You are on page 1of 9

Backscattering cross sections of live fish: PDF and aspect

Kung HuangClarence S. Clay

Citation: The Journal of the Acoustical Society of America 67, 795 (1980); doi: 10.1121/1.383954
View online: http://dx.doi.org/10.1121/1.383954
View Table of Contents: http://asa.scitation.org/toc/jas/67/3
Published by the Acoustical Society of America
Backscattering cross sections of live fish: PDF and aspect
Kung Huang
Gulf Researchand DevelopmentCompany,H. T.S.C., 11111 South Wilcrest,Houston,Texas 77099

Clarence S. Clay
Geophysical
and Polar ResearchCenter, Universityof Wisconsin,Madison, Wisconsin53706
(Received17 May 1979;acceptedfor publication26 November 1979)

The probabilitydensityfunction(PDF) of the peaksof the envelopesof sonar echo from live fish were
measuredat beam aspect.The measurements were made at 220 kHz and in a waveguide.The fish was
the common shiner(Notropiscornutus)and was about 120 mm (about 18 acousticwavelengths)long.
The PDF of the echoes was approximately Rayleigh when the fish was moving gently. The
backscattering
crosssectionequaled4.2X 10-Sm. Transformation
of the PDF's to a targetstrength
display in decibels displaced the maximum of the PDF to the target strength equaling 101og0
(abs/Ao)
+ 3 dB whererrbs
is themeanbackscattering
crosssection
andA0= 1 m2.Thetargetstrengths
of
the common shiner (120 mm) and the mummechog(Fundulus heteroclitus,100 mm) were measuredas a
function of aspectangle. Comparisonof the experimentalmeasurementsand Love's empirical target
strengthsfor any aspectshowedthat the measuredtarget strengthsat broadsideaspectwere about the
sameand the target strengthsat other aspectangleswere severaldecibelslessthan Love's values.Linear
arraysof point scattererswere chosento match the grossaspectdependenceof the target strengthsof the
fish. The lengthsof the arrays were 6.5 mm for the common shiner and 16.5 mm for the mummechog.
Theselengthswere lessthan the lengthsof the correspondingfish'sswim bladders.

PACS numbers: 43.30.Dr, 43.30.Gv, 43.30. Vh, 43.80.Ev

INTRODUCTION pute fish density. -5 When doing these computations


with the sonar equation, one uses the target strength,
The measurements of the backscattering cross section TS-- 10logxo(Crb,/Ao)
, (2)
of a fish involve one way or another the probability
density function (PDF) of a set of measurements be- whereAois the referencearea 1 m'.x6
cause the fish is alive and can move between each ob- Some of the literature gives values of target strength
servation. Correspondingly, a sequence of echoes from and (b, that are calculated from the largest fish echo
the fish often appears to have a sequence of random am- from a sequence of observations. xx-x4 The amount of
plitudes. The PDF depends on many factors and among error this causes in fish density estimates depends
them we list the length of the fish in acoustic wave- upon the breadth of the fish echo PDF. For example,
lengths X, behavior of the fish, motion of the sonar if the breadth of the fish echo PDF is very narrow
transducers relative to the fish, and aspect of the fish. about a mean value, the error is small. However, if,
In addition to the problem of random echo amplitudes as in situ data indicate, ' the breadth of the fish echo
from the fish, the fish can be at some random position PDF is broad, then the bias is large because a too
in the sonar beam and this gives an additional random large value of cry,gives a too small value of fish den-
factor that alters the echo amplitude at the receiver. -8 sity. In addition, a broad fish echo PDF can bias anal-
To eliminate the second variability, we measure the ysis procedures that reject echoes that are less than
PDF of fish echoes in a controlled experiment. arbitrary threshold.
A brief review of the sound scattering process gives We can gain some insight into the nature of the fish
a few fundamental relationships and a nomenclature. echo PDF by considering the fish as being a flexible
Our experiments are for the high-frequency sound scat- line array of small scatters. The overall echo is the
tering when the length of the fish is many acoustic sum of the contributions from each of the small scat-
wavelengths and well above the resonance frequency of terers. As the fish flexes and/or changesits aspect
the swim bladder. From the discussion in Clay and relative to the sonar, the phases and amplitudes of the
Medwin (pp. 228-229), 4 the amplitude of the sound pres- contributions from the scatterers change. The sums of
sure signal (echo) from a fish is proportional to an ef- the contributions as received by the sonar, that\is the
fective sound scattering length l of the fish. A compu- echoes, fluctuate in shape and amplitude from ping to
tation of the integral echo squared for the echoes gives ping. Recalling Rayleigh's derivation of the PDF of the
the backscattered cross section as being the mean amplitude of the sum of many harmonic oscillators
square of the observations of l: having random phases,?.8we expect the PDF of the
(:rb,
= </) and lo--(z', (1) peaks of the envelopesof the echoes, Ill, to have a
Rayleigh PDF:
where < > is averaging operation, o is in m', and l is
in m. The backscattering cross section and fish den- - ll/(2lo)], (3)
sity (numberof fish/m a) are usedto computethe vol- where lo is the rms value of l. The Rayleigh PDF is
ume reverberation, or one can use volume reverbera- also the PDF of the envelope of l when l has a Gaussian
tion data and the backscattering cross section to corn- PDF and the rms value of l is lo. The maximum value

