*Clupea harengus*) in the Gulf of Maine during the Ocean Acoustic Waveguide Remote Sensing 2006 Experiment

^{1}, Mark Andrews

^{1}, Srinivasan Jagannathan

^{2}, Ruben Patel

^{3}, J. Michael Jech

^{4}, Nicholas C. Makris

^{5}and Purnima Ratilal

^{6}

### Abstract

The low-frequency target strength of shoaling Atlantic herring (*Clupea harengus*) in the Gulf of Maine during Autumn 2006 spawning season is estimated from experimental data acquired simultaneously at multiple frequencies in the range using (1) a low-frequency ocean acoustic waveguide remote sensing (OAWRS) system, (2) areal population density calibration with several conventional fish finding sonar (CFFS) systems, and (3) low-frequency transmission loss measurements. The OAWRS system’s instantaneous imaging diameter of and regular updating enabled unaliased monitoring of fish populations over ecosystem scales including shoals of Atlantic herring containing hundreds of millions of individuals, as confirmed by concurrent trawl and CFFS sampling. High spatial-temporal coregistration was found between herring shoals imaged by OAWRS and concurrent CFFS line-transects, which also provided fish depth distributions. The mean scattering cross-section of an individual shoaling herring is found to consistently exhibit a strong, roughly /octave roll-off with decreasing frequency in the range of the OAWRS survey over all days of the roughly experiment, consistent with the steep roll-offs expected for sub-resonance scattering from fish with air-filled swimbladders.

This research was supported by the National Oceanographic Partnership Program, the Office of Naval Research, the Alfred P. Sloan Foundation, and the National Oceanic and Atmospheric Administration and is a contribution to the Census of Marine Life. We would like to thank the science parties, officers, and crew of the research vessels *Oceanus, Endeavor, Delaware II*, and *Hugh Sharp*. We would also like to thank Olav Rune Godø and Redwood W. Nero for the many interesting discussions on the topic, and the research assistants who helped in the data analysis.

I. INTRODUCTION

II. MULTI-SENSOR EXPERIMENT DESIGN AND RESOURCES

III. DATA PROCESSING AND ANALYSIS

A. Generating instantaneous wide-area OAWRS images of the ocean environment

B. Estimating areal fish population density from instantaneous OAWRS imagery

C. Estimates of areal fish population density from CFFS

D. Estimating low-frequency target strength by matching OAWRS and CFFS population densities

E. Frequency dependence of target strength estimated by differencing OAWRS scattering strength images over wide areas

IV. RESULTS AND DISCUSSION

A. Measured abundance

B. Measured low-frequency target strength

C. Using measured low-frequency target strength to infer swimbladder properties

D. Space, time, and frequency dependencies

V. CONCLUSION

### Key Topics

- Scattering measurements
- 17.0
- Aggregation
- 10.0
- Marine vessels
- 10.0
- Transmission measurement
- 10.0
- Acoustical measurements
- 9.0

## Figures

Location of OAWRS 2006 experiment on the northern flank of Georges Bank in the Gulf of Maine. Plus indicates location of moored OAWRS source array deployed on Oct 1–3 at 42.2089N, 67.6892W, the coordinate origin for all OAWRS images in this paper. Circle shows typical area imaged by OAWRS, diameter and wider than Cape Cod, in . Geographic locations of trawls deployed by NOAA FRV *Delaware II* are overlain. Dots indicate trawls where herring were predominant species. In contrast, diamond indicates a trawl where silver hake and squids dominated. The gray dashed box bounds the area of OAWRS imaging during the OAWRS 2006 experiment.

Location of OAWRS 2006 experiment on the northern flank of Georges Bank in the Gulf of Maine. Plus indicates location of moored OAWRS source array deployed on Oct 1–3 at 42.2089N, 67.6892W, the coordinate origin for all OAWRS images in this paper. Circle shows typical area imaged by OAWRS, diameter and wider than Cape Cod, in . Geographic locations of trawls deployed by NOAA FRV *Delaware II* are overlain. Dots indicate trawls where herring were predominant species. In contrast, diamond indicates a trawl where silver hake and squids dominated. The gray dashed box bounds the area of OAWRS imaging during the OAWRS 2006 experiment.

