Silent research vessels are not quiet

Egil Ona, Olav Rune Godø, Nils Olav Handegard, Vidar Hjellvik, Ruben Patel, and Geir Pedersen

Citation: The Journal of the Acoustical Society of America 121, EL145 (2007); doi: 10.1121/1.2710741
View online: https://doi.org/10.1121/1.2710741
View Table of Contents: http://asa.scitation.org/toc/jas/121/4
Published by the Acoustical Society of America

Articles you may be interested in

Comment on “Silent research vessels are not quiet” [J. Acoust. Soc. Am. 121, EL145–EL150]
The Journal of the Acoustical Society of America 123, 1831 (2008); 10.1121/1.2839134

Radiated noise characteristics of a modern cargo ship

The Journal of the Acoustical Society of America 107, 118 (2000); 10.1121/1.428344

Underwater radiated noise from modern commercial ships

The Journal of the Acoustical Society of America 131, 92 (2012); 10.1121/1.3664100

Underwater radiated noise measurements of a noise-reduced research vessel: comparison between a US Navy
noise range and a simple hydrophone mooring.
Proceedings of Meetings on Acoustics 12, 070003 (2011); 10.1121/1.3660550

Acoustic Ambient Noise in the Ocean: Spectra and Sources

The Journal of the Acoustical Society of America 34, 1936 (1962); 10.1121/1.1909155

Noise characteristics of Japanese fisheries’ research vessels

The Journal of the Acoustical Society of America 103, 3036 (1998); 10.1121/1.422582
Ona et al.: JASA Express Letters 关DOI: 10.1121/1.2710741兴 Published Online 15 March 2007

Silent research vessels are not quiet

Egil Ona, Olav Rune Godø, Nils Olav Handegard, Vidar Hjellvik,
Ruben Patel, and Geir Pedersen
Institute of Marine Research, P.O. Box 1870 Nordnes, 5817 Bergen, Norway
egil.ona@imr.no, olavrune@imr.no, nilsolav@imr.no, vidar.hjellvik@imr.no,
ruben.patel@imr.no, geir.pedersen@imr.no

Abstract: Behavior of herring 共Clupea harengus兲 is stimulated by two

ocean-going research vessels; respectively designed with and without regard
to radiated-noise-standards. Both vessels generate a reaction pattern, but,
contrary to expectations, the reaction initiated by the silent vessel is stronger
and more prolonged than the one initiated by the conventional vessel. The
recommendations from the scientific community on noise-reduced designs
were motivated by the expectation of minimizing bias on survey results
caused by vessel-induced fish behavior. In conclusion, the candidate stimuli
for vessel avoidance remain obscure. Noise reduction might be necessary but
is insufficient to obtain stealth vessel assets during surveys.
© 2007 Acoustical Society of America
PACS numbers: 43.80.Nd, 43.80.Ev, 43.30.Sf, 43.40.Rj 关CM兴
Date Received: December 12, 2006 Date Accepted: January 11, 2007