795 J. Acoust.Soc.Am. 67(3), Mar. 1980 0001-4966/80/030795-08500.80 (D1980 AcousticalSocietyof America 795
of the RayleighPDF occursat Ill = lo. The meanvalue ment proceduresare in the companionpaper.z Briefly,
of the envelope(Ill) is the shaded transducer array excites the waveguide in
the first mode and as a receiver it receives the first
<lzl>
--zd* I/1'exp[-1112/(21o2)]dl mode. The sound pressure field is proportional to
sin(2/h), whereh is thewaterdepthandis 22 cmfor
=(rr/2)/21o
' (4) these measurements. The excitation of the first mode
is extremelysensitiveto slight tilts of the transducer.
The mean (lll) is 1.2531o.
To define the scattering region, we use the criteria
Again, consideringthe fish as a line array of scatter- thatthepressure
amplitude
variationbelessthan10%;
ers, the sum of the scattered componentsis largest the width shouldbe within the first Fresnel zone; the
when the fish (array) is broadside and small when the distance from the source shouldbe several waveguide
fish is in line with the sound beam. As the fish is
depths. The dimensions
of the scatteringregionare
turned, one wouldexpecta sequenceof maxima and 6 cmhighat mid-depth,10 cm wide, andbetween75
minima that are due to the interference of the contribu-
and150cm range. This regionis big enough for a 120-
tions of the individual scatterers along the array (fish).
mm-length fish.
Two types of measurements are given in the paper' The signalprocessingsystemis shownin Fig. 2. It
the PDF of the peaks of the envelope of fish echoes at
is partlyanalogandpartly digital. The time gateis
broadside aspect and the target strengths as a function
adjustedto passthe echoandeliminateotherscatter-
of aspectangie. The PDF of the echoesis comparedto ed signals. The bandpassfilter passedthe 220-kHz
the RayleighPDF. We showthat the transformationof carrier frequency.Thecomputer
hasa 10-bitA/D
the Rayleigh PDF to the correspondingPDF of target converter that samplesthe signal after envelopeand
strengthsgives a curve havingits maximumat a target peakdetection.Sincewe samplethe peakof oneecho
strengththat is 3 dBhigherthan101ogm(ab,/Ao).
The for eachping, the digitizingrate is the sameas the
target strengthsas a functionof aspectare compared pingrepetitionfrequency,2 Hz. The computercom-
to Love's formula TMand the scattering functions of
paresthe peakamplitude[ll as
arrays of point scatterers.
(n)Al <Ill< + 1)A1, (s)
I. APPARATUS AND TECHNIQUE where Al is the height of each amplitude bin and adds
The sound scattering measurements are made in one count to the nth bin for each echo amplitude satis-
waveguide, Fig. 1. The upper interface is air and the fying(5). The systemhas256amplitudebins. A pro-
lower interface is a 2.54-cm styrofoam sheet. The gram, that usesthe calibrationparameters,converts
construction is like a child's wading pool in which a the data to the PDF, the mean backscattering cross
large plastic sheet inside a frame holds the water. The section,andthe target strength. We use256pi.ngsfor
transducer and fish holding frame are suspended from the PDF and backscattering cross-sectional measure-
the heavy woodenframe over the tank. The transducer ments.

