Low-Frequency Ambient Noise in the Deep Sound Channel -The Missing Component

SACLANTCEN Rep SR-51

SACLANT ASW RESEARCH CENTRE REPORT

LOW-FREOUENCY AMBIENT NOISE IN THE DEEP SOUND CHANNEL - THE MISSING COMPONENT

RONALD A. WAGSTAFF

15 JULY 1981

NORTH ATLANTIC LA SPEZIA, ITALY TREATY ORGANIZATION

This document is unclassified. The information it contains is published subiect to the conditions of the legend printed on the inside cover. Short quotations from it may be made, in other publications if credit is given to the author(s). Except for working copies for research purposes or for use in official NATO publications, reproduction requires the authorization of the Director of SACLANTCEN. This document is releamcd to a NATO Government at the direction of the SACLANTCEN subject to the following conditionr:

1. The reciplenl NATO Government agree8 to use its best endeavours to ensure that the informmtion herein disclosed, whether or not it bears a security classification, 1s not dealt with in any manner (a) contrary to the intent of the provisions of the Charter of the Centre, or (b) prejudicral to the rights of the owner thereof to obtain patent, copyright. or other like statutory protection therefor.

2. If the technical :nformation was originally releared to the Centre by a NATO Government subject to rertrictionm clearly marked on this document the recipient NATO Government agrees to ure Itm beet endeavours to abide by the terms of the restrictbnm RO imposed by the releasing Government. SACLANTCEN REPORT SR-51

NORTH ATLANTIC TREATY ORGANIZATION SACLANT ASW Research Centre Viale San Bartolomeo 400, 1-19026 San Bartolomeo (sP), Italy. national 0187 560940 tel: international + 39 187 560940

LOW-FREQUENCY AMBIENT NOISE IN THE DEEP SOUND CHANNEL -THE MISSING COMPONENT

by Ronald A. Wagstaff

(Reprinted from J. AcousticaZ Society America 69, 1981: 1009-1014)

15 July 1981

This report has been prepared as part of Project 21.

B.W!&+$L B.W!&+$L Direct I # Low-frequency ambient noise in the deep sound A channel-The missing component I R. A. Wagstaff I SACLANT ASW Research Centre, Lu S~ia,I&ly (Received 23 June 1980; qtedfor publication 26 November 1980)

There is an important wmponht of the undersea ambient noise which has generally been overlooked or ignored. It is the noise which amvb at a sensor, located in a deep sound channel, by way of ducted sound propagation. It is generated at the surface and becomes channeled m a result of a gradually sloping sound channel axis and through repeated reflections from a sloping bottom. It can be of sufficient level to dominate the local measurement. Examples are given in which neglecting it leads to incorrect interpretation of results and erroneous wnclusions, including agreement between modeling and measurement.

