An extended calibration target has been developed for calibrating the intensity output of a multibeam echo sounder (MBES). The target was constructed of chain links arranged similar to a curtain, providing an extended surface target with a mean scattering strength of −17.8 dB at 200 kHz. The target was used to calibrate a 200 kHz MBES, and the MBES was subsequently used to collect seafloor backscatter over sand and gravel seafloors. Field results were compared with calibrated split-beam echo sounder measurements at an incidence angle of 45°. The results suggest that the chain target is a viable MBES calibration tool.

Measurements of the scattered intensity from the seafloor made with high-frequency (nominally greater than 100 kHz) multibeam echo sounders (MBES) are often used to infer characteristics of the seabed (Todd et al., 1999; Pickrill and Todd, 2003; Brown and Blondel, 2009; Lurton and Lamarche, 2015). Whether revisiting an area to detect changes in the seafloor (Rattray et al., 2013; Clarke, 2012), mapping a region of the seafloor with multiple MBES systems (Greenaway and Rice, 2013), or attempting to invert seafloor scattering models for characteristics of the seafloor such as roughness and acoustic impedance (Fonseca and Mayer, 2007), it is essential to calibrate the MBES so that its recorded intensity (or other analogous value such as intensity level or pressure amplitude) can be converted to acoustic values such as target strength (TS) or seabed scattering strength (Sb). Narrow beamwidths, typically between 0.5° and 2°, and imperfect knowledge of transmit and receive sensitivities, beamformer gains, etc., make calibrating MBES in terms of the recorded intensity difficult in practice. Methods proposed for conducting these types of calibrations utilize either standard target spheres (Foote et al., 2005) or reference hydrophones and projectors (Lanzoni and Weber, 2010). These methods have different inherent limitations including difficulties in reliably positioning calibration spheres in comparison to echo sounders that use split-aperture correlation techniques (Burdic, 1991) on two axes [i.e., split-beam echo sounders (SBESs)] for within-beam target positioning, as opposed to only one axis as is the case for the typical hydrographic MBES; cumulative errors associated with the calibration of individual system components, as would be the case when errors associated with individual transmitting and receive sensitivities add; and requirements for long test-times associated with navigating a target sphere or hydrophone through all beams with a resolution in both along-track (i.e., in the typical direction of motion for a MBES used for mapping the seafloor) and across-track (i.e., perpendicular to the direction of motion in the horizontal plane) angles high enough to accurately measure both the beamwidths and the overall system sensitivity at the beam maximum response angles. Furthermore, these methods ultimately do not provide a calibration that is entirely analogous to the measurement of an extended (i.e., large compared to the resolution in an individual MBES beam) surface target like the seafloor. Although not required for calibration, using a calibration target that is morphologically similar to the final measurement objective (in this case, the seafloor) may decrease the impact of calibration errors. For example, slight errors in estimation of the ensonified area made during calibration with an extended surface target and in estimates of seafloor scattering strength could potentially cancel.

In this work, we explore the use of an extended surface target for MBES calibration. The motivating factor behind the design of an extended surface calibration target was to simulate the scattered response of the seafloor in a controlled and repeatable manner, but without some of the complicating phenomena associated with the seafloor including angle-dependence in the backscattering cross section and scattering from within the sediment volume. The target was also designed so that it did not require precise positioning (including relative orientation) of the target. We describe the design and construction of the extended target, its acoustic characterization, the methodology for its use in MBES calibration, and provide results for one such MBES. We include a brief description of field trials with the calibrated MBES for completion.

The extended target was designed to be similar to a simple conceptual model of the seafloor, where the scattered pressure can be considered as the sum of a large number of random elements dispersed throughout the ensonified area. In such a model, the scattered pressure ps can defined as

ps=iaieji,
(1)

where ai and i are the amplitude and phase of the ith element and are both considered to be random variables. If the number of elements in Eq. (1) is large, then the central limit theorem applies and ps is distributed as a complex Gaussian and |ps| is Rayleigh distributed.

