Propagation of anthropogenic noise in the ocean

PropAnthNoise-Hovem: June 13, 2012

Propagation of anthropogenic noise in the ocean

Jens M. Hovem and Tron Vedul Tronstad

The Norwegian University of Science and Technology (NTNU), and SINTEF ICT Trondheim Norway

Abstract— Acoustic modeling technique is used to study long range propagation of airgun signals with emphasis on the low frequencies having impact on fish species of commercial interest, the frequency range from 25 Hz to 300 Hz. The acoustic propagation model is based on ray theory and can deal with range-dependent bathymetry and depth-dependent sound speed profiles. The bottom is modeled with a sediment layer over a solid rock and requires input values for geoacoustic parameters of the sediments layer and the rock with compressional speed, shear speed and absorptions coefficients. The source level and directionality is modeled for an airgun array with an arbitrary number of airguns of different sizes.

The model calculates the complex spectrum and the full-waveform time response of the sound field to long distances from which a number of useful field descriptors and measures are derived. The sound pressure level at selected frequencies and the sound exposure level are calculated as function of range and compared with the threshold reaction level based on auditory and startle response levels for cod. A critical distance is defined as the range when the sound levels is lower that the reaction level. The environmental impact on the sound propagation is studied and discussed on the basis of a number of scenarios with different bathymetry, bottom properties and seasonal variations of sound speed profiles. Using realistic values for source level and directivity of an airgun array, the critical distances were found to vary from 8 km to more than 50 km, depending on the sound speed profiles and the degree of bottom interaction, but the values of the critical distance are very sensitive to the assumed value of reaction threshold. The model has been tested and verified on data obtained at a real seismic survey conducted in the summer of 2009 at Vesterålen – Lofoten area (Nordland VII). In this experiment signals were recorded at fixed hydrophone positions as the seismic vessel approached from a maximum distance of 30 km toward the receiving positions. The same situation was modeled using available geological and oceanographic information as input to the acoustic model. The agreement between the real and recorded signals and the model results is good. This indicates that acoustic-biological modeling may be useful in the design and planning of seismic surveys to minimize the conflict between surveying and fishing.

Paper presented at the 35^th Scandinavian Symposium on Physical Acoustics, 29^th January - 1^th February 2012, Geilo, Norway.

1

Int 1

Oceans e low atten noise is anthropo oil/gas p purpose, seismic the low f Hawkins emission their natu in certain and the demonst increased resulting This has the impa or to rest to defin knowled The biol behavior controlle natural r found in signal. A away fro than the fish is li response The acou long-ran the envir and the p generally signals f

roductio

environment nuation of so

mainly gene ogenic noise rocessing eq such as by exploration.

frequency ba s, (1973), Po n coincides w ural activitie n areas where

Norwegian S trated the so

d fish activi g in reduced c

raised quest act on fishing

trict the surv e the minim dge and exper

logical issue r. Biological ed condition reactions typi n the literatur A startle thres om the sound auditory leve kely to show es, Pearson et

ustic issue is nge propagati ronmental co physical prop y a sound sp from a shallo

on

are noisy be ound in wate erated by win sources, ship quipment and y active sona Anthropoge and from app opper, et al.

with the frequ s, or even ca e there also i Sea with on-

und of air-g ty and to ch catches.

tions and dem g is minimiz veys to times mum distanc rience in sou is to unders information with fish in ical for free re for differe shold can be d source. The

els, Eaton et w other types t al. (1992).

s to understa ion depends onditions, in perties of the peed minimu ow source pr

ecause acoust er compared, nd-driven oce

pping noise, d wind mills.

ar, especially nic noise oft proximately f

. (2003), Sim uencies of pe ause physical

is important c -going confl guns signals

hanges in sc

mands for im ed. This cou of the year w ce. Evidently und propagati stand the feat may be obta n captivity in swimming fi ent species a defined as t e values that al. (1995), K s of behavior

and the physi on the envir particular th e bottom. Fo um near the ropagate with

tic waves can , for instance eans waves a noise genera There are als y high powe ften dominate from 10 Hz a mmonds and erception of l damages. A commercial f

icts between may exceed chooling and

mposing rules uld be by req with no fishin y this must ion in the oce tures of the ained from m n pens or tan fish. Auditory and these val

the sound lev exists the sta Karlsen et al.

ral changes a

ics of low fr ronments. Th he oceanograp

r instance, d sea surface h little attenu

n propagate t e, to sound i and rain. In a ated from pil so the acoust er military so

es over the n and upwards d MacLenna sea mammal An example is fishing intere n seismic exp

d the thresho d water posit

and restricti quiring a min ng. The critic

be based o eans.

sound that tr measurements nks, with the y threshold v lues are obta vel the cause artle thresho (2007). How at lower sou

requency pro he propagatio aphic parame during winter resulting in uation to lar

to long dista n air. The na addition ther ling and drill tic signals de onar, and ai natural ambie

to 1000 Hz, an (2004). T ls and fish an s seismic exp est. This is th ploration and old for alarm tion up to a

ion on the se nimum distan

cal issue is to on both biol

riggers react s and observ e obvious ris values as fun ained by exp strong and r ld levels are wever, it is im nd levels tha

opagation in on of underw ters, the topo r seasons in n an acoustic rge distances

ances due to t atural ambie re are a wide

ling, noise fr eliberately sen

r guns used ent noise, es

or more, Ch his frequenc nd may there ploration for he case in the d fishing. St m responses

distance of

ismic survey nce to fishing o establish th logical know

tions or chan ation of a few sk of not obs nction for fre posing fish to

rapid reactio about 60-70 mportant to no

an those cau

the ocean an water sound d ography of th northern wat

surface chan . In other ar

the relative nt acoustic e variety of rom subsea nt out for a in marine specially in hapman and cy band of efore affect

oil and gas e North Sea tudies have leading to f 30-50 km

ys such that g activities, he rules and wledge and

nges in fish w fishes in serving the quency are o harmonic ns directed 0 dB higher ote that the sing startle

nd how the depends on he seafloor ers there is nnel where reas, and at

2

other seasons, the conditions may be entirely different, perhaps with downward refraction and strong bottom interactions with a significant portion of the energy disappearing into the bottom. Most of the knowledge we have today on these issues is developed by military research and development in connection with passive sonar to detect submarines at large distances, which involves about the same frequency band and distances as relevant to disturbances and effects on fish behavior.

