M arek Moszynski , Andrzej Stepnowski

1

The influence of fish morphological and behavioural parameters on acoustic data in

algorithmic reconstruction of fish length distribution

Marek Moszynski, Andrzej Stepnowski

Gdansk University of Technology Poland

ICES ASC 17-21 September 2007, Helsinki, FinlandICES CM 2007/H:08Effects of environmental changes on the biology, physiology, and behaviour of pelagic fish

2

The influence of fish morphological and behavioural parameters on acoustic data in

algorithmic reconstruction of fish length distribution

ICES ASC 17-21 September 2007, Helsinki, FinlandICES CM 2007/H:08Effects of environmental changes on the biology, physiology, and behaviour of pelagic fish

Abstract The paper investigates the algorithm for estimation the fish length distribution from acoustic target strength data. The theory of scattering from a tilted cylinder is used for modelling the fish directivity pattern of swimbladdered fish. The model allows formulating the dependence of target strength on two main components: fish maximum target strength and the fish directivity pattern. As both terms depend on fish length, the inverse technique could be used to reconstruct unknown fish length distribution from acoustic data, when morphological parameters of fish are properly assumed. However, as it is shown, the algorithmic approach is very sensitive to some of behavioural parameters of swimming fish. Thus, although the effect of unknown fish tilt angle could be partially removed by statistical processing, the mean value of fish tilt angle still may produce large errors. The method and its results are verified on actual data acquired during the survey and compared to trawl catches.

3

Introduction (1)

Echo Level E

Target Strength

TS

Fish length

L

FishBiomass

Q

acoustical measures physical measures

Catchdata

Fish echo processing chain:

regression modelsmeasurements:• ex situ• in situ

Ei = SL+RS + TSi(li, i , zi,, fo ) + 2B(i ) - 2TL( Ri, α)

4

Introduction (2)

Backscattering modelTilt angle statistics

INVERSE PROCESSING

Sample catchRegression relation

MEAN VALUE PROCESSING

pTSFish

lengthL

< l >

pl

5

Fish length estimation

pTS plpTS0

-tilt angle statistics-backscattering model

backscattering model

problems:• unknown titl angle during ensonification• unknown fish directivity pattern

6

Fish backscatter models for swimbladdered fish

• tilted cylinder - Haslett (1962)

• finite bent cylinder model - Stanton (1989)

• low resolution acoustic model - Clay (1991)

• Kirchhoff ray mode model (KRM) - Clay, Horn (1994)

• boundary element model - Foote, Francis (2002)

simple

precise

7

Haslet model for swimbladdered fish

Haslett, 1962• swimbladder is approximated by a combination of: a hemisphere, a short cylinder, a cone of fixed dimensions relative to the fish fork length. • then this shape is modified to: a cylinder maintaining their geometrical cross section.

lecb=0.24L

2aecb=0.049L

0.2L0.125L

8

Methods

i f z = x + y t h e n dxxzxpzp yxz ),()( , ( i . e . E = B + T S o r T S = T S 0 + D f ) i f x y i n d e p e n d e n t r a n d o m v a r i a b l e s t h e n

dxxzpxpzp yxz )()()( i f x y d e p e n d e n t r a n d o m v a r i a b l e s t h e n

dxxxzpxpzp xyxz ),()()( |

9

Methods (2)

i f z = x + y t h e n dxxzxpzp yxz ),()( , ( T S = T S 0 + D f ) i f x y d e p e n d e n t r a n d o m v a r i a b l e s t h e n

dxxxzpxpzp xyxz ),()()( | f o r T S 0 a n d D f

000|0 ),()()(00

dTSTSTSTSpTSpTSp TSDTSTS f

10

Backscatter theory (1)T h e a m p l i t u d e o f a c o u s t i c b a c k s c a t t e r i n g l e n g t h o f a g a s - f i l l e d

c y l i n d e r i n w a t e r m a y b e e v a l u a t e d f r o m H e l m h o l t z - K i r c h h o f f i n t e g r a l( M e d w i n a n d C l a y ) :

