Species Identification of Antarctic Krill Euphausia superba Using the 2-frequency Difference Method

한수지 50(6), 788-798, 2017

788

Copyright © 2017 The Korean Society of Fisheries and Aquatic Science pISSN:0374-8111, eISSN:2287-8815 Korean J Fish Aquat Sci 50(6),788-798,2017

Original Article

서 론

남극크릴

(Euphausia superba)

,

,

,

(Everson, 2000),

,

(Hewitt and Demer, 1993; Atkinson et al., 2009; Jarvis et al., 2010; Fielding et al., 2014).

,

,

.

1982

(Commission for the Conservation of Antarctic Marine Living Resources, CCAMLR)

,

(Hewitt and Low, 2000; Hewitt

et al., 2004).

25

,

1985

4

28

CCAMLR

.

이와같이남극크릴을관리하기위해시공간분포및밀도를 파악하는것이매우중요한데

,

(Lawson et al., 2008; Cox et al., 2011; La et al., 2016).

선박에서 음향조사를수행하는과정에서는배경잡음

,

,

,

,

.

.

.

,

주파수차법을 이용한 남극크릴(Euphausia superba)의 종 식별에 관한 연구

최석관·한인우 ¹ *·황두진 ¹ ·김태호 ¹ ·안두해·이경훈 ¹

국립수산과학원 원양자원과, ¹전남대학교 수산과학과

Species Identification of Antarctic Krill Euphausia superba Using the 2-frequency Difference Method

Seok-Gwan Choi, Inwoo HAN

¹

Distant Water Fisheries Resources Research Division, National Institute of Fisheries Science, Busan 46083, Korea

1Division of Fisheries Science, Chonnam National University, Yeosu 59626, Korea

Antarctic krill Euphausia superba are important components of the Antarctic marine ecosystem both economically and ecologically; to manage this species effectively, their distribution and abundance must be understood. Using the Kwang Ja-Ho (3,012 tonnage), a commercial fishing vessel, we conducted acoustic surveys during April 13-24, 2016, to estimate the distribution and population size of krill around the South Shetland Islands of the Antarctic Continent, We used acoustic techniques based on the dB-difference, a method used mainly to classify of marine species. We found that Antarctic krill were present in numbers over 99% at six survey stations, with the exception of Station 3, where we only found Electrona carlsbergi . There was no difference in cell size due to frequency differences, but echo signals differed between species: 4.7-12.0 dB for Antarctic krill, and -4.1~0 dB for Electrona carlsbergi .

Key words: Antarctic krill, Euphausia superba, 2-Frequency difference method, Species identification

This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial Licens (http://creativecommons.org/licenses/by-nc/3.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

https://doi.org/10.5657/KFAS.2017.0788 Korean J Fish Aquat Sci 50(6) 788-798, December 2017

Received 6 September 2017; Revised 18 October 2017; Accepted 30 October 2017

*Corresponding author: Tel: +82. 61. 659. 7124 Fax: +82. 61. 659. 7129

E-mail address: [email protected]

주파수차법을 이용한 남극 크릴의 종 식별

789

별하는것이필요하다

.

,

,

2

(Miyashita et al., 1997; Kang et al., 2002; Han, 2017).

본논문에서는음향자료를수집하는과정에서발생하는다양 한배경잡음을제거하는방법을이용하여남극크릴의밀도추 정의정도를높이고

,

38

120 kHz

. 2

(EK60, 38 and 120 kHz)

South Shetland island (Subarea 48.1)

.

재료 및 방법

조사해역 및 표본 채집

본 연구의조사해역은 남극해의

48.1

South Shet- land island

Elephant island

,

200 m

.

2016

4

13

24

(3,012 tonnage, Insung, Korea)

, Fig. 1

7

. 7

.

,

,

200

선택하여

CCAMLR (2011)

진크릴측정법인

’Discovery’

꼬리까지의길이를

mm

. 조사 시스템의 구성

음향조사 시스템은 광자호의 선저에 부착된 과학어군탐지 기

(EK60, Simrad, Norway)

,

1982

에발효된

CCAMLR

Table 1

같이설정하였다

. 38 kHz

120 kHz

ES80B, ES120-7C

, Table 1

38

120 kHz

1024 ms, ping

2

초

,

0-1100 m

집하였다

.

과학어군탐지기의 교정은 음향조사 실시 전에 연안 해역

(62°28.7’S, 59°42.4’W)

, 38 kHz

60 mm,

120 kHz

23 mm

시스템의주요교정사항은

Table 2

(Foote et al., 1987).

