Macrozooplankton/Micronekton Acoustics

Chair: Tim Stanton

Participants: Steve Bollens, Dezhang Chu, Clarence Clay, Chuck Greene, Charles Greenlaw (briefly), Lee Gordon, Roger Hewitt, Van Holliday (briefly), Bob Miyamoto, Dave Potter, Doug Sameoto, Yvan Simard, Sharon Smith, Peter Wiebe, Alan Wirtz.

Introduction

Animals ranging in size from approximately 0.5 to 5.0 cm were considered in this working group. This range overlaps with the sizes discussed in the small zooplankton group, and hence, there will be overlap of the sonar frequencies used to study the two classes of animals.

Acoustical methods can support several major science objectives involving the macrozooplankton and micronekton studies within GLOBEC. In general, the acoustic techniques can help quantify the spatial and temporal distribution, abundance, and associated size distribution of the macrozooplankton and micronekton, and their predators. By varying the deployment scheme, acoustic methods can be used in studies examining biological processes in ship, Eulerian, and Lagrangian coordinate systems. Acoustic techniques can be used to examine population growth, mortality, and physical dispersion. It should be stressed that the acoustic techniques should be used simultaneously (when possible) with other methods involving: 1) nets and pumps for direct species identification; and 2) measurements of physical properties of the ocean such as salinity, conductivity, temperature, etc. for quantitative tests of coherence.

The complexity of the biological processes that need to be measured require a variety of sonar* configurations and deployment schemes. Much time was spent discussing these requirements with the recommendation that versatile modular sonars be developed that can be deployed in many ways. In order to improve reliability and simplicity, the sonars should have as many identical components as possible.

Overview of Sonar Techniques

A variety of techniques are available which affect the design of sonars and their deployment. The techniques can be split into two broad categories: one beam per frequency, where more emphasis is on the statistical interpretation of the echoes; and multiple beams (2 or more) per frequency, where more hardware is involved so that direct measurements of target strength are obtained when the animals are resolved as individuals. The "split-beam" sonar is placed in this latter category. The output of this phase-sensitive system provides both range and arrival angle of echoes from individually resolved animals.

All methods assume the availability of an acoustic scattering model for the transformation of echo data from voltage levels to estimates of animal size. When individual animals are not resolvable by sonar, an inversion of multifrequency, volume scattering strength data can sometimes produce an estimate of the size-frequency distribution. The assumed scattering model (or models) are implicit in the inversion algorithm. The inversion method will only work when the transition point between Rayleigh and geometric scattering is within the range of frequencies used in the acoustic system. For frequencies typically used in the assessment of macrozooplankton (10's to 100's of kHz), the inversion method would only apply to populations involving macrozooplankton of a few centimeters in size or smaller.

When the individual animals are resolved, deconvolution, dual-beam, and split-beam methods can be used to produce distributions of echo amplitude or target strength. With the use of an appropriate scattering model, that data can then be used to estimate size-frequency distributions. When individual animals are not resolved, systems involving the deconvolution, dual-beam, or split-beam methods can produce echo integration data, scaled by target strengths of individuals (perhaps measured at a nearby location where the animals are similar), to estimate biomass.

The advantages of the different methods were discussed. Clearly, when individual animals can not be resolved, echo integration techniques need to be used. One beam per frequency is sufficient in that case. If the individual animals can be resolved, tradeoffs arise between the use of one beam and multiple beam per frequency systems. One beam per frequency systems require less hardware per frequency, but more effort in the development of processing algorithms. Whereas, multiple beam per frequency systems require more hardware per frequency, but less effort in the development of algorithms. We concluded that comparative studies between the approaches should be conducted as soon as possible.

One Beam per Frequency

Two inversion techniques have been developed for extracting animal size information from acoustic backscattering data. One uses a multifrequency sonar, the MAPS (Multifrequency Acoustic Profiling System; Holliday et al., 1989; Pieper and Holliday, 1984), while the other, which involves a deconvolution, can be used on single frequency or multiple frequency systems (Clay, 1983; Stanton and Clay, 1986). Both methods involve "accepting" scattering data from the animals in all parts of the acoustic beam. The MAPS is particularly successful when individual animals are not resolved. It can, however, work equally well in a uniform distribution of resolved targets, if given enough time at each point in space. Beam effects are taken into account once the data are averaged. The deconvolution approach requires that the individual animals be resolved. Beam effects are removed from the data using this technique and the result is a set of echo amplitude histograms. Both techniques rely on applying mathematical inversion methods in the post-processing software. While the software that contains the algorithms currently resides in the laboratories of Holliday, Pieper, and Clay, the math is well known (Lawson and Hansen, 1974; Holliday, 1977; Leih and Holliday, 1982; Clay, 1983) and is standard in other disciplines such as seismology.