)cos(

)sin()sin(sin

)( 00

00

ecb

ecbBSBS kl

klll ( 1 )

l B S 0 = l e c b ( a e c b / 2λ ) 1 / 2 - m a x i m u m b a c k s c a t t e r i n g l e n g t h ,a e c b , l e c b - r a d i u s / l e n g t h o f t h e e q u i v a l e n t s w i m b l a d d e r a s a c y l i n d e r ,χ - f i s h a n g u l a r c o o r d i n a t eχ 0 - t i l t a n g l e o f t h e s w i m b l a d d e rk = 2π / λ - w a v e n u m b e r

+0

lecb

aecb

k

11

Backscatter theory (2) I n t h e l o g a r i t h m i c f o r m :

),,,(),,( 00 flDfalTSTS ecbfecbecb

T S = 2 0 l o g | l B S | T S 0 m a x i m u m t a r g e t s t r e n g t h

2log200

ecbecb

alTS

B f ( . ) l o g a r i t h m i c f i s h a n g u l a r p a t t e r n i n d o r s a l a s p e c t

)cos()sin(

)sin(sinlog20),,,( 0

0

00

ecb

ecbecbf kl

klflD

12

Tilt angle/Angular fish position PDF

Multiple echo statistics: moving vessel and stationary fish model

equally probable distance

from the centre to the trace of the fish

moving fish and stationary vessel model

equally probable crossing angle

rm ax

zm ax

r

Rz

t

M o d e l 2 - : U ( 0 , /2 )

r

t

1

r

t

M o d e l 1 - 1 : U ( 0 ,r )

cos = 1 / r 1: U (0,r) 1 = r u

p()=sin

p()=2 / (0,/2)

13

Tilt angle dependance (1)

)cos()sin(

)sin(sinlog20),,,( 0

0

00

ecb

ecbecbf kl

klflD

f = 38kHz0=8°lecb=L/4

14

Conditional fish beam pattern PDF

Df [dB]

TS0 [dB]

15

randomgenerator

pTS

TS

Statistical processing

0|ˆ TSD fp

0ˆTSp

Lp̂inversionEMSInversion

16

Processing example

a) acoustically measured target strength TS at 200kHz b) conditional PDF of the fish directivity pattern assuming swim bladder tilt angle 5 c) estimated maximum target strength PDF d) reconstructed fish length distribution along with the

catch histogram (in cm)

a) b) c) d)

17

Case study 1 • NOAA/Alaska Fisheries Science Center - summer 2002 - Bering Sea• provided by Neal Williamson (PMEL - Seattle)

18

Survey data

•Simrad EK500 v.5.30 echosounder• 38kHz split beam transducer• logged w/ Sonardata's Echolog 500• 14-07-2002 8:57 – 11:22 am• 6776 pings (540MB) • 2002 tracks of walleye pollock (Theragra chalcogramma)

19

Survey data analysis

30 40 50 600

10

20

30

40

pL1 14-07

30 40 50 600

10

20

30

40

50

pL2 14-07

-80 -60 -40 -200

100

200

300

400

pTS 14-07

30 40 50 600

0.2

0.4

0.6

0.8

1

pL,p'L

f=38kHz

[dB]

[cm]

20

Survey data analysis (2)

-80 -60 -40 -200

100

200

300

400

pTS(f=38kHz) 14-07

-80 -60 -40 -200

500

1000

1500

pTS(f=120kHz) 14-07

30 40 50 600

10

20

30

40

pL1 14-07

30 40 50 600

10

20

30

40

50

pL2 14-07[dB]

[cm]

21


30 40 50 600

0.5

1f=38kHz

0=5

30 40 50 60

0=6

30 40 50 60

0=7

30 40 50 60

0=8

30 40 50 60

0=9

30 40 50 60

0=10

30 40 50 600

0.5

1f=120kHz

0=5

30 40 50 60

0=6

30 40 50 60

0=7

30 40 50 60

0=8

30 40 50 60

0=9

30 40 50 60

0=10

Reconstruction of fish length PDF for different mean swimbladder tilt angle 0 along with estimate from catch data.

Upper sequence for 38kHz and lower for 120kHz. X-axis represents fish length in [cm].