It is important to distinguish between stationary broad band noises from for instance a passing ship and the impulsive noise from an airgun or other similar sources. First it is not clear to translate a fish reaction to a steady harmonic tone or a burst of tone to the reaction of the impulsive sound from a nearby airgun.

At longer ranges we are dealing with the propagation an oceanic waveguide with multiple reflections from the sea and bottom causing a significant time spread of the received signal that may last several seconds. In such situations the peak pressure is a time-space varying consequence of the coherent effects of multi-path contributions and bandwidth. Furthermore there are often confusions and misunderstanding with the respect to the characterization of sound and source levels, in particular for transient and impulsive sound. Carey (2006) has written a summary paper on this problem and defined a number of matrices, units and recommended practices.

To assess these factors quantitatively it is necessarily to use mathematical /numerical modeling tools that can take all the effects of the oceanography, the bathymetry and the geophysical properties of the bottom into account. Modeling of propagation conditions has always been an important issue in underwater acoustics and there exists a wide variety of mathematical/numerical models based on different approaches. The most common models are based ray theory, expansion in normal modes, models based on wave number integration technique and models based on the solution of the parabolic equation. For an overview of these models and for further references, see Jensen et al. (1993). Although the effects of anthropogenic noise on marine life is an active field of research and has high public interest, it is surprising that the propagation of sound to large distances are hardly mention in the works dealing with bioacoustics. Noteworthy exceptions are the two companion papers by Erbe and Farmer [2000(a) and 2000(b)] where they first presented a software model to estimate impact zones on marine mammals and then applied the model to case of icebreakers affecting Beluga whales.

This paper describes the use acoustic modeling technique to study long range propagation of impulsive sound with emphasis on the low frequencies having impact on fish species of commercial interest, the frequency range from 25 Hz to 400 Hz. After first describing the model and the required inputs, modeled results are compared with results obtained from a joint seismic-acoustic survey conducted in the summer of 2009 at Vesterålen – Lofoten area (Nordland VII). In this experiment signals were recorded at a fixed hydrophone positions as the seismic vessel approached from a maximum distance of 30 km toward the receiving positions. The same situation was modeled using available geological and oceanographic information as input to the acoustic model. The agreement between the real and recorded signals and the model results is good, Tronstad and Hovem (2011), Hovem and Tronstad (2012).

3

Mo 2

PlaneRay eigenray eigenray initially times an turning p calculati the eigen routine.

The cont function by multi feature o of ray h PlaneRay found to interest t articles b fields fro The inpu (1) A

t (2) A (3) A c (4) A Items (1 interactio descripti acoustic

C 2.1.

The prog depth. In have a co

odel desc

y is a ray tr y contribution ys connecting to launch a h d informatio points. This ons. Subsequ nrays, within

tributions of for propaga iplying with of sorting and history, whic y model has be sufficien to commerci by Hovem (2 om at differen uts to the mod A sound spe therefore ran A bathymetry A descriptio coefficients f A record or m 1) and (2) a

onal data b ions such as values for di

haracteriza gram calcula n order to un ommon refer

cription

racing propa n. The cruci g a source p high number n on where t s information uently, and u n the span o

f all the eigen tion between

the frequen d interpolatio ch enables t

been extensi nt accurate fo ial fishing. M 2008, 2011). I

nt environme del are:

eed profile i.

nge dependen y chart from on of the s for each laye model of the are normally banks. Seaflo gravel, sand ifferent types

ation of sou ates the transm nderstand and

rence. In free

agation mod ial task in th position with

of rays and the rays have n, the ray his using the stor

of initial ang

nrays are coh n the source ncy function

on scheme is the user to ively tested a or frequencie More inform In this paper ental conditio

.e. sound spe ncy in the sou which depth seafloor, pos er.

sound sourc easy to ob oor informat

or silt etc. T s of bottom m

und fields a mission loss d compare th e space the g

del that calcu his type of r the receiver calculates ca e been reflect story, is stor ed ray histor gles rays, are

herently sum and receiver

of the sourc s the automat study the va against the O es above 25 H mation on the

r we concentr ons.

eed as functi und speed can h versus rang

ssible layere e as function btain, either tion can be There exist se materials, Ha

and propaga as function he transmissi geometrical lo

ulates the so ray modellin r positions.

alculate their ted from the red in the com ry the trajecto

e found by

mmed to prod rs. The time r ce signal and tic sorting an arious multip OASES (SAF

Hz, which is e PlaneRay M

rate on the us

ion of depth nnot be treat e can be extr ed, with sou n of frequency by in situ m e more diffi

everal papers amilton (1987

ation condi of frequency on loss for th oss is that of

ound field by ng is to find

The approac trajectories t surface and b mputer for u ories, travel t

a special so

duce the freq response is s d Fourier tra nd separation path contrib FARI) model sufficient fo Model can b

se of the mod

h. Only one p ed.

