Field Calibration of a 3D Scientific Fisheries Sonar

Sea Technology, Feb 2008 by Ona, Egil, Andersen, Lars Nonboe

Accuracy Matters When Measuring Sensitive Pelagic Fish Schools

One of the most serious causes of error in acoustic surveys for abundance estimation of pelagic fish stocks is vessel interference. The noise from the measuring research vessel, or other relevant stimuli, is sensed by the fish long before the vertical echo sounder beam has measured their abundance. Acoustic density dilution can occur both by horizontal movements and diving. In both cases, the echo sounders on board the research vessel will experience a lower mean density on their transecting than what was available in the undisturbed situation.

The sonar used in this article was specifically designed to overcome some of the problems of measuring pelagic fish close to the sea surface. Traditional fishery sonar has been used both qualitatively and quantitatively for many years, but often suffered from low data output resolution and lack of proper calibration procedures. The observation volume has also been limited to one horizontal slice through the water column, often showing only the top of the schools, like the top of an egg. Therefore, the design criteria of the new Simrad MS70 research sonar were accurate calibration procedures as established for echo sounders, high data output quality and larger observation volume. The idea of obtaining the entire school in one ping was conceived during the sonar planning stage.

The standard targets calibration method used for split-beam echo sounders was, therefore, expanded to cover calibration of each of the 500 beams of the sonar, but also to simultaneously cover a frequency range from 70 to 120 kilohertz. Since practical sonar calibration must be made at sea several times a year, selected, sheltered fjords were used, where the vessel can be anchored or fixed to the shore during the calibration exercise. Experience has shown that large, powerful reference targets perform better than small, weak targets in these situations. This is partly due to the overall reverberation level in the measurement volume, but also to individual fish targets visiting the measurement volume.

Two large, specially designed standard targets were made for this sonar. These were the 75 and 84-millimeter-diameter tungsten carbide spheres, weighing 3.3 and 4.6 kilograms, respectively. Since the sonar use frequency-rotated sector transmission, each horizontal sector of 25 beams, covering 60� on the port side of the vessel, uses a certain part of the frequency band. Furthermore, each sector in the vertical direction has its specific center frequency, decreasing down to the lowest sector number, 20. Calibration, therefore, is achieved by computing the effective target strength of the standard target from the target strength spectra of the spheres, but weighed with the received bandwidth used in each sector.

Sonar Calibration

On February 7, 2007, the sonar was calibrated with the new spheres on the research vessel G.O. Sars in Sandviksflaket, just outside Bergen Harbor, with the vessel anchored with one stern anchor and one bow anchor. Prior to the calibration, sea temperature, salinity and pressure were measured by a conductivity, temperature, depth probe, from which the sound speed in the sonar observation volume was derived. Target strength as a function of sound velocity for the two standard targets at the specific frequencies used was computed prior to the calibration.

The temperature at a transducer depth of eight meters and downward was quite constant, between 8� and 9� C, with sound velocity varying between 1,485 meters per second at eight meters1 depth to 1,489 meters per second at 25 meters' depth. The sound-speed profile was further entered into the sonar to compute the frequency-dependent absorption coefficient in the system time-varied gain.

The large weight of the standard target demanded the normal motorized calibration winches to be replaced by two hand winches on this particular calibration and positioned the target about 13 meters from the transducer face, using two six-meter-long rods to hold the nylon lines. From this two-point rig, the target could be slowly moved across the entire cross section of the 500-beam array, ensuring multiple detections in each individual beam. Some selected beams were inspected and calibrated more accurately, enabling a full split-beam calibration.

Several new features were implemented in the sonar, relevant for the calibration and data collection. Firstly, the transducer is a modern composite 800-element transducer, forming the 500 beams. Four extra overlapping synthetic beams are formed electronically for each of the sonar beams in order to facilitate splitbeam phase measurements. New algorithms now enable the sonar to measure the target position relative to the transducer face in the entire observation volume. Secondly, detailed tank measurements have shown that with the new composite transducer technology, almost perfect agreement between theoretical computation and the actual beam pattern is recorded. It is therefore assumed that to actually measure the beam pattern of the main lobe for each individual beam is not required-rather, theory is used for beam compensation. This simplifies the calibration to the straightforward procedure of guiding the calibration target through all 500 beams, measuring the target strength of the sphere and adjusting the receiver gain for the difference between the measured target strength and the computed target strength of the standard target. Similarly, the same assumption is held for the equivalent beam angle of each individual beam when measuring the volume backscattering strength, Sv, or the echo energy of multiple targets, like schools or layers of fish. The integrated echo energy from the standard target is measured for each target location, corrected for beam gain and compared with the theoretically expected echo energy, identical to calibration of modern split-beam echo sounders. If needed, these assumptions may be checked for individual beams by detailed split-beam calibrations.