drive is a 0.2-ms-length ping having a carrier frequen- An exampleof measurements of echoesfrom a steel
cy of 220kHz. Therepetitionrate is 2 pings/s.
TM sphere0.8 cm in diameteris shownin Fig. 3. The
The details of the excitation calibration and measure- amplitudebinsare 1 dB. The targetstrengthis -54
: 1 dB, where 1 dB is the standarddeviationof the
measurements. Since the steel sphere is a stationary
target, this is a measureof the amplitudefluctuations
in the system. The 1-dB fluctuationsare muchless
than breadth of the PDF of fish target strengths.

- ENV
ELOPE
_
DETECTOr.,
J IL)LfLI'
1

S 0 N A R SCOPEQ
PD
P8/E I
comPUTE.I
S Y S T E M
ji - AID I
DATA STORAGE
I
& PROCESSINGJ

Txo

FIG. 1. Apparatusfor measuringthe backscattered


signals
from live fish. Small hooks and monofilament line are att-
ached to the mouth and tail of the fish. The tank is filled to
a depthof about22 cmandis 2.4tax2.4 m.tg'' FIG.2. Electronic
systemblockdiagram)
9

796 J.Acoust.
Soc.Am.,Vol.67,No.3,March
1980 K.Huang
andC.S.Clay:Backscattering
cross
sections
offish 796
FIG. 3. Relative target strength
O_ of a steel ball (0.8-cm diameter).
This shows the width of the PDF
I I i I I I I I I I I I
of a stationarytarget.ls.20
10 5 Ocm

I
-50dB

sb
The causes of most of the fluctuations appear to be
external to the electronic system. For example, tiny
ripples on the surface of the water cause noticeable
fluctuations of echo amplitudes. Background electrical
I I I I I I I, I
and acoustic noise add to the fluctuations. Tiny air
cm 5 0
bubbleson the tank bottom, steel sphere, and trans-
ducer are a major cause of fluctuations and extraneous- FIG. 4. Sketches of the fish. The sketches are based on x
ly scattered signals. At resonance, air bubbles have rays of the fish. The swim bladders sb, are indicated by the
. backscattering cross sections that are orders of mag- shaded areas. Neither of these fish were the ones used for the
nitude larger than their cross-sectional area (Ref. 4, acoustic measurements. (a) Common shiner, (b) Mumme-
p. 200). Sometimes the bubbles appear to grow while ehog.
one is watching a run of measurements. We clean the
bubbles from the surfaces by wiping them off with a flexibility of the lines permit the live fish to move so
sponge. that aspect of the fish and its body shape change from
moment to moment. We identify three behaviors,
After the calibration of system and scattering region,
"calm," "alive," and "wild." Examples of the PDF's of
we measure the backscattering echo response by in-
target strengths for each of these behaviors are shown
serting the fish. We constrain the fish by hooking its
in Fig. 5. Each measurement takes a little over 2 min
mouth and tail with small fishing hooks, then tying
for 256 pings at the 2-Hz-ping repetition frequency.
the fish loosely to the ends of the framework using a
monofilament line. The frame rotates the fish horizon- 1. Calm behavior. The fish is very still without any
tally around a vertical axis. The angular increments apparent movement in the scattering region. The tar-
for the backscattering measurements as a function of get scatters like the steel sphere or a dead fish. Usually
aspect are 5. The loosely tied fish is able to move a a minor disturbance in the tank causes the fish to be
few degrees from each aspect setting. The mean back- more active.
scattered cross section for the aspect angle is the
2. Alive behavior. The fish moves and changes'its
mean of 40 echoes and takes 20 s. The target strengths
aspect a little during the measurements. We believe
as a function of aspect are for a single fish as it is that this behavior is most like a free fish.
turned through 360 .
3. Wild behavior. The fish was wild when it strug-
The ambient noise level and backscattered response
gled while trying to get off the hooks. Its horizontal
from the fishing hooks were measured for comparing
and vertical displacements are large. This behavior
the signal-to-noise ratio. The equivalent target
generally occurred just after the fish was hooked to
strength for the noise levels were from -60 to -75 dB
the lines and placed in the tank. We waited until it
and were lower than the fish's target strengths.
quieted down for measurements.
Measurements were made on two species of fish, An effort was made to take data when the fish had
the common shiner (Notropis cornutus)and the common
alive behavior. If the fish was calm it was stimulated,
mummechog (Fundulus heteroclitus, from the east and if it was wild we waited for the fish to calm down.
coast). A freshly killed commonshiner was x rayed.
We hoped to measure the fish when it was behaving as
The swim bladder was identified as a dark area on the
near as possible to a free fish.
x ray and this was confirmed after dissecting the fish,
Fig. 4(a). The x ray of a preserved museumspecimen III. PDF OF FISH ECHOES AT BROADSIDE ASPECT
of a mummechog was sketched for Fig. 4(b). Since the ANGLE
swim bladder was not gas filled, it was questionably
The fish was loosely tied in the fish holding frame
identified as a very faint shadowon the x ray (the fish
was not dissected for inspection). Neither of these fish and placed in the broadside aspect. The PDF of the
echoes were calculated from 256 echoes from a 12-cm
were the fish used in the sound scattering experiments.
fish, the common shiner. The fish echo PDF, in ar-
II. FISH BEHAVIOR bitrary amplitude units, is shown in Fig. 6(a). The
dashed line is the least-squares fit of the Rayleigh PDF,
The target fish is hooked and loosely tied to the frame Eq. (3), to the data. The rms fitting error was 2.2%.
to hold it in the scattering region. The looseness and The same set of data were regrouped into 1-dB-ampli-