[(A)–(C)] OAWRS images of areal fish density zoomed-in around massive herring shoals, with densities exceeding in population centers. Measured during evening to midnight hours of October 4, 2, and 1, respectively. (A) The total population of herring in the large dense shoal is roughly , and that in the diffuse cloud outside the large shoal is roughly . Imaged shoal populations of herring are approximately and respectively for (B) and (C). Uncertainty in the abundance estimate is 17–20%. Note that the figures are plotted on different scales, and the coordinate origin is the source location shown in Fig. 1.

[(A)–(C)] OAWRS images of areal fish density zoomed-in around massive herring shoals, with densities exceeding in population centers. Measured during evening to midnight hours of October 4, 2, and 1, respectively. (A) The total population of herring in the large dense shoal is roughly , and that in the diffuse cloud outside the large shoal is roughly . Imaged shoal populations of herring are approximately and respectively for (B) and (C). Uncertainty in the abundance estimate is 17–20%. Note that the figures are plotted on different scales, and the coordinate origin is the source location shown in Fig. 1.

Profiles of water-column sound speed from XBT and CTD measurements made from all four research vessels on the Northern Flank of Georges Bank and Georges Basin during OAWRS 2006.

Profiles of water-column sound speed from XBT and CTD measurements made from all four research vessels on the Northern Flank of Georges Bank and Georges Basin during OAWRS 2006.

Herring target strength at 950 and estimated by matching areal fish density in OAWRS and CFFS data acquired during midnight hours of October 2. [(A)–(C)] A sequence of instantaneous OAWRS scattering strength images zoomed into the region containing a massive herring shoal with overlain CFFS line-transect (solid line) made at nominal tow-speed of . (D) CFFS time-depth echogram provides local depth distributions of fish aggregations. Dashed lines at 23:30 EDT and 01:00 EDT correspond to transect start and end points and , respectively. (E) The areal fish population densities inferred from CFFS measurements following Eq. (6) are plotted as a function of time in black, and the corresponding areal fish population densities in dB, , are plotted in gray. (F) The OAWRS scattering strength measurements and (G) instantaneous target strength estimates along CFFS line-transects at 950 and . Target strength estimates near the edge of shoals are not accurate because of nonstationarity. (H) Population of herring within the area shown in (A)–(C) determined with various OAWRS fish density thresholds. Solid line gives population above the threshold and dotted line gives population below the threshold.

Herring target strength at 950 and estimated by matching areal fish density in OAWRS and CFFS data acquired during midnight hours of October 2. [(A)–(C)] A sequence of instantaneous OAWRS scattering strength images zoomed into the region containing a massive herring shoal with overlain CFFS line-transect (solid line) made at nominal tow-speed of . (D) CFFS time-depth echogram provides local depth distributions of fish aggregations. Dashed lines at 23:30 EDT and 01:00 EDT correspond to transect start and end points and , respectively. (E) The areal fish population densities inferred from CFFS measurements following Eq. (6) are plotted as a function of time in black, and the corresponding areal fish population densities in dB, , are plotted in gray. (F) The OAWRS scattering strength measurements and (G) instantaneous target strength estimates along CFFS line-transects at 950 and . Target strength estimates near the edge of shoals are not accurate because of nonstationarity. (H) Population of herring within the area shown in (A)–(C) determined with various OAWRS fish density thresholds. Solid line gives population above the threshold and dotted line gives population below the threshold.

The intensity of scattered returns from shoals is highly frequency-dependent. The histograms illustrate that it is easier to detect shoals over background regions at higher frequencies. Simultaneous trawls show shoals are overwhelmingly comprised of herring while background regions yield negligible herring (Table IV, Fig. 12). [(A), (C), and (E)] OAWRS images of herring shoal acquired simultaneously at three distinct frequency bands centered at 415, 950, and at 00:41:15 EDT on October 2. The colorscale used in (A), (C), and (E) is the same as in Figs. 4(A)–4(C). [(B), (D), and (F)] Histograms of scattering strength values at locations within the shoal (areas inside the dashed box) and in a background region (areas inside the solid box) plotted for comparison. The data are ambient noise limited in background areas due to weak source level and is not shown.

