THE JOURNAL, OF THE ACOUSTICAL, SOCIETY OF AMERICA VOLUME 34.

NUMBER 12 DECEMBER 1962

AcousticAmbient Noise in the Ocean: Spectraand Sources

Gom)o• M. WE•z

U.S. Navy Electronics

Laboratory,San Diego52, California
(ReceivedMay 25, 1962)

The resultsof recentambient-noise investigations,

after appropriate processing,are compared on the
basisof pressurespectrain the frequency band1 cpsto 20 kc. Severalpossible sources are discussedto
determinethemostprobable originof the observednoise.It is concluded
that, in general,
theambientnoise
is a compositeof at least three overlapping components: turbulent-pressure
fluctuationseffectivein the
band1 cpsto 100cps;wind-dependent noisefrombubbles andsprayresulting, primarily,fromsurface agita-
tion,50cpsto 20kc; and,in manyareas,oceanic traffc,10cpsto 1000cps.Spectrum characteristics
ofeach
component and of the composite are shown.Additionalsources, includingthoseof intermittentand local
effects,
arealsodiscussed.Guidelines for theestimation of noiselevelsaregiven.

INTRODUCTION
decreased
as the frequency
decreased,
and at 100cps
HE summary
workof Knudsen,
Alford,and and below,little or no dependence
wasseen;whileother
Emling
•.2discussed observershave reported a substantialwind-speed
the natureof underwater
acousticambient noise in the frequencyrange from dependence
extendingto frequencies
as low as 50 cps.
100cpsto 25 kc. While a few resultsfromremoteopen- As might be expected,differingprocedures have been
oceanareas were available, a large part of the source usedin obtainingand processing data, and in defining
results.Data from the varioussources
data for their stud)' was taken in off-shoreareas,and in cannotalwaysbe
the vicinity of ports and harbors.Three main sourcesof compareddirectly but must be given additionaltreat-
underwaterambientnoisewereidentified:watermotion, ment in many cases.
includingalsothe effectsof surf,rain, hail, and tides; This reviewis designedto bring togetherfor com-
manmadesources, includingships;and marine life. The parison,after beingappropriately processed,the results
"Knudsen"curvesshowingthe dependence of recentinvestigations;
of the noise to showthat manyof the ob-
from water motion on wind force and sea state are well serveddifferences aswellassimilaritiescanbeexplained
known.Increasedlevelsdueto nearbyshippingand in- by certainplausibleassumptions asto sourceandsource
dustrialactivity havebeenobserved.For severalmarine- characteristics; and to indicatehow to apply this in-
a-se.g., snappingshrimpand croakers,the formationin estimatingthe ambient-noise
life sources, levelsfor a
noisecharacteristics and the timesandplacesof occur- given situation.
rence have been indicated.
A number of ambient-noise studies has been made 1. AMBIENT-NOISE SPECTRA

since1945,including
someinvestigation
of thefrequency
The mainpurposeof thispaperis to discuss the more
rangebelow 100 cpsand someadditionalmeasurements
widespreadand prevailingcharacteristics of ambient
in deep-wateropen-ocean areas.A great deal of the
underwater ambient-noise information is the result of noisein the ocean.Obviousnoisefrom marine life,
nearbyships,and other sourcesof intermittent and local
investigationsconductedby U.S. Navv Laboratories, noise is not included in the data considered in this
and by universityand commerciallaboratoriesoperat- section.
ing under contractwith governmentagencies,usually
with the Office of Naval Research. As has been shown,•.2 in the absenceof soundsfrom
ships and marine life, underwater ambient-noiselevels
While mostof the later resultshave beenin general are dependent on wind force and sea state, at least at
agreementwith the Knudsenel al. data, thereappearto
frequencies between100 cpsand 25 kc. Therefore,wind
be somesignificantdifferences from the earliersummary
dependence wasmadethe startingpointfor theanalysis
dataandamongtherecentdata.For example,aswill be
shown, in several studies it •vas observedthat, at fre-
of the resultsof recentinvestigations.
quenciesbelow 500 cps, the dependenceof the under- The processing of data reportedin differingterms
water ambient-noiselevelson wind speedand seastate included the following: conversionof levelsto dB re
0.0002dyn/cm• and a 1-cpsbandwidth;estimationof
1 v. O. Knudsen, R. S. Alford, and J. W. Eraling, "Survey of wind force from stated sea states (see Table I); the
UnderwaterSound,Report No. 3, Ambient Noise," 6.1-NDRC- derivationof spectracorrespondingto the meansof the
1848 (September26, 1944) (P B 31021).
'-'V. O. Knudsen, R. S. Alford, and J. W. Emling, J. Marine Beaufort-scalewind-speedranges,fromgivenequations
Research7, 410 (1948). or graphs relating level to wind speedat various fre-
a E. O. Hulburt, J. Acoust.Soc.Am. 14, 173 (1943). quencies;and the computationof averagespectrum
4 D. P. Loye and D. A. Proudfoot, J. Acoust. Soc. Am. 18, 446
(1946). levelscorresponding to Beaufort-scale
groupings.
When
• M. W. Johnson,F. A. Everest, and R. W. Young, Biol. Bull. the samplingwassmall,graphicalsmoothingand inter-
93, 122 (1947).
• F. A. Everest, R. W. Young, and M. W. Johnson,J. Acoust. polation were often employed.
Soc. Am. 20, 137 (1948). Each datum point usedin determiningthe ambient-
1936
 ACOUSTIC AMBIENT NOISE IN THE OCEAN 1937

TAnr.E I. Approximaterelationbetweenscalesof wind speed,waveheight,and seastate.

Wind speed 12-h wind Fully arisensea

Beau- Range Mean Fetchb.½ Sea-
Sea fort knots knots Wave height'.b Wave height•.bDurationb.½ naut. miles state
criteria scale (m/sec) (m/sec) ft (m) tt (m) h (kin) scale
Mirror-like 0 < 1 0
(<0.5)
Ripples 1-3 2
I (0.5-1.7) (1.1)
Small wavelets 4-6 5 <1 <1
2 (1.8-3.3) (2.5) (<0.30) (<0.30) 1
I.arge wavelets, 7-10 8• 1-2 1-2 <10
scatteredwhitecaps 3 (3.4-5.4) (4.4) (0.30-0.61) (0.30-0.61) <2.5 (<19) 2
Small waves, 11-16 13« 2-5 2-6 10-40
frequentwhitecaps 4 (5.5-8.4) (6.9) (0.61-1.5) (0.61 1.8) 2.5 6.5 (19-74) 3
Moderate waves, 17-2l 19 5-8 6-10 40-100
many whitecaps 5 (8.5-11.1) (9.8) (1.5-2.4) (l.8 3.0) 6.5-1l (74-185) 4
l.arge waves,
whitecapsever)- 22-27 24[ 8-12 10-17 106-200
where, spray 6 (11.2-14.1) (12.6) (2.4-3.7) (3.0-5.2) 11 18 (185-370) 5

Heaped-up sea,
hlown spray, 28-33 30« 12-17 17-26 200•t00
streaks 7 (14.2-17.2) (15.7) (3.7 5.2) (5.2-7.9) 18-29 (370-740) 6

Moderatelyhigh, long 3440 37 17-24 26-39 400-700

waves,spindrift 8 (17.3-20.8) (19.0) (5.2-7.3) (7.9-11.9) 29-42 (740-1300) 7

The average height of the highest one-tlfird of the waves (significant wave height).
I';stimated from data given in U. $. Navy Hydrographic Office (Washington. D.C.) publications HO 604 (lOll) and HO 603 (1955}.
The minimum fetch and duration of the wind needed to generate a fully arisen sea.

noisespectrais an averageof severalsamplesfrom a maxima, the highestvalue occurringat a frequency

singlelocality. In many cases,shipbornesystemswere between400 and 800 cps.None of the other spectra
usedandusuallyonlya very fewsampleswereobtained demonstratesthis spectrumshape clearly, although
at eachof severalstationsin the samegeneralarea.The thereare suggestions of it in some,beingindicatedby a
mean spectra derived from such measurementscom- flattening between 200 and 1000 cps. The maxima
priseaveragevaluesof data from all stationsin the same appearingin Figs.1(b) and 1(d) andFigs.2(b) and2(d)
generalarea. occurat frequencies 100cpsand below,and the levelsin
the neighborhoodof the spectrummaximum are not
VIAl 20 cps to 104cps wind-dependent.
The wind-dependentaspectsof the ambient noise
The spectra* resultingfrom measurementsmade in
appearto be greatlyinfluencedby non-wind-dependent
five differentshallow-waterareasare shownin Fig. 1. components.In Ihe spectraat the right in each figure,
[Shallowwaterisdefinedaswaterlessthan 100fathoms little wind dependence is evident belowabout 200 cps
(183m) in depth.•lDeep-waterambient-noise spectra and the levelsof the non-wind-dependentnoiseare as
from five differentareasare presentedin Fig. 2. It is high as or higherthan the levelsshownfor the highest
evidentthat spectracorresponding to the samescaleof wind speedsin the left-handgraphsin eachfigure.The
wind speedor sea state can exhibit considerablediffer-
levels of this non-wind-dependentnoise decrease
encesin spectrumshapeand level. However, as is indi-
rapidly, 8 to 10 dB per octave, at frequenciesabove
catedby the arrangement
of the figures,groupings
can 100 cps.
bemadewithinwhichthe spectraare roughlythesame,
and between which distinct differences are evident. In the areasassociated with Figs. l(a) and l(c), a
relativelyhigh residualnoiselimits winddependence to
The wind-dependentspectra at the left in Fig. 1
wind speedsmore than 5 to 10 knots (Beaufort 2 to 3),
[parts (a), (c), and (e)• are characterizedby broad
dependingon frequency.
ßUnlessotherwiseindicated,all spectraare given in termsof
pressure-spectrumlevel in dB re 0.0002 dyn/cm •, the reference At frequenciesabove 500 cps, where the noiselevels
bandwidthbeing Icps as includedin the definition o1 the term showwinddependence in nearlyeverycase,the spectra
"spectrum level" in Sec. 2.8, American Standard Acoustical
Terminology,ASA-SI.l-1960 (American StandardsAssociation, have the samegeneralshapeand approacha spectrum
May 25, 1960). slope of about --6dB per octave above 1000 cps.
1938 GORDON Mo WENZ

GO

4O 40---

I lb - o. *. •'. 'o\
ß .. _-%•x4-0. •.
I •--..':,,,.,,%0. ,•
' I' %o.ø v•
I o?....
10: 103 104

• GD•-

Fro. 1. Shallow-water am-

bient-noise
spectra,showing
averagespectrumlevels for
4O each of several Beaufort-
scalewind-speedgroupings,
as measured in five different
areas. The dotted curves
define component spectra
accordingto an analytical
interpretation of the ob-
servedspectra.

m I0 I0'

$o

AVERAGEWiND SEA STATE

SYMBOL , SPEED,
KNOTS $CALE:•
37 7
40-- 6
4

13« 3

8% 2
5 I
2
2O
©- RESIDUAL
ß -- COINCIDENTDATA

ß :-- SEE ALSO TABLE I

IO 10•
FREQUENCY
- CPS

However, the shallow-waterlevels (Fig. 1) are in isassumed to havea spectrumwitha broadmaximum

deep- between
generalabout5 dB higherthan the corresponding 100cpsand10130 cps,likethose
in Figs.1(a),
water levels(Fig. 2) at the samefrequencyand wind l(c), and l(e). In general,
the spectrumof the non-
speed.The figures•how averagevalues. The variability wind-dependent
component
is assumed
to peakat 100
is such that the higher deep-waterlevelsfor a given cpsor lower,and to fall off steeplyabove100cps,as
condition are about the same as the lower shallow-water seenin Figs.l(b), l(d), 2(b), 2(d),a and2(e). As the
levels; that is, the distributionsoverlap. figures
illustrate,
quitereasonable combinationsofsuch
For frequencies between20 cpsand 10kc, the rangeof spectraresultin spectralike the observed
spectra.
the data in Figs. 1 and 2, the diverse ambient-noise In Figs.1(a) and1(c), thespectrum of theresidual
spectra can be explained by assumingvarious com- noise doesnotdecrease rapidlywithfrequency, andthe
binationsof a •vind-dependentcomponentand a non- winddependence isalteredat thehigherfrequenciesas
wind-dependentcomponent.The application of this wellas at lowfrequencies.
In thiscasethe observed
interpretationisdemonstrated
in Figs.1 and 2 by dotted 8Thepeaknear60cpsin Fig.2(d)isnotcaused
by a self-noise
curveswhichindicatethe probablespectraof the com- "hum"component fromsystem-power sources.
The maximum
ponentswhich have combinedin eachcaseto produce appearsto be real but may be accentuated
by variationsin the
response
of the measurement
systemnot revealed
by the cali-
the observedspectrum.The wind-dependentcomponent bration data.
 ACOUSTIC AMBIENT NOISE IN THE OCEAN 1939

60

I'tc,. 2. Deep-water am-

bient noisespectra,showing
average spectrumlevels for 40
each of several Beaufort-
scalewind-speedgroupings,
as measured in five different
areas. The dotted curves
define component spectra
accordingto an analytical
interpretation of the ob-
served spectra.