797 J. Acoust.Soc.Am., Vol. 67, No. 3, March1980 K. HuangandC. S. Clay: Backscattering


crosssections
of fish 797
WILD
CALM ALIVE %

30 -- 15 15--

FIG. 5. Probability density func-


tions of different fish behaviors.
Three different behaviors of
fish during the measurements are
(a) calm, (b) alive, and(e) wild.t9
10 --

I I I ! I
--50 --40 --30 --40 --30 --50 --40 --30

T S IN D B

rude bins and plotted on Fig. 6(b). Sec. V, we use the dependenceof the target strength on
aspect to estimate the effective length of the fish.
The fit of the Rayleigh PDF to the data gives a single
parameter specification of the rms scattering length 1o Some experimenters use the PDF of the target
and the breadth of the PDF. Computations using the strengths in their work. The transformation of the
technique given in the companionpaper2yield Rayleigh PDF (3) to the correspondingtarget strength
PDF uses the followingtransformation of PDF's:
l=6.5 x10 -s m,
(v)dv = (uYiu , ()
Cb,= 4.2 X 10-5 m2 ,
v=f(u), u=g(v),
TS =-43.7 dB.

The rms scattering length is proportional to the length


of the fish and the impedance contrast of the fish rela-
=
tive to water ,X(pc)/(pc)' where xVt(v) is the transformed PDF. The transforma-
tion of (3) to the target strengthPDF XVTs(S)uses
lo--liA(oc)/(pC), (6)
s =btn(lll/Lo), Izl e/, (8)
where If is the effective length of the fish. Later in
b = 20 logjoe= 8.686 dB,
where s is in decibels and Lo is the reference length
1 m. The substitution of (8) and (3) into (7) gives

XVrs
(s)=(L/lob)es/bexp[-(L/21o)eS/b]
. (9)
The limits of the target strengths are infinity. The
lO maximum of XVTS(S)
occurs at

10',,
Sm=bln(lo/Lo)+ (0/2) ln2

I t or

s= =20log(lo/Lo)+3 riB.
(10)

I xx
The mean value of s is

O0 I I
o. ol I 0.02 - 50 - 40 - 30 <s)= s XVs(s)ds. (11)
(a) I01,
m (b) TS,dBrem
z
FIG. 6. PDF of echoes from '*live" fish, beam aspect. The It is easy to compute Ts (s) and (s) numerically be-
fish is a common shiner, 120 min. The measurements are causexVs (s) tendsto zero on each side of the maxi-
for 256 echoes. Using the p.r.f. of 2 Hz, the duration of the mum. A numerical computationof Ts (s) for lo =6.5
measurementis over 2 min. The range is 75 era.19 (a) PDF x10 - m is the dashed curve on Fig. 6(b). The experi-
versus amplitude in linear units. The dashedline is the
mental target strengths are
Rayleigh PDF. Theoretical Rayleigh PDF is a least-squares
fit. Themaximum
is at/oor(%s)
1/2.(b)PDFversus
target. s =20og,o(IZl/to), (x2)
strength in decibels. The dashedline is the transformation
of the RayleighPDF to the equivalentPDF of target strengths. where the Ill are from the same data set as on Fig. 6(a).
The maximumof this PDF is at 10 log10(%s/A0)+3 dB. As indicated in (10), the peak of (s) is at -40.7 dB