The intensity of scattered returns from shoals is highly frequency-dependent. The histograms illustrate that it is easier to detect shoals over background regions at higher frequencies. Simultaneous trawls show shoals are overwhelmingly comprised of herring while background regions yield negligible herring (Table IV, Fig. 12). [(A), (C), and (E)] OAWRS images of herring shoal acquired simultaneously at three distinct frequency bands centered at 415, 950, and at 00:41:15 EDT on October 2. The colorscale used in (A), (C), and (E) is the same as in Figs. 4(A)–4(C). [(B), (D), and (F)] Histograms of scattering strength values at locations within the shoal (areas inside the dashed box) and in a background region (areas inside the solid box) plotted for comparison. The data are ambient noise limited in background areas due to weak source level and is not shown.

(A) Schematic of scattering strength levels at two distinct frequencies and as a function of local fish density and (B) their difference at a given OAWRS pixel. The scattering strength difference equals the mean background level difference for low fish densities, while at high fish densities, the scattering strength differences equals the fish target strength difference.

(A) Schematic of scattering strength levels at two distinct frequencies and as a function of local fish density and (B) their difference at a given OAWRS pixel. The scattering strength difference equals the mean background level difference for low fish densities, while at high fish densities, the scattering strength differences equals the fish target strength difference.

OAWRS scattering strength level differences for the indicated frequency pairs as a function of areal fish density in dB for data acquired between 22:00 and 22:45 EDT on October 3. The scattering strength difference at high fish densities equals the target strength difference for the frequency pair shown.

OAWRS scattering strength level differences for the indicated frequency pairs as a function of areal fish density in dB for data acquired between 22:00 and 22:45 EDT on October 3. The scattering strength difference at high fish densities equals the target strength difference for the frequency pair shown.

Fork length distributions of most frequently caught species, Atlantic herring, Acadian redfish, Haddock, and Silver hake, from trawls deployed on Georges Bank (Fig. 1). The mean fork length of herring is with a standard deviation 6.8% of the mean. The equation (Ref. 7) is used to convert herring’s fork length to the total length, where and are in cm. The mean fork length of redfish is with a standard deviation 15% of the mean. The equation (Ref. 72) is used to convert redfish’s fork length to total length. Silver hake’s fork length is the total length.

Fork length distributions of most frequently caught species, Atlantic herring, Acadian redfish, Haddock, and Silver hake, from trawls deployed on Georges Bank (Fig. 1). The mean fork length of herring is with a standard deviation 6.8% of the mean. The equation (Ref. 7) is used to convert herring’s fork length to the total length, where and are in cm. The mean fork length of redfish is with a standard deviation 15% of the mean. The equation (Ref. 72) is used to convert redfish’s fork length to total length. Silver hake’s fork length is the total length.

Atlantic herring length-weight regression calibration. The dots are the length-weight data obtained from the trawl-survey conducted by U.S. National Marine Fisheries Service in conjunction with OAWRS 2006 experiment, and the gray solid line indicates the derived best-fit length-weight regression, which can be expressed as , where is the weight of herring in kg, is the fork length of herring in cm, and and are empirical regression parameters. For this trawl dataset, and .

Atlantic herring length-weight regression calibration. The dots are the length-weight data obtained from the trawl-survey conducted by U.S. National Marine Fisheries Service in conjunction with OAWRS 2006 experiment, and the gray solid line indicates the derived best-fit length-weight regression, which can be expressed as , where is the weight of herring in kg, is the fork length of herring in cm, and and are empirical regression parameters. For this trawl dataset, and .