racted.

und speed y and radiati measurement ficult to obt s in the litera 7), Hamilton

itions y and range t

he various co f spherical sp

y coherent a d the all the ch used in P together with bottom or go use in all the times and am orting and in

quency-doma subsequently ansformation ns of the diff butions separ l by Schmidt or application be found in r

del for predic

profile is ac

density and on direction ts or from n tain except ature that giv and Backma

to receivers a onditions it i preading with

addition of significant PlaneRay is h the travel one through e following mplitudes of nterpolation

ain transfer y calculated n. A useful ferent types rately. The t (1987 and n to fish of reports and ction sound

cepted and

d abortions

national or for verbal ves relevant

an (1982).

at the same is useful to h which the

4

sound intensity decays proportionally to distance squared. In an ideal waveguide with constant water depth and sound speed the geometrical loss follows a cylindrical spreading law with the sound pressure decays proportionally with distance. It follows that the geometrical spreading in an ideal waveguide may be approximated with a spherical spreading law at short distance and cylindrical spreading at longer distance. A combination of the two spreading laws may be expressed by the equation

1

2 2

2

10 2

TL 10log 1 .

t

r r r

  

 

 

     

(1)

This expression gives the asymptotic behavior

10

10 10

20log ( ),

TL 20log ( ) 10log , .

t

t t

t

r r r

r r r r

r

 

     

  

(2)

In these equations rt is a transition range where the transmission loss goes from spherical close to the source and to cylindrical at long ranges. A reasonable value for rt is a value close to the water depth.

Equation (2) may be used for rough calculation of the transmission loss, but is not included and neither are effects of oceanography and bathymetry. Frequency dependent acoustic absorption in the water is implemented using the expression of Francois and Garrison (1982).

The sound exposure level SEL is an energy (E) measure obtained by integrating the square of the sound pressure p²(t) and normalize with respect a reference sound pressure and an exposure time tref

 

10 2

2

10log

ref ref

SEL E

p t E p t dt

 

  





(3)

The reference sound pressure pref is chosen to be 1Pa and the reference time tref equal to one second.

The SEL values and the spectral levels of selected frequency components may be used to characterize the loudness of airgun sound and to compare with assumed reaction thresholds of different species of fish.

Since the model calculates the complete waveform of the sound other measures can easily be implemented.

5 So 2.2.

Informat source s measurem absence frequentl frequenc

F

A 2.3.

A typica direction different waveleng calculati With ref Qn,m().

and the (

In the in-

ource pulse tion on the s signal may ments witho

of such info ly used in s cy spectrum i

Figure 1. T d

irgun array al layout use n and Ny gun

t sizes distrib gths it is cle ons of the ac ference to Fi The signal f (angular) freq

S -line directio

e and array source signa be availabl out contamin ormation, th seismology, is shown in F

Time pulse a discussion.

y directivit ed in comme ns in the cro bute in over ear that the d coustic field.

igure 2 the from the arra quency ω, ca

( , , ) S    



on where the

-1 0 1

Amplitude

0 5 10 15 20

Spectrum amplitude

y directivity al and direct

le for the s nation by ref e model use Sherif and Figure 1.

and frequenc

y

ercial seismic oss direction an area of 2 directivity o

positions of ay in the far an be express

 

, _{n m}, ex

n mQ 

azimuth ang

0 50

1 0 1

0 50

0 5 0 5 0

y

tivity is requ surveyor, bu flections from es a syntheti

Geldhart (1

cy spectrum

c surveys is n with. Typi 20 x 25 m. C f the airgun

f the guns ar field as func sed by



xp i x_{n m}c c





gel  is zero,

100 15

Time – ms

100 15

Frequency –

uired to dete ut the progr m the sea su ic source sig

959). An ex

of the 50 H

shown in Fi ically, there Considering t array is imp

re denoted b ction of eleva

cos cos y_n this reduces

50 200

50 200

Hz

ermine the so ram requires

urface or fro gnal in the f

xample of a

Hz Ricker sou

igure 2 with may be aro the size of th portant and m

by (xn,yn ) an ation angle 



, cos sin

n m  

to a line arra

250

250

ound field. R s far field om the botto form of a Ri a Ricker pul

urce signal u

Nx guns in und 30 activ he airgun arr

must be incl

nd source str

, the azimut



ay of sources

Records of directional om. In the icker pulse lse and its

used in the

the towing ve guns of ray and the lude in the

rengths are th angel ,

(4)

6

   

, ,

( , , 0) _{n m n m} exp _ncos .

S Q i x

c

   



In the across direction perpendicular to the towing direction  = /2 with

   

, ,

( , , / 2) _{n m n m} exp _mcos ,

S Q i y

c

    



Figure 2. Air gun array with Ny lines each with Nx guns

Equations (4) and (5) are general expressions valid for any number of sources with different frequency spectrum and locations, but requires that the source functions are known for each of the guns in the arrays. In cases that such detailed information is not available the following approach is used to estimate the directivity function of the airgun array. The approach is based on the assumption that all the guns in the array produces pulses of the same shape, but with a peak pressure amplitude that depends only on volume of the pressure chamber, given that the pressure in the all guns are the same. Thereby the source strengths in the above equations for the beam pattern are replaced by volume weights

   

, 0 _{n m} .