798 J. Acoust.Soc.Am., Vol. 67, No. 3, March1980 K. Huangand C. S. Clay: Backscattering


crosssectionsof fish 798
and is 3 dB higher than the peak of the Rayleigh PDF (measurements -2 dB). Rotation of the fish gives the
and the target strength of lo =-43.7 dB. Numerical dependence on aspect angle. Comparisons of the sound
evaluation of (11) for /o=6.5x10 -a m gives (s) =-43.2 scattered at the right and left side give an indication of
riB. Equation (10) and numerical evaluations of (11) the reproducibility of scattering measurements of a
can be used to calculate lo or Vb, from target strength single fish. The main lobes,, O =0 and 180, have
measurements. about the same shape. The side lobes vary because the
directional locations of the maxima change. The en-
The Rayleigh PDF describes the results of a process
velopes of the side lobes are about the same. Compari-
in which the scatterers change their position and attitude
son of the widths of the main lobe of the scattered sound
relative to the position of the sonar system and direc-
tion of insonification between each observation. Pre- from the common shiner and that of the mummechog
show that the main lobe of the common shiner is wider
sumably one has many observations of nonoverlapping
than the main lobe of the mummechog. While this ob-
echoes. If the sonar moves (as on a ship) and the fish
servation applies to one common shiner and one mum-
are still, the relative changes exist and we expect the
mechog, it may relate to anatomical differences of the
PDF of echoes to be Rayleigh. It is important to notice
fish.
that one parameter, lo, specifies both the breadth and
the maximum of the Rayleigh PDF. This is also true A comparison of the measurements of Love's em-
for the corresponding PDF of the target strengths. pirical tarKet strength functions, the dashed lines, is
shown on the figure. Love states that the target
strengths are maximum values and we assume that the
IV. ASPECT ANGLE DEPENDENCE
functions are smoothed maxima of echoes for a given
number of observations. To explore this, we use (3)
The dependencies of the sound backscattered by a
to calculate the probability of an echo having a peak
120-mm common shiner and a 100-mm mummechog as
amplitude less than a as follows:
a function of aspect angle are shownon Figs. 7(a) and
7(b). The measurementsare 201Ogxo<[/[>
and using(4)
are approximately 2 dB larger than the target strength.
Each value of <[/1>is average value of 40 echo peaks
fOal
P([ll<alo)= W([ll)d[ll=l-e-J/2. (13)

and takes 20 s. The angular increments are 5 and, Since the measurementsare for {Ill), we use (4) and the
probability of I/I less than 1.253 lo is
since the loosely tied fish was able to flex its body, the
fish movement is a few degrees. P([/[ <1.253 lo)=0.54.
At side aspect, O =0 , the target strengths of both This is the probability that Love's curves exceed lo by
the common shiner and the mummechog are -44 dB at least 2 riB.

TAIL

SIDE

HEAD

(o) FIG. 7. Mean target strength versus


aspect. (a) Common shiner, 120 min.
(b) Mummechog, 100 min. The wave-
TAIL length of the sound was 6.8 min. The
overall lengths were 18 and 15X. The
target strengths were measured at 5
increments and each measurement was
the averageof 40 echoes)9 The dashed
lines were calculated from Love.t4
SIDE