, 38 kHz

120 kHz

Parameter Setting

Transducer ES38B ES120-7C

Power (W) 2000 250

Pulse duration (microsecond) 1024 1024

Ping interval (second) 2 2

Data collection range

(minimum-maximum) (m) 0-1100 0-1100

Bottom detection range

(minimum-maximum) (m) 5-1100 5-1100

Display range

(minimum-maximum) (m) 0-1100 0-1100

Table 2. Calibration result of frequency 38 kHz and 120 kHz for scientific echosounder

Frequency (kHz) 38 120

Two-way beam angle (dB) -20.6 -21.0

Receiver bandwidth (kHz) 2.43 3.03

Transducer gain (dB) 26.82 27.64

3-dB Beam angel (athwart/along) (deg.) 7.08/7.03 6.47/5.60 Absorption coefficient (dB km^-1) 9.8 24.7

Sound speed (m s^-1) 1448.9 1448.9

Target strength (dB)

Length (mm)

dB difference (dB)

y = 0.0435x + 318.35 R² = 0.43

0 200 400 600 800 1000 1200

0 2000 4000 6000 8000 10000 12000

Catch (kg) NASC (m

/nmi

)

SV raw data

NOISE SV removed

noise Data range bitmap 1 Mask 1

Erosion filter 3×3

Dilation filter 5×5 Dilation filter 7×7 Data range

bitmap 3

Mask 2

Select Median filter 7×7

Data range bitmap 2

Wang et al. (2015) De Robertis and Higginbottom

(2007)

38 kHz raw data Noise 38 kHz

Cleaned 38 kHz Resample by ping

38 kHz

SV difference 120-38 kHz (dB)

120 kHz raw data

Noise 120 kHz Cleaned 120 kHz

Resample by ping 120 kHz

Data range bitmap

Mask (Krill) Resample vertically

120 kHz Match ping times

Elephant Island Sou th S hetla nd I slan d

ANTARTICA Subarea 48.1

Longitude

Latitude

1.0

0.8 0.6

0.4

0.2

0 -20 -10 0 10 20

MVBS

(dB) 10 ping×2 m

T1 T2 T3 T4 T5 T6 T7

Cumulative Distribution Function

Fig. 1. Acoustic survey station for collection of Antarctic krill Eu- phausia superba sample. Trawl collection station were shown from T1 to T7.

최석관

ㆍ

ㆍ

ㆍ

ㆍ

ㆍ

790

각은

-20.6, -21.0

, Receiver bandwidth

2.43, 3.03

, Transducer gain

26.82 dB, 27.64 dB

.

9.8 (dB/km), 24.7 (dB/km)

,

수온과염분의영향을받는음속은두주파수모두

1448.9 m/

s

.

남극크릴의 채집어구

조사대상인남극크릴은중층트롤을이용하여채집하였다

.

롤조업에사용된그물의설계도를

Fig. 2

.

중층트롤의총그물의길이는

167.6 m

,

40 m, 72 m

.

32 m

,

1.5 m,

폭은

3 m

.

15 mm

,

100 mm

.

,

.

,

.

,

.

.

현용 중층트롤 어구의로프 부분은

PP

Super Danline 3st.

(Danlin

Fig. 2. Arrangement of the Antarctic krill Euphausia superba trawl net used in this study.

주파수차법을 이용한 남극 크릴의 종 식별

791

강도가일반로프에 비해

1.3

)

,

PE

(Fig 2). PP Danline 8st

(91.3m),

(86.1 m),

.

, PP

(Compound rope)

.

Super 12st.

PP

,

Eurofix br. netting

. Dyneema SK75 12st

(88.0 m)

,

(64.4 m)

.

,

4

14

4

23

10

7

,

1

시간기준으로하였으며

,

30,

은

300 m

.

크릴의트롤채집예망속도는약

2.0-3.0 knot

. 잡음 제거

음향조사의 자료 분석은 후처리소프트웨어

(Echoview Ver.

4.7, Echoview Software Pty. Ltd, Australia)

.

,

,

,

제거하는것이매우중요하다

. De Robertis and Higginbottom

(2007)

,

Wang et al. (2015)

Fig. 3

.

, De Robertis and Higginbottom (2007)

Wang et al. (2015)

Target strength (dB)

Length (mm)

dB difference (dB)

y = 0.0435x + 318.35 R² = 0.43

0 200 400 600 800 1000 1200

0 2000 4000 6000 8000 10000 12000

Catch (kg) NASC (m

/nmi

)

SV raw data

NOISE SV removed

noise Data range bitmap 1 Mask 1

Erosion filter 3×3

Dilation filter 5×5 Dilation filter 7×7 Data range

bitmap 3

Mask 2

Select Median filter 7×7

Data range bitmap 2

Wang et al. (2015) De Robertis and Higginbottom

(2007)

38 kHz raw data Noise 38 kHz

Cleaned 38 kHz Resample by ping

38 kHz

SV difference 120-38 kHz (dB)

120 kHz raw data

Noise 120 kHz Cleaned 120 kHz

Resample by ping 120 kHz

Data range bitmap

Mask (Krill) Resample vertically

120 kHz Match ping times

Elephant Island Sou th S hetla nd I slan d

ANTARTICA Subarea 48.1

Longitude

Latitude

1.0

0.8

0.6

0.4

0.2

0 -20 -10 0 10 20

MVBS

(dB) 10 ping×2 m

T1 T2 T3 T4 T5 T6 T7

Cumulative Distribution Function

Fig. 3. Flow chart for the noise removal using acoustic data analy- sis software.