Qn m  Q  V^

(7)

In this expression Vn,m is the normalized volume of the individual guns and the exponent  has the value of 1/3 [Caldwell and Dragoset (2000)]. Q0(ω) is a frequency function of the common source signal, which here is taken as a Ricker pulse as described in the previous paragraph. This approach is a relatively poor approximation to the signature from a single air gun, but a reasonable approximation to the signature of all the guns combined since the firing times and the sizes of the individual guns are selected and synchronized to produce a single sharp pules without with a minimum of bubble oscillations.

In the in-line direction the transmitted signal from the array is

x y

Ny

Nx

Qn,m

(xn ,m)

7

   

0 1

( , , 0) ^x exp cos .

N

n n

n

S Q q i x

c

     



 

 



Where qn represent the relative strengths of the individual airgun sources and xn are their positions along the x-axis. The far field directivity function as function of elevation angle and frequency is then

 

1

( , ) ^x exp cos .

N

n n

n

B q i x

c

   



 





As an example we the values from an airgun array used in 2009 in a seismic survey at Nordland VII field are

(10)

Figure 3 shows the resulting beam pattern for the frequencies of 50 Hz and 100 Hz. For 50 Hz the reduction in transmitted level in the horizontal directions is about -5 dB compared to level in the vertical direction, At 100 Hz there is a side lobe reaching a level of about – 15 dB compared with the level in vertical direction. The maximum amplitude is in the vertical direction and in the modeling of the time and frequency responses the maximum peak amplitude is scaled up to the level specified by the surveyor, which is 255 dB re 1 Pa in the case of the Nordland VII survey. From Figure 3 is evident that the array directivity is important for the accurate calculations of the sound field.

Figure 3. Modeled directivity of the airgun array for 50 Hz and 100 Hz calculated for the weights and positions given by equation (10)

 

 

0.2308 0.1963 0.1679 0.1656 0.1350 0.1044 , 0 5.0 7.5 11.0 13.0 16.0

n n

q x





-20dB 30

210

60

240

90

270 120

300 150

330

180 0

Array beam pattern for 50 Hz and 100 Hz Freq. = 50 Hz Freq. = 100 Hz

8 Th 2.4.

The min or chang species i MacLenn Figure 4 about 51 threshold and rapi behavior behavior response a reactio threshold dependen discussio

F

he bio-acou imum distan ge in behavio

is difficult to nan (2004)].

4 showing the 10³ Pa and d level in Fig

d reactions d ral changes a ral changes a es. To give an on threshold d. We have nce is weak on and must n

Figure 4. A C K

ustic model ce between t or is defined o ascertain an However, t e auditory an

the startle re gure 4 is abo directed awa at lower soun

and alarm re n indication d of 310⁷ P

e also negle in the freque not in any ca

Auditory and Chapman and Karlsen et al.

10¹ 10³ 10⁴ 10⁵ 10⁶ 10⁷ 10⁸ 10⁹

Sound pressure (rms) – Pa

l

the source an d as the criti nd is an area the problem nd startle thr esponse thres out 60-70 dB ay from the

nd levels tha esponses at s of possible m

Pa which tr ected the fre

ency band of ase be consid

d startle thre d Hawkins ( . (2007).

nd a fish popu ical distance a of active re with definin reshold for A shold is abou B higher than sound sourc an those caus

sound levels maximum int ranslates into equency dep f primary int dered as more

eshold for A (1973) and C

10² Frequency –

ulation requir e. The actual esearch not d ng a reaction Atlantic cod.

ut 510⁷ Pa.

n the auditory e, but the fis sing startle r s 20-25 dB l teraction ran o 150 dB re pendence in

terest. This t e than an illu

Atlantic cod.

Chapman (19

Hz

Auditory thre Startle thresh

red to avoid l reaction thr discussed her n threshold c At 50 Hz th Note that th y level. Start sh is likely t response. Som lower than th nges, we have e 1Pa., wh the reactio threshold val

stration.

Auditory th 73). Startle t

10³ eshold hold

any significa reshold for t re [Simmond an be illumi he auditory th

e startle soun tle responses

to show othe me studies h he threshold e in this stud hich is close on threshold

lue is only in

hreshold data threshold dat

ant reaction the various ds, J and D.

inated with hreshold is nd pressure s are strong er types of have shown d for startle dy assumed e to startle

since the ntended for

a are from ta are from

9

Exp 3

The Nor the sum observat observe t the analy recording The inpu the bathy along the Accurate indicated where th suggeste bottom d wave att Figure 6 values q agreeme with the compare km. Obs distances

F

perimen

rwegian Petro mmer of 2009 tions and me

the degree to ysis, line 134 g the sound r ut for the mo ymetry extrac e seismic line e information d before. In t he bottom in ed the follow

density 2500 enuation 1.0 6 shows the m quoted above

nt between t e increasing ed with the a

serving the f s than this

Figure 5. M

ntal verif

oleum Direct 9. This was asurements w o which comm 44, is about 3

received from deling is the cted for offic e and the ray n about the a this case we u

n the area i ing values to kg/m³, comp dB/λ.

measured an e with no fu the measured

depth at ra ssumed react fish behavio

Modeled sou

1480 0 50 100 150 200 250

Sound speed

Depth – m

Norland VII Lin

fication

torate (NPD) s a regular c were conduc mercial fishe 30 km long a m the approac average sou cial sea maps y traces from acoustic prop used informa is characteri o use in the m

pressional w

nd modeled s further attem

d and modele anges over a

tion threshol or and fishin

und field from

d – m/s ne 1344

0

Nordlan

) conducted commercial cted to obtain es was affecte and passing d ching seismi und speed pro

s. Figure 5 sh a source at 6 perties of the ation from th ized as grav modeling of l wave attenuati

sound expos mpts to obtai ed levels is g about 7 km.