Experimentally determined low-frequency target strength corresponding to the average scattering cross-sections of shoaling herring observed from OAWRS imagery acquired from five shoals on (Table VI) at 415, 735, 950, and 1125 (circles) with standard deviations (error bars). Comparison with Love-model mean target strength for shoaling herring, with physical parameters tabulated in Table III, of different swimbladder semi-minor axes over the shoals’ depth distributions (lines). The best least-squares fits shown are obtained only using target strength estimates of each shoal (Table IV) with standard deviations less than . Arrows indicate the target strength uncertainties due to potential masking from background scattering (Sec. III E) for given frequencies. The best-fit means and standard deviations of inferred swimbladder volume, swimbladder semi-minor axes, and corresponding neutral buoyancy depths of each shoal were tabulated in Table VI. (A) Shoaling herring, distributed between 120 and (Fig. 4), imaged with the OAWRS system from 23:30 EDT October 2 to 01:00 EDT on October 3. (B) Shoaling herring, distributed between 135 and , imaged with the OAWRS system from 18:55 to 19:50 EDT on October 3. (C) Shoaling herring, distributed between 120 and , imaged with the OAWRS system from 22:00 to 22:45 EDT on October 3. (D) Shoaling herring, distributed between 120 and , imaged with the OAWRS system from 06:10 to 06:50 EDT on September 27 (Fig. 13). (E) Shoaling herring, distributed between 150 and , imaged with the OAWRS system from 07:25 to 07:50 EDT on September 29.

Experimentally determined low-frequency target strength corresponding to the average scattering cross-sections of shoaling herring observed from OAWRS imagery acquired from five shoals on (Table VI) at 415, 735, 950, and 1125 (circles) with standard deviations (error bars). Comparison with Love-model mean target strength for shoaling herring, with physical parameters tabulated in Table III, of different swimbladder semi-minor axes over the shoals’ depth distributions (lines). The best least-squares fits shown are obtained only using target strength estimates of each shoal (Table IV) with standard deviations less than . Arrows indicate the target strength uncertainties due to potential masking from background scattering (Sec. III E) for given frequencies. The best-fit means and standard deviations of inferred swimbladder volume, swimbladder semi-minor axes, and corresponding neutral buoyancy depths of each shoal were tabulated in Table VI. (A) Shoaling herring, distributed between 120 and (Fig. 4), imaged with the OAWRS system from 23:30 EDT October 2 to 01:00 EDT on October 3. (B) Shoaling herring, distributed between 135 and , imaged with the OAWRS system from 18:55 to 19:50 EDT on October 3. (C) Shoaling herring, distributed between 120 and , imaged with the OAWRS system from 22:00 to 22:45 EDT on October 3. (D) Shoaling herring, distributed between 120 and , imaged with the OAWRS system from 06:10 to 06:50 EDT on September 27 (Fig. 13). (E) Shoaling herring, distributed between 150 and , imaged with the OAWRS system from 07:25 to 07:50 EDT on September 29.

(A) An example of Love-model target strength corresponding to the average scattering cross-section of an individual for mixed species content and swimbladder semi-minor axes shown. Comparison of Love-model target strength with experimentally determined mean target strength estimates of shoaling herring, distributed between 135 and , imaged with the OAWRS system from 18:55 to 19:50 EDT on October 3. Presence of redfish in trawl-determined percentage (dashed gray line) has negligible effect on best-fit target strength compared to herring alone (solid black line), while including an unrealistic percentage of redfish yields far worse fits (dashed black line). The silver hake, found at shallower water depth in trawls, have resonance peak above (dash-dot curve) making their contribution negligible. Herring target strength with a resonance frequency at (dashed gray line) based on Love’s model using length and depth distribution obtained from CFFS and trawl measurements is found to be neutrally buoyant at , and is lower than those measured by the OAWRS system. (B) Same as (A) but for shallower herring neutral buoyancy depths. Arrows indicate potential target strength uncertainties for given frequencies.