22


5 6 7 8 9 100.1

0.2

0.3

0.4

0.5

Root mean square error function obtained from 38kHz and 120 kHz estimates versus assumed swimbladder tilt angle

23


Estimates of length PDF for mean swimbladder tilt angle 0=7 along with catch data

30 35 40 45 50 55 600

0.2

0.4

0.6

0.8

1

24

Tilt angle dependance (3) Target strengths as a function of tilt angle for a 31.5cm pollock

at dorsal aspect at 38kHz and 120kHz Foote (1985)

Walleye pollock Theragra chalcogramma (Horne - Radiograph Gallery)

25

Case study 2

• R/V “G. O. Sars” • March 17 to April 5, 2004• Lofoten 2004 survey• Lofoten islands, from 67oN to 70oN, • spawning grounds of North East Arctic Cod • shelf between 500 m to about 50 meters • sea temperature 6.8 – 7.1oC from 40–300m • 5 Simrad EK60 split beam echosounders

26

Experiment• standard sphere calibration methods CU64 (18 kHz), CU60 (38 kHz) , WC38.1 (70, 120 and 200 kHz)• transducers mounted in one of the instrument keels of the vessel • full half-power beam widths 7o, except for the 18 kHz (11o)• the transmitted pulse duration was identical on all frequencies - 1.024 ms• the Bergen Echo Integrator, BEI. • heave, roll, pitch and yaw Seatex MRU 5 -Simrad EM 1002 at 10 Hz • CTD observations (Sea-Bird SBE9).• trawling partly on fixed locations, mostly on registrations for identification of the targets and for biological sampling.• Campelen 1800 bottom survey trawl • Åkratrawl, a medium sized midwater trawl• Standard biological parameters were measured on all catch samples, • individual total length, weight, gonad and liver index, age and stomach content.

27

Trawl data

28

29

38kHz 70 kHz

120kHz 200 kHz

30

Survey data • provided by Institute of Marine Research - Bergen

Norwegian cod echoes at depth range 100-160m acquired with 18kHz system

31



32



33



34



35

Target strength data

-60 -40 -200

50

100

150

pTS 18kHz (20-03-2004)

-60 -40 -200

100

200

300

400

500

pTS 38kHz (20-03-2004)

-60 -40 -200

50

100

150

200

250

pTS 70kHz (20-03-2004)

-60 -40 -200

100

200

300

400

pTS 120kHz (20-03-2004)

-60 -40 -200

100

200

300

400

500

pTS 200kHz (20-03-2004)

36

Results

40 60 80 100 1200

0.5

1

40 60 80 100 1200

0.5

1

40 60 80 100 1200

0.5

1

40 60 80 100 1200

0.5

1

40 60 80 100 1200

0.5

1

40 60 80 100 1200

0.5

1

40 60 80 100 1200

0.5

1

40 60 80 100 1200

0.5

1

40 60 80 100 1200

0.5

1

40 60 80 100 1200

0.5

1

40 60 80 100 1200

0.5

1

40 60 80 100 1200

0.5

1

38kHz

70kHz

120kHz

200kHz

2° 5° 8°

37

Tilt angle dependance (3) TS/length relationship on tilt angle for atlantic cod

TS = 20log L + B20 , McQuinn, Winger (2002)EK500 38kHz SB 7

B20

Atlantic codGadus morhua(Horne - Radiograph Gallery)

38

Problems Proposed method can be applied only for one species in the data

set. The knowledge of simplified morphological parameters of

swimbladder (aecb, lecb, χ0) are required . The statistics of fish orientation changes are also required

(normal distribution is assumed).

39

Conclusions The estimated PDF of acoustic backscattering length of fish

differs from actual fish length PDF. The transformation of physical fish length distributions is a result

of combined effect of random fish length and its scattering pattern.

The process of removing fish directivity pattern effect requires application of inverse technique as fish length information is included in maximum fish target strength TS0.

The knowledge of mean fish swimbladder tilt angle (χ0) can be estimated by multifrequency approach using simple comparison of their fish length rms estimates.

M arek Moszynski , Andrzej Stepnowski

Documents