The MAPS technique has been used in a number of field studies involving zooplankton while the deconvolution technique has been applied almost exclusively to echoes from larger sized targets such as fish. The latter method has been used at least once to extract zooplankton size distributions from echo data (Stanton and Clay, unpublished). The agreement between size distributions estimated acoustically and from net/pump collected samples has been very encouraging with both acoustic methods.

Multiple Beams per Frequency--Target Strength Estimation Techniques

Two techniques have been developed to directly estimate the target strengths of fish: the dual-beam technique (e.g., Ehrenberg, 1974; Traynor and Ehrenberg, 1979); and the split-beam technique (Ehrenberg, 1979; Traynor and Ehrenberg, in press). The methods involve comparing the outputs of each of the two (dual-beam) or four (split-beam) beam channels so that the target strengths can be determined directly. While both methods were originally developed for detection and quantification of fish schools, the dual-beam technique has recently been used in studies of zooplankton and micronekton (Richter, 1985a,b; Greene et al., 1989; Wiebe et al., 1990), and a split-beam system has been used to study deep-sea micronekton (Smith et al., 1989). Recent technological advances in the dual-beam technique have made it feasible to analyze in situ the single echoes returning from individual zooplankters as small as several millimeters. Finally, it is important to stress that, although the dual- and split-beam systems produce a single value of target strength for each echo, the target strength is highly variable with each animal. This variability mandates the collection of a statistical ensemble of echoes for the animals to be studied. For a given sampling volume, the same number of pings are required for all methods, single or multiple beam.

Overview of Acoustic Sampling Methods

The design of acoustic measurements is tied intimately to the length of the organism relative to the acoustic wavelength and the density (number/unit volume) relative to the resolution of the sonar. The specific instrumentation packages and deployments change for the different sizes and densities (number/unit volume) of objects and biological processes being studied. Because of the wide variety of organism sizes (including predator and prey) and their spatial and temporal variability, there is no standard mode of deployment or commercially available sonar system that can meet all GLOBEC science objectives concerning macrozooplankton and micronekton. We therefore recommend the following sonars with a variety of deployment modes.

Modes of Deployment

The following three deployment modes are recommended as particularly useful in GLOBEC studies.

Ship-mounted or towed survey systems

This system would provide survey data during transects taken in conjunction with physical measurements of the ocean (CTD, etc.). These data could loosely be called "synoptic". We concluded, however, that given the finite amount of time it takes to conduct a transect, the features may change enough so that the data are not a true acoustic snapshot of the region, hence the terminology "survey".

Ship-based cast or tow-yo systems

These systems, deployed from the ship, would provide a data set to complement the survey by providing a closer look at the organisms via the cast or tow-yo methods. Higher spatial resolution is obtainable in this mode. Furthermore, by use of a side scan configuration at the deeper depths, one can measure horizontal spatial distribution.

Remotely deployed systems

These systems would involve a sonar(s) mounted on freely drifting buoys, freely drifting neutrally buoyant platforms, remotely operated vehicles (ROV's), bottom mounted/moored platforms, or yo-yo platforms (standalone systems that periodically move up and down in the water column). The drifting systems address the requirement for performing times series analysis in Lagrangian coordinates while the latter two conduct times series measurements in Eulerian coordinates.

The details of the above modes of deployment will be discussed in later sections in the context of the acoustic measurements.

Recommendations for Sonar Design

We envision a number of sonars for eventual use in the GLOBEC program. In order to improve on the reliability and flexibility of the systems, we recommend that most components of the above sonars be modular and identical. Naturally, the remotely deployed systems require changes in power and sampling strategies. The sampling or ping rate, for example, could be a programmable feature.