To construct a target that matched the conceptual model described by Eq. (1), a curtain of stainless steel, type 18, jack-chain was assembled. Type 18 jack-chain is constructed of a loop of 1.2 mm diameter wire formed into a figure-eight and then twisted so that the loops of the figure-eight are approximately oriented at right angles. When assembled into a chain the spacing between the individual loops is 1 cm. Lengths of jack chain (200, 2 m long) were assembled in the style of a curtain, with spacing between chains of 1 cm (Fig. 1). The orientations of the individual double-loops are randomly oriented over 360° (Fig. 1, right). The 2-m wide extended target covered only a portion of the beams at the test ranges described here (2–8 m), requiring a rotation of the MBES to conduct a full calibration. The top and the bottom of the target were fixed using fiberglass supports, and the chain was designed to hang vertically in the water with the fiberglass supports out of the field-of-view of the acoustic system being calibrated.

Fig. 1.

The extended target constructed from stainless-steel, type 18 jack chain at 1 cm spacing, including a close up of the chain (right).

Fig. 1.

The extended target constructed from stainless-steel, type 18 jack chain at 1 cm spacing, including a close up of the chain (right).

Close modal

A 200 kHz Simrad EK60 SBES (Kongsberg Maritime, Horten, Norway) was used to acoustically characterize the extended target in an acoustic test tank measuring 18 m long × 12 m wide × 6 m deep at the University of New Hampshire Center for Ocean Engineering. The SBES operates using a single piston transducer, an ES200-7CD, with a nominal one-way beamwidth measured from the −3 dB points, of 7° according to manufacturer specifications. The SBES was initially calibrated following the method of Demer et al. (2015) by collecting echoes from a 38.1 mm tungsten carbide sphere while controlling the relative orientation between the sphere and the transducer in order to generate a beam pattern, resulting in a 2-way equivalent beamwidth, θeq, of 5.0° (similar to the results in Weber and Ward, 2015).

The calibrated SBES was used to characterize the extended target in two different tests. The first test explored potential angle-dependence in the extended target between ±50° in 5° increments. To do so, the chain target was hung vertically in the test tank so that it was parallel to the longest wall (18 m) of the tank, at a distance of approximately 2 m away from the wall, and then moved to 21 fixed positions along the wall. The SBES was mounted on a rotating pole 1.5 m from the opposite wall, and manually aligned using a sighting laser so that its beam would approximately intersect the center of the chain target. The location of the rotating pole was held constant throughout the experiment, resulting in ranges between the SBES and the target center of between 6 and 9.5 m. At each fixed target position, acoustic echoes were collected with the SBES at a ping repetition rate of 0.25 s, which was qualitatively observed to be low enough to avoid biasing the target response with reverberation in the tank. The process of moving the chain target and aligning the SBES was repeated 10 times to generate an ensemble of random realizations of the target, with the assumption that for each realization the chain target would be in a slightly different position.

Recorded SBES data were converted to Sp (apparent TS, uncorrected for effects related to beam pattern) and two split-beam phase angles. At the phase zero-crossing, corresponding to the beam maximum response axis (MRA) in the across-track (horizontal plane in the tank) direction, the Sp is equivalent to the TS of the chain target and these values corresponding to the MRA were extracted from the data and used for subsequent analysis. The TS can be converted to a measure that is analogous to Sb (despite being the scattering strength of the chain target rather than the seafloor) by accounting for the ensonified area, A (Kinsler et al., 2000)

Sb=TS10log10A.
(2)

A can be calculated as the lesser of two values which depend on the angle of incidence on the chain target, θi: the area of an ellipse whose minor axis can be approximated as Rtan(θeq/2) and whose major axis can be a approximated as Rtan(θeq/2)/cosθi

A=πR2tan2(θeq/2)/cosθi,
(3)

or by the limit of the pulse projection in the across-track (horizontal) direction and the beam projection in the along-track (vertical) direction

A=Rθeqcτ/2sinθi,
(4)

where c is the speed of sound and τ is the pulse length. For all of the SBES tests conducted in this work a pulse length of 128 μs was used.

No angular dependence between ±50° in the chain target Sb was detected from the SBES measurements (Fig. 2), which was the expected result given the random orientation of links within the chain target together with the conceptual model described by Eq. (1). Averaging all measurements for the test of angle dependence, the estimate of the mean chain target Sb was −18.0 dB with a 95% confidence interval of [−18.6, −17.4].