PACS numbers: 43.30.Nb, 43.30.Bp

winds and shippingg noise source characteristics, pro- Ambient noise in the ocean is a complex phenomenon. cessing technique^,'^*^^ and various forms of noise To begin to understand it, one must have a knowledge models12-l5 to name but a few. In some cases, the in- of acoustic propagation and noise-source distributions vestigations are attempting to account for just? few and characteristics. Further, to understand mea- decibels. It tends to imbue a feeling of confidence, sured noise data requires knowledge of the influence of that the scientific community understands the first- the measurement system on the data. For practical order effects and simply needs to "chip away" on the reasons, past investigations of the noise have concen- second-order effects to achieve a complete understand- trated on some aspects that are considered interesting ing of the ambient-noise field. (In this context, second- or important, while limiting consideration of the others. order effects are those which, if omitted, would affect This being the case, it is not surprising that a mrijor the noise by only a few decibels.) component of the noise can remain unrecognized while There has, unfortunately, been one first-order com- the general appearance is that only relatively minor ef- ponent of the noise which has generally been overlooked fects remain to be investigated. or ignored. It is the noise which arrives at a sensor, located in a deep sound channel, by way of continuously This paper discusses a major component of the noise # refracted (RRR) ray paths. This noise is usually gene- which has either been overlooked or ignored by the bulk rated a considerable distance from the measurement of the scientific community investigating the undersea location. It can be of sufficient level to dominate the ambient noise environment. The component being re- local measurement. There is evidence that some ex- ferred to is the noise from distant sources which ar- perimentalists understood the importance of this com- rives at a sensor, located within a deep sound channel, ponent. 4316 The modelers, however, being among the by way of ducted sound propagation. The discovery major recipients of new knowledge and their models, of this component is not the result of new theory, nor the barometers with which to measure the total accumu- can it be attributed to one individual. The "germ of the lation and utilization of that knowledge, have generally ideaworiginated more than 20 years ago when it waS ignored it. The only exception which comes to mind is observed that the signal level received from a source at Ref. 12, in which it is explicitly treated as a separate a considerable distance could be more than if the source was much closer in range. The application of this prin- component. ciple to establish the existence of noise which travels At frequencies below a few hundred hertz, where throughout the oceans by way of the deep sound channel shipping is the major source, the noise can be domi- had not been made until recently. Its omission in the nated by the noise which travels from afar by way of interpretation of noise measurement results, and the SOFAR channel (deep sound channel) propagation. The lack of documented evidence of its theoretical or ex- noise from nearby and less distant sources may con- perimental investigation, indicates it is not presently tribute only a fraction of a decibel to the total noise recognized nor understood by the scientific community level. This situation could well exist in most areas as a whole. On a limited basis, this paper attempts to of the oceans in the northern hemisphere. Neglecting change this. it can be a serious error, leading to misguided experi- ment design and incorrect interpretation of the noise I. BACKGROUND environment. When one considers how specialized our theoretical There are several reasons why the "SOFAR channel and experimental investigations of the undersea am- noiseu has been overlooked or ignored. Among them bient-noise environment have become, since the clas- are the following: sic paper by Wenz in 1962,' he cannot help being im- pressed. The investigations include the horizontal2 and (1) All areas do not possess a SOFAR channel. vertical directionality of the noi~e,~~~depth dependen~e,~ (2) It is not a significant contribution above a few coherence characteristic^,^ attenuation coefficient^,^ hundred hertz where attenuation is severe.