ld given in e ng indicate th

m a source at

5 10

Ra Sd=6 m, Rd=83

nd VII

a seismic sur marine seis n calibrated r

ed, Løkkebor directly abov c vessel towi ofiles measur hows the soun 6 m depth out e sea floor ca he Geologica vel. The tabu

line 1344. Co ion: 0.1 dB/λ

ure level (SE in a better f

good, both s The measu quation (3), hat there we

6 m depth to

15 20

ange – km 3 m, Angles= -13: 1

rvey off the mic survey, records of th rg et al., (201 ve a hydroph

ing an airgun red at the day nd speed pro t to distance an be more d al Survey of N

ulated value ompressional λ, shear wave

EL) as funct fit with the showing that ured and mo

indication a ere no or lit

o a distance u

25 0 50 10 15 20 25 13

coasts of Ve but in add he airgun sou

10). The chos hone at a dep

n array at 6 m y of the expe ofile and the b

of more than difficult to o Norway (NG es of Hamilt l wave speed e speed: 600

tion of range measured v

level deceas odeled SEL critical dista ttle reactions

up to 30 km.

0 00 50 00 50

esterålen in dition other

unds and to sen line for pth of 83 m m depth.

eriment and bathymetry n 30 km.

obtained, as GO), (2010) ton [1987]

d 2000 m/s, m/s, shear

e using the values. The ses sharply

values are ance of 5-6

s at longer

10

Figure 6. Measured and modeled sound exposure level (SEL) as function of range for line 1344 in Nordland VII with comparison with the assumed SEL threshold level of 150 dB re 1Pa²s.

The sharp decrease in level is partly caused by the increasing water depth, but also losses associated with the bottom reflections. The values for the sound speed and attenuation yield the bottom reflection loss shown in Figure 7. The sound speed of 2500 m/s gives a critical angle of 42, but as can be seen from the figure there is a significant bottom reflection loss at lower angles. This propagation loss is caused by loss of energy due to sound absorption and conversion to shear waves in the bottom. Most important is the relatively high value of the shear speed and attenuation (600 m/s) and dB/λ).

Figure 7. Bottom reflection loss for values used to model line 1344. Compressional wave speed: 2000 m/s, bottom density 2500 kg/m³, compressional wave attenuation 0.1 dB/λ, shear wave speed 600 m/s and shear wave attenuation 1.0 dB/λ

10^-1 10⁰ 10¹

120 130 140 150 160 170 180 190 200

Norland VII Line 1344: Sd=83 m, Rd=6 m

Range – km SEL – dB re 1 Pa2 s

PlaneRay Measured Reaction threshold

0 20 40 60 80

0 2 4 6 8 10

Grazing angle – deg

Bottom reflection loss – dB

11

This is further illustrated by Figure 8 showing the transmission loss as function of range and for the cases when the shear speed in the bottom is as 600 m/s compared with the loss of zero shear speed in the bottom. The differences between the curves displayed in the two figures is the shear wave conversion loss and at ranges longer than 10 km the loss is about 20 dB. The dotted line is the geometrical transmission loss of equation (1), which gives a slope of 10 log(r) for ranges longer than the water depth at the source.

Figure 8. Modeled transmission loss for Nordland VII as function of range and for the frequencies of 50 Hz and 100 Hz: Left with for 600 m/s shear speed, and right with zero shear speed.

10⁰ 10¹

20

40

60

80

100

120

Range – km

Transmission loss – dB

Norland VII Line 1344: Sd=83 m, Rd=6 m Freq = 50 Hz Freq = 100 Hz TL_geo

10⁰ 10¹

20

40

60

80

100

120

Range – km

Transmission loss – dB

Norland VII Line 1344: Sd=83 m, Rd=6 m Freq = 50 Hz Freq = 100 Hz TL_geo

12

The 4

The purp distances use of th structure different fixed to consider

4.1 The The soun 10 E) at we prese the extre (Februar

F

e enviro

pose of this w s relevant for he model and e and comp t sound speed

6 meter as red.

e sound spee nd speed pro t different se ent only the r eme variation ry) and summ

Figure 9. M F

onment

work is to de r eliciting fis d to illustrate position of bo

d profiles, ba s typical for

ed profiles ofiles used in easons at diff results from t n that can be mer (July) con

Measured so February, Ma

1460 1470 0 50 100 150 200 250 300

So

Depth – m

1460 1470 0 50 100 150 200 250 300

So

Depth – m

velop a mod sh interaction

how the env ottom we co athymetry an r a seismic

n the simulat ferent years. T

the February e expected. T nditions are F

ound speed p ay, July and N

1480 1490 150 ound speed – m/s

Halten February

1480 1490 150 ound speed – m/s

Halten July

del for predict n and disturb vironmental f onstructed a nd bottom pro

airgun, but

ions were m The selected y and the July The ray trace

Figure 14.

profiles at in November

00 1510 14

0 50 100 150 200 250 300

Depth – m

00 1510 14

0 50 100 150 200 250 300

Depth – m

tion the soun bances for fis factors of the set of hypo operties. In a

both deep

measured at H profiles are y profiles sin es for the two

the Norweg

60 1470 1480 Sound spee

Halten

60 1470 1480 Sound spee

Halten Nov

nd field from shing activiti e oceanograp thetic, but r all the scenar

and shallow

Halten in the shown in Fig nce these two

o selected so

gian Sea (Hal

1490 1500 1510 ed – m/s

May

1490 1500 1510 ed – m/s vember

m an impulsiv ies. To demo phy, the batho

ealistic, scen rios the sour w receiving

Norwegian gure 9. In the o profiles are ound profiles

lten) for the

0

0

ve source to onstrate the ometry and narios with rce depth is depths are

Sea (64N, e following typical for s for winter

months of

13

Figure 10. Ray traces from a source at 6 m depth for winter and summer conditions at Halten in the Norwegian Sea, Only a few rays are shown to give a simple impression of the impact of the sound profiles variation,