IO2 10: lO• IO 102 I0 • 10•

FREQUENCYCPS

8EAUFORT AVERAGE
WiND SEA
STATE
SYMBOL SCALE
,• 8
SPEED,
KNOTS
SCALE
*: 4O
I 37 7
o 5 19 4
•3!5 3
ß 3 8• 2
5 1

* COINCIDENT DATA

• SEEALSOTABLE1
I0 - I•: ' IO•
EREQUENCYCPS

spectraare a combinationof the wind-dependent spec- When measuredin the samearea, or even in the same
tra with the spectrumof a residual-noise component placewiththesamesystem
andat thesamewindspeed,
whichprevailsat the lowerwind speeds. there is often considerable variation in the observed
In Figs. l(e), 2(a), and 2(c), wind dependence ap- levelsof the wind-dependent
noiseas measuredat
pearsto be universal.Also,minimaor inflectionpoints differenttimes.Suchdifferencesare illustratedin Fig. 3,
appear in the spectrabetween100 and 500 cps.This whichshowsspectracomprising levelsaveragedover
spectrumshapesuggests the possibility,at least,of two two differenttime periodsat each of two different
differentwind-dependent sources or mechanisms. In the locations.While the spectraare of the samegeneral
s•-stem used for the measurements from which the data shape,for the samewindspeeds, the wind-dependent
in Figs. 2(a) and 2(c) were obtained, the hydrophones September levelsin Fig. 3(a) run about8 dB below
were not well isolated from the effects of surface fluctu- thosefor Januaryin Fig. 3(b), and thewind-dependent
ations,and there is a strongpossibilitythat the low- June-Julylevelsin Fig.3(c)areabout5 dB belowthose
frequencywind-dependent noiseis a form of system for September-October in Fig. 3(d). If it is assumed
self-noise.
Thisdoesnot applyto the datain Fig. 1(e), that the sourceof the wind-dependent noiseis in the
ho•vever. surfaceagitationresultingfromtheeffects of thewind,
 1940 GORDON M. WENZ
5O 5O

ß ß ....... ß ....

3O

Strp
LOGATION A

102 103
10 TTT,:,,,, , ,,,,,,
10 102 103

Fro. 3. Ambient-noise
spectra,
illustrating differences in the
-- ø-ø--ø•ø-ø'40
. t.-•• -•-••, averages of levels measured in the
same area and at the same wind
speeds,
butduringdifferentperiods
of time.

JUN'JUL [ I SEP-OCT I I -•

20
LOCATION
a/ LOCATION
B
•o 102 103 10 102 103
FREQUENCY
- CPS FREQUENCY
- CPS

BEAUFORT AVERAGEWIND SEA STATE

SYM60L SCALE SPEED,KNOTS SCALE
•
• 8 37 7
0 5 19 4
v 4 13t/2 3
ß 3 8« 2
o 2 5 1
e 1 2

ß -- RESIDUAL

•--SEE ALSOTABLE1

10o

JAN
ir[8
MAR
APR
NOV

• g60

• •40
I 10 162 103
FREQUENCY- CPS

Fro. 4. Low4requcncy
ambient-noise
spectra,comparing
the
averages of a number of measurements made in each of five differ- FIO. 5. Low frequencyambient-noise
spectra,comparing'the
entareas.Theopentriangles
andtheinvertedtrianglesrepresent averages of levelsmeasured
duringdifferentmonthsof theyearat
datatakenin thesamegeneralareabutat differentdepths. the same location.
 ACOUSTIC AMBIEXIT NOISE IN THE OCEAN 1941

WIND

WIND
FORCE
71011 [1[AUfOIIT
SCALE
BEAUFORT
SCARE

i
- C[11NCIDENT
DATA
Fw,. ½J. Low-frequency
ambient-noise
spectrashow-
ing win(I-dependence iu
very shallow water (less
than 25 fathoms or 46 ml at
two widely separated loca-
tions.

1 111 ID• I0 ] 10 lO' 1D]

EREflUENCYCPS

variationsof this uature are not entirely unexpected. 1.3 Minimum Levels
Windspeedaloneisonlya crudeandincomplete meas- The lowest levels encountered in the data available to
ureof the surfaceagitationwhichdepends alsoon such the author are shown in Fig. 7. The solid symbols
factor, as the duration, fetch, and constancy of the
representmeasurements made in an inland lake. The
wind,and its directionin relationto localconditionsof
remainderof the data pertainsto measurementsin the
swell, current, and, in near-shoreareas, topography. ocean. Data from the same set of measurements are
Subjective
estimates
of seastatearenotnecessarily
an connectedby dashedlines.Symbolsshownwith down-
improvement over wind speedas a measureof the ward-pointing
arrowsdesignate
equivalentsystem-noise
pertinent-urfaceagitation. levelswhichmark an upperlimit to the ambient-noise
levels existingat the time of the measurement.The
[-1.2] 1 cps to 100 cps solid curve defines levels which will almost always be
exceededby observedlevels.
The amount of data available from measurements at
As indicated by the individual sets of measurements,
very low frequencies
is relativel
b-small.No consistent it is unlikelythat the very low levelswill be encountered
wind dependence has beenreported,exceptfor very in every'part of the spectrumat the sametime. This may
shallowwater. Someobviouseffectsfrom nearby ship-
pinghavebeenobserved.
Exclusiveof theseobvious
effects.rather wide variationsin the very low-frequency
noiselevelshavebeenexperienced.
In Fig.4, theindividualcurvescompare
theaverages
of a number of measurements made in each of five
differentareas.All of thedatain Fig. 5 wereobtainedat
the .,amelocation,the different curvesshowingaverages
of datatakenduringdifferentmonthsof the)'ear.Wind
dependence in verb'shallowwater (lessthan25 fathoms g I %.%'- ....
or 46 m} at two widelyseparatedlocationsis demon-
:traled in Fig. 6.
Somegeneralizations can be made about the results ,%-:
shownin Figs.4 6. The very low-frequency noisemay o• '• ' •"'"•:>• I
(lifter in level by 20 to 25 dB from oneplace to another,
and from one time to another. The spectrumshape
below 10 cpsis nearIx'always the same,and has a slope Fro. 7. The empirlcailowerlimit of ambient-noisespectra(solid
of --8 to --10 dB per octave. Between 10 cps and I00 curve), as determined by the lowestof observedlevels.The solid
symbols refer to measurementsmade in an inland lake, the open
cps, the spectrumoften flattensand may even showa symbols to those io the ocean. Symbols with downward-pointing
broad maximum, but in someinstancesthe spectrum arrows designateequivalent system-noiselevels which mark an
slopeshowslittle or no changefrom the slopebelow upper limit to the ambient-noiselevel existing at the time of the
measurement. The sea state 0 curve from references 1 and 2 is
10 cps. shownfor comparison.
1942 GORDON M. WENZ

beinterpretedasanindicationthat thedifferentregions dyn/cmø'for a 1-cpsbandwidth,andf is the frequency

of the spectrumare dominatedby differentcomponents in cps.According to Eq. (I), thethermal-noise spectrum
which combineto producethe observedspectra,but hasa slopeof +6 dB/octaveand a level of -- 10dB at
which are not all at a minimum at the same time. 35 kc. At the upper frequencylimit of the ambient-
The solid curve is an estimate of the minimum levels noisedata shownin Fig. 7, around 20 or 30 kc, the
existing in the ocean. This estimate is probably high minimum ambient-noise levels are about the same as the
since the curve is to some extent an indication of the thermal-noise levels.It is obviousfrom Fig. 7, however,
state of the measurement art. In the results of one in- that, at frequencies
below 10 kc, even the lowestof the
vestigation,it wasstated that duringa periodof meas- observed ambient-noise levels are well above the
urementcovering44 h of data, 40% of the time the thermal-noise limit, and other noise sourcesmust be
noiselevelsat 200 cps were too low to measure.The found to explainthe observedspectra.
limiting equivalentsystem-noisespectrumlevel at 200
cpswas10 dB re0.0002dyn/cm•. The data represented 2.2 Hydrodynamic Sources
by the s3mabols with downward-pointing arrowsare A wide variety of hydrodynamicprocesses is con-
alsoan indicationof theneedfor improvedmeasurement tinually taking placein the ocean,evenat zeroseastate.
techniques. It is known that the radiation of sound often results
1.4 Interpretation from theseprocesses.
In the absenceof noisefrom marine life and nearby 2.2.1 Bubbles
ships,the underwaterambient-noise spectrumbetween
1 cps and 10 kc may be resolvedinto several over- An oscillatingbubble is an effectivesoundsource.
lapping subspectra:A low-frequency spectrumwith a Both free and forced oscillations of bubbles occur in the
--8 dB to --10 dB per octavespectrum-levelslope,in ocean, particularly in the surfaceagitation resulting
the range1 to 100cps;a "non-wind-dependent" spectrum from the effects of wind.
in the range10 cpsto 1000 cpswith a maximumbetween Pertinent information concerningthe radiation of
20 and 100cps and falling off rapidly above 100cps sound by air bubbles in water has been given by
(sometimes not observed); and a wind-dependent spec- Strasberg?n The soundpressuresassociatedwith the
trumin the range50 cpsto 10 kc with a broadmaximum highermodesof oscillationof the bubblesare negligible
between100 cpsand 1000cpsand a --5-dB- or --6-dB- so that only simplevolume pulsations(zeroth mode)
per-octaveslopeabove1000cps. needbe considered.In the caseof forcedoscillations,the
At lowfrequencies,the ambientnoiseis dominatedby soundenergytends to be concentratedat the natural
the component,or components,characterizedby a frequencyof oscillationof the zerothmodealso,but this
-- 8-dB- to -- 10-dB-per-octaveslope,whichextendsin tendencymay be altered if the frequenciesassociated
somecasesto frequencies ashighas 100cps.Above500 with the environmentalpressurefluctuationsare much
cps,wind-dependent noisenearlyalwaysprevails.The belowthenaturalfrequencyof oscillationof the bubbles.
frequencybandbetween10 cpsand 1000cps,beingthe The natural frequencyof oscillationfor the zeroth
regionof overlap,is a highlyvariableonein whicheach mode is
observedspectrum dependson a combinationof the f0 = (3vP•p-•) •'(2•rR0)-I, (2)
three overlappingcomponentspectra,each of which
where3' is the ratio of specificheatsfor the gas in the
may vary independentlywith time and place.
bubble, p• is the static pressure,p the density of the
Minimum levelsare determinedin somecases,mostly
liquid, and R0 is the mean radius of the bubble. The
in shallowwater, by local residual-noisecomponents,
amplitudeof the radiatedsoundpressureat a distance
suchas thoseillustratedin Figs. 1(a) and 1(c), whose d from the center of the bubble is
levelsexceedthelevelsof themoregenerallimitingnoise
indicatedin Fig. 7. po= 3?p•rod
-•, (3)
2. SOURCES OF NOISE IN THE OCEAN r0 being the amplitude of the zeroth mode of oscillation.
It is assumedthat the amplitude of the bubble oscilla-
2.1 Thermal Agitation tion is relatively small so that the various modes are
The effectsfrom the thermal agitation of a medium independentof each other.
determine a minimum noise level for that medium. For The natural frequencyis inverselyproportionalto the
the ocean,the equivalentthermal-noisesound-pressure bubblesize,and the radiatedsound-pressure amplitude
levelisgiven,for ordinarytemperatures
between0øand is directly proportional to the bubble-oscillation
ampli-
30øC,by the relationS: tude. There is a practical limit to bubblesize,and it is
quite probable,also,that in many casesthere wouldbe
L•-- I01-3-20logf, (1) a predominanceof bubblesof nearlyonesizeonly. It is
where L• is the thermal-noise level in dB re 0.0002 •oM. Strasberg,J. Acoust.Soc. Am. 28, 20 (I956).
n H. iV[. Fitzpatrick and M. Strasberg,David Taylor Model
9 R. H. Mellen, J. Acoust.Soc. Am. 24, 478 (1952). Basin Rept. 1269 (January I959).
 ACOLISTIC AMBIENT NOISE IN THE OCEAN 1943