The various functions of the sonars such as logic or transceiver electronics should be constructed on separate electronic "cards" that plug into a card cage or rack. The size of the cage and number of cards would depend upon the number of frequencies used. The sampling strategy or sequence of pings that is specific to the particular sonar configuration, deployment, and biological process can be programmed into the logic card that all systems would have in common. For example, a shipboard survey system may be acquiring data on all sonar channels continuously and simultaneously, hence requiring a continuous supply of power. Because of power constraints, the remotely deployed Systems would need to turn off automatically between pinging sequences. As a result, the system would be turned off most of the time to minimize depletion of battery power. There would be a difference in power supply sections of the cages depending upon whether the system would receive AC shipboard power or DC battery power in a remotely deployed system.

We listed all commercially available sonars and came to the conclusion that no such system as described above exists. Furthermore, there is no system available that could be easily modified to fulfill the needs of the GLOBEC program. The system that comes closest to meeting the needs of this program is one currently under development by Clay at the University of Wisconsin (NSF funding). It is modular and can provide acoustic signals at a variety of frequencies, although it requires AC power. As a minimum, the system would need to be modified so that it could: 1) accept DC power at low consumption rates (e.g., by use of low power integrated circuit components); and 2) have adaptive sampling modes to allow for variable sampling schemes that depend upon the available power. With all of these factors included, we recommend that a sonar system be developed, with attention initially being paid to the design of prototype systems, such as that under development at the University of Wisconsin.

Since deployment of sonars are specific to each biological process under investigation, we recommend that the "platform" for each type of deployment be addressed on a case-by-case basis by the user (science PI) and be built by, or subcontracted to be built for, the user.

Acoustic Frequencies

It is clear from the scattering behavior of animals, that in order to discriminate various sizes of the animals, one must use a broad range of frequencies. Practical limits on the size of transducers provide the lower limit of frequency, while absorption of sound determines the upper limit of frequency. We recommend frequencies between 38-420 kHz be used. These end points correspond to the frequencies of systems that are available commercially. Although the system described here should be built from the ground up, we recommend that some of the frequencies be identical to those of off-the-shelf units to allow comparisons between historical data and data collected with the new system. The number of frequencies required depends upon interpretation methods used and power requirements (when the units are remotely deployed). We recommend that a minimum of 3 frequencies be used and a maximum of approximately 8-10.

Sonar Resolution

Since animals may sometimes occur at densities much greater than one per cubic meter, it is not practical for all sonars to be able to resolve individual animals in all situations. Some systems under certain deployment schemes, however, should be able to resolve individuals so that direct estimates of density and statistical properties of patchiness can be made.

Survey System

Acoustical surveys from surface-deployed systems can rapidly map in three-dimensions the distributions of biological sound scatterers within large volumes of water Surveys such as these can be used to assess patchiness on the scales of 10's to 100's of meters, the spatio-temporal coupling of predator and prey populations, and the effects of physical and topographic features on animal distributions.

This system, as well as the use of other remote sonar systems, could provide water column acoustic data (echograms) during transects and help direct the location and timing of "point" sample methods (nets, etc.) The echogram would also help place the various point measurements into the context of the complex ocean structure during later analysis. The system would include the maximum number of acoustic frequencies, store massive amounts of raw data at high speed, and provide high quality acoustic data for sophisticated interpretation.

We recommend that the system be operated at fixed depth with the option of being lowered to depths near scattering layers to examine the layers at higher resolution. All sonars on the tow body should transmit their signals simultaneously (as opposed to sequentially) and the echoes should be acquired simultaneously so that the various echoes can be associated with the same sampling volume. The beamwidths should be similar or identical to help achieve this goal. The approach of using simultaneous transmissions and acquisition also minimizes the time between acoustic transmissions for each frequency and allows for maximum resolution of the patch structures.

ROV, Profiling, and Towed Systems

While the survey system can provide a picture of the distributions of the sound scatterers over a large volume of the ocean, much of the data collected represents organisms located at great distance from the sonar. The spatial resolution of each sonar decreases with increasing distance from the sonar, and as a result there is a loss of the lower end of spatial scale that can be measured at those ranges. The survey system can occasionally be lowered to increase the resolution, but this may result in the loss of survey data. We recommend that other sonars be deployed in or near a region of scatterers in order to improve the spatial resolution of the measurements.