Fig. 2.

Observations of the potential angle-dependence in Sb from the chain target made using a calibrated 200 kHz SBES. The average Sb is given by the filled circles, and the 95% confidence intervals for the individual estimates of the average (i.e., at each θi) are given by the x's.

Fig. 2.

Observations of the potential angle-dependence in Sb from the chain target made using a calibrated 200 kHz SBES. The average Sb is given by the filled circles, and the 95% confidence intervals for the individual estimates of the average (i.e., at each θi) are given by the x's.

Close modal

The estimates for the ensonified area in the tests described by Fig. 2 ranged from 0.10 to 0.23 m2, or between 1000 and 2300 chain loops contributing to the scattered field. A narrower beam system, such as a MBES, might have a smaller footprint with few chain loops contributing to the scattered field (e.g., a 1° normal incidence beam at a range of 10 m would ensonify approximately 240 chain loops). Too few scattering elements may alter the underlying statistics of the scattering field, violating the assumptions in the central limit theorem. To test this possibility, the experiment above was repeated (N = 14 repetitions with 15 independent measurements at each angle) at a fixed θi of 45° and at ranges from 2 to 8 m, corresponding to between 240 and 950 chain loops. The results (Fig. 3) exhibit no apparent range-dependency. Averaging all measurements for this range-dependent test, the estimate of the mean Sb was −17.5 dB with a 95% confidence interval of [−18.2, −16.5].

Fig. 3.

Observations of the potential range-dependence in Sb from the chain target made using a calibrated 200 kHz SBES. The average Sb is given by the filled circles, and the 95% confidence intervals for the individual estimates of the average (i.e., at each target range) are given by the x's.

Fig. 3.

Observations of the potential range-dependence in Sb from the chain target made using a calibrated 200 kHz SBES. The average Sb is given by the filled circles, and the 95% confidence intervals for the individual estimates of the average (i.e., at each target range) are given by the x's.

Close modal

In both tests (Figs. 2 and 3), a chi-squared goodness-of-fit test was used to test the hypothesis that the sample cumulative distribution function was Rayleigh, using the Rayleigh parameter estimated from the data. In each test, the hypothesis was accepted at the 5% significance level. The pooled data from both tests (N = 98 for the range test and N = 210 in the angle test) show a mean Sb of −17.8 dB with a 95% confidence interval of [−18.3, −17.3]. This pooled mean was taken to be the true Sb of the chain target.

A 200 kHz Reson T20-P MBES was calibrated in the same tank used for characterizing the chain target. At 200 kHz, the manufacturer states that this system has a 2° transmit beamwidth in the along-track direction, and a 2° receive beamwidth at broadside. In the configuration chosen for this effort, 256 beams were generated by the MBES and spread over a 140° swath. A 130 μs pulse length was used, and the pulse repetition rate was 1 s. The sound speed in the fresh-water tank was estimated to be 1480 m/s, based on a temperature of 19.2  °C.

The MBES was mounted on the bottom of a pole so that the transmit beam was oriented with its main, larger axis horizontal in the tank (i.e., with the “along-ship” direction oriented vertically). The chain target was suspended from a floating platform at a distance of 7.3 m from the MBES (see example MBES image in Fig. 4). The target subtended an angle of approximately 15° at this range, and to avoid errors associated with partial ensonification of the target only the data corresponding to the middle 5° of the target were analyzed. The MBES was rotated about its vertical axis in 1° increments through ±80°. At each increment of rotation, the MBES was used to record the scattered field from the chain target while the target was manually moved back and forth in a gentle but random manner (approximately ±10 cm with a 1 s period). The chain target motion allowed several independent realizations of the scattered field to be collected in a time-efficient manner.

Fig. 4.

An example MBES ping showing two walls of the test tank and the chain target.

Fig. 4.

An example MBES ping showing two walls of the test tank and the chain target.