4 1009 J. Acoust. Soc. Am. 69(4), April 1981 0001 -4966181104 1009-06$00.80 O 1981 Acoustical Society of America 1009 (3) It is ma~kedby high density local shipping. channel and become trapped. Once in the channel, it (4) The most significant noise sources are gene- travels with relatively low loss through cylindrical rally believed to be within a few hundred miles spreading. Northrop et al.," give just such an ex- of the sensor. ample for noise (explosive sources) originated in shal- (5) Consideration of distant noise sources and range low water off Point Arena, California, being recorded dependent acoustic propagation are required. by hydrophones at SOFARdepths near Eniwetok and Mid- (6) Low-frequency vertical directionality measure- way. The maximum levels received correspond to shots ments, which give a direct measure of it, are over the edge of the continental shelf, with levels de- relatively recent. creasing by about 10 dB seaward and about 5 dB shore- ward. The shoreward reduction was attributed to the increased number of reflections necessary to get the II. MECHANISMS sound into the channel. A seaward reduction was attri- There are several mechanisms by which sound can buted to the bottom becoming deeper than channel depth. become trapped within the deep sound channel, includ- For sound generated over the slope, they state ". . . the ing diffraction and scattering of sound originating near first bottom reflection is from the continental slope, the surface and the source being physically located where the effect of the steeper bottom slope (3i0)be- within the channel. The contributions from these comes important in channeling BR/SR ("bottom-re- mechanisms, however, can probably be considered flected/surface-reflectedv) rays into RRR ("continuously second- or third-order effectswhen compared to the refracted") rays. For example, for a surface shot in total amount of noise Which is trapped in the channel. 300 m of water a ray that is initially the 0" ray steepens to 11" before it strikes the bottom at a depth of 475 m. There are at least two first-order effects which, to- and the ray becomes continuously refracted. Steeper gether can account for the levels that are measured. rays become RRR after one or more bottom reflections The first is the axis of the deep sound channel approach- ing the surface with increasing latitude where large on the lower continental slope. " An enhancement in concentrations of shipping exist. This happens in the level (less propagation loss) is gained as a result of the RRR propagation having a more favorable geometric North Pacific and the North Atlantic. Figure 2 of Ref. spreading law than either RSR (reflected-surface re- 7 illustrates this characteristic in the northeast Paci- fracted) or BR/SR with no boundary interactions and fic. Ships at high latitudes (about 45 "-50 "N) would corresponding losses. Similar experimental results, radiate sound to the south which could become trapped for noise due to shipping, were reported by Morris.'' after only a few surface reflections, as a result of the Officerz1also discusses this mechanism from a theore- gradually changing sound speed profile. Once the noise tical "point of view." A range-dependent propagation is confined to the channel, a cylindrical spreading law would apply. The importance of this mechanism can be model, which can handle a variable bottom pr~file'~"~ 6 could calculate the propagation loss for such a case. appreciated when one realizes that for economic rea- sons, ships tend to follow great circle paths across the Gaining an understanding of the importance of the ocean basins. During the summer this puts large con- L down-slope conversion mechanism for introducing centrations of ships at high latitudes. During the win- noise into the deep sound channel is straightforward. ter the ships tend to be farther south to avoid the more Assuming an average continental slope of 5"," extend- extreme sea and weather conditions. The surface wa- ing 1000 m into the channel, there exists an offshore ter is also cooler. This may change the location of the surface band which surrounds the continents and ex- region for good sound coupling into the channel with a tends to about 11.4 km beyond the continental shelf, corresponding effect on the amount of noise coupled. Fig. l(a). A similar, probably thinner, band surrounds The relative magnitude of the contribution of this com- all islands. Sound from any generator operating within ponent of the noise to the total SOFAR channelnoise is these bands would be expected to eventually find its way difficult to visualize mentally. However, it involves a into the deep sound channel (assuming favorable condi- "straightforwardfl range-dependent propagation calcu- tions). The received level would depend on the distance lation, which can be handled by current propagation to the source, the number of reflections, and the bot- models. "-19 tom loss characteristics of the slope and shelf. In the The second mechanism, which is important for the case of seamounts or guyots (an isolated submarine introduction of noise into the deep sound channel is the peak with a flat top), a similar ring or annulus would '6downslope conversion phenomenon. " This refers to be expected [Fig. l(b)] such that, the sound from hny the conversion of bottom reflected/surface reflected surface source within the annulus would be reflected (BR/SR) raypaths into continuously refracted (RRR) into the deep sound channel. However, since an aver- raypaths, when a sloping bottom is encountered at aged slope for the upper part of a seamount or guyot is depths within a deep sound channel. Upon reflection, about 15°,23only the steeper rays (about-30" on a direct the angle of the reflected ray is less than the angle of intersection and somewhat less on an oblique intersec- ' the incident ray by twice the slope angle. After re- tion) will enter the sound channel after reflection. As- peated reflections, the incident ray angle can become suming 1000 m of height within the channel gives a sufficiently reduced to eliminate additional boundary thickness of the annulus equal to about 4 km and radius interactions. By this mechanism, sound generated in somewhat less than that of the guyot or seamount at shallow water, by the usual sources (ships, winds, channel depth. With the largd number of seamounts, waves, surf, biologics, etc. ), can enter the deep sound guyots, and islands within an ocean, the extensive

1010 J. Acoust. Soc. Am., Vol. 69, No. 4, April 1981 R. A. Wagstaff: Low-frequency ambient noise 1010 L * ASSUME UPPER 1000 m --I1 4 km OF SLOPE OF CHANNEL WITHIN SOUND CHANNEL- *

CHANNEL AT NEAR HORIZONTAL ANGLE FIG. 1. Conversion of high- angle raypaths to nearby hor- izontal by reflection off a sloping bottom.