1. Introduction
Vessel-induced fish behavior during acoustic density estimation1,2 and trawl sampling3,4 may
bias survey estimates of stock abundance. As noise has been considered a major stimulus, fish-
eries research institutions worldwide are investing in new silent research vessels5 in accordance
with recommendations from the International Council for the Exploration of the Sea 共ICES兲.6
The lack of fish avoidance observed from a stealth vessel7 has been considered a result of re-
ducing vessel noise,7,8 but no direct comparison with a traditional research vessel has demon-
strated this. Nevertheless, the scientific community has tacitly accepted that the major avoid-
ance stimulus originates from the sound characteristics of the vessels, which is also the basis for
the ICES recommendations.1,6,9
In 2003 the new Norwegian diesel-electric propulsioned research vessel “G. O. Sars”
共GS兲 关Gross Registered Tonnage 共GRT兲 4067 tons, Length Overall 共LOA兲 77, 5 m兴, fulfilling
the ICES demands for a silent vessel, was put into operation. Several earlier reports have docu-
mented vessel avoidance of Norwegian spring spawning herring 共Clupea harengus兲, showing
that this stock is underestimated acoustically when distributed in the upper 100 m of the water
column.1,2 Currently the acoustic observations are used to establish an abundance index. As-
suming similar or randomly varying conditions for observation among years, the time series of
indices gives a relative change in abundance from one year to the next that is utilized in the
stock evaluation. A vessel comparison was therefore an absolute necessity before the new GS
could be used in data collection for the official annual stock assessments. Vessel comparison
experiments were carried out in December 2004 in the Ofotfjord in northern Norway between
GS and the previous standard vessel “Johan Hjort” 共JH兲 共GRT 1828 tons, LOA 64, 4 m兲. JH is
a traditional research vessel with sound emission above the ICES standard6 and thus far noisier
than GS.
2. Methods
The two vessels followed each other at standard cruising speed and maximum distance along
the exact same triangular cruise track 关Fig. 1共a兲兴. Both vessels collected acoustic data according
to the standard protocol for acoustic surveys, and, along the pursue track, an upward-looking
echosounder and an acoustic Doppler current profiler 共ADCP兲 were placed. This allowed us to
record herring density by depth as well as the mean swimming velocity of the fish layer during

J. Acoust. Soc. Am. 121 共4兲, April 2007 © 2007 Acoustical Society of America EL145
Ona et al.: JASA Express Letters 关DOI: 10.1121/1.2710741兴 Published Online 15 March 2007

Fig. 1. Results from the pursuit experiment and mooring passages. 共a兲 The numerical-density vessel-transects
integrated over intervals of 1 nmi, shown as numbered light and dark sequences of points. Each point indicates the
start of a 0.1 nmi sampling distance. Gray points indicate bottom depth 共lightest: about 60 m, darkest: about 520 m兲.
The cross at sampling distance no. 11 indicates the position of the moored platforms. Arrows indicate cruising
direction. 共b兲, upper panel. Numerical density along the transects. Each point represents the average of the log-
transformed numerical densities for one vessel at one interval. The error bars show ±2 standard errors of the
averages. 共b兲 lower panel: The corresponding vessel differences in numerical density, and the number of passages for
each vessel. Dashed vertical lines indicate the corners of the triangle in 共a兲. 共c兲 Vertical swimming velocity compo-
nent when GS 共red curves兲 or JH 共blue curves兲 passed the moored ADCP. Thin lines denote single passages, and
thick lines denote the averages over all passages. Vertical dotted lines are drawn 2 min before and after the point of
passage. 共d兲 Vertical fish distribution 共median depth兲 when GS 共red curves兲 or JH 共blue curves兲 passed the moored
echosounder. Thin lines denote single passages and thick lines denote the averages over all passages. Vertical dashed
lines are drawn 2 min before and after the point of passage. 共e兲 Average sA when GS 共red curves兲 or JH 共blue curves兲
passed the moored echosounder. Thin lines denote single passages and thick lines denote the averages over all
passages. Vertical dotted lines are drawn 2 min before and after the point of passage.

and between passages of the two vessels. The mooring was passed four times by JH and three
times by GS at moderately dense recordings of herring at 40– 80-m depth during the night. The
two vessels were completely darkened during the experiments. First, we present the method and
protocol for collecting and analyzing the data from the vessel-mounted echosounders. Then, the
method to collect and analyze the data from the bottom-mounted platforms is presented, includ-
ing noise level confirmation for the vessels.
Raw echosounder data were recorded using the Simrad EK500 共Kongsberg Gruppen,
Kongsberg, Norway兲, 18 kHz on JH, and the Simrad EK60, 18 kHz on GS. The calibrated raw
data were directly transferred to the format of the postprocessing system BEI, and scrutinized
by standard procedures for herring surveys by the same two operators. In fjord surveys with
high densities, this basically involves the removal of bottom detection errors and isolating the
herring layers by integration boundaries. The processed herring density data were stored in a
database with 0.1 nmi⫻ 10-m depth bins in absolute, linear units for the “nautical area scatter-
ing coefficient,” sA 共m2 / nmi2兲, a standard unit in fisheries acoustics.10
The sA values from the two vessels were compared using a standard method,11 slightly
modified since the method is designed for two vessels following parallel transects, whereas in
our experiment each part of the survey transect was traversed 3–4 times by each vessel. Also, we
have chosen to replace ln共x兲 in the published method with 10 log共x兲. The modified method is as