Close modal

The data recorded by the MBES corresponded to integer values representing the beamformer output, hereafter referred to as digital value or DV (unreferenced) when converted to decibels. DV is proportional to the echo level (EL) observed at the MBES receiver, but also includes the unknown receive sensitivity and beamformer gains and the known user-applied gains, G, noting that the MBES “power level” was set to 210, which corresponds approximately to the MBES source level in dB re 1 μPa @ 1 m. The approach taken here to calibrate the system is to develop a “catch-all” calibration coefficient, C, that incorporates all of the unknown quantities (assumed to be constant) such that

DV=EL+G+C.
(5)

Using the active sonar equation (Urick, 1983), Eq. (5) can be expanded in terms of EL and rearranged to isolate the C

C=DVG(SL40log10R2αR+TS),
(6)

where R is the range to the target in meters, α is the absorption in dB/m, and TS is defined by Eqs. (3) and (4) using estimated beamwidths of 2.0°. The specific DV used in this work correspond to the automated “bottom” detections from the MBES, which reliably assumes that the chain target is a feature of the non-existent seafloor in the tank. C is estimated as a function of beam steering angle, θs with the result shown in Fig. 5. The calibration results show a 1 dB ripple that may be due to variations in transmit or receive beam pattern, or some other unknown effect, and for simplicity these ripples are smoothed by fitting a fourth order polynomial to the data such that C can be estimated as

Ĉ(θs)=107.44.65×103*θs6.44×104*θs21.81×106*θs33.15×107*θs4,
(7)

with the result shown in Fig. 5.

Fig. 5.

The calibration factor, C for the MBES presented as a function of beam steering angle, θs. The dashed line bounded by the dotted lines represents the average C and the upper and lower two standard deviation bounds; the thick solid line represents a polynomial fit to the data, and the diamonds represent the results of calibration checks using a standard target sphere.

Fig. 5.

The calibration factor, C for the MBES presented as a function of beam steering angle, θs. The dashed line bounded by the dotted lines represents the average C and the upper and lower two standard deviation bounds; the thick solid line represents a polynomial fit to the data, and the diamonds represent the results of calibration checks using a standard target sphere.

Close modal

To serve as an independent comparison of the chain target calibration results shown in Fig. 5, a 38.1 mm tungsten carbide sphere with an estimated TS of −39.4 dB (Chu, 2012) was suspended in the beam and used to estimate C using Eq. (6) for clusters of beams near θs of −45°, 0°, and +45°. Care was taken to locate the sphere on the MRA by lowering the sphere through the beam in small increments, and subsequently taking only the maximum DV for each beam in which the sphere was located. The results are overlaid on the chain target data in Fig. 5, and agree with the chain target response within approximately 1 dB. The 1 dB ripple that appeared in the chain target result appears to also be present in the calibration sphere data, suggesting that this ripple is a feature of the system rather than a measurement artifact.

As a further test of the chain target calibration methodology, the calibrated MBES was used to conduct a seafloor backscatter survey in lower Portsmouth Harbor, NH (see Weber and Ward, 2015 for a more complete description of this location). The MBES was mounted on a small vessel along with the same 200 kHz SBES used to acoustically characterize the chain target in the acoustic test tank. The SBES was mounted at a 45° elevation angle in the roll plane, pointing toward the port side of the vessel. Both the MBES and the SBES were triggered so that they alternated pings, with each system pinging twice per second. The same operational settings (e.g., pulse lengths, power settings) used in the tank for both systems were also used in the field.

Data collected with the SBES were processed identically to that used for characterizing the chain target when generating Figs. 2 and 3, using a combination of Eqs. (2)–(4). Data collected with the MBES were processed using a re-arranged Eq. (6) to solve for TS, and then the smaller of Eqs. (3) and (4) to convert TS to Sb, with the C given by Eq. (7).

A comparison of the Sb collected from the MBES and the SBES over sand, gravel, and bedrock substrates, including the entire survey line shown in Fig. 1 of Weber and Ward (2015), show that the MBES estimate for Sb is approximately 0.5 dB higher than the SBES estimate, and that the lower and upper bounds for 95% of the differences (MBES Sb - SBES Sb) are −1 dB and 2 dB, respectively.