to be expected, therefore,
that in generalthe spectrum spectrumslopeis 12dB per octave,but at high fre-
has a maximumat somefrequencyassociatedwith quencies thespectrumis determined by detailsof very
eithera predominantbubblesizeor a maximumbubble rapid changesin soundpressurewhichare not given
.qze,the exact shapedependingon the distributionof correctlybY the acoustictheory.
bubblesizesand amplitudesof oscillation. The noiseproducedby a stirringrod 2 in. longand
FranzTM has measuredthe soundenergy'radiated by • in. in diameterrotating at 4300rpm in the Thames
air bubbles formed when air is entrained in the water River (New London, Connecticut) was measuredby
followingthe impact of water dropletson the surfaceof Mellen.n His resultsare given in the form of a sound-
the water. His resultsare given in the form of one-ha]f- pressure spectrum which shows a maximum near
octave-bandsound-energyspectrawhich exhibit maxi- 1000cpsand slopeof approximately-- 6 dB per octave
ma. The declinetowardlowerfrequencies is sharp(8 to at higherfrequencies.
The spectraof noisefrom cavitat-
12dB per octave,in termsof energy-spectrum level) ing submergedwater jets as reportedby Jorgensen •.'•
and is attributed to an almost completeabsenceof showa slopeof approximately12 rib per octaveat low
bubbleslargerthan a certainsize.A moregradualde- frequencies, in agreementwith the aeonstictheory,and
cliuetowardhigherfrequencies (--6 dB to --8 dB per a slopeof about--6 dB per octaveat highfrequencies,
octave) was found and was interpretedas being the in agreementwith Mellen'sdata. Observedspectraof
resultof a decreasein the radiatedsoundenergyper noise radiated by submarinesexhibit characteristics
bubble rather than a decreasein the prevalenceof which are in generalagreementwith thesedata and
bubbles. which have been attributed to cavitation effects2 •
l)ata on bubble size and environmental conditions are
The spectrumshapeof cavitationnoiseis similar to
not availablein sufficientdetail for makingexact pre- that of the air-bubblenoise,which,as hasbeenpointed
dictionsconcerning the bubblenoisein the ocean.How- out, resemblesthe spectrumshapeof the wind-depend-
ever,a roughappraisalcan be made.Accordingto Eqs. ent ambientnoise(seeFigs. 1 and 2). For cavitiesof
(2) and (3), a sphericalair bubbleof meanradius0.33 comparablesize one would expecthigher noiselevels
cm, in water at atmosphericpressure,oscillatingwith from cavitation than from the simplevolumepulsations
an amplitudeone-tenththe mean radius (r0---0.1R0), of gas bubblessince the amplitude of oscillationis
has a simplesource-pressure leveP'• referredto 1 m, of usually greater.
about 133dB above0.0002dy'n/cm•-at a frequencyof From the foregoing,it is concludedthat air bubbles
approximately1000cps.For a frequencyof 500 cps,the and cavitation producedat or near the surface,as a
mean-bubbleradius is about 0.66 cm, and, for the same result of the action of the wind, could very well be a
amplitude-to-size ratio, the sourcelevel is 6 dB higher. sourceof the wind-dependentambient noise at fre-
These source levels are some 75 to 100 dB above the quenciesbetween50 cpsand 10 kc.
observedambient-noisespectrum levels at these fre• Bubblesare presentin the sea (or lakes)evenwhen
quencies.The noisefrom suchbubblesourcescouldbe the wind speedsare belowthat at whichwhitecapsare
observed at a considerable distance. The maxima in the produced.Bubblesare created,not only by breaking
observedwind-dependent ambient-noise spectra(see waves,but alsoby decay'ingmatter, fishbelchings,and
Sec.1.1and Figs.1 and 2) occurat frequen- gasseepage
preceding from the seafloor.Furthermore,
thereis
ciesbetween300 cpsand 1000cps,whichcorrespond
to evidence of the existence of invisible microbubbles in
bubble sizes of 1.1 cm to 0.33 cm in mean radius, a the sea,and of the occurrenceof gassupersaturationof
reasonableorder of magnitude. varying degreenear the surface.Theseconditionspro-
The characteristic broadness of the maxima in the vide a favorableenvironmentfor the growth of micro-
wind-dependent ambient-noisespectracan be explained bubble nuclei into bubblesas a result of temperature
by lhe reasonableassumptionthat in the surfaceagita- increases,pressuredecreases, and turbulenceassociated
tion the bubble size and energy distributionsare not with currentsand internal waves, as well as with surface
sharply concentratedaround the averages.The am- waves2s--øø As the bubblesrise to the surface,growingin
bient-noisehigh-frequencyspectrumslope above the size (becauseof the decreasing hydrostaticpressure),
maximum,approximately--6 dB octave,agreeswith they are subjectedto transientpressureswhich induce
that of the bubble noise. the oscillations
whichgeneratethe noise.Even on quiet
The nature of cavitation noise has been described by
• R. H. Mellen, J. Acoust.Soc.Am. 26, 356 (1954).
Fitzpatrick and Strasberg.n Accordingto the acoustic 'a D. W. Jorgensen,David Taylor Model Basin Rept. 1126
theory, the sound-pressure spectra have maxima at (November 1958).
• D. W. Jorgensen,J. Acoust.Soc.Am. 33, 1334 (1961).
frequenciescorrespondingapproximately'to the re- •* NDRC SummaryTech. Repts. Div. 6, Vol. 7, Principlesof
ciprocalof the time requiredfor gro;vthand collapseof Underwater:Sound,
:ffec.12.4.,5.(Distributedby ReaearchAnalysis
Group, Committee on Undersea Warfare, National Research
the vapor cavities.At low frequenciesthe predicted Council.)
•s E. C. LaFond and P. V. Bhavanarayana,J. Marine Biol.
•-•G. J. Franz, J. Acoust.Soc.Am. 31, 1080 (1959). Assoc.India 1, 228 (1959).
•aSource-pressure levelis definedas the soundspressure
levelat t• W. L. Ramsey,Limnologyand Oceanography7, I (1962).
a specifiedreferencedistancein a specifieddirectionfrom the •oE. C. LaFond and R. F. Dill, NEL TM-259 (1957) (un-
effective acoustic center of the source. publishedtechnicalmemorandum).
1944 GORDON M. WENZ

daysand in the absenceof wind, bubbleshave beenseen

to emergefrom the water, sometimespersistingfor a
time as foam, and then to burst. Theseoceanographic
data supportthe hypothesisthat bubblenoisemay still
be an important componentof underwater ambient
noise,evenwhenthereis little or no surfaceagitation
from the wind. FIG. g. Surface-wavepres-
sure-levelspectra,derivedfrom
Thus,thereis evidenceof the presence of bubblesin Neumann-Pierson surface-
theoceanbothwhenwindsarehighandwhenwindsare wave elevationspectra(seeref-
erences22 and 23).
low, or even during a calm. Oscillatingand collapsing
bubblesare efficientand relatively high-levelnoise
sources.Both the leveland shapeof the observedwind-
dependentambient noise can be explainedby the
characteristics of bubble noise and cavitation noise.

2.2.2 WaterDroplets
The underwaternoiseradiatedby a spray of water The spectraof surfacewaves,that is, the spectrum
dropletsat thesurfaceof thewaterhasbeeninvestigated densities of the time variation of the surface elevation
by Franz? The noisefromsuchsplashes appearsto be at a fixedpoint, accordingto Neumann• and Pierson, "-a
madeup of noisefrom the impactand passage of the are representedby the relation
droplet through the free surface. In many cases,air
bubbles are entrained so that the total noise includes /t•(co)
= Cco
-• exp(-- 2g-*co-•-v-'•), (4)
contribution from the bubble oscillations as •vell. The
where•.o(co)
is themean-square
elevation
of thesurface
sound-energyspectrum has a broad maximum near a per unit bandwidthat the angularfrequencyco,g is the
frequencyequalto twicethe ratio of the impactvelocity accelerationof gravity, and • is the wind speed.In cgs
to theradiusof thedroplets.Towardslowerfrequencies, units the constantC, determinedfrom empiricaldata,
the spectrumdensitydecreases graduallyat a rate of 1 is equal to 4.8X 10• cm•-sec-a.
or 2 dB per octave.At frequencies
abovethe maximum, Surface-waveelevationspectrafor windsof force3, 5,
the slopeapproaches--5 or --6 dB per octave.The and8 (about5, 10,and 20 m/sec)werecomputed using
impactpart of theradiatedsoundenergyincreases with Eq. (4). The spectrashownin Fig. 8 arein termsof the
increase
in dropletsizeandimpactvelocity.The relation pressure-spectrum levels correspondingto the mean
is modifiedsomewhat by the bubblenoise,particularly square of the variation in the surface elevation in
at intermediate velocities.
referenceto a 1-cpsbandwidth. The maximum of spec-
Franz estimatedthe sound-pressure spectrumlevels trum energy occursat frequenciesbelow 0.5 cps, and
to be expectedfrom the impactof rain uponthe surface the band of maximum energymovesto lower frequen-
of thewater.He concluded that rain exceedinga rate of ciesas wind speedincreases.
0.1 in./h wouldbeexpectedto raiseambient-noise levels Equation (4) appliesto a fulh' developedsea.When
and flatten the spectrumat frequencies above 1000cps the seais not fully developed,the high-frequencypart of
under sea-state-I conditions. The measurements of
the spectrumis unchanged,but the largerlow-frequency
ambientseanoisemadeby Heindsmannet al.*-tduring waveshave not yet beenproduced,and the spectrumis
periodsof rainfallare in fair agreementwith the esti- cut off at the lower end as roughly exemplifiedby the
matesmadeby Franz. dashedcurvein Fig. 8. The cutofffrequencydependson
The noisefromsplashes of rigid bodies,suchas from the duration and fetch of the wind.
hail or sleet,is in generalsimilar to that from water For frequenciesabove1 cps,the valueof the exponen-
droplets,but is modifiedby the effectsof resonantvibra- tial functionin Eq. (4) is ver\' nearly unity, and the
tions of the bodies.
spectrum densties decreaseas f -• (-- 18 dB peroctave).
In addition to the effectsof precipitation,there is the However, the relation was derived from measurements
possibility that noticeablecontribution to the ambient of the larger wavesof frequenciesbelow 0.5 cps, and
seanoisemay comefromsprayandspindrift,especially extrapolationto frequencies above0.5 cpsis not certain.
at the higherwind speeds. The higher frequencysurfacefluctuationsare in the
form of small gravity wavesand capillaries.Phillipsu
2.2.3 Surface Win'es discusses an equilibrium region for the small gravity
The fluctuations in the elevation of the surface of a wavesfor whichthe spectrumis givenby the relation
body of water causesubsurface pressurefluctuations
which, whethercontrolledbv compressibility or not, •- (co)
-.•7.4X 10-ag'•co
-a. (5)
affect the transducerof an underwater system. •' G. Neumann, Department of the Army, Corps of Engineers,
Beach Erosion Board Tech. Mere. No. 43 (December 1953).
2•T. E. Heindmnann, R. H. Smith, and A.D. Arneson, J. -*aWillard J. Pierson,Jr., Advancesin Geophysics2, 93 (1955).
Acoust. Soc. Am. 27, 378 (1955). • O. M. Phillips, J. Marine Research16, 231 (1957-1958).
 ACOUSTIC AMBIENT NOISE [N THE OCEAN 1945