(A) An example of Love-model target strength corresponding to the average scattering cross-section of an individual for mixed species content and swimbladder semi-minor axes shown. Comparison of Love-model target strength with experimentally determined mean target strength estimates of shoaling herring, distributed between 135 and , imaged with the OAWRS system from 18:55 to 19:50 EDT on October 3. Presence of redfish in trawl-determined percentage (dashed gray line) has negligible effect on best-fit target strength compared to herring alone (solid black line), while including an unrealistic percentage of redfish yields far worse fits (dashed black line). The silver hake, found at shallower water depth in trawls, have resonance peak above (dash-dot curve) making their contribution negligible. Herring target strength with a resonance frequency at (dashed gray line) based on Love’s model using length and depth distribution obtained from CFFS and trawl measurements is found to be neutrally buoyant at , and is lower than those measured by the OAWRS system. (B) Same as (A) but for shallower herring neutral buoyancy depths. Arrows indicate potential target strength uncertainties for given frequencies.

[(A)–(C)] Locations of trawls over simultaneous OAWRS images. Trawls 137 and 139 were made directly through shoals as shown in (B) and (C). In contrast, trawl 134 was made in a region with no shoal (A), but one shoal would form in the vicinity later. The OAWRS source locations are the coordinate origin in all OAWRS images. On October 3, the OAWRS source ship was moored at 42.2089N, 67.6892W. The trawls in (A)–(C) were deployed and towed along the solid lines starting at and ending at . The dashed lines indicate the contours of 100, 150, 180, and water depth.

[(A)–(C)] Locations of trawls over simultaneous OAWRS images. Trawls 137 and 139 were made directly through shoals as shown in (B) and (C). In contrast, trawl 134 was made in a region with no shoal (A), but one shoal would form in the vicinity later. The OAWRS source locations are the coordinate origin in all OAWRS images. On October 3, the OAWRS source ship was moored at 42.2089N, 67.6892W. The trawls in (A)–(C) were deployed and towed along the solid lines starting at and ending at . The dashed lines indicate the contours of 100, 150, 180, and water depth.

Herring target strength at estimated by matching the areal density of OAWRS and CFFS data acquired during the early morning hours of September 27. Similar to Fig. 4 but for a contiguous shoal segment starting at 06:10 EDT on September 27.

Herring target strength at estimated by matching the areal density of OAWRS and CFFS data acquired during the early morning hours of September 27. Similar to Fig. 4 but for a contiguous shoal segment starting at 06:10 EDT on September 27.

(A) Experimentally determined mean and standard deviation of 677 measured instantaneous broadband one-way TL data after matched filter with center frequency. Plotted as a function of range, with modeled one-way TL overlain. Transmission data acquired by a single desensitized hydrophone on OAWRS receiver array on October 1–3 2006. Modeled TL computed by Monte-Carlo simulation with parabolic equation inputing measured oceanography and bathymetry of Georges Bank environment. Modeled TL (dashed gray line) in a perfectly reflective waveguide with no water-column attenuation by normal-mode model. (B) After six-sample averaging, the rms error is reduced to , consistent with theory for stationary averaging of intensities of circular complex Gaussian random fields (Refs. 16 and 26).

(A) Experimentally determined mean and standard deviation of 677 measured instantaneous broadband one-way TL data after matched filter with center frequency. Plotted as a function of range, with modeled one-way TL overlain. Transmission data acquired by a single desensitized hydrophone on OAWRS receiver array on October 1–3 2006. Modeled TL computed by Monte-Carlo simulation with parabolic equation inputing measured oceanography and bathymetry of Georges Bank environment. Modeled TL (dashed gray line) in a perfectly reflective waveguide with no water-column attenuation by normal-mode model. (B) After six-sample averaging, the rms error is reduced to , consistent with theory for stationary averaging of intensities of circular complex Gaussian random fields (Refs. 16 and 26).

## Tables

OAWRS receiving array angular resolution at broadside and endfire , and aperture length as a function of imaging frequency . A Hanning spatial window is applied in the beamforming.

OAWRS receiving array angular resolution at broadside and endfire , and aperture length as a function of imaging frequency . A Hanning spatial window is applied in the beamforming.

Conventional fish finding sonars, SIMRAD EK60 and EK500 specifications. The angular beamwidth is denoted by , the pulse duration by PD, and repetition rate by RR. The resolution diameter, Res, is calculated for water depth.