HEAD T S, dB
(b)
799 J. Acoust.Soc.Am., Vol. 67, No. 3, March1980 K. Huangand C. S. Clay: Backscattering
crosssectionsof fish 799
V. MODELING OF A FISH BY AN ARRAY OF do not think it is important to match the detailed loca-
SCATTERERS tions of the maxima and minima of the experimental
curve because side lobes from opposite sides of the
The numerical modeling of a fish by an array of point fish do not match each other. The spacing of 1.5 mm
scatterers requires the matching of the theoretical between scatterers is approximately ?./4 at 220 kHz.
target strength as a function aspect to the experimental We mention that the array models were made by using
measurements. We use a linear array of noninteracting the acoustic data and prior to taking the x rays and
point scatterers for the model. Using the geometry on making the sketches in Fig. 4.
Fig. 8 and ignoring the range dependence, the scattered
sound pressure is
Comparisons of the target strengths for the models
and the fish are shown in Fig. 8. The model matching
the common shiner has a length of 7.5 mm. This is the
Ps-eWt bmexp(-i2ky,,sinO), k =2r/X, (14)
effective scattering length of the fish'. Recalling (6)
and lo =6.5 mm, the impedencecontrast A(pc)/(pC)
__-o
bm
=1, (15) =0.87. This large contrast suggests that an organ such
as the gas-filled swim bladder is the major scatterer
TS= 10log[psiz +C, (16) of sound. Referring to Fig. 4(a), the common shiner's
swim bladder has a 10-mm segment and a 30-mm seg-
where b, is the relative amplitude of the rnth scatterer;
ment. Either of the segments are longer than the length
y mis the position of the scatterer along the array;
of the scattering array, 7.5 mm.
2y,,k sin0 is the phase shift of the scattered wave; C is
an adjustable constant to match the model TS to the The equivalent array of scatterers that models the
fish TS at =0. In the matching process, we choose mummechog is 16.5 mm long for a 100-mm fish, Fig.
the M, Ym, and bmso that the main lobes of the theo- 7(b). From the probable length of the swim bladder of
retical curve and the experimental curves match and the fish sketched in Fig. 4(b), 30 mm for a 60-mm-
so that the side-lobe levels are about the same. We
length fish, a 16.5-mm array would fit inside the swim
bladder.

Our lack of knowledge of the anatomy of the live fish


in their normal attitudes precludes adding more details
to the models.

Vl. DISCUSSION

All of our results and this discussion use the mea-


surements of a few fish from two species. The fish are
alive and are many acoustic wavelengths long. Rather
than repeat these qualifications over and over, we urge
the reader to keep them in mind.

These experiments display the variability of the


echoes from one trial to the next and the dependence of
the average echo amplitude on the aspect angie of the
fish. Compared to the echoes from a steel sphere, the
-50__..- .=_ echoes from a live fish that is free to move and flex its
body has a large range of amplitudes. A very still fish
has a narrow range of amplitudes and is somewhat like
the steel sphere. The Rayleigh approximately fits the

(b)
FIG. 8. Comparisons of scatter
head by arrays of "int scat-
echoes from a modestly moving fish. Our laboratory
measurements
field measurements.
confirm
2'4
the less direct results of our

terers" and the rget strehs of live fish versus aspect.


The Rayleigh PDF is rather broad, and a single
The sold is cident and the backscattered sil is obseed
the =0 direction. The arrays of scatterers are along the parameter, the rms amplitude of the echoes, specifies
les of e fish; 9 is normal to the le of t scatterers. both the amplitude and breadth of the function. For ex-
The source frequency is 220 z and the acoustic wavelenh ample, a third of the peak amplitudes are less than
is 6.8 min. For both arrays, the spaces beeen the 0.90lo anda third are greater than 1.48lo. Thus
scatrers are equal and 1.5 min. The bm'sare e relaUve echoes from the same fish can have a wide range of
scatr streh of each t scatterer. The oreUcal values as the fish moves. Since the PDF of the echoes
array scatrg fction is multiplied by a consent to tch
is broad, one needs the average of many measurements
the rget streths from Fig. 7. The solid les are the
and the shed les are the t scatterer rget strehs. to determine the mean backscattering cross section.
Array rameters follow. (a) Common shiner, fish leah We need to know the mean backscattering cross section
120 ram, M=6; arrayle L =6.5 ram; b0 =b =0.105; bl = ba because it is the quantity that relates the integral echo
=0.184; b =bs=0.211. (b) Mmechog, fish leith of 100 squared measurements to the number of fish per cubic
M= 12; array leh=16.5 ram; bm=. meter.