The February profile is a typical winter profile where colder water nearest the surface gives a positive sound speed gradient, increasing sound speed with depth, over the whole water column. The consequence is upward refraction with concentration of sound in the upper part of water column. The bottom interaction is weak since the rays are bent upwards and the propagation is only weakly dependent on the acoustic properties of the bottom

The July profile is governed by a relative high surface temperature decreasing with depth and resulting in a sound speed minimum at a certain depth, in this case at about 50 m. A sound speed minimum creates a sound channel where the sound is trapped and propagates to very long distances. Except for propagation in the sound cannel most of the sound is reflected many times from the bottom giving a high degree of sensitivity to bottom parameters.

4.2 The bottom model

As already demonstrated the bottom reflection losses are important for low frequency sound propagation and in many cases is the determining factor for how far the noise may affect the behavior of fish. The bottom model used in this study has a sediment layer over a solid rock half space as shown in Figure 11.

The sediment layer with thickness D is modeled as fluid with sound speed cs and density s. The rock has sound speed crp, shear speed crs and density r. All waves are attenuated with absorption coefficients s,

rp and rs, measured in dB wavelength, respectively for three wave types.

Figure 12 shows a contour plot of the bottom reflection loss (dB) as function of incident grazing angle and frequency. The parameters are, cs= 1700 m/s, crp= 3000 m/s, crs= 600 m/s, s = 1800 kg/m³r = 2500 kg/m³, s = rp = rs =0.5/. The reflection coefficient and the reflection loss are functions of the product of the acoustic frequency f and layer thickness D represented as the vertical axis in the figure. The reflection loss Figure 12 can therefore be scaled to any layer thickness and frequency.

The reflection loss changes with frequency and for very low frequencies approaches the reflection loss of a homogenous rock bottom, but approaches the reflection loss of a uniform sedimentary bottom at high

14751485 0 50 100 150 200 250

Sound speed – m/s

Depth – m

Halten-Winter

0 10 20 30 40 50

0 50 100 150 200 250

Range – km Sd=6 m, Angles= -60 : 60

1480 1500 0

50 100 150 200 250

Sound speed – m/s

Depth – m

Halten-Summer

0 10 20 30 40 50

0 50 100 150 200 250

Range – km Sd=6 m, Angles= -10 : 10

14

frequencies. The two critical angles for the sound speeds of cs =1700 m/s and crs = 3000 m/s for the sediment and the rock, respectively 28 and 60 are clearly recognizably in the graph.

With zero shear speed in the rock and zero wave absorptions in the sediment and the rock, the reflection loss is zero for angles smaller than the critical angles, which in this example is 28 and 60 for high and low frequencies, respectively. However, the reflection loss of Figure 11 shows significant losses also in areas of low frequencies and low angles. The losses in these areas are consequences of energy lost by wave absorption and conversion to shear waves in the bottom. Figure 12 (left) shows the low-frequency and low angle area extends up to frequency- layer thickness product of 500 m/s. For a frequency of 50 Hz the propagation is significantly affected by properties down to about 10 m into the bottom.

Propagation of sound at low frequencies are very dependent on bottom reflection losses and this issue has been treated extensively in the literature, as for instance Hovem, Richardson and Stoll (1991.

This study compares the propagation using two bottom types

 Bottom type (I) Homogenous sediment with sound speed 1700 m/s, density 1800kg/m³ and attenuation 0.5 dB/ wavelength.

 Bottom type (II) Sediment layer with thickness 2 m over solid rock with compressional speed 3000 m/s, shear speed 600 m/s, density 2500 kg/m³. Both wave types with attenuations equal to 0.5 dB/wavelength.

Bottom type (I) results by setting the thickness of the sediment D to infinity and type (II) is with D=2 m.

Figure 11. Sea bottom model with a sediment layer over a solid rock half space Solid rock

Thickness = D cs= 1700 m/s

s = 1800 kg/m³

s =0.5/

crp= 3000 m/s, crs= 600 m/s,

r = 2500 kg/m³,

rp = rs =0.5/.

Sediment layer:

15

Figure 12. Bottom reflection loss (dB) as function of frequency and incident grazing angle for a layered bottom with a sediment layer over hard rock. The parameters are given in the text.

Figure 12 shows the bottom reflection loss for bottom type(I) and (II) as function of incident grazing angle and for the frequencies of 50 Hz and 100 Hz. For bottom type (I) the reflection loss is independent of frequency and the critical angle is 28 In the following discussion, bottom type (I) is referred to as a low loss bottom and bottom type (II) as a high loss bottom.

Figure 13. Expanded view of the bottom reflection loss (dB) as function of frequency and the incident grazing angle. Left: Bottom type (II), Right: Bottom type (I), which is independent of frequency. The parameters are given in the text.