Conventional fish finding sonars, SIMRAD EK60 and EK500 specifications. The angular beamwidth is denoted by , the pulse duration by PD, and repetition rate by RR. The resolution diameter, Res, is calculated for water depth.

Physical parameters of modeled fish species and their measured target strength at with a CFFS.

Physical parameters of modeled fish species and their measured target strength at with a CFFS.

Mean low-frequency target strength estimates. The estimates are obtained by correlating OAWRS with CFFS data along CFFS transect. This approach is only applied to OAWRS data at 950 and . For the other frequencies, the estimates are obtained by the approach of differencing OAWRS images. The Diff is the expected target strength difference between the given frequency and .

Mean low-frequency target strength estimates. The estimates are obtained by correlating OAWRS with CFFS data along CFFS transect. This approach is only applied to OAWRS data at 950 and . For the other frequencies, the estimates are obtained by the approach of differencing OAWRS images. The Diff is the expected target strength difference between the given frequency and .

Minimum detectable fish density (M.D.D) in OAWRS imagery.

Minimum detectable fish density (M.D.D) in OAWRS imagery.

Experimentally inferred means and standard deviations of swimbladder volume, and , semi-minor-axis, and , over the depth distributions of the shoals, and corresponding means and standard deviations of neutral buoyancy depth, and , where neutral buoyancy depth is restricted to water-column depths of in the least squares fit. All three parameters are assumed to be Gaussian random variables completely characterized by their respective means and standard deviations.

Experimentally inferred means and standard deviations of swimbladder volume, and , semi-minor-axis, and , over the depth distributions of the shoals, and corresponding means and standard deviations of neutral buoyancy depth, and , where neutral buoyancy depth is restricted to water-column depths of in the least squares fit. All three parameters are assumed to be Gaussian random variables completely characterized by their respective means and standard deviations.

Concurrent high-speed rope trawl deployed by NOAA FRV *Delaware II* within or in the vicinity of large herring shoals imaged by OAWRS during the OAWRS 2006 experiment in the Gulf of Maine and Georges Bank at shoal depth. The number of most frequently caught species in each trawl deployment, including Atlantic herring, Acadian redfish, Silver hake, and Haddock are tabulated. Trawls 134, 137, and 139 were made with simultaneous OAWRS imagery, and trawls 137 and 139 were made directly through shoals as shown in Figs. 12(B) and 12(C). In contrast, trawl 134 was made in a region with no shoal formed (Fig. 12(A)), but one shoal would form in the vicinity later. Trawl 105 was made through shoals imaged by OAWRS before, and trawl 106 was made through shoals imaged by OAWRS later.

Concurrent high-speed rope trawl deployed by NOAA FRV *Delaware II* within or in the vicinity of large herring shoals imaged by OAWRS during the OAWRS 2006 experiment in the Gulf of Maine and Georges Bank at shoal depth. The number of most frequently caught species in each trawl deployment, including Atlantic herring, Acadian redfish, Silver hake, and Haddock are tabulated. Trawls 134, 137, and 139 were made with simultaneous OAWRS imagery, and trawls 137 and 139 were made directly through shoals as shown in Figs. 12(B) and 12(C). In contrast, trawl 134 was made in a region with no shoal formed (Fig. 12(A)), but one shoal would form in the vicinity later. Trawl 105 was made through shoals imaged by OAWRS before, and trawl 106 was made through shoals imaged by OAWRS later.

Concurrent high-speed rope trawl deployed by NOAA ship FRV *Delaware II* within or in the vicinity of large herring shoals imaged by OAWRS system during the OAWRS 2006 experiment in the Gulf of Maine and Georges Bank at shoal depth. The dates, times (Eastern Daylight Time), deploy depths, and geographic locations of the trawls are tabulated.

Concurrent high-speed rope trawl deployed by NOAA ship FRV *Delaware II* within or in the vicinity of large herring shoals imaged by OAWRS system during the OAWRS 2006 experiment in the Gulf of Maine and Georges Bank at shoal depth. The dates, times (Eastern Daylight Time), deploy depths, and geographic locations of the trawls are tabulated.

Article metrics loading...

Full text loading...

Commenting has been disabled for this content