There are a variety of platforms that can be used to deploy sonars from a ship --submersibles, remotely operated vehicles, vertically profiling instruments (e.g., MAPS), towed vehicles (e.g., Batfish, MAPS, V-fin, etc.), nets (e.g., MOCNESS or BIONESS), and trawls. The first few platforms can be used to acoustically explore concentrations of interest while deploying a sonar on a net or trawl, thereby providing some degree of ground truth and species identification information in concert with the acoustical data set.

Free-Drifting Buoy and Bottom Mounted/Moored Systems

Measurements need to be performed in coordinate systems besides that of a ship. It is important to study distributions of animals at fixed locations (Eulerian) and on platforms that are allowed to drift with the flow of the water (Lagrangian).

The changes in acoustic scattering levels at the proposed GLOBEC study sites will vary on time scales ranging from daily to seasonal to yearly. While in some of these areas the general levels of volume reverberation are predictable to some extent, the biological and physical factors which affect the levels are not well understood. As a result, the reverberation levels are not sufficiently predictable. Most studies of acoustic backscattering in the oceans have been conducted from ships which limit the duration, areal coverage, and vertical extent of the data. Autonomous free-drifting buoys and bottom mounted moorings can solve this problem. They should be equipped with echo sounder electronics and transducers, a digital signal processor, data storage, and satellite and radio communications systems. These systems can be used to periodically and frequently collect high frequency backscattering information (for example, 120 kHz and 420 kHz) from remote locations and relay the information to a ship or shore location in real-time. Because the systems are autonomous, with finite electrical energy and computer memory, the sampling strategies must be carefully adapted to the biological process of interest to make the most of those quantities.

Envisioned in both the surface free-drifting system and the moored instrument is an acoustic system which includes two or three frequencies which can be sequentially activated. Profiles of acoustic backscattering would be obtained at depth intervals (nominally 1 m) to a maximum range of operation which is frequency dependent (typically 10's of meters to ca 200 m). In order to adapt the sampling protocols to the phenomena being studied, the instrument package needs to be programmable so that a duty cycle of choice can be selected and echo sounder parameters such as pulse length and processing criteria can be altered. Data to be collected could include individual target strengths as a function of range and average backscattering strength for each depth interval. Data would be stored on a mass storage unit (e.g., an optical disk) in the buoy or mooring unit for post processing. Reduced data in the form of a target strength histogram, and integrated intensity for a reduced set of depth intervals at each frequency averaged over some time interval (i.e., 2 hrs), would be produced for daily transmission via satellite to shore. Real-time, two-way radio telemetry should also be available.

Bottom mounted acoustic systems can be used to measure the acoustic scattering by animals located at and near the bottom. Depending on the application, the systems could be used to look vertically (up or down), horizontally, or at other angles. Because of the difficulty of bottom interference with measurements in the horizontal or downward directions, new acoustic techniques must be applied. These new techniques should utilize vertical split beams to resolve bottom versus near bottom animals and make comparisons between pings to determine slight changes in the acoustic scattering due to bottom animals. This ability to measure macrozooplankton on, or very close to, the bottom will be an exceedingly important capability in areas such as Georges Bank where very large demersal shoals occur during the daytime. For maximum coverage of the surrounding volume of water, one may consider mounting the sonar on a rotating vertical shaft so that the region may be scanned much like a radar system (electronic steering via beamforming is also a possibility, although the complexity of the electronics is increased).

Team Responsible for Acoustic Systems

The maintenance, calibration, refinement, and further development (especially of software) of the above acoustic systems is beyond the experience of most biologists and end users of such systems. Therefore, it is essential that a team of experts (2 or 3) in acoustics, electronics, and software be assigned responsibility of ensuring that the system is calibrated and operating to the proper specifications before and after each cruise. This type of facility-level support is certainly available in other disciplines, such as geoacoustics, where the operation and maintenance of the Sea Beam multibeam bathymetry system is provided.

Specific tasks of the team would include: 1) ensuring that the mechanical terminations of the cable and the various wire connections are maintained during the cruises; 2) developing the basic software for recording and storing the data which would be modified to suit the various users; 3) matching the acoustic system to the different ships before each cruise; 4) responsibility for trouble shooting problems during the cruise; 5) maintenance and operation of the winch and care of the tow cable and the towed body; 6) ensuring that acquired data is of the highest quality, and that the researcher is notified of any malfunction during the survey. This means that the team must be responsible for data acquisition at all times; and 7) documenting the protocol for calibration and maintenance so that changes in personnel will not affect calibration and maintenance procedures.