EL146 J. Acoust. Soc. Am. 121 共4兲, April 2007 Ona et al.: Silent research vessels are not quiet
Ona et al.: JASA Express Letters 关DOI: 10.1121/1.2710741兴 Published Online 15 March 2007

Fig. 2. Noise signatures of JH 共blue兲 and GS 共red兲, respectively, measured in the keel aspect of both ships using the
Naxys hydrophone. The maximum recommended levels of noise from a free-running survey vessel at any speed up
to and including 11 knots according to ICES CRR 2096 is also included in the figure 共black curve兲. Due to a possible
inaccuracy in the hydrophone positioning, the absolute levels may be 3 dB off. This does not affect the relative
difference between the vessels.

follows: Let ␳i be the true fish density at elementary sampling distance unit i, and xij
= ␣j␳i10␴␧ij the density measured at sampling distance i by vessel j, j = 兵1 , 2其, averaged over all
passages 共there is no visible trend in the measured densities over the time period used in the
analysis兲. Here, ␣j is a vessel-dependent bias, ␴2 is the variance, and ␧ij is random noise. De-
fining di = 10 log共xi1兲 − 10 log共xi2兲, we have di = ␦ + ␴␧i⬘, where ␦ = 10 log共␣1 / ␣2兲 and ␧i⬘
= 10共␧i1 − ␧i2兲. Testing the null hypothesis that H0 : ␣1 = ␣2 is equivalent with testing H0 : ␦ = 0,
and if the di are independent, this can be done using a two-sided t-test. The ratio ␣1 / ␣2 is
¯
estimated by 10d/10, where d̄ = 兺idi.
The horizontal resolution of the data is 0.1 nmi, but using this resolution in the analy-
sis would yield autocorrelated di, and the t-test could not be used. We have therefore aggregated
the data so that each sampling distance is 1.0 nmi 关Fig. 1共a兲兴. This has another advantage as
well: the sampling distances for the two vessels do not match exactly, but the mismatch is
smaller relative to the sampling length when this is increased 共or resolution is decreased兲.
At a selected position in both surveys 关Fig. 1共a兲兴, the vessels passed directly above a
bottom-moored platform, carrying a calibrated, upward-looking EK60, 38 kHz echosounder,
and a calibrated underwater hydrophone 共Naxys A/S, Bergen, Norway兲 with a computer for
digital sound recordings. The hydrophone was used to verify the noise levels from the two
vessels 关Fig. 2兴. An upward-looking bottom-moored 75-kHz acoustic Doppler current profiler
共Teledyne RD Instruments, San Diego兲 50 m perpendicular to the track line was used to mea-
sure the swimming speed of the herring layer. The raw backscattering data from all of the four
ADCP beams were used to create a mask to isolate the herring layers from the surrounding
water, and to calculate the resulting vertical and horizontal velocity components.
Avoidance reactions were analyzed using data from seven vessel passages over the
moored platforms, four by JH and three by GS, including data from 10 min before each passage
to 10 min after. The acoustic backscatter from the moored echosounder was integrated over the
fish layer and is presented in sA units. Herring depth distribution and sA were available with a
resolution of about 3 pings per second, and ADCP data with one recording every fifth second.
To remove some of the random noise, the depth and sA data were smoothed using a moving
average with a window of 10 s. The ADCP data were smoothed using a window of 15 s for the
vertical component and 25 s for the horizontal component. Vessel differences and effects of
vessel passages were tested for using two-sided t-tests with n = 3 共JH兲, 4 共GS兲, or 3 + 4 共both