Grazing angle – deg

Frequency–thickness product – m/s Bottom reflection loss – dB

20 40 60 80

500 1000 1500 2000 2500 3000

0 5 10 15 20

Grazing angle – deg

Frequency–thickness product – m/s Bottom reflection loss – dB

20 40 60 80

500 1000 1500 2000 2500 3000

5 10 15 20

0 20 40 60 80

0 1 2 3 4 5

Grazing angle – deg

Bottom reflection loss – dB

Bottom reflection loss – dB 50 Hz 100 Hz

0 20 40 60 80

0 1 2 3 4 5 6

Grazing angle – deg

Bottom reflection loss – dB

Bottom reflection loss – dB

16

4.3 The bathymetry

The effects of the bathymetry are investigated for the downslope and upslope propagation cases depicted in Figure 14. The water depth changes with 150 m over a distance of 30 km having a slope of about 0.5 %. As will be demonstrated, even a gentle slope of the magnitude may significantly influence the propagation and the critical distance

Figure 14. Examples of scenarios for discussing bathymetric effects on long range propagation.

The ray diagrams in Figure 14 shows several interesting and important features. Downslope propagation leads to a thinning of rays causing faster decay of the sound level with range than propagation in waters with constant depth. Upslope propagation yields a concentration of rays with range such that the geometrical propagation loss is initially reduces and the sound level increases with range until a point may be reached where the grazing angles of the rays reach 90, which signifies a cutoff in propagation.

(Back-reflected rays are ignored in the propagation model). In addition there is effect of bottom reflection losses, most pronounced for the winter conditions as observed earlier.

1480 0 50 100 150 200 250

Sound speed – m/s

Depth – m

Downslope - Winter

0 10 20 30 40 50

0 50 100 150 200 250

Range – km

1480 0 50 100 150 200 250

Sound speed – m/s

Depth – m

Upslope - Winter

0 10 20 30 40 50

0 50 100 150 200 250

Range – km

1480 1500 0

50 100 150 200 250

Sound speed – m/s

Depth – m

Downslope - Summer

0 10 20 30 40 50

0 50 100 150 200 250

Range – km

1480 1500 0

50 100 150 200 250

Sound speed – m/s

Depth – m

Upslope - Summer

0 10 20 30 40 50

0 50 100 150 200 250 Range – km

17

Sim 5

This sect demonst bathyme

5.1 Th The first type bot This case Figure 1 and (II).

source an

where tre

The actu response The red This tim propagat reflection higher re be estim

This esti homogen depth an calculate be observ Figure 1 Direction medium.

mulation

tion presents trate the imp etry may imp

he reference t case is the s ttom modeled

e is often ref 5 shows the The time s nd receiver h

tr

eal and tredar ual value of c es.

dotted lines me duration i

te at angles n loss and m eflection loss ated directly

tre

imate of the nous and flat nd with solid

ed on the bas ved, the time 16 shows th nality is not . The dotted

n results

s the modelin portance the act low frequ

e case, the P simple case w d as a homo ferred to as th time respon cale is in re has been subt

red real re

t r

 c re the real an

cred is not imp

in Figure 15 s mainly giv

closer to the may therefore ses and die o by the ratio

0

1

ed cos

crit

r

c 

 

time duratio t, but the esti or layered b sis of the high

e duration or he transmissi included in line is the g

ng results for e three facto uency propag

Pekeris wa with constan ogenous fluid he Pekeris wa nses as funct

duced time tracted. The r

,

ed

r

nd reduced tim portant as lo

5 represent a ven by the cr e horizontal e propagate t out more rap

of sound spe

2 0

1 ^b

t

c c

r c

  

  

n of the chan imate may al bottom. In the

hest wave sp responses m ion loss as the plot sinc geometrical t

r the environm ors of sound gation both se

veguide t sound spee d, characteriz aveguide and

ion of reduc with the gro reduced time

mes, respect ong as the ch

an estimate o ritical angle plane than t to long dista idly with ran eeds in the w

0 . c 



nnel impulse lso be useful e figure the e

eed in the bo may be very lo function of ce this is con transmission

ments descri d speed pro eparately and

ed and consta zed with sou d may serve a ed time and oss time dela e is defined a

tively, r is ra osen value r

of the durati of the botto the critical a ances. Rays w

nge. Thus the ater and the b

e response as in other case estimated tim ottoms of (I) ong for a har

range for th nsidered a pr

loss of equa

ibed in sectio ofile, the bo d in combina

ant water dep und speed, d as a reference

range for th ay of the tra as

ange and cred

esults in a go

on of the ca om reflection angle experie with steeper e time durati bottom as

ssumes that t es with mode me duration o 1700 m/s an rd bottom.

he two botto roperty of the ation (1), wh

on 4. The inte ottom compo

ation.

pth and with density and a

e case.

he two bottom ansmission b

is the reduc ood display

annel impulse n coefficient.

ence almost angles will ion of the im

the bottom is erately range of the impuls nd 3000 m/s (

om types (I e source and hich gives a s

ention is to osition and

a sediment attenuation.

m types (I) etween the

(11)

tion speed.

of the time

e response.

. Rays that no bottom experience mpulse may

(12)

s fluid-like, e dependent se response (II). As can

I) and (II).

d not of the slope of 10

18

log(r) for ranges longer than the water depth at the source. The transmission loss for bottom type (II) with shear wave conversion is significantly higher than for bottom type (I), the difference approaching 10 to 15 dB at the longer ranges. This is as expected from plots of bottom reflection loss in Figure 13

The critical range where the sound level drops below the threshold value of fish reaction can be defined in various ways. We have chosen to use either the SEL values or the levels of selected spectral 50 and 100 Hz. Figure 17 shows the SEL values as function of range calculated for the bottom types (I) and (II) and compared with the assumed threshold value of fish reation. In Figure 18 the spectra lines of 50 Hz and 100 Hz are compared with the same threshold. The difference between the sound level at 50 Hz and 100 Hz is mainly caused by the directivity and the higher source level at 50 Hz than for 100 Hz. The critical range is defined by the crossing with the assumed reaction threshold indicated by the dashed line. For propagation over the low loss bottom (I) the critical range is about 40 km for both criteria whereas the critical range is about 15 km for the bottom type (II).