Accordingto Eq. (5), the pressurelevel corresponding

to•(27r) (frequency
1 cps)is 132.6dB, essentially
the
sameas the corresponding valuesderivedfrom Eq. (4) 160
I
•ß .^ /'
'[' •
OCEAN
WAVE
*"MEASUREMENTSI
(seeFig. 8). A relationis givenfor capillaries
•5in which
the surface-displacement spectrumis proportionalto
co7/a.Details of the capillary-gravitywave systemare • •r 140 .....
not known,althoughCox• haspresentedevidencefrom
wave-slope observations whichsuggests that capillary
wavesbecomeimportantonly for wind speedsabove
6 m/sec (11 to 12 knots). •q 120
Kinsman" has smnmarized a nmnber of surface-wave
spectrumlneasurements andhisresultsindicatethat the
slopeis between-- 13.5and - 16.5dB per octavein the
frequencybandfrom0.7 to 2.1 cps. g OO AMBIENT•
NOISE• ',',
'
The first-orderpressurefluctuationsinducedby the [ MEASUREMENTS
surfacewavesare attenuatedwith depth, the attenu-
ation being frequency-dependent? • aoThe character-
isticsof the "depth filter" are shownin Fig. 9. The depth IO4 I0-: I IO
FREQUENCY-
filter isa low-passfilter with a sharpcutoffandgenerally
limits significantfirst-orderpressureeffectsfromsurface I:[G, 10. Pressurelevel spectracomparingresultsof oeean4v•ve
wavesto frequencies below0.2 to 0.3 cps,and to depths measurements (derived from •eference 31) and ambien[-noise
measurements.The dashedgutyesare extrapolations,
less than a few hundred feet.
In Fig. 10 low-frequency pressurespectraobtained
from "ocean-wave"measurements are comparedwith depth filter, it is doubtful that the ambientnoiseat
spectra resulting from shallow-water"alnbient-noise" frequencies above1 cpsincludesany significantcontri-
•neasurements. The ocean-wavespectrawere derived bution from the first-orderpressurefluctuationsinduced
fromdata reportedby Munk e! al.a•The ambient-noise by surfacewaves.At frequencies below0.3 cps,approxi-
spectrawereselectedfrom the sources for Fig. 6 (of this mately, thesepressurefluctuationswill very likelycom-
paper).The effectof the depthfilter is indicatedby the prise a large part of the "ambientnoise"observedat
ocean-wavespectra. Consideringthe difference in veryshallowdeplhs(<300 ft or 100m) with pressure
measurement methods and the variety of environmental transducers.
conditionsinvolved, the two sets of data merge re- Longuet-Higgins aø-hascalledattentionto a second-
markably well. orderpressurevariationwhichis not attenuatedwith
Becauseof the steepnegativeslopeof the surface- depth. These second-order pressurevariationsoccur
wave spectra,and becauseof the steep cutoff of the whenthe wavetrainsof the samewavelength travelin
oppositedirections.The resultingpressurevariation is
of twicethe frequencyof the two waveswith an ampli-
tude proportionalto the product of the amplitudesof
the two waves.When the depth is of the sameorderas,
or greater than, the length of the compression wave,
Fro. 9. Depth-filter char-
compression wavesare generated.
acteristics,showingthe attenu- The conditionsfor theLonguet-Higgins effectaremet
ation of first order pressure in the open ocean where the winds associatedwith a
fluctuations as a function of
frequency at three selected cyclonicdepression producewavestravellingin opposite
depths. directions.Opposingwavesoccurwhen wavesare re-
fiectedfrom the shore.The vagariesof localwindsmay
alsoproducethe requiredpatternsin the high-frequency
capillary-gravitywave system, which, though short-
lived, may be numerousand frequent.
One may postulatethat surfacewavesare a sourceof
2.•O. M. Phillips,J. Marine Research16, 229 (1957 1958). low-frequencyambient noise by way of these second-
•aCharlesS. Cox, J. Marine Research16, 244 (195%1958). orderpressurevariations.A COlnparison
of the observed
-'•Blair Kinsman, J. Geophys.Research66, 24ll (1961).
•-sS. Rauch, University of Calif. Department of Engineering, ambient-noise levels in Fig. 10 with the estimated
Fluid Mechanics
Lab.'I•ech.Rept.HE-116-191(November29, surface-wavespectrumlevelsin Fig. 8 indicatesthat the
1945).
m R. L. Wiegel, University of Calif. Department of Engineering,
pressurevariations are of sufficientmagnitude.The
Fluid MechanicsLab. Mere. HE 116-108 (September8, 1948). observedambient-noiselow-frequencyspectrumslope
•øW. H. Munk, F. F.. Snod•rass, and M. J. Tucker, Bull.
ScrippsInst. of Oceanog.Univ. Calif. 7, 283 (I959), Fig. 5. 5I. S. Longuet-Higgins,Phil. Trans. Roy. Soc. (London)
s• Reference30, charts 2.1, 4.1, and 5.1. A243, 1 (1950).
1946 GORDON M. WENZ

of -- 8 to -- 10 dB per octavecouldbe accountedfor by violent turbulence,suchas the caseof a turbulent jet,a7

the assumptionof a suitablecombinationof the capillary the level of the radiated pressurefluctuationsis low
and gravity wavesin the wavesystemfor the frequency comparedto ambient-noise
levels.It is concludedthat
range0.3 to 10 cps. the radiated noise from turbulence does not contribute
It is concludedthat the second-order pressurevarœ- to the observedambientnoise,exceptpossibly'under
ationsresultingfrom surfacewavesmay sometimes be a specificlocalconditions.
significantpart of the ambient noiseat frequencies The pressurefluctuationsof the turbulenceitself are
below10 cps (seeSec.1.2 and Figs.4-6). of muchgreatermagnitudethan thoseof the radiated
noise.asA pressure-sensitive
hydrophone TMin the turbu-
2.2.4 Turbulence lent regionrespondsto thesepressurefluctuationsas it
doesto an5' pressure
fluctuations,whethertheyare those
The state of turbulence is mainh- one of unsteady of propagatedsoundenergyor not.
flow with respectto bothtime and spacecoordinates. Accordingto experimentalresults and the generally
Whenthe fluid motionis "turbulent,"irregularitiesexist acceptedtheor5' of turbulence, aathe followingrelations
relativeto a point movingwith thefluid,as wellas rela- may be used for rough estimates of the turbulent
tive to a fixed point outsidethe flow. velocityand pressurefluctuations:
Turbulence may occur in a flukl as a result of current
flow alonga solidboundaryand alsowhen layersof the a•0.05 C, (6)
fluid with different velocities flow past or over one
aø-(k)• t•4 (k<<0.01cm •), (7a)
another. Turbulencemay be expectedin the oceanat
lhe water ocean-floorboundary,particularly in coastal ff-'(k)• k (k <0.01), (7b)
areas,straits, and harbors;at the seasurfacebecauseof
the movementand agitation of the surface;and within •"(k) = maximum (0.01<k<0.1), (7c)
the medium as a result of the horizontal and vertical a'-'(k)o•k-•ta (k>0.1), (7d)
water movements,such as advection,convection,and
density currents. ff•(k) • k* (k>>0.1), (7e)
Noise resultingfrom turbulencecreatedby relative f= kl?(2•r)-', (8)
motion between the water and the transducer is con-
sidered to be self-noise of the system rather than (9)
ambient noise in the medium.
The first of these relations states that, if and when
hluch of the energy input into the ocean occursat turbulenceexists,the rms turbulent velocity • is on the
frequenciestoo low to be of direct consequenceto the average
aboutfivepercentof themean-flow
velocity•?.
ambient noise at frequenciesabove 1 cps. In the pro- The general features of the turbulent-velocity spectra
cessesof turbulence,aathe largest-scaleeddiesand the are indicatedby the set (7a)-(7e), showingthe approxi-
lowestwavenumbers correspondto the regionof energy mate dependenceof the turbulent-velocityspectrum
input. The largest eddiesbreak up into smaller and densityfunction/•"k) on the wavenumberk in different
slnallereddies,someof the energybeing transferredto parts of the wavenumberrange.Equation (8) for the
higherand higherfrequencies.This is preciselythe kind frequencyf is a reminderthat it is the meanflaw velocity
of mechanismwhich could transfer low-frequency that relateswavenumberand frequencyin this case.An
energyin the oceanto higherfrequencies. estimateof the rms turbulent-pressure fluctuation may
However, the generationand radiation of noisefrom be derived from the mean-squareturbulent velocity
turbulenceare a very inefficientproce-s.Radiated noise accordingto expression(9) in whichp is the densityof
levelsderived from the relationsformulated by Light- the fluid.
hill,a4usingvaluesof turbulent-velocityfluctuationsand
Pochapsky
aahasestimatedthat the magnitudeof/• is
dissipation rates estimated by Pochapsky, a• for the
no morethan 2 cm/secfor the horizontalvelocitycom-
oceanare many ordersof magnitudebelowthe observed ponentgof the "ambient" oceanicturbulence.By Eq.
ambient-noiselevels.The levelscorrespondingto experi-
(6), the correspondingcurrent speed is 40 cm/sec
mentalvaluesfoundfor regionsof strongcurrentsin an
(0.8 knot). Theseestimatesare indicativeof an upper
inland passageby Grant et al.a• are alsolow by many limit to the backgroundturbulence.Flow rates of as
ordersof magnitude.Even whenthereis comparatively nmch as 200 cm/sec (3.9 knot) are observedin the
aaSeveralof the classicalpapersmay befoundin the book,Turb*t- swifter oceancurrents,and of 600 cm/sec (11.7 knot) or
lence,edited by S. K. Friedlander and L. Topper (Interscience more in straits and passageswhere very strong tidal
Publishers,Inc., New York, 1061). More recent experimentaland
theoretical resultsare included in the book,J. O. Hinze, in Turbu- a*RefereI•ce 11•p.268.
lence(McGraw-Hill Book Company, Inc., New York. 1959). a•The possiblesignificanceof the pressurefluctuationsnear and
s• M. J. Lighthill, Proc. Roy. Soc. (London) A222, 1 (1954). within the turbulent regionswas suggestedto the author by Paul
saT. E. Pochapsky,Columbia University Hudson Laboratories O. Laitinen.
Tech. Rept. 67 (March 1, 1959). • The so-called"velocity" transduceris not exceptedsincesuch
aaH. L. Grant, R. W. Stewart, and A. Moi!liet, Pacific Naval devicesare sensitiveto pressure
gradientsandwill respondto the
Laboratory,Esquimalt,B.C., CanadaRept. 60-8 (1960). pressuregradients of the turbulence.
 ACOUSTIC AMBIENT NOISE IN THE OCEAN 1947