SOUND SPEED

REGION WITHIN CHANNEL (ASSUME 1000 rn) above-shelf surface belt surrounding the ocean, and the introducing noise into the deep sound channel and to the large number of ships transiting at high latitudes, it is presence of "channeledm noise of sufficient level to evident that the amount of noise that could get trapped dominate the measurement. What remains to be esta- in the deep sound channels could be considerable. blished is that this component is important to the un- derstanding of the noise field, and neglecting it could be One need not rely entirely on the establishment of a serious error. This can best be done by example. mechanisms for channeling the noise to suggest the possible existence of a major component. Its exis- First, consider what can happen when it is neglected. tence has been established by direct measurement, at Weston, for example, presents a model for noise depth least in the case of the Northeast Pacific Ocean. This dependence in Ref. 15. He compared results from his was accomplished by measuring the vertical direction- model with data measuredin the Northeast Pacifi~.~By ality of the noise with an array having beamwidths small a convenient "trickn15 he manufactured one below-axis compared to the channel-limiting ray-angles (less than datum from another datum that was above the channel about 15" or 20" on channel axis). These measure- axis. This improved the agreement between modeling ments, reported by Anderson4 were for four depths and measurement which was then pronounced '

101 1 J. Acoust. Soc. Am., Vol. 69, No. 4, April 1981 R. A. Wagstaff: Low-frequency ambient noise 101 1 model predicted the maximum, between 3000 and 4000 without the contribution from the noise which is gene- * m. (Note: FLIP was the.measurement platform in rated at adistance (chiefly above the continental shelves) September 1973 for the measurements reported by both and eventually becomes trapped within the channel Anderson4 and Morris. 5, The measured result, unlike The open circles are for calculations which include the the modeled result, showed a monotonic decrease with channeled noise. The solid and dashed horizontal lines depth. I indicate measured levels with error bounds. The dashed lines correspond to the FLIP data reported in Now consider a case inwhichthe SOFAR component Ref. 5. The solid lines correspond to data measured of the noise is included by the noise model. The loca- , by an Acoustic Data Capsule (ACODAC) in the same tion for this example is the Northeast pacific' in the general area but to the north of the FLIP location. The vicinity of the measurements reported by Anderson4 profile at the FLIP location, as well as bottom and and Morris. The average distribution for all ships critical depths at both measurement locations, are also in the Northeast Pacific ocean given in Ref. 24 was used given. as an estimate of the shipping during the measurement time period. The ocean was then divided into several Figure 2 illustrates that when the SOFAR channel noise "pie sliceyJsectors about the modeled location. This is not included, the shape of the noise versus depth permitted adequate representation of the variations in profile is similar to that obtained by WestonI5 but not range to the continental shelves surrounding the basin in agreement with measured data, Inclucling the SOFAR and in the shipping density above them. The RANDI component, on the other hand, brings the profile in model12 was then used to calculate the noise level at line with the measured data. The maximum differences several frequencies and depths. This was done two between the two modeled results in Fig. 2 is about 7 dB different ways. The first excluded all noise which ar- at the channel axis. Below critical depth, the two rived by way of the deep sound channel (RRR compo- modeled results were identical as a result of RRR pro- nent). The second calculation included it. In addition, pagation being confined to the channel. since RANDI calculates the noise as a function of verti- The vertical array responses to the modeled field for cal arrival angle, the response of a vertical array, I~ 100 Hz are compared in Fig. 3 to the "averagedu data having the same hydrophone spacing and aperture as of Ander~on.~The solid lines are the modeled results that used by Ander~on,~was calculated. The ambient- for five different depths. The erratic behavior of these noise calculations were then compared to measured responses for angles greater than 20" from the hori- data. zontal (0 declination/elevation angle) is due to grating Figure 2 presents the omnidirectional noise level lobes caused by being above the half-wavelength design results obtained for 100 Hz. These results are typical frequency of the array. The dashed curves are the of similar comparisons at other frequencies. The corresponding averaged measured results. Because solid circles correspond to the modeled 100-Hz levels of the grating lobe problem, the measured results were

SOUND SPEED. m/s

measured C--l ACODAC rile A

c--4 FLIP s~teB

FIG. 2. Comparison of modeled am- bient-noise depth-dependence re- sults at 100 Hz for two cases (with and without SOFAR channel noise) with data measured in the North- eastern Pacific Ocean.