Estimates of Sb averaged in 5° bins and collected within 25 m for six locations with substrates ranging from fine sands to gravel are shown in Fig. 6. These six locations, labeled A–F, correspond to the identically labeled locations described by Weber and Ward (2015). At an incidence angle of 45°, the agreement between MBES and SBES for the same locations ranges from 0.5 to 1.5 dB. The angle dependence in Sb observed by the MBES is in general agreement with previous empirical studies and models of seafloor scattering from similar seabeds (e.g., Jackson and Richardson, 2007), exhibiting a greater separation at oblique incidence angles than near normal incidence (a total range of 12 and 3 dB, respectively).

Fig. 6.

Field observations of Sb collected with the 200 kHz calibrated MBES at incidence angles between 0° and 64°, and comparison to a calibrated 200 kHz SBES at an incidence angle of 45°. The data are collected at six different locations which are characterized by Weber and Ward (2015) as follows. A,B: medium sands with high shell hash content and bedforms ranging from ripples to sand waves; C,D: very poorly to poorly sorted sandy pebble gravels or pebble gravels; E,F: very fine sands to pebbly fine sands with 88%–99% sand content and sometimes abundant sand dollars.

Fig. 6.

Field observations of Sb collected with the 200 kHz calibrated MBES at incidence angles between 0° and 64°, and comparison to a calibrated 200 kHz SBES at an incidence angle of 45°. The data are collected at six different locations which are characterized by Weber and Ward (2015) as follows. A,B: medium sands with high shell hash content and bedforms ranging from ripples to sand waves; C,D: very poorly to poorly sorted sandy pebble gravels or pebble gravels; E,F: very fine sands to pebbly fine sands with 88%–99% sand content and sometimes abundant sand dollars.

Close modal

A new MBES calibration method has been proposed and tested. This method uses a “curtain” constructed of jack-chain to simulate an extended surface target containing random scatterers (i.e., the individual links). This chain target shows no evidence of angle dependence in its scattering strength, which is the expected result given the randomness in chain-link alignment within the curtain. Tests with the chain target support a Rayleigh scattering model, consistent with simple models for the seafloor or any other extended surface target with distributed, random scattering “centers” in sufficient number to satisfy the central limit theorem.

Comparisons between MBES calibrations conducted using the chain target and a methodology using standard target spheres show agreement to within 1 dB. Field tests with the calibrated MBES show measurements of Sb that are consistent with an easier-to-calibrate (but narrower field-of-view) SBES, with differences of approximately 1 dB.

The chain target is thought to be a viable tool for calibrating MBES, requiring no knowledge of several individual MBES components (e.g., receiver gain, source level, beamformer gain) and only approximate knowledge of the transducer beam patterns and some understanding of the system power and gain settings (note that these were held fixed throughout this work). There are some advantages of the chain target calibration methodology over other methods, including time-efficiency, less precise positioning of targets or test hydrophones, and a requirement for only one rotation of the MBES through the angle of interest. Perhaps more importantly, the chain target calibration offers the potential to be a more appropriate calibration for MBES systems that will be used for seafloor scattering measurements because it inherently incorporates the MBES beamwidth, whereas other calibrations require a separate calibration for beamwidth and risk a higher cumulative error. The chain target described here must hang vertically, forcing a specific MBES mounting geometry that is most-useful in a tank facility. Further development of this method may be needed for field calibration.

This work was supported under NOAA Grant No. NA10NOS4000073. Initial discussions with Xavier Lurton regarding MBES calibration methods are gratefully acknowledged for their contributions to the initial idea of an extended calibration target.