decreaseswith depth and becomesnegligibleat a depth,

dependenton the surfacevelocity,of the orderof 45 to
>•12o 200 m (25 to 110fathoms).The speedof the elementar
current,therefore,doesnot changegreatly with depth
exceptnearthe surfaceand the bottom,and in shallow
water, where the bottom is near the surface.The char-
acteristics of the elementar current are modified by the
effects of density currents resulting from internal
gg forces.
sø'Vertical circulatorypatternsmay occurwhich
,•.• 80 producevelocitymaximaand minimabetweenthe sur-
face and the bottom?
The theoreticaland conjecturalaspectsof the pre-
ceding discussionhave experimental support. For
60--
example, measurements • made in water 45 m (25
fathoms) in depth showeda logarithmicdecreaseof
I0' I
FREQUENCY
I0
- CPS
I0 z velocitywith depth from about 40 cm.secto 10 cm sec
between160 cm and 20 cm above the bottom, and con-
FIG. 11.Turbulent-pressure-level
spectra,derivedfrom theoretical ditionsof turbulencewerefoundto exist.Recentdeep-
andexperimental
relations[seeEqs. (6) through(9)-]. water measuremenls • • made with the Swallow neu-
trally buoy'antfloat indicate that deep currents are
currentsare experienced. Corresponding rms turbulent fasterand morevariablethan wasanticipated,with no
velocitiesare 10 cm/sec and 30 era/sec.
evidencefor a decreasein speedwith depth. Average
Equations(6) through(9) wereusedto deriveturbu- speedsof 6 cm/sec were observedat 2000-m depth
lent-pressurespectrafor eachof the threeflow condi- (approximately 1100fatholns),whileat 4000-mdepth
tions mentionedin the precedingparagraph.These (2200fathoms)the averagewas12 cm sec,withasmuch
spectraare shownin Fig. 11 in terms of turbulent- as 42 cm/secbeingobservedin two cases. s?The exis-
pressurespectrumlevels.The estimatesare roughand tenceof eddieswith a typical diameterof as nmch as
variationsof at leastoneorderof magnitudeare prob- 185km (100 nauticalmiles) was impliedby the ob-
able.A comparison with thespectrain Figs.4 6 reveals servedfluctuationsof the deepcurrents.
that the estimatedoceanicturbulent-pressure spectra
FronsEqs. (6) and (9), onewouldordinarilyexpect
agreequite well in both slopeand level with the am-
the turbulent-pressure
levelsto increase
aml decrease
as
bient-noisespectrabelow I0 cps, and betweenI0 and the flow velocities increase and decrease. In the limited
100cpsin someinstances.
amount of low-frequencyambient-noisedata available,
For a hydrophone to respond to the turbulent-
there is no evidenceof consistentdepth dependence.
pressurefluctuationsit must be in the regionof turbu-
lence. There is evidence that the movement of the waterThis couldbe explainedin part by the probablevariety
of patterns in the vertical velocity structure of the
massesin the oceanis mostly chaoticand turbulent.so
currents.Sincethe speedof the drift currentdecreases
The dimensions of flow are nearlyalwayslargeso that
with depth,a decrease in turbulent-pressure levelwith
turbulentconditionsmay prevail,even thoughthe flow
depth shouldbe observedin shallowwater and at shal-
velocity is often small. Turbulenceis spreadout and
lowdepthsin deepwater,exceptwhenstrongdensit5'or
maintainedin the oceanvolumeby a continualsourceof
localcurrentsare present.
turbulent energy at the boundaries,particularly the
The characteristicsof the "drift" current, which is
sea-surfaceboundary.There is reasonto believe that
there exists in the ocean an essentiallyuniversal wind-dependent, could quite reasonablyexplain the
wind dependence of the ambientnoisein very shallow
"ambient" turbulencewhichvarieswidely in intensity
with both time and place. water, which is illustrated in Fig. 6, and also,possibly-,
Accordingto the "elementar"currenttheory,*• which the wind dependenceal frequenciesbelow 100 cps in
Fig. l(e).
assumes a homogenous ocean,the variationof velocity
with depth dependson the mutual effect of wind- Wheneverlocalboundariesare involved,particularly
induced"drift currents"and "gradientcurrents"which if there are sharp edgesand rough surfaces,the local
result from the pressuredifferencesproducedby sea- scalesof motionare generally'small,but the velocitiesof
surfaceslopes.The magnitudeof the pure gradient
currentis constantwith depth, exceptnear the bottom 4• Reference40, Vol. I, Chap. XV.
boundarywherefrictionalforcesdecreasethe magnitude 4aReference40, Vol. I, Chaps. XVI and XXI.
4• R. M. Lesser,Trans. Am. (leophys. Union 32, 207 (1951).
logarithmically'to zero. The speedof the drift current •sj. C. Swallow,Deep SeaResearch4, 93 (1957).
4• j. C. Swallow and L. V. Worthington, Nature (London) 179,
• Albert Derant, PhysicalOceanography (PergamonPress,New 1183 (1957).
York, 1961),in particular VoL I, part II. • M. Swallow, Oceanus (Woods Hole OceanographicInstitu-
•t Reference40, VoL I, p. 413. tion) VII (3), 2 (1961).
19'48

the localturbulencemay be large,and the localeffects • 60J-... .... TF SOURCE

SPECTRUM
SHAPE
may be intense.
A great deal of the low-frequencyambient-noisedata • o -- IOTO100GPS
has been acquired using systems employing fixed
bottom-mountedhydrophones. With suchsystemsit is
sometimesdifficult to separatethe self-noiseproduced
by turbulenceresultingfrom water motion past the
stationarytransducerfrom noisewhich is characteristic
of the medium. SimiLarconsiderationsapply to ship-
borne systemswhen differential drift between the ship
and the hydrophonecausesthe hydrophoneto be towed
by the ship,or whenany forces,suchasbuoyancyand
gravity, producerelative motion between the hydro-
phoneand the water.
On the basisof spectrumshapeand level, there is
goodsupportfor the h.xpothesisthat one componentof 1D 10J 1DI 1O•
FREfiUENCY-gPS
1o•v-frequencyambient noise is turbulent-pressure
F[o. 12. Traffic-noi• s•ctra deducedfrom ship-noiresource
fluctuations.There is goodreasonto believethat turbu- characteristics and attenuation effects. Several variations are
lence of varying degree is a general circumstance shown. For example, •e curve lB4 definesthe expectedspatrum
throughoutthe ocean. By the mechanismsof turbu- shapeat 1• nautical miles (185 kin) from a sourcewho• noise
lence,a portionof the energywhichis introducedinto spectrum is flat up to 1• •s and decreases-6 dB per octave
above 1• cps, the effectivesourcedepth •ing 20 ft (6m).
the oceanat very low frequencies(large eddies,low
wavenumbers) is transferred
to rangeof higherfrequen- concentrationof shipsat relativelycloserange,suchas
cies,which,accordingto the curvesof Fig. 11, extends might be the casein shallowwater near a harbor and
above1 cps.Someaspectsof depth and wind depend- coastalshippinglanes.The traffic-noisecharacteristics
ence(and the lackof it) areexplainedby the hypothesis, are determined bv the mutual effect of the three factors.
but the data are inconclusive.
Traffic-noisecharacteristicsalso dependon the kinds
The conclusionis that, while noise radiated by turbu-
of shipsinvolved,that is, upon the nature of the source.
lencedoesnot greatly influencethe ambient noise,the In generalthe baseis broad sothat individual differences
turbulent-pressurefluctuationsare probablyan impor- blend into an averagesourcecharacteristic.
tant componentof the noisebelow 10 cps, and some- A study of the noisefrom surfaceshipM ø indicates
times in the rangefrom 10 to 100 cps. that on the average,when measuredat distancesof
2.3 Oceanic Traffic
about20 yards,the sound-pressure-level spectrahavea
slopeof about --6dB per octave. The spectrumis
The ambient noisemay includesignificantcontribu- highlyvariableat frequenciesbelow1000cps,and,under
tionsfrom two typesof noise •sfrom ships.Ship noiseis somecircumstances, the slope tends to flatten in the
discussedbrieflyin Sec.2.6. Tra•c noiseis the subject neighborhood of 100cps.This source-spectrum shapeis
of this section. altered in transmissionby the frequency-dependent
The degreeto which traffic noiseinfluencesthe am- attenuation part of the transmissionloss.Accordingto
bient noisedependson the particular combinationof Sheehvand Halley,aøthe attenuationis 0.033fl dB per
transmission
loss,num.
bero.fships,and the distribution
of kilorard or 0.066fl dB per nauticalmile,wheref is the
shipspertaining to a given situation. For instance, frequencyin kc. At long ranges,the attenuationin-
a significant contribution could result from widely creasesrapidly with frequencyabove500 cps.
scatteredshipsif the averagetransmission lossper unit For most surfaceships, the effectivesourceof the
distancewere relatively small suchas lnight be the case radiated noise is between ten and thirty feet below the
at a deep-waterlocationin an open-oceanarea crossed surface.Up to frequenciesof about 50 cps, the source
by transoceanicshippinglanes. A significantcontribu- and its image from surfacereflectionoperate as an
tion couldalsoresult even when transmissionlossesper acousticdoublet radiatingnoisewith a spectrmnslope
unit distanceare high if therewerea comparativelylarge of +6 dB per octave relative to the spectrmnof the
simple source.
• Ship noiseis the noise from one or more ships at closerange. To obtain some notion of the probable shape of
It may be identified by short-term variations in the ambient-noise
characteristics,such as the temporary appearanceof narrow-band traffic-noisespectra,the foregoinginformation was used
componentsand a comparativelyrapid rise and fall in noiselevel. in derivingthe curvesshownin Fig. 12. Variationsin
Ship noi• is usually obviousand thereforegenerallycan be and is
deleted from ambienbnoise data. *• M. T. Dow, J. W. Emling, and V. O. Knudsen,"Survey of
Trajlic noiseis noiseresultingfrom the combinedeffect of all UnderwaterSound,ReportNo. 4, Soundsfrom SurfaceShips,"
shiptraffic.exceptingthe immediateeffectso[ shipnoiseasdefined 6.1-NDRC-2124 (1945).
in the precedingparagraph.Traffic noiseis usuall.vnot obviousas •oM. J. Sheehy and R. Halle.v, J. Acoust. Soc. Am. 29, 464
such. (1957).
 ACOUSTIC AMBIENT NOISE IN THE OCEAN 1949

thespectracaused by differencesin sourcedepth,differ- dependentcomponentof the ambientnoiseare between