J. Acoust. Soc. Am. 121 共4兲, April 2007 Ona et al.: Silent research vessels are not quiet EL147
Ona et al.: JASA Express Letters 关DOI: 10.1121/1.2710741兴 Published Online 15 March 2007

Fig. 3. The echogram for a single passage over the moored echosounder when passed by GS 共a兲 and JH 共b兲,
respectively. The black line is the median depth distribution for this passage, and the blue and red lines are the means
of the median depth distributions for all passages for JH and GS, respectively.

vessels兲. For example, when testing for a vessel difference in vertical displacement, the differ-
ence DJH,i = q̄50,i,before − q̄50,i,after between the median depths averaged over the time periods
5 – 2 min before and 2 – 5 min after passage i , i = 兵1 , . . . , 4其, was calculated for JH, and similarly,
DGS,j , j = 兵1 , . . . , 3其 was calculated for GS. A difference between DJH and DGS was then tested for
using a standard t-test with n = 3 + 4, not assuming equal variance.
3. Results and discussion
First, we show that the herring numerical densities measured from the two vessels were similar.
The average numerical density recorded by GS during the pursuit experiment was 97% of the
average recorded by JH, which is a nonsignificant difference 共p = 0.75 under the null hypothesis
of no vessel differences兲.
Next, we show that GS initiated a more intense and prolonged avoidance reaction than
JH by analyzing the results from the moored instrumentation. The vertical mean swimming
velocity as estimated by the ADCP during the period of passage 共from 2 min before to 2 min
after兲 was significantly higher for GS 共−70.5 cm/ s兲 than for JH 共−42.7 cm/ s兲 关t-test, p
= 0.009, n = 3 + 4, Fig. 1共c兲兴. For GS the mean vertical velocity of the fish layer corresponds to

EL148 J. Acoust. Soc. Am. 121 共4兲, April 2007 Ona et al.: Silent research vessels are not quiet
Ona et al.: JASA Express Letters 关DOI: 10.1121/1.2710741兴 Published Online 15 March 2007