The above mentioned responsibilities, which do not represent all of the anticipated tasks, are a consuming job and justify a dedicated team.

Format of Data/Data Management

Data must be archived and easily exchanged between programs to meet GLOBEC objectives. To facilitate this exchange we recommend the following actions. Data should be digitized and in scientific units (e.g., m2, dB//ÁPa, etc.). The type of data should be explicitly described (e.g., raw acoustic, biomass, target strength). All relevant information should be fused to the data and should include synchronization of date, time, position, instrument values (e.g., sample rates, noise levels, bandwidths), calibration information, processing information (ping averages, bin depths), and any other ancillary information (temperature, salinity, etc.). Finally, a data interchange format should be developed that will permit archiving and exchange of acoustic data. An interchange format is desirable since many commercial instruments will have proprietary data formats and it is unlikely that one format can be defined for all data acquisition systems.

Data should be archived and merged with non-acoustic information at a central data management facility. This facility is essential to maintaining information for the climatic time scale experiments.

Other Systems

While we recommend the use of the modular sonar approach described above, which can perform a suite of measurements under a wide variety of conditions, there are other systems that should be considered. These systems can provide types of data not otherwise obtainable.

3-D Imaging Sonar

This planar sonar array can be used to produce three-dimensional images of the volume of interest. The "images" indicate the location of animals and their echo levels so that the spatial density and inter-animal separations can be directly measured. No indirect interpretation technique is required to provide such information. By tracking the animal through the acoustic beam and recording the statistics of its time-varying echo levels, classification would be possible by use of scattering models. This type of system is currently under development both at SIO (Jaffe) and WHOI (Stanton).

Acoustic Doppler Current Profiler

The Acoustic Doppler Current Profiler (ADCP) estimates water motion by measuring the frequency of reverberation from different depths in the water column. In comparison to most acoustic instruments used to study oceanic organisms, this instrument transmits very long pulses so that the doppler shift induced by motions of the animals can be detected and resolved. An implementation of this technology is commercially available from RDI. This system is used routinely to obtain estimates of water current profiles under the assumption that the composite motion of the animals which dominate the sound scattering reflect the motions of the water mass.

The motions measured include both active swimming and passive advection. Researchers could potentially use doppler information to study animal behavior, e.g., vertical migration. A quantity related to volume backscattering strength can also be measured with currently available ADCP's. With absolute system acoustic calibration and careful attention to a variety of stability issues (Flagg and Smith, 1989a, 1989b), such information is potentially useful to biological oceanographers.

Passive Localization

While all of the above-mentioned techniques involve active acoustic methods in which a burst of sound is transmitted into the water and the resultant echoes are detected for localization and classification purposes, one can also take advantage of the fact that some animals (fish included) generate their own sound. In particular, some species of fish generate sound during the time of spawning. A grid of hydrophones could be used to pick up sounds from individuals or schools. Triangulation or even tomographic methods could then be used to estimate the location of the animals. Classification of animals by the nature of the sounds they make could be possible if the properties of the sounds that are generated by various species were known.

Summary of Recommendations

We must address the complex needs of the GLOBEC science objectives with a versatile set of sonars and deployment schemes. This requires:

  1. The development of a sonar whose components are modular and interchangeable so that the same design can be used to meet all needs of the program. Ultimately a number of duplicate system components would be constructed so that a variety of acoustic systems could be assembled;

  2. The establishment of a facility(ies) where a permanent team of specialists maintains, operates, and refines the acoustic hardware and software. Such a facility would be similar to what is available in other disciplines and would allow the user to concentrate on achieving the science objectives;

  3. Use of other acoustic systems such as a 3-D imager, doppler profiler, or passive localizer to provide other forms of data; and

  4. The establishment of data archive protocols to facilitate long term (10's of years) use of the data.
The development and construction of sonars with the attributes discussed above was considered to be appropriate for inclusion within future "Calls for Proposals" from the GLOBEC program.

Finally, it was agreed that any hardware associated with the deployment of the sonars such as tow-body, drifting buoy, etc. should be treated on a case-by-case basis by the users in their respective science proposals.