encesin the shapeof the source-noise spectrum,and 40 and55dB, according
to Figs.l(b), l(d), 2(b), 2(d),
differencesin the attenuationat differentrangesare and 2(el.
indicatedby the composite set of curves.The effectof It is apparentfromthisevaluation that the effective
the sourcedepth at low frequencies is shownby the distancefor traffic-noise
sources
in the deep-open ocean
curvesnumbered(1) for a depthof 20 ft, and (2) for can be as much as 1000 miles or more.
10ft. The choicesof source-noisespectrumshape,based The spectrain Figs.1 and 2 werearrangedoriginally
on the data reportedby Dow,4•are describedin termsof on the basisof spectrumshape.A similararrangement
the slopesof the sound-pressure-level spectra,and the would result from traffic-noise considerations. At the
resulting curves are identified as follows: (A) --6 dB locationsfor Figs. l(a) and l(e), someship noiseis
per octave; (B) 0rib per octave up to 100cps and encountered(deletedfrom the data whendetected),but
--6 (lB per octave above 100cps; and (C) 0dB per usual transluissionrangesare too short for the com-
octave up to 300cps and --6rib per octave above positeeffectof traffic noise.Figure 1(c) pertainsto an
3{•1cps.The changein spectrumshapeas the range isolatedarea whereship noiseis infrequent.The data in
varie:, a consequence of attenuation, is shownby curve Figs. 1(b) and 1(d) wereobtainedin the midstof both
(3) representing a rangeof 500 miles (926 km), curve coastaland transoceanic shippinglanes,and illustrate
(4) a rangeof 100miles(185kin), andcurve(5) a range the case of a comparatively'large concentrationof
of 10 miles(185kin). The spectrmn corresponding to a sourcesat relativelycloserange.While Figs.2(a) and
particularset of conditions may be foundby following 2(c) representmeasurements in deepwater,i.e., greater
the curves identified by the relevant nmnbers and letter. than 100 fathoms in depth, the locationswere not in
For example,the curvelB4 is the spectrumform which the openoceanin the senseof beingopento long-range
would be observed at 100 nfiles from a source located at transmission. The spectrashownin Figs. 2(b), 2(d),
a depthof 20 ft, and whosenoisespectrumis flat up to and 2(el werederivedfrom measurelnents madein the
100cp: anddecreases at 6 dB peroctaveabove100cps. deepoceanopento long-rangetransmission andcrossed
There is a remarkablesimilarity betweenthe syn- by transoceanic shippinglanes.
thetictraffic-noise spectraof Fig. 12 and the spectraof The evidenceis strongthat the non-wind-dependent
the non-wind-dependent componentof the observed componentof the ambientnoiseat frequencies between
ambient noise,which are discussedin Sec. 1.1. In each 10 cpsand 1000cpsis trafficnoise.It is concluded that,
case,the maximumis in the vicinityof 100cps,and lite while thereare lnany placeswhichare isolatedfrom this
spectrmn fallsoffsteeply above 10•)cps. noise,in a largeproportionof the ocean,trafficnoiseis a
The high-frequency "cutoff" occurs at lower fre- significant element of the observedambient noise and
quenciesin the deep-waterspectraof Figs. 2(b), 2(d), often dominatesthe spectrabetween20 and 500 cp•.
and 2(el thanin the shallow-water spectraof Figs.l(b)
and l(d). This effectcan be explained,or evenantici- 2.4 Seismic Sources
pated,by asstuning that the averagerangeof the effec- As a result of volcanic and tectonic action, waves are
tive traffic-noise
sourcesis generallylessfor shallow- set up in the earth. Even when the point of origin is
than for deep-waterlocations. distant from the oceanboundary,appreciableamounts
Measurements of the noiseradiatedby surfaceships of the energymay rind their way into the oceanand be
have been reportedby Dow e! al.• Corresponding to propagated as compressionalwaves in the water
thesedata, for surfaceships the equivalent simple (T phase)? •- •a (Similareffectsoften resultfrom artifi-
source-pressurelevels in a 1-cpsband at 100cps at a cial causessuchas manmadeexplosions.)
distance of 1 yard are between 125dB and 145rib
When observedat closerange, waterbornenoiseof
(re 0.0002 dvn cn¾-')in most cases.Hale• has shown
seismic(volcanic)originhasbeenreported aaas includ-
that experimentalresultsfrom long-rangetransmission ing observableenergy at frequenciesup to at least
in deepwater do not fit the free-field,sphericaldiverg- 500 cps. The spectrum characteristicsdepend on the
ence law very well, and that better agreement with magnitudeof the seismicactivity, the range,and details
experimentdoesresultif boundariesand sound-velocity of the propagationpath, includingany land or sea-floor
structure are taken into account. Accordingto this segments. Experimental dataa•.'•s indicate that in
theory and experiment,105 dB is a reasonableestimate
az1. Tolstoy, M. Ewing, and F. Pre.*$. Columbia University
of the averagetransmission
lossat 100cp,-for a rangeof Genvhysicall,ab. Tech. Rept. 1 (1949). or Bull. Seismol.Soc.Am.
500 miles. Accordingly,the spectrumlevel (re 0.0002 40. 25 (1950).
dvn 'cm•) at 100cpsfrom one"average"shipsourceat asR. S. Dietz and M. J. Sheehy,Bull. Geol. Soc. Am. 65. 1041
(I954).
5(10
miles
is20to40dB;from10"average"
ships
allat a• D. H. Shutbet, Hull. Seisintfi.Soc. Am. 45, 23 (1955).
500 miles,30 to 50 dB (assumingpoweraddition); and • D. H. Shutbet and Maurice Ewing, Bull. Seismoi.Soc. Am.
47. 251 (1957).
from 100 ships,40 to 60 riB. At a rangeof 1000milesthe aaJ. 5I. Shodgrassand A. F. Richards,Trans. Am. Geophy•.
levels would be only 3 to 6rib lower. The spectrum Union 37, 97 (1956).
levels at 100cps of the aforementionednon-wind- • Allen R..Milne. Bull. Seismol.Soc. Am. 49, 317 (1959).
aaJ. Nnrthrop.M. Blaik. and [. Tolstoy,J. Geophys.Research
a• F. E. Hale, J. Acoust.Soc.Am. 33, 456 (1961). 65. 4223 (1960).
1950 GORDON M. WENZ

generalthe spectrumhasa maximranbetween2 and 20 face is assumed. The velocities which were measured
cps and that noticeablewaterbornenoisefrom earth- fall within the range of the resultsof seismicmeasure-
quakesmay be expected at frequencies froin 1 to 100 ments made on land,•o-•t which were lnentionedin the
cps.The noiseis manifestas a singletransient,or a precedingdiscussion.However, the sea-floorvelocity
series of transients, of relatively short duration and spectrashoweda slopeof --5 or --6 dB per octave.
infrequentoccurrence. However, in someareas and The experimentaldata indicate a closerelation be-
duringsomeperiodsof timethe frequency of occurrence tween the seismicbackgroundand the nearby pressure
may be as oftenas severaltimesan hour. fluctuationsin the water. A pertinent questionis:
A seismicbackground of continuous disturbanceof \Vhich is the cause and which the effect ? Possibly some
varyingstrengthis alsoobserved, beingattributedto equilibriumprocessexists,the directionof net energy
the aftereffectsof the more transientevents,and to the transferdependingon the relativeenergylevelsexisting
effectsof storms,and of wavesand swellat the coastal in the two media in a particular situation.
boundary', with localcontributions
from winds,water- Fromthis brief survey,it is concluded that noisefrom
falls, traffic, and machinery.The spectrumof the earthquakesdoes dominate the ambient noise at fre-
vertical ground-particle displacementsof the back- quenciesbetween 1 and 100 cps, but such eflects are
groundnoiseasobserved noise transientand highlydependenton time and location.
onland(andexcluding
Significantnoisefrom lesser,but more or lesscontinu-
from obvious local and transient sources)has a maxi-
nmm between0.1 and 0.2 cps, with amplitudesfrom ous,seismicdisturbances is possibleparticularly-when
2X 10--øto 20/z.so•øThe amplitudesdecreaseapproxi- current velocities and turbulence are at a mininmm,
matelyin inverseproportionto frequency between1 but additional data are needed for a more definite
and 100cps.At 1 cps,the amplitudes rangefrom 10-a evaluation.
to 10-t it, and at 100 cpsfrom 10-s to 10-a it. Vertical The seismicdisturbancesmay have a direct effecton
and horizontalvelocity spectrashowmaxima at the bottom-mounted transducers. For this reason, the
•ame frequenciesas the displacementspectra? In vibration sensitivities of the transducer should be known
the neighborhood of 0.5 cps, the upper frequency and taken into considerationin the design of such a
limit of the data, the velocityspectrabeginto flatten, sy'stem,and in the interpretationof results.
suggesting
a flatvelocity
spectrumabove1cps,asmight
be expected
fromthe inversefrequency
dependenceof 2.5 Biological Sources
the displacementspectra.
A crude estimate of noisein the ocean associatedwith Many speciesof marinelife have beenidentifiedas
the continuousseismicdisturbancesmay be obtainedby
noiseproducers.Noise of biologicalorigin has been
observedat all frequencieswithin the limits of the
assuming, in the absence of specificdata, that the
systemsused,which,in aggregate,have coveredfrom
seismicspecuumcharacteristics on the seafloorare 10 cps to above 100kc.1-•.•a.6aThe individual sounds
about the same as those on land and that the vertical
are usually of short duration, but often frequently
components
of theparticledisplacements
andvelocities repeated,andincludea widevarietyof distinctivety'pes
of the water at the boundaryare the sameas thoseof
such as cries, barks, grunts, "awesomemoans," mew-
the sea floor. For such conditions,the pressure-level
ings,chirps,whistles,taps,cracklings,
clicks,etc. Pulse-
spectrumis essentially
fiat between1 and 100cps, typesoundswhichchangein repetitionrate, sometimes
varying in level between45 and 95dB re 0.0002 very quickly,havebeenidentifiedwith echolocationby
dyn/o•. The peaklevelsbetween 0.1 and0.2cpsare porpoise.• •s Repetitive pulse soundshave also been
froin 65 to 120dB. The spectrumshapedoesnot agree
attributed to whales.Continuous(in time) biological
with theobserved ambient-noisespectra(seeSec.1), but
noiseis frequentlyencountered
in someareaswhenthe
the levels are of sufficientmagnitudeto suggestthe
soundsof many individualsblendinto a potpourri,such
possibility
thatsome
ofthevariabilityin ambient-noise
as the crac'ldingof shrimp and the croakerchorus)-s
spectra
maybea consequence
of theseismic background The contributionof biologicalnoiseto the ambient
activity. noisein the oceanvaries with frequency,with time, and
Measurementsat frequencies between4 and 400 cps with location,so that it is difficult to generalize.In some
directly comparingbackground seismicvelocitycom- casesdiurnal, seasonal,and geographicalpatterns may
ponents andwaterborne soundpressures as measuredbe predictedTM from experimentaldata, or from the
at the sea bottom in shallow water•-•'•ashow order-of-
habits and habitats, if known, of known noisemakers.
magnitude
agreement
whencontinuity
across
theinter-
a•J. N. Bruneand J. Oliver,Bull. Seismol.Soc.Am. 49, 349 • W. N. Kellog, R. Kohler, and H. N. Morris, Science117, 239
(t9•). (1953).
•0 G. E. Frantti, D. E. Willis, and J. T. Wilson, Bull. Seismol. • M.P. Fish, University of Rhode Island. Narragansett Marine
Soc.Am. S2. 113 (1962). Lab., Kingston,Rhode Island Reference58-8 (1958).
s•R. A. Haubrichand H. M. Iyer, Bull. Seismot.Soc.Am. 52, • W. N. Kellog, Science128, 982 (1958).
87 (1962). •?W. N. Kellog,J. Acoust.Soc.Am. 31, 1 (1959).
•*-E.G. McLeroy (unpublished dataL •aK. S. Norris. J. H. Prescott, P. V. Asa-dorian, and Paul
oaR. D. WorleyandR. A. Walker(unpubli-•hed data). Perkins, Biol. Bull. 120, 163 (1961).
 ACOUSTIC AMBIENT NOISE IN Till.; OCI-;.\X 19.51

Noiseshavingthe distinctivenatureof biologicalandprobable causes in thevariouspartsoi tilespectrmn

sounds are readilydetectedin the ambientnoise,but between1 cpsand 100kc.
the biologicalsourceis not alwaysocrlain. (In thebasisof spectrulncharacteristics.
theprincipal
The data presentedin Sec.1 excludenoiseof known componenlsof the prevailing ambient noi.,c in the
or suspectedbiologicalorigin. ocean are:

(a) a 1ow-frequenc3 componentcharacterizedby a

2.6 Additional Sources
spectrum-level slope of approximately--1(I dB per
Various other source• of intermittent and local effects octaveand wind dependence in very shallowwater,
includeship.,,industrialactivity,explosions, precipita- tile most probal)le sourcebeing ambient turbulence
lion, and sea itc. (lurbulenl-pres.sure Iluctuations);
Shipnoiseis thenoisefromoneor moreshipsat close lb) a high-frequency component characterized by
range,and, as usedhere, is differentiatedfrom traltic wind dependence, a broad maximumbetween100 and
noised • Ship noisecausesshort-termvariationsin the 10(10oilsand a slopeof approximately--6dB per
ambientnoisecharacterized by the temporaryappear- or'laveat frequencies abovethe maximunbwith levels,
anceof narrow-bandcomponents at frequencie., below on the average, some 5 dB lessin deepwater than in
1001)cps, and broad-bandcavitation noiseextending shallow (most probable source--bubbles andsprayfrotax
wellinto the kilocycleregion,oftenwith low-frequencysurfaceagitation);
modulationpatterns. (c) a medium-frequency component characterizedby
htduslrialaclivilyon shore,such as lille driving, a broadmaxi,num between 10and200cps,anda steep
hammering, riveting,and mechanical activity of many negativeslopeat frequencies abovethe maximulnhand
kinds, can generatewaterbornenoise.The character- (most probable source oceanic traffic);and
isticsof the noisedependon the particular.qtuation. (d) a thenhal noisecomponentcharacterized by a
Noisefromindustrialactivitymaypredominate at times +6 dB per oclaveslope.
in particular near-shoreareas. ('omponent (a) nearIx' always predominatesat fre-
Noise from explosions is very much like that from quenciesbelow 10 cps,and its influencemay be evident
earthquakes(seeSec.2.4). At closerange,the effects up to frequencies as high as ll}0 cpsin tile absenceof
covera wide rangeof frequencies,but at longerrange component(c). Componentlb) nearly alwav>pre-
thespectrumhasbeenmodifiedby propagation,and the dominates at frequencies above500cps.Component (c)
larger part of the energy is usually at frequencic.sfrequently predominates in thebandfrom20 to 200cps,
below 100cps. lint is not obse,'ved in isolated•øareas.('omponent(d)
PrecipiIation noiseis basicallynoisefi'oma sprayof is iNfactorat frequencies above20 kc.
waler droplets(rain) and rigkl bodies(hail) (seeSec. The general similarity between the observedlevels
2.2.2). The effectsof precipitationare most noticeahle and lhe spectrumshapeof the ambientnoiseand of the
at frequencies above500 cps,but max extendto ils low spectrae.-timatedfor the turbulent-pressure fluctua-
as l(10cps if heavv precipilation occum when wind lions leadsto the conclusion that tile low-frequency
speedsare low. ambientnoise,component (a), consistsprimarilyof the
The various kinds of set, ice movements are a source
turbulent-pressure fluctuations (whicharemuchgreater
of noisewhichat timescoversa wide rangeof frequen- in magnitudethan the radiatedsoundpressures gene-
cies at high level. The noiseoriginatesin Ihe straining rated by turlmlence). It may be said further that turbu-
and crackingof tile ice from therlnal effects,and in the lenceis a process
I)v which.•omeof the energyintro-
grinding,slkling,crunching,and Iraropingof 11oes
and duced into, or originating in, the ocean at very low
bergs. h'eqnencics
is transformed into energyof consequence
It is difficult to generalizeon the characteristicsof the to theambientnoiseat frequenciesabove1 cps.The not
variouskinds of intermitlent and local noise,sincesuch unreasonable assumptionof a widespreadambient
noiseis dependent to a great degree on the parlicular turbulencein the oceanis required.(SeeSec.2.2.4.)
time and place of concern. A small amount of low-frequencydata taken in very
shallowwaterindicates a dependence of level,but not of
3. COLLIGATION spectrumslope,on windspeed(seeFig. 6). This wind
dependence
is explainedby the influenceof wind-caused
The experimentaland theoreticaldata whicb have drift currents on the turbulence. The data are not of
beenpre•entedin the foregoinglead to the conclusion
that the generalspectrmncharacteristicsof the prevail sufficient
quantityto establish
evena tentativequanti-
ing ambient noisein the ocean are detemfined by tile tative relationshipbetweenlt.vel and windspeed;so
noneis shownin l:ig. 13.
combinedeffect of seve,'alcomponents,which, though
Componentsof tile seismicI)ackgroundnoiseill the
of widespreadand continual occurrence,vary each in
itsownwaywithtimeandlocation.
Acomposite
picture "Isolated"ellher because
. geographic
l,•calionor because
is given in Fig. 13, whichsummarizesthe characteristics lm•l,agation c,mdilions. or ho/h.
1952 GORDON M. WENZ

INTERMITT•NTAND LOCAl'EFFECTS
EARTHQUAKES I
AND EX_PLOSIONS
4. BIOLOGICS
PRECIPITATION
SHIPS, INDUSTRIAL ACT VITY
SEA ICE•.-•
IOO
KEY
LIMITS OF PREVAILING NOISE
-- WIND-DEPENDENT BUBBLE AND SPRAY NOISE

LOW-FREQUENCY VERY-SHALLOW-WATER
! WIND DEPENDENCE
!
--• HEAVY PRECIPITATION

..... HEAVY TRAFFIC NOISE

C&::::::•: USUAL TRAFFIC NOISE- SHALLOW WATER
= USUAL TRAFFIC NOISE- DEEP WATER
-- -- -- THERMAL NOISE

........ GENERAL PATTERN OF NOISE FROM Fro. I3. A compositeof ambient-

EARTHQUAKES AND EXPLOSIONS
noisespectra, summarizingresults
............ EXTRAPOLATIONS
and conclusionsconcerning spec-
trum shapeand level and prohable
I murces and mechanisms of the
ambient noise in various parts of
WIND FORCE the spectrum between 1 cps and
BEAUFORT)
100 kc. The key identifies com-
ponent spectra. Horizontal arrows
show the approximate frequency
:• 40 hand nf influence of the various
sources. An estimate of the am-
I- bient noise to be expected in a
particular situation can be made
by selecting and combining the
pertinent componentspectra.

• 2o
•J•

o..
PREVAILING NOISES
Z TURBULENT-PRESSURE
0
m 0
l, OCEANIC TRAFFIC

I BUBBLES AND SPRAY

WAVES-SECO'NO-ORDER
PRESSURE
EFFECTSI
(SURFACE
AGITATION)
/MOLECULAR
SEISMIC BACK'.r4ROUND)• / AGITATION
-20
I0 102 I0" 104 I0
FREQUENCY- CPS

earth, includingthe seafloor,are oftenattribuled to the where current velocities and turbulence are at a mini-
effectsof pressurefluctuationsin the oceanrather than mum. (See Sec. 2.4.)
vice versa.There is probablysomesort of a give-and- Becauseof the steepnegativeslopeof surface-wave
take processbv which in someareasthe seismicwaves spectra, and because of the rapid attenuation with
are producedby waterbornepres>urefluctuations,such depth,
froin surface
it is doubtful
wavesthat
are the
of much
first-order
significance
pressureto
effect>
the
as the Longuet-Higginssecond-order pressureeffects
from surface-wave trains, while in other areas the ambientnoiseat frequencies above1 cps.However,it
seismicactivity causeswaterbornepressurefluctu- is probable lhat the Longuet Higgins second-order
ations. Such a process would allow hydrodynamic effectscontributeto the noiseat frequenciesup to
pressureeffects to be transmitted from one place to l(I cps, particularly in or near storm areasand certain
another via seismicprocesses.
The estimatednoisefrom coastalareas. (See Sec. 2.2.3.)
seismicbackgroundactivity differsin spectrumshape The wind dependenceand the "on-the-average"--5
from that of the low-frequencycomponent (a), but to --6 dB per octaveslopeof the high-frequency
com-
seismic noise could account for some of the variability ponent (b) are well established.The high-frequency
observedal lowfrequencies
and couldbecomesignificant wind-dependent
curvesshownin Fig. 13 representover-
 ACOUSTIC AMBIENT NOISE IN THE OCEAN 1953

all averages,includingboth deep- and shallow-water occurat frequencies above10cps and often include
data. (The comparable shallow-water
averages are 2 to narrow-bandcomponents belowI000 cps.Noisefroin
3 dB higherthan the over-allaverages,the deep-water industrialactivity occursin particularnear-shoreloca-
averages2 to 3 dB lower.) The hypothesisthat gas tions.Noiseresultingfrom sea-icemovementis dis-
bubbles,cavitation,and sprayin the surfaceagitation tinctive in character,coveringa wide range in fre-
are the sourcesof tbe high-frequencywind-dependent quenc_v, but is restrictedto particulargeographic
areas.
component(b) is well supportedby the observedspec-
trum levels,spectrumshape,and band of maximran 4. APPLICATION NOTES
levels.(SeeSecs.2.2.1 and 2.2.2.)
The provisional pictureof ambientnoisein theocean
The assumptionthat the medium-frequency com- whichhasbeendeveloped showsthat theambientnoise
ponent(c) is the resultof the combinedeffectof many dependson severalvariables,someof which,unfortu-
shipsat relativelylongrangeisshownto bea reasonable nately, are as difficult to estimateor measureas the
one bv the general agreement between the observed
ambient noiseitself. However,a few generalguidelines
spettra and the spectraestimatedfor traffic noise(see
can be givenfor estimatingand predictingthe ambient
Sec.2.3). The high-frequency"cutoff" beginsat lower
noisewhichprevailsin the absence of intermittentand
frequenciesin deep water, as comparedto shallow local effects.
actordingto the observeddata (Sec.1). This effectis to
be expectedfrom the curvesin Fig. 12, if one assumes [-4.1] 500 cps to 20 kc
that, while in both casesthe rangesare comparatively
long,the averagerangeof the effectivenoisesourcesis E•timatesof theprevailingambient-noise levelsin the
greaterfor deepwater than for shallow. oceanat frequencies between500 cpsand 20 kc can be
The shapeof the 1ower-linfitcurvein Figs.7 and 13 is madefroma knowledge of thewindspeed.Estimates of
not greatlydifferentfrom that whichwouldresultfrom windspeedcanbe obtainedfromthe Atlasof Climatic
a combinationof a turbulent-pressure spectrum(Sec. Chartsof theOceans issuedby tbeU.S. \\'eatherBureau
2.2.4), a low-levelbubble-noisespectrumdue to a (1938),if morespecificinformationis lacking.Table I,
reinanentdistributionof bubbles(which,as was shown or similar tables,may be usedto estimatewind speed
in Sec.2.2.1, may existevenat zeroseastate), and the from wave-heightor sea-stateinformation.Knowing
thefinal noise spectrum.It is quite likely that the the wind speed,one may determine"on-the-average"
empMcallower-limitcurve is, to somedegree,an indi- levelsby referenceto tbe high-frequency wind-depend-
cation of the state of the art of measuringvery low- ent curvesin Fig. 13 (which,includingthe influenceof
pressurelevels in the ocean. \Vith a sensitivesystem, morerecentdata, are 2 to 3 dB lowerin level than the
one might expectto encountereven lower levels,parti- Knudsen•.'-'averages).
cularlr at low frequenciesin a regionof minimum cur- For example,a wind speedof 7 knots (3.6m/see)
rent and turbulence, where the probable source of fallsat the lowerlimit of the rangefor Beaufortwind
residualnoiseis seismicbackgroundactivity. force3. According to thewind-force-3curvein Fig. 13,
The relatively high residual or thresholdnoiseob- the estimatedspectrumlevel at 1 kc (to the nearest
servedin someshallow-waterlocations[-Figs.1(a) and dB) is 32 dB (re 0.0002dyn/cmø).The curvesrepresent
1fb)] apparentlyarisesfromlocalconditions. The shape arcragelevelsand may be considered as corresponding
of the residualspectrasuggests that the noiseis a com- to the mean of the Beaufort wind-speedgrouping.
bination of the average elIects from such sourcesas Interpolations may be madefor estimates of levelsat
local turbulenceand cavitalion (particularlyif rough otherwindspeeds. For theexampleof 7 knots,an inter-
boundaries and high currentsare involved),industrial polatedlevelestimateat 1 kc is 30dB. Similarinter-
activity, marginal traffic noise,and the lnore subtle polationsmay be madefor wind speedsfallingin the
typesof self-noise, suchas that whichmay resultfrom wind-force-4,-6, and -7 groups,for which curvesare
relativemotionbetweenthe hydrophoneand the water. not shownin Fig. 13.
The prevailing features of the ambient noise are The curvesin Fig. 13 are averagesover both deep-
altered by intermittent and local effects. The noise from and shallow-waterdata. For a closerdeep-wateresti-
earthquakes
andexplosions
perturbs mate, the over-all average level should be lowered by
the spectrum
mostly at frequenciesaround 10 cps, the extent of the 2 or 3 dB. Thus, for deep water and a wind speedof
bandof influencebeingdetermined,to a largedegree,by 7 knots, the estimatedlevel at 1 kc is 27 or 28 dB. For
the range. Disturbance from biological sourcesmay shallow water, the over-all levels are raised by 2 or
show up at almost any frequency, often conformingto 3 riB. (The shallow-water estimates are about the same
diurnal, seasonal,and geographicpatterns, but with as those of the Knudsen curves, while the deep-water
distinctivecharacteristics.The effectsof precipitation estimatesare about 5 dB lower in level.)
are most apparent at frequenciesabove 500 cps, but As it happens,the wind-dependentspectra follow
mav extendIo lowerfrequencies whenheavy precipita- approximately anempirical"ruleoffives,"whichisstated
tion occursat low wind speeds..Majorship-noiseeffects as follows:
1954 GOR DON M . \V E N Z