2 – 3 bl s−1 共body lengths per second兲, while the maximum recorded corresponds to about
4 bl s−1. Since the horizontal velocity component is not taken into account, and as variation is
expected for individuals over the depth range of the layer, this is a strong reaction compared to
the maximum individual swimming speed for herring of 6 – 7 bl s−1.12 The horizontal swim-
ming speed in the same period was similar for the two vessels.
From the moored echosounder 共Fig. 3兲, the diving is observed as a vertical displace-
ment of the herring layer 关Fig. 1共d兲兴. From about 2 min before passage and to 2 min after pas-
sage a median displacement of about 20 and 40 m is seen for JH and GS, respectively 关Fig. 1共d兲
and Fig. 3兴. The difference between the vessels in terms of change in median depth distribution
from 5 – 2 min before passage to 2 – 5 min after passage is significant on a 1% level 共t-test, n
= 3 + 4, p = 0.003兲. The fish also needed more time to return to its original distribution after being
disturbed by GS 关Fig. 1共d兲兴.
The numerical density when passing the mooring was on average similar to that before
vessel passage 关Fig. 1共e兲兴. The variability in numerical density was large during the experiment,
and it is difficult to separate vessel-induced effects from natural variations on a detailed level
based on only seven passages. Nevertheless, there was a clear decrease in numerical density
after the vessels had passed. Comparing the intervals 2 – 5 min before passage and 2 – 5 min
after passage, the decrease is significant on a 5% level both for GS 共t-test, p = 0.022, n = 3兲 and
JH 共t-test, p = 0.045, n = 4兲.
4. Conclusion
Although the moored echosounder did not record any significant difference in numerical den-
sity before and during passage for any vessel, the fish reaction pattern is strong and clearly
vessel dependent. The differences in behavioral response and the magnitude of the response
demonstrate the potential to cause severe bias, particularly seen in the perspective of earlier
experience.1,2 This illustrates the complexity of the vessel avoidance behavior, but more impor-
tant, the results show that a stimulus other than noise, as defined by ICES, must be responsible
for the reaction. Silent vessels have many advantages for reliable acoustic surveys, e.g., improv-
ing signal-to-noise ratio. Reducing the vessel noise may be necessary but is not a sufficient
measure, and as long as influential candidate stimuli for fish avoidance remain obscure, the
ICES goal of a establishing a stealth vessel design appears unrealistic.
Acknowledgments
This work was partially supported by grants from the Norwegian Research Council. We are
grateful to the engineers Ronald Pedersen, Ingvald Svellingen, Atle Totland, and Terje Torkelsen
who established and operated the instrumentation. Captains and crews on G. O. Sars and Johan
Hjort are thanked for their patience and cooperation during the experiments.
References and links
1
K. Olsen, J. Angell, F. Pettersen, and A. Løvik, “Observed fish reactions to a surveying vessel with special
reference to herring, cod, capelin and polar cod,” FAO Fish. Rep. 300, 131–138 共1983兲.
2
R. Vabø, K. Olsen, and I. Huse, “The effect of vessel avoidance of wintering Norwegian spring spawning
herring,” Fish. Res. 58, 59–77 共2002兲.
3
E. Ona and O. R. Godø, “Fish reaction to trawling noise: The significance for trawl sampling,” Rapp. P.-V. Reun.-
Cons. Int. Explor. Mer 189, 159–166 共1990兲.
4
N. O. Handegard and D. Tjøstheim, “When fish meets a trawling vessel: Examining the behaviour of gadoids
using a free floating buoy and acoustic split-beam tracking,” Can. J. Fish. Aquat. Sci. 62, 2409–2422
共2005兲.
5
J. Moore, “Stealth ship sets sail for a quiet life fishing for data,” Nature 共London兲 423, 7 共2003兲.
6
R. B. Mitson, “Research vessel standards: Underwater noise of research vessels, Review and
Recommendations,” ICES Co-op. Res. Rep. 209, 1–61 共1995兲.
7
P. G. Fernandes, A. S. Brierley, E. J. Simmonds, N. W. Millard, S. D. McPhail, F. Armstrong, P. Stevenson, and
M. Squires, “Fish do not avoid survey vessels,” Nature 共London兲 404, 35–36 共2000兲.
8
P. G. Fernandes, A. S. Brierley, E. J. Simmonds, N. W. Millard, S. D. McPhail, F. Armstrong, P. Stevenson, and
M. Squires, “Fish do not avoid survey vessels 共addendum兲,” Nature 共London兲 407, 152 共2000兲.
9
R. B. Mitson and H. P. Knudsen, “Causes and effects of underwater noise on fish abundance estimation,”
Aquat. Liv. Res. 16, 255–263 共2003兲.
10
D. N. Mac Lennan, P. G. Fernandes, and J. Dalen, “A consistent approach to definitions and symbols in

J. Acoust. Soc. Am. 121 共4兲, April 2007 Ona et al.: Silent research vessels are not quiet EL149
Ona et al.: JASA Express Letters 关DOI: 10.1121/1.2710741兴 Published Online 15 March 2007

fisheries acoustics,” ICES J. Mar. Sci. 59, 365–369 共2002兲.

11
R. Kieser, T. J. Mulligan, N. J. Williamson, and M. O. Nelson, “Intercalibration of two echo integration
systems based on acoustic backscattering measurements,” Can. J. Fish. Aquat. Sci. 44, 562–572 共1987兲.
12
J. H. S. Blaxter and W. Dickson, “Observations of the swimming speeds of fish,” J. Cons., Cons. Int. Explor.
Mer 24, 472–479 共1959兲.

EL150 J. Acoust. Soc. Am. 121 共4兲, April 2007 Ona et al.: Silent research vessels are not quiet