1.
Brown
,
C. J.
, and
Blondel
,
P.
(
2009
). “
Developments in the application of multibeam sonar backscatter for seafloor habitat mapping
,”
Appl. Acoust.
70
(
10
),
1242
1247
.
2.
Burdic
,
W. S.
(
1991
).
Underwater Acoustic System Analysis
, 2nd ed. (
Prentice Hall
,
Englewood Cliffs, NJ
), pp.
328
331
.
3.
Chu
,
D.
(
2012
). (private communication).
4.
Clarke
,
J. E. H.
(
2012
). “
Optimal use of multibeam technology in the study of shelf morphodynamics
,” in
Sediments, Morphology and Sedimentary Processes on Continental Shelves: Advances in Technologies, Research and Applications
, International Association of Sedimentologists Special Publication 44, pp.
3
28
.
5.
Demer
,
D. A.
,
Berger
,
L.
,
Bernasconi
,
M.
,
Bethke
,
E.
,
Boswell
,
K.
,
Chu
,
D.
,
Domokos
,
R.
,
Dunford
,
A.
,
Fässler
,
S.
,
Gauthier
,
S.
,
Hufnagle
,
L.
,
Jech
,
J.
,
N.
Bouffant
,
Lebourges-Dhaussy
,
A.
,
Lurton
,
X.
,
Macaulay
,
G.
,
Perrot
,
Y.
,
Ryan
,
T.
,
Parker-Stetter
,
S.
,
Stienessen
,
S.
,
Weber
,
T.
, and
Williamson
,
N.
(
2015
). “
Calibration of acoustic instruments
,”
ICES Cooperative Research Report 326
, p.
133
.
6.
Fonseca
,
L.
, and
Mayer
,
L.
(
2007
). “
Remote estimation of surficial seafloor properties through the application Angular Range Analysis to multibeam sonar data
,”
Mar. Geophys. Res.
28
(
2
),
119
126
.
7.
Foote
,
K. G.
,
Chu
,
D.
,
Hammar
,
T. R.
,
Baldwin
,
K. C.
,
Mayer
,
L. A.
,
Hufnagle
,
L. C.
, Jr.
, and
Jech
,
J. M.
(
2005
). “
Protocols for calibrating multibeam sonar
,”
J. Acoust. Soc. Am.
117
(
4
),
2013
2027
.
8.
Greenaway
,
S.
, and
Rice
,
G.
(
2013
). “
A single vessel approach to inter-vessel normalization of seafloor backscatter data
,” in
Proceedings of the US Hydrographic Conference 2013
,
New Orleans, LA
(March 24–28).
9.
Jackson
,
D.
, and
Richardson
,
M.
(
2007
).
High-frequency Seafloor Acoustics
(
Springer Science & Business Media
,
New York, NY
).
10.
Kinsler
,
L.
,
Frey
,
A.
,
Coppens
,
A.
, and
Sanders
,
J.
(
2000
).
Fundamentals of Acoustics
, 4th ed. (
Wiley & Sons
,
New York
),
461
pp.
11.
Lanzoni
,
J. C.
, and
Weber
,
T. C.
(
2010
). “
High-resolution calibration of a multibeam echo sounder
,” in
OCEANS 2010
, pp.
1
7
.
12.
Lurton
,
X.
, and
Lamarche
,
G.
(eds.). (
2015
). “
Backscatter measurements by seafloor-mapping sonars
,”
Guidelines and Recommendations
,
200
pp. http://geohab.org/wp-content/uploads/2014/05/BSWG-REPORT-MAY2015.pdf (Last viewed March 1, 2017).
13.
Pickrill
,
R. A.
, and
Todd
,
B. J.
(
2003
). “
The multiple roles of acoustic mapping in integrated ocean management, Canadian Atlantic continental margin
,”
Ocean Coastal Manage.
46
(
6
),
601
614
.
14.
Rattray
,
A.
,
Ierodiaconou
,
D.
,
Monk
,
J.
,
Versace
,
V. L.
, and
Laurenson
,
L. J. B.
(
2013
). “
Detecting patterns of change in benthic habitats by acoustic remote sensing
,”
Mar. Ecol. Prog. Ser.
477
,
1
13
.
15.
Todd
,
B. J.
,
Fader
,
G. B.
,
Courtney
,
R. C.
, and
Pickrill
,
R. A.
(
1999
). “
Quaternary geology and surficial sediment processes, Browns Bank, Scotian Shelf, based on multibeam bathymetry
,”
Marine Geol.
162
(
1
),
165
214
.
16.
Urick
,
R.
(
1983
).
Principles of Underwater Sound
, 3rd ed. (
McGraw-Hill
,
New York
), p.
23
.
17.
Weber
,
T. C.
, and
Ward
,
L. G.
(
2015
). “
Observations of backscatter from sand and gravel seafloors between 170 and 250 kHz
,”
J. Acoust. Soc. Am.
138
(
4
),
2169
2180
.