The SEL value variation with range is important to sensitivity of the critical distance. In this case the low loss bottom type (I) give a slope with range of 10 log (r), the high loss bottom type (II) varies 20 log (r) at the crossings with the reaction threshold.

Figure 15. Time responses in a Pekeris waveguide as function of reduced time and range from source receiver calculated for the two bottom types (I) and (II).

0 5 10 15

0 10 20 30 40 50

c_red=1510 m/s

Reduced time – s

Range – km

Pekeris-250(I): Sd=6 m, Rd=15 m Direct Bottom Surface Refracted Surface-Bottom

0 5 10 15

0 10 20 30 40 50

c_red=1510 m/s

Reduced time – s

Range – km

Pekeris-250(II): Sd=6 m, Rd=15 m Direct Bottom Surface Refracted Surface-Bottom

19

Figure 16. Transmission loss as function of range for the two bottom types (I) and (II).

The dotted line is the geometrical transmission loss of equation (1)

Figure 17. SEL values and spectral valus for 50 Hz as function of range for the Pekeris waveguide calculated for the bottom types (I) and (II). The dashed line is the assumed threshold value of fish reation.

10⁰ 10¹

20 40 60 80 100 120

Range – km

Transmission loss – dB

Pekeris-250(I): Sd=6 m, Rd=15 m Freq = 50 Hz Freq = 100 Hz TLgeo

10⁰ 10¹

20 40 60 80 100 120

Range – km

Transmission loss – dB

Pekeris-250(II): Sd=6 m, Rd=15 m Freq = 50 Hz Freq = 100 Hz TLgeo

10⁰ 10¹

100 120 140 160 180 200

SEL - dB re 1 Pa2 s

Range – km

Pekeris-250(I): Sd=6 m, Rd=15 m SEL

50 Hz line

Reaction threshold

10⁰ 10¹

100 120 140 160 180 200

SEL - dB re 1 Pa2 s

Range – km

Pekeris-250(II): Sd=6 m, Rd=15 m SEL

50 Hz line

Reaction threshold

20

Figure 18. Spectral line values at 50 Hz and 100 Hz as function of range for bottom types (I) and (II). The dashed line is the assumed threshold value of fish reation.

5.2 Propagation under different seasonal condition

The effects of seasonal variations of the environment are illustrated by simulation of propagation at typical summer and winter conditions in the Halten area in Norwegian Sea using the speed profiles of July and February shown in 0Figure 9. This case may serve as example of the propagation effects caused by the combination of sound speed profiles and bottom properties over a flat bottom.

Figure 19 shows the time responses for winter and summer conditions and for the two bottom types (I) and (II). In the winters, the signals from a shallow source propagates mainly in the surface channel being repeatedly reflected from the sea surface and refracted at different depths without striking the bottom.

Consequently the bottom composition is not an important factor for the propagation. Under summer conditions the most propagation paths are with bottom reflections and therefore the transmission is strongly dependent on the bottom properties. The influences of the bottom types are also clearly visible in the same way as can see in Figure 15.

Figure 20 shows the SEL values and spectral values at 50 Hz for propagation under summer and winter conditions and for bottom types (I) and (II). Winter conditions give strong transmission to receivers at shallows depth with critical range in excess of 50 km, almost independent of the bottom proerties. For the summer conditions the critical distance is about 20 km for both bottom types (I) and (II).

10⁰ 10¹

100 120 140 160 180 200

Sound level – dB re 1 Pa

Range – km

Pekeris-250(I): Sd=6 m, Rd=15 m Freq. = 50 Hz Freq. = 100 Hz Reaction threshold

10⁰ 10¹

100 120 140 160 180 200

Sound level – dB re 1 Pa

Range – km

Pekeris-250(II): Sd=6 m, Rd=15 m Freq. = 50 Hz Freq. = 100 Hz Reaction threshold

21

Figure 19. Time responses at Halten as function of reduced time and range for summer and winter conditions calculated for the two bottom types (I) and (II).

Figure 20 shows the SEL values and spectral values at 50 Hz for propagation under summer and winter conditions and for bottom types (I) and (II). Winter conditions give strong transmission to receivers at shallows depth with critical range in excess of 50 km, almost independent of the bottom proerties. For the summer conditions the critical distance is about 20 km for both bottom types (I) and (II).

Figure 21 show the transmission loss as function of range for the frequencies of 50 Hz and 100 Hz. In these plots the directionality and the level of the source are not included. The model results of the propagation loss are significantly different from that of the simple equation (1).

0 5 10 15

0 10 20 30 40 50

c_red=1510 m/s

Reduced time – s

Range – km

Halten-Winter(I): Sd=6 m, Rd=15 m Direct Bottom Surface Refracted Surface-Bottom

0 5 10 15

0 10 20 30 40 50

c_red=1510 m/s

Reduced time – s

Range – km

Halten-Winter(II): Sd=6 m, Rd=15 m Direct Bottom Surface Refracted Surface-Bottom

0 5 10 15

0 10 20 30 40 50

c_red=1510 m/s

Reduced time – s

Range – km

Halten-Summer(I): Sd=6 m, Rd=15 m Direct Bottom Surface Refracted Surface-Bottom

0 5 10 15

0 10 20 30 40 50

c_red=1510 m/s

Reduced time – s

Range – km

Halten-Summer(II): Sd=6 m, Rd=15 m Direct Bottom Surface Refracted Surface-Bottom