1012 J. Acoust. Soc. Am., Vol. 69, No. 4, April 1981 R. A. Wagstaff: Low-frequency ambient noise 1012 4 FIG. 3. Comparison of stim- ulated vertical array res- ponse to the modeled 100-Hz ambient noise, at five depths, with the corresponding mea- sured resu~ts.~

40 30 20 10 0 10 DE ANGLE ldegreer from hornrot.

reported only for angles less than about 20" from the noise with depth in the North Pacific," J. Acoust. Soc. Am. horizontal. The agreement between the measured and 66, 1446- 1452 (1979). modeled results is generally good, within two or three 5~.B. Morris, "Depth dependence of ambient noise in the northeastern Pacific Ocean," J. Acoust. Soc. Am. 581- decibels. Without the near horizontal arrivals from 64, 590 (1978). the deep sound channel, the modeled and measured re- 6~.J. Urick, "Correlative properties of ambient noise at sponses would have deep notches near horizontal at all Bermuda." J. Acoust. Soc. Am. 40, 1108-1111 (1966). channel depths. Even if the noise source levels were 'A. C. Kibblewhite, J. A. Shooter, and S. L. Watkins. "Exam- increased in the model results to compensate for the ination of attenuation at very low frequencies using the deep- lack of the SOFAR component near the horizontal, the water ambient noise field," J. Acoust. Soc. Am. 60, 1040- t vertical directionality results could not agree. It is 1047 (1976). 'J. H. Wilson, "very low frequency (VLF) wind-generated also obvious from Fig. 2 that the depth dependence re- noise produced by turbulent pressure fluctuations in the sults, as well, cou1.d not agree. An increase in source atmosphere near the ocean surface," J. Acoust. Soc. Am. 66, level merely shifts the depth profile in level; it does not 1499-1507 (1979). change the shape. 9~.M. Gray and D. S. Greeley, "Source level model for pr* peller blade rate radiation for the world's merchant fleet," IV. CONCLUSIONS J. Acoust. Soc. Am. 67, 516-522 (1980). 'OR. A. Wagstaff, "Horizontal directionality estimation consid- A simplistic concept of the noise field may be ade- ering array tilt and noise field vertical arrival structure," quate for some cases, but it is not when the noise field J. Acoust. Soc. Am. 67, 1287-1294 (1980). is complex. Unfortunately, unless one has a basic "J. H. Wilson, "Application of the Fourier Series method to understanding of noise fields which are complex, he the detection and localization of signals embedded in a noise," cannot distinguish one from the other. Considering J. Acoust. Soc. Am. 64, 1064-1068 (1978). "R. A. Wagstaff, "RANDI: Research Ambient Noise Direc- range-variable acoustic propagation to the basin boun- tionality Model,'' Naval Undersea Center, TP 349, April daries coupled with distributions of distant-noise 1973. sources is a complexity which accounts for low-fre- ''R. J. Talham, "Ambient-sea-noise model," J. Acoust. Soc. quency noise in the deep sound channel. This compo- Am. 36, 1541-1544 (1964). nent of the noise has been generally overlooked or ig- 14science Applications, Inc., "Review of Models of Beam-Noise nored. This can result in misguided experiment de- Statistics," SAI-78-696-WA, November 1977. sign and incorrect interpretation of the results. 15~.E. Weston, "Ambient noise depth-dependence models and their relation to low-frequency attenuation," J. Acoust. Soc. Am. 67, 530-537 (1980). I%. B. Morris, "Preliminary Results on Seamount and Con- tinental Slope Reflection Enhancement of Shipping Noise," SIO 'G. M. Wenz, "Acoustic ambient noise in the ocean: spectra Reference 75-34, November 1975. and sources," J. Acoust. Soc. Am. 34, 1936-1956 (1962). H. Watson and R. W. McGirr, "RAYWAVE 11: A Propaga- ' "w. '5. H. Wilson, "Spatial correlation of wind-generated noise at tion Loss Model for the Analysis of Complex Ocean Environ- very low frequencies," J. Acoust. Soc. Am. 60, 315-319 ments," Naval Undersea Center, TN 1516, April 1975. (1976). "B. G. Roberts, Jr., "Horizontal-Gradient Acoustical Ray- 3~.H. Axelrod, B. A. Shoomer, and W. A. Von Winkle, "~erti- Trace Program TRIMAIN," Naval Research Laboratory Re- cal directionality of ambient noise in the deep ocean at a site port 7827, 1974. near Bermuda," J. Acoust. Soc. Am. 37, 77-83 (1965). 19~.S. Perkins and R. N. Baer, "A Corrected Parabolic-Equa- 4~.C. Anderson, "Variation of the vertical directionality of tion Program Package for Acoustic Propagation," Naval Re-

