for orange roughy or co-occurring species is beyond the scope of this paper. A future goal is to use the AOS to build a reference set of visually validated, fish-TS measurements by species. This will facilitate explorations of a larger AOS dataset and historic measurement- and model-based estimates of TS.
Further applications of the AOS
The experiments also demonstrated an application of the AOS to identify the species composition in a deep-water community over Cameron’s seamount. In the past, trawls near the seabed yielded large catches of orange roughy there. Acoustic surveys of the area also mapped strong and extensive backscatter well above the seamount. Consequently, there was contention among scientists and fishers whether the pelagic scatterers were mainly orange roughy or another species.
An analysis of the dual-frequency Sv data, using a multifrequency technique (Kloser et al., 2002), indicated that a small number of orange roughy was present close to the seabed in our study. An analysis of the two-frequency TS indicated that gas-bladdered fish of various sizes were located above the orange roughy. Video images of the fish in the water column revealed Johnston’s cod, a large gas-bladdered species with a large TS (McClatchie and Coombs, 2005), which comprised 80% of the net catch, by weight. Apparently, the scatter in the water column above the seamount was not caused by orange roughy.
The methods and results in this paper are specific to the measurements made near New Zealand on the commercially important orange roughy and their co-occurring species. More generically, however, this study highlights the versatility of the AOS for measuring and validating TS of multiple fish species and for investigating deep-water ecosystems. Further improvement should allow quantification of smaller fish species and thereby permit a more comprehensive characterization of the species present.
|
Acknowledgements |
|---|
We thank the CSIRO Wealth from Oceans Flagship and the New Zealand Ministry of Fisheries, and its Deepwater Working Group, for their support and input into this project. We also thank the officers and crew of RV "Tangaroa" and Matt Sherlock, Matthew Horsham, and the CSIRO Marine Technology and Engineering Group for their efforts. The experimental use of the AOS, and subsequent analyses, was funded by Ministry of Fisheries Project ORH200601. We thank Jess Tyler for her editorial assistance. Jock Young and Sally Wayte of the CSIRO and three anonymous referees are thanked for their helpful reviews.
|
References |
|---|
|
TopIntroductionMethodsResultsDiscussionReferences |
|---|
-
Anon. Workshop to improve biomass estimates of orange roughy. (2005) Ministry of Fisheries Deepwater Working Group report DWORHAC005/14 rev1. 8.
Clark M. R. Experience with management of orange roughy (Hoplostethus atlanticus) in New Zealand waters, and the effects of commercial fishing on stocks over the period 1980–1993. In: Deepwater Fisheries of the North Atlantic Oceanic Slope, pp. 251–266—Hopper A. G., ed. (1995) Kluwer Academic, Dordrecht, The Netherlands.
Dalen J., Nedreaas K., Pedersen R. A comparative acoustic-abundance estimation of pelagic redfish (Sebastes mentella) from hull-mounted and deep-towed acoustic systems. ICES Journal of Marine Science (2003) 60:472–9.
Demer D. A., Renfree J. S. Variations in echosounder-transducer performance with water temperature. ICES Journal of Marine Science (2008) 65:1021–1035.
Ehrenberg J. E. A review of in situ target strength estimation techniques. (1983) Symposium on Fisheries Acoustics, Bergen, Norway, 21 June 1982.
Foote K. G., Knudsen H. P., Vestnes G., MacLennan D. N., Simmonds E. J. Calibration of acoustic instruments for fish density estimation: a practical guide. ICES Cooperative Research Report (1987) 144:69.
Horne J. K. Acoustic approaches to remote species identification: a review. Fisheries Oceanography (2000) 9:356–71.[CrossRef][Web of Science]
Jaffe J. S., Ohman M. D., De Robertis A. OASIS in the sea: measurement of the acoustic reflectivity of zooplankton with concurrent optical imaging. Deep Sea Research II (1998) 45:1239–53.[CrossRef]
Kloser R. J. Improved precision of acoustic surveys of benthopelagic fish by means of a deep-towed transducer. ICES Journal of Marine Science (1996) 53:407–13.
Kloser R. J., Ryan T. E., Macaulay G. J., Lewis M. E. Acoustic measurements of Cascade Plateau orange roughy. Final report to Australian Fisheries Management Authority. (2007) Castray Esplanade, Hobart, Australia: Copy held at CSIRO Marine and Atmospheric Research Laboratories.
Kloser R. J., Ryan T., Sakov P., Williams A., Koslow J. A. Species identification in deep water using multiple acoustic frequencies. Canadian Journal of Fisheries and Aquatic Sciences (2002) 59:1065–77.
Kloser R. J., Williams A., Koslow J. A. Problems with acoustic target strength measurements of a deepwater fish, orange roughy (Hoplostethus atlanticus, Collett). ICES Journal of Marine Science (1997) 54:60–71.
Koslow J. A., Kloser R., Stanley C. A. Avoidance of a camera system by a deepwater fish, the orange roughy (Hoplostethus atlanticus). Deep Sea Research I: Oceanographic Research Papers (1995) 42:233–44.[CrossRef]
McClatchie S., Coombs R. F. Low target strength fish in mixed species assemblages: the case of orange roughy. Fisheries Research (2005) 72:185–92.[CrossRef][Web of Science]
Myriax. Echoview 4.30 Help file. (2007) Hobart, Tasmania, Australia: Myriax Pty Ltd.
Simmonds E. J., MacLennan D. N. Fisheries Acoustics: Theory and Practice, 2nd edn. (2005) Oxford: Blackwell Science. 437.
Warren J. D., Stanton T. K., Benfield M. C., Wiebe P. H., Chu D., Sutor M. In situ measurements of acoustic target strengths of gas-bearing siphonophores. ICES Journal of Marine Science (2001) 58:740–9.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  R. J. Kloser, T. E. Ryan, J. W. Young, and M. E. Lewis Acoustic observations of micronekton fish on the scale of an ocean basin: potential and challenges ICES J. Mar. Sci., July 1, 2009; 66(6): 998 - 1006. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||