In thefrequencybandbetween500 cpsand 5 kc the high windsand stormsbecauseof a generalincreasein

ambientsea-noise spectrumlevelsdecrease 5 dB per the turbulent energy, and because of the Longuet-
octavewith increasingfrequency,and increase5 dB Higginssecond-order pressureeffects.
with each doublingof wind speedfrom 2.5 to 40 When observations are madeat very shallowdepths,
knots;the spectrumlevel at I kc in deepwater is lessthan 25 fathoms (roughly 50 m), one shouldbe
equalto 25 dB (5>(5) re 0.0002dyn/cm-øwhenthe preparedfor high and wind-dependentlevels,as indi-
wind speedis 5 knots,and is 5 dB higherin shallow cated by the shaded(upper left) area in Fig. 13. The
water. data are insufficientto establisha quantitative relation
betweenlevel and wind speedwhich can be applied
The "rule of fives"is fairly accurateup to a frequency generally.
of 20 kc.
Although these qualitative approximationsdo not
To illustrate the use of the rule of fives: The estimated
permit very definite predictionsof the ambient noiseat
spectrum level at 4 kc in deep water a hen the wind the low frequencies.they should be useful in defining
speedis 20 knots (10.3m/see)is equalto 25 dB (1-kc specificareasfor investigation.
levelfor 5 knots)plustOdB (windspeeddoubledtwice
from 5 knots) minus10dB (2 octavesabove 1 kc), [4.3• 10 to 500 cps
whichaddsup to 25dB. The corresponding shallow-
water level is 30 dB, and the meanbetweenthe deep In the frequencyband between10 and 500 cps,the
and shallowvaluesis 27« dB. This value is about the noiseis influencedby at least three components.To
sameas the valueindicatedat 4 kc by the wind-force-5 predict the noise,one nmst know or assumethe spec-
curve in Fig. 13, showingthat the rule-of-five•result trum for a low-frequencycomponent,a traffic-noise
agreesreasonablywell with the averagedata. component,and a high-fre.'luencywind-dependentcom-
It is believedthat the procedureswhichhave been ponent, and then combinethe componentspectra.The
given will lead to useful estimatesof the ambient sea high-frequencywind-dependentcomponent and the
noisein the frequencybandfrom500cpsto 20 kc. But 1owdrequencycomponentwere discussedin Secs.4.1
it shouldbe remembered that considerabledeparture and 4.2, respectively.
may sometimesbe expected,sincewind speedi.- not a The experimental data indicate that the traffic-noise
precisemeasureof the actual surfaceagitation (see levelsusuallyfall within the medium-frequency shaded
Sec.1.1 and Fig. 3), nor are estimatesof seastate. If areasshownin Fig. 13. The differencebetweenshallow-
informationis knownaboutduration,felch,and topo- waterlevels(areashadedby largedots)anddeep-water
graphy,a judicioususeof the kind of informationgiven levels (area shadedby lines) is to be noted.The more
in Table I will improvethe reliabilityof the estimate. extremehigherand lower traffic-noiselevelscorrespond
Ideally, information is neededas to the details of the respectively to relatively nearby concentrationsof
surfaceagitation,suchas the relationbetweenmeteoro- shippingand to the more remote© areas. The classifi-
logicaland oceanographic conditions and the distribu- cationas a "remote" area may be time-dependent,the
tionsof thesizesofbubbles anddroplets, andthedegree result of daily, seasonal,or other variationsin propa-
of cavitation.:\t low •ind speeds, the effectsof other gation or traffic patterns. As was brought out in Sec.
influences
may be expected the variability. 2.3, the effectivedistancefor traffic-noisesourcesin the
to increase
deep oceancan be as much as 1000 milesor more, and
a ratherwidelyobservedtraffic-noise component is to he
[-4.2• 1 cps to 10 cps
expected.A comparisonof the traffic-noisespectrawith
An "on-the-average" spectrum-level slopeof --8 to the other spectra in Fig. 13 indicatesthat the iraliSt-
--10 dB per octavein the frequencyrange from 1 to noisecomponentgenerallypredominatesat frequencies
10 cpsmay be predictedwith someassurance,sincethe between 20 and 200 cps.
observeddata are quite consistentin this respect.This In shallmy-waterareasisolatedfrom generalshipping
slopepersistsdown to 0.05 cps,and perhapsto lower activity in very remote deep-waterareas,and in some
frequencies,exceptwhen the first-ordereffectsof surface deep-water areas which are isolated by underwater
wavesare encounteredat shallowdepths. topographyor oceanographicconditions,the effect of
It is to be generallyanticipatedthat the levelswill the traffic-noisecomponentis very small, and levels
shouldbe estimatedby combiningonly the appropriate
fall between the lower-limit curve and the bottom of the
shadedarea (upperleft) in Fig. 13. Accordingto the low-frequencycomponentand the high-frequencywind-
reachedin Sec. 2.2.4, the spreadin the dependentcomponent.
conclusions
observedlow-frequencyambient-noise levelsis princi-
pally the resultof the variation in the oceanicturbulent- 4.4 Example
pressurefluctuations,and the higherlevelsare to be The guidelinessuggestedby the foregoingremarks
expectedin or nearthe major oceaniccurrents,or any- were usedto estimate three spectra,which are shownin
whereelsewhererelativelyhighwater motionis known Fig. 14, corresponding to three frequentlyencountered
to exist.Onemightalsoexpecthighlevelsto accompany situations.The three spectraare: (1) that expectedin
 A CO UST i C A M BI E X T N O i .%I-; I N T II E O C F..\ X 1955

80
X I i CONDITIONS
• i I WINDFORCE
3 {BEAUFORT
SCALE)
• ' I AVERAGE
LOW-FREQUENCY
NOISE
o
X . DEEP
WATER.
AVERAGE
TRAFFIC
NOISE
o
• I , SHALLOW
WATER,
AVERAGE
TRAFFIC
NOISE
o '• i - SHALLOW
WATER.NO TRAFFICNO,SE
d •0

•n
:.-..
.....
COM.ON.T
SPECTRA
l' Ic,. 14..\ml•icnt-noiseSl,CC
tnt estimated for three fre- .....
quently encounteredsitnations.
The dotted-line extensions indi-
40 .......
ß __ \
cate. in the regionsof nverlal,,
the imlividual c,mponent st,ec
tra from which the estim;ttcd
spectrawere derived.
0.

LG
2O
r•

Z 0
0

I0 I0 • I0 a 104 I0 s
FREQUENCY- CPS

dcel)waterwith averagetratticnoise,(2) that expected These componentswere then combinedby power

in shallowwaterwith averagetraliic noise,and (3) that addition.The resultingdeep-waterspectrmnis shown
expectedin isolatedsh011ow water (no traffic noise); by the•olidcnrve,the twoshallow-water
spectraby the
whenin eachcaseaveragelow-frequency noisecondi- dashed lines.
tions(averageoceanicambientturbulence)and h)rce-3
windsprevail. S. CONCLUDING REMARKS

The shapesof the individualcomponents, in the re-

The scopeof thisreviewhasbeenlimited,primarily,
gionof overlapbeforecombination, areindicatedby the to comparisons basedon averagedspectrmncharacter-
dotted extension.-. The low-frequencycomponent, 7øin
istics.(lther features,for whichexistingdata are very
Fig. 14,is at a levelapproximately
halfwaybetweenthe meager,and which shouldbe investigatedfurther,
t.wer-limit curve and the bottom of the low-frequency
include:temporalcharacteristics suchas hourly,daily,
shadedarea (upper left) in Fig. 13. The medium-
seasonal, and like time patterns;qpatialcharacteristics,
frequency
curvesin Fig. 14aretheapproximate
median.• suchas the directionalpropertiesof the ambient-noise
of the deep- and shallow-walershaded traflic-nnise
soundfield; other .,tatisticalt)ropertiessuchas amI)li-
areas in Fig. 13. The shalhm'-waler high-frequency
rude distributi,ns; and the factors influencing such
cmnl)onentis about 2• (lb abovethe wind-force-3curve characl eristics.
•hown in Fi,z. 13. the deep-watercomponent2• dB
below.
The accumulation of ambient noise data is not yet
large in amount nor comprehensive
in scopewhen con-
;' It is of interest to note that this approximatinn of the average sidereel in relation to the number and range of the
low frequencynoise-levelspectrumderivedfrom observed;tm- variables.The generalizationswhich have been made
hient-noisedata is abnostidenticalto the turbulent-pressure
level
spectrumestimatedfr)r ambientoceanicturbulenceas shownin are subjectto further test. The discussion of sources
Fig. lt. This nearly exact agreementis fortuitous,but is a good and mechanismshas someverisimilitude, and patterns
imlication of the simiktrity between observed ambient m,ise illustratedin Sec. l are frequentlyrepeated.But more
spectraat the lowfrequencies
and theprobableoceanicturbulent-
pressure-levelspectra. directexperimentaldata and moreprecisequantitative
1956 GORDON M. WENZ

are needed.The resultsand conclusionsDietz,

relationships F. T., Kahn, J. S., and Birch, W. B., "Effect of Wind on
Shallow Water Ambient Noise," J. Acoust. Soc. Am. 32,
shouldbe usefulfor immediate needsand for guidance 9•5(A) (1960).
in future investigations. Dietz, F. T., Kahn, J. S., and Birch, W. B., "NonrandomAssocia-
tions between Shallow Water Ambient Noise and Tidal Phase,"
ACKNOWLEDGMENTS
J. Acoust.Soc.Am. 32, 915(A) (I960).
The writer thanksElaine A. Kyle, Mel A. Calderon, Frosch, R. A., "How to lX[akean Ambient Noise in the Ocean,"
and H. M. Linnette for assistancein processing
the data J. Acoust. Soc. Am. 32, 915(A) (1960) (Hudson Laboratories
Contribution No. 8I).
and in preparingthe figures,and Dr. Robert W. Young
Lomask, Morton R. and Saenger,R. Alfred, "Ambient Noisein a
for comments and suggestionsregarding the nmnu-
DeepInland Lake," J. Acoust.Soc.Am. 32, 878 (.1960).
script. Lomask, Morton and Frassetto, Roberto, "Acoustic Measure-
An essentialpart of the sourcematerialwasfoundin mentsin Deep Water Using the Bathyscaph,"J. Acoust. Soc.
unpublisheddata resultingfrom the work of the follow- Am. 32. 1028 (1960).
ing: Arthur D. Ameson, Alan Berman, D. Gradher, Milne, Allen R., "Shallow Water Under-Ice Acousticsin Barrow
R. Halley, W. G. Houser, Henry R. Johnson,R. It. Strait," J. Acoust.Soc.Am. 32, 1007 0960).
Nichols, Jr., F. R. Nitchic, Elliot Rhian, JosephD. Walkinshaw,H. M., "Low-FrequencySpectrumof Deep Ocean
Richard, Jr., Donald Ross,A. J. Saur, R. A. Walker, AmbientNoise,"J. Acoust.Soc.Am. 32, 1497(A) 0960).
and R. I). Worlev. Milne, A. R., "Ambient Xoise Under Old Shore-FastSea Ice,"
J. Acoust.Soc.Am. 33, 1675(A) (1961).
ADDITIONAL REFERENCES Saenger,R. Alfred, "An Estimate of the OffshoreAmbient Noise
SpectrumProducedby PoundingSurf," J. Acoust.Soc.Am. 33,
In addition to the referencescited in the footnotes, 1674(A) (1961).
the followingarc alsopertinent: Willis, Jack and Dietz, Frank T., "Preliminary Study of Low-
Berman, Alan and Saur, A. J., "Ambient Noise as a Function of FrequencyShallow Water Ambient Noise in Narragansett Bay,
Depth," J. Acoust. Soc. Am. •2, 915(A) (1960) (Hudson R. I.," Narragansett Marine Laboratory, University of Rhode
LaboratoriesContributionXo. 77). Island Reference614 (November 1961).