101 3 J. Acoust. Soc. Am., Vol. 69, No. 4, April 1981 R. A. Wagstaff: Low-frequency ambient noise 1013 search Laboratory Memorandum Report 3688, 1978. "F. P. Shepard, Submarine Geology (Harper and Row, New 20~.Northlop, M. S. Loughridge, and E. W. Werner, "~ffect York), Chap. 10. of near-source bottom conditions on long-range sound propa- "5. Northrop and R. A. Frosch, "Seamounts in the North gation in the ocean," J. Geo. Res. 73, 12 (1968). American Basin," Deep-sea Research, Vol. 1 (1954). 'ID. B. Officer, Introduction to the Theory of Sound Transmis- 24~.Ross, J. Mahler, and L. P. Solomon, "World shipping sion with its Application to the Ocean (McGraw-Hill, New distributions," J. Acouet. Soc. Am. Suppl. 1, 58, S122 (19751, York, 1958), p. 159. and an unpublished manuscript (1974).

. I 1014 J. Acoust. Soc. Am.. Vnl 69, No. 4. Aoril 1981 R. A. Wagstaff: Low-frequency ambient noise 1014 INITIAL DISTRIBUTION

Copies Copies

MINISTRIES OF DEFENCE SCNR FOR SACLANTCEN

MOD Belgium SCNR Belgium DND Canada SCNR Canada CHOD Denmark SCNR Denmark MOD France SCNR Germany MOD Germany SCNR Greece MOD Greece SCNR Italy MOD Italy SCNR Netherlands MOD Netherlands SCNR Norway CHOD Norway SCNR Portugal MOD Portugal SCNR Turkey MOD Turkey SCNR U.K. MOD U.K. SCNR U.S. SECDEF U.S. SECGEN Rep. SCNR NAMILCOM Rep. SCNR NATO AUTHORITIES 6 NATIONAL LIAISON OFFICERS Defence Planning Committee NAMILCOM NLO Canada . SACLANT NLO Denmark SACLANTREPEUR NLO Germany CINCWESTLANT/COMOCEANLANT NLO Italy COMIBERLANT NLO U.K. CINCEASTLANT NLO U.S. COMSUBACLANT COMMAIREASTLANT NLR TO SACLANT SACEUR NLR Belgium C INCNORTH NLR Canada ' CINCSOUTH NLR Denmark COmAvSOUTH NLR Germany COMSTRIKFORSOUTH NLR Greece COMEDCENT NLR Italy COMMARAIRMED NLR Netherlands CINCHAN NLR Norway NLR Portugal NLR Turkey NLR UK NLR US Total initial distribution SACLANTCEN Library Stock Total number of copies