800 J. acoust.Soc.Am., Vol. 67, No. 3, March1980 K. Huangand C. S. Clay: Backscattering


crosssectionsof fish 800
The intrinsic breadth of the amplitude PDF of fish the echoes from many fish (assuming all of the fish are
echoes presents the following limitations. Many in situ a single species and have the same size). The second
echo measurements are needed to determine a back- is the equivalence of repeated measurements of a mov-
scattering cross section or target strength. If several ing fish and repeated measurements of a stationary
sizes of fish are present, the PDF's of all of their fish from a moving platform. If, as we believe, these
echoes overlap and identification is difficult unless the equivalences are valid, then laboratory descriptions of
peaks of the individual PDF's are well separated. the scattering processes at the fish apply to field mea-
Other information, such as species behavior, net hauls, surements.

etc., may be crucial in interpreting the acoustic data.


ACKNOWLEDGMENTS
The dependence of the backscattering cross section
and target strength on aspect angle is another variabil- Professor John Magnuson supplied the fish. Mr.
ity of the scattering process. For the simple models Barry Heist made the x rays and dissected the fish.
in this paper we use noninteracting point scatterers. Mrs'. Alison Mares edited the manuscript. This re-
Recognizing that the scatterers are likely to be gas search was funded by the National Oceanographic and
bubbles, the interactions between bubbles are appreci- Atmospheric Administration's Office of Sea Grant, De-
able and need to be included in more sophisticated partment of Commerce, through an institutional grant
modeling efforts. '' Models of the fish as linear arrays to the University of Wisconsin at Madison and the Office
of point scatterers give dependencies on aspect angle of Naval Research. This is contribution No. 377 of the
that match.the major features of the experimental mea- Geophysical and Polar Research Center, University of
sure me nts. Wisconsin at Madison.

The lengths of the arrays that model the fish are


much shorter than the lengths of the fish, i.e., 7.5 mm
1R. E. Craig and S. T. Forbes, "Design of a sonar for fish
for a 120-mm common shiner, and 16.5 mm for a 100- counting," Fisk. Dir. Set. Vav. Unders. 15, 210-219 (1969).
mm mummechog. The model arrays are less than the 2M. L. Peterson, C. S. Clay, and S. B. Brandt, "Acoustic
length of the swim bladders. Since the combination of estimates of fish density and scattering function," J.Acoust.
the effectice scattering length of the fish and length of Soc. Am. 60, 618-622 (1976).
the model array give a large acoustic contrast, we 3j. E. Ehrenberg, J. H. Green, and A. R. Wirtz, "A dual-beam
believe that most of the sound is scattered at the swim acoustic system for measuring the target strength of individ-
bladder. To relate sound scattering characteristics and ual fish," Oceans '76, 16C-1 Mar. Technol. Soc. J. IEEE
(1977).
the anatomy of the fish, we suggest that x rays be made
4C. S. Clay andH. Medwin, AcousticalOceanography:
Princi-
of the live fish in as near the normal attitude of the fish
ples and Applications (Wiley-Interseienee, New York,
as possible. Our reason for suggesting this is that the 1977), Chaps. 7 and A7. (The numerical values in Table
shortness of the model arrays compared to the lengths A7.1.1 are about 15% low. Please write the authors for this
of the swim bladder implies that the swim bladder does and other errata.)
not scatter coherently as a unit. A sequence of x rays 5L. M. Deuser, D. Middleton, and T. D. Plemons, "On the
classification of underwater acoustic signals II. Experiment-
may show the fish changing its shape and inflation of its
al applications involving fish," J. Aeoust. Soe. Am. 65, 444-
swim bladder. The effective scattering length is then 455 (1979).
an average of sound scattering for the various configur- Manualof Methods
for FisheriesResourceSurveyandApp-
ations. raisal, Part 2: The use of acoustic instruments for fish de-
tectionand abundance estimation, edited by S. T. Forbes
It is obvious that the attitude or aspect of the fish
and O. Nakken (FAO, Rome, 1972).
relative to the direction of the sonar beam affects the
?R.E. Thorne, "Investigationsinto the relationshipbetween
average size of the echoes in two ways. For a stabil- integrated echo voltage and fish density," J. Fish. Res.
ized transducer having a vertical sonar beam, the fish Board Can. -8, 1269-1273 (1971).
can have a distribution of dips of their bodies and this 8j. E. EhrenbergandD. W. Lytle, "Acoustictechniquesfor
gives a corresponding distribution of echoes? For a estimating fish abundance," IEEE Trans. Geosei. Electron.
hull-mounted sonar, the sonar beam swings with the GE-10, 138-145 (1972).
9R. W. G. Haslett, "Determination of the acousticbackscatter-
ship's roll and the aspect angle varies. Depending on
ing patterns and cross sections of fish," Br. J. Appl. Phys.
the amount of roll, the transducer can insonify the fish 13, 349-357 (1962).
from directions that are far from the direction for lH. W. Volberg, "Acoustic target strengthof several species
maximum backscatter. Since this effect depends on of fish," Straza Industries Rep. R-101, E1 Cajon, California
sea conditions and ship roll, we expect a bias toward (1963).
smaller average backscattering cross sections when liB. S. McCartneyandA. R. Stubbs,"Measurementsof the
swimbladder resonance," in Proceedings of an International
the seas are rough. It may be particularly necessary
to use stabilized sonar transducers when the sonar Symposiumon Biological Sound Scattering in the Ocean,
edited by G. B. Farquhar (U.S. Government Printing Office,
beamwidths are narrow and the main lobes of the fish's
WashingtonDC 1970), pp. 180-211.
backscattering cross sections are narrow. l'R. H. L)ve, "Maximumside-aspecttarget strengthof an in-
dividual fish," J. Acoust. Soc. Am. 46, 746 (1969).
In applying the results of these laboratory measure-
13R.H. Love, "Measurementsof fish target strength:a re-
ments to field measurements of sound scattered fish,
view," Fish. Res. Bull. 69(4), 703-715 (1971).
there are two equivalences to consider. The first is 14R.H. Love, "Target strengthoran individualfish at any as-
the equivalence of many repeated measurements of the pect," J. Acoust. Soc. Am. 62, 1397-1403 (1977).
echoes from a single fish and a few measurements of isj. D. PenroseandG. T. Kaye, "Acoustictarget strengths

801 J. Acoust.Soc.Am., Vol. 67, No. 3, March1980 K. HuangandC. S. Clay: Backscattering


crosssections
of fish 801
of mrine organisms," J. Acoust. Soc. Am. 65, 374-380 problem gives the Rayleigh PDF of the envelope. J. L. Law-
(1979). son and G. E. Uhlenbeck, Threshold Signals (Boston Tech-
l%Veuse bs, the backscatteringcross section, becausethat is nical Publishers, Boston, 1964), pp. 124-126.
what we measure and use. Some experimenters measure b 9Theexperimentalmeasurementsare from the M.S. thesis
and then convert the measurement to (s--4r%s. For the tar- of Kung Huang: "PDF of soundscattered from live fish,"
get strengthcalculation,they use TS=10 log101%/(4A0)]. M.S. thesis, Geology and Geophysics, Weeks Hall, Uni-
This operation implies that % is the total scattering cross versity of Wisconsin at Madison (1977).
section and that the scattering is isotropic. Of course, the 2C.S. Clay andK. Huang, "Singlemodetransmissionand
actual s is not 4Obsfor a complicated scattering object. acoustic backscattering measurements in a laboratory wave-
The preferred practice is to report the measured quantity. guide," J. Acoust. Soc. Am. 67, 792-794 (1980).
If one measures backscattered sound signals, the report bs 2y. W. Lee, StatisticalTheoryof Connuuicatiou
(Wiley, New
and if one measures the total scattered sound, then report York, 1960), pp. 190-193.
D. E. Weston, "Acousticinteractioneffectsin arrays of
l?Lord J. W. S. Rayleigh (J. W. Strutt), The eory of Sound small spheres," J. Acoust. Soc. Am. 39, 316-322 (1966).
(Dover, New York, 1945), Vol. 2, pp. 35-42. K. G. Foote, 'rheoretical estimation of the mean echo in-
18Theradar clutter problem is aboutthe same as the scatter- tensity-fish number density for encaged saithe in the dorsal
ing of sonar signals by fish and the solution of the radar aspect," Fisk. Dir. Ser. Vav. Unders. 16, 457-464 (1979).

802 J.Acoust.
Soc.Am.,Vol.67,No.3, M.
arch1980 K.Huang
andC.S.Clay:Backscattering
cross
sections
offish 802

You might also like