(a) 120 kHz raw

(b) TVT

(c) Minimum SV

(d) Erosion filter 3×3

(e) Dilation filter 5×5

(f) Dilation filter 7×7

(g) Median filter 7×7

(i) Select (dB)

38kHZ 120kHZ

SV raw data

De Robertis and Higginbottom

(2007)

Wang et al.

(2015)

(a) 5 ping x 2 m (b) 10 ping x 2 m

(c) 20 ping x 2 m (d) 30 ping x 2 m

(e) 40 ping x 2 m (f) 50 ping x 2 m (dB)

Fig. 4. Example of echogram for the noise removal based on acoustic data.

최석관

ㆍ

ㆍ

ㆍ

ㆍ

ㆍ

792

의해제안된방법을사용해서잡음제거의정도를비교하였다

.

,

raw data

TVT (Time Varied Threshold)

.

data range bitmap

S

(volume back- scattering strength)

S

mask

.

, data range bitmap

,

.

위의과정을거쳐잡음은상당수제거되었지만

,

Erosion filter 3×3

. Erosion filter

,

Di- lation filter 5×5, Dilation filter 7×7

. Dilation filter

data range bitmap

S

, mask

.

Median filter 7×7

, Select

mask

Median filter

,

data range bitmap

(Fig. 4).

주파수 차이를 이용한 남극크릴의 신호 추출

남극크릴의 에코를추출하기위해서는

38 kHz

120 kHz

.

(mean volume backscattering strength, ΔMVBS)

Fig. 5

.

해수면및해저면

,

.

CCAMLR

ping inter-

val

2

.

, resampled

.

이가명확하게결정되면

,

data

range bitmap

120 kHz

mask

,

ping

120 kHz

.

음향을이용하여음향산란층에존재하는남극크릴의밀도를 추정하기 위해서는 채집된 표본에대한 후방산란강도

(target

strength)

.

톤과같은작은어류및부레가형성되지않는치어들은크기가 작을뿐만아니라

,

.

,

distorted wave Born approximation (DWBA)

stochastic distorted wave Born approximation (SDWBA)

.

SDWBA

3

(1)

(Stanton et al., 1996; McGehee et al., 1998).

f

=∫

rpos

k

a (γ

-γ

)e

J

(2 k

a cos β

) ｜ d r ^pos ｜ ···(1)

4 k

cos β

여기서

, k

(m

, k=2π/λ), λ

(m, λ=c/f ); c

(m/sec), f

(Hz)

.

1

,

2

. γ

γ

γ

=(κ

-κ

)/κ

γ

=(ρ

-ρ

)/ρ

,

κ

(κ=1/ρc

)

, ρ

(kg/m

), c

(m/s)

. a

(m)

, r

(

), β

(degree), J

1

1

. f

(σ

)

,

(2)

.

TS=10logσ

=10log│ f

│

··· (2)

Target strength (dB)

Length (mm)

dB difference (dB)

y = 0.0435x + 318.35 R² = 0.43

0 200 400 600 800 1000 1200

0 2000 4000 6000 8000 10000 12000

Catch (kg) NASC (m

/nmi

)

SV raw data

NOISE SV removed

noise Data range bitmap 1 Mask 1

Erosion filter 3×3

Dilation filter 5×5 Dilation filter 7×7 Data range

bitmap 3

Mask 2

Select Median filter 7×7

Data range bitmap 2

Wang et al. (2015) De Robertis and Higginbottom

(2007)

38 kHz raw data Noise 38 kHz

Cleaned 38 kHz Resample by ping

38 kHz

SV difference 120-38 kHz (dB)

120 kHz raw data

Noise 120 kHz Cleaned 120 kHz

Resample by ping 120 kHz

Data range bitmap

Mask (Krill) Resample vertically

120 kHz Match ping times

Elephant Island Sou th S hetla nd I slan d

ANTARTICA Subarea 48.1

Longitude

Latitude

1.0

0.8

0.6 0.4

0.2

0 -20 -10 0 10 20

MVBS

(dB) 10 ping×2 m

T1 T2 T3 T4 T5 T6 T7

Cumulative Distribution Function

Fig. 5. Flow chart of acoustic data processing using dB difference method at 38 kHz and 120 kHz.

주파수차법을 이용한 남극 크릴의 종 식별

793

DWBA

한몸체변형을확률적

(stochastic)

SDWBA

,

리하기위해서먼저

SDWBA

도를분석하였다

(Demer and Conti, 2005; Conti and Demer, 2006; CCAMLR, 2010). SDWBA

,

38

120 kHz

,

(g=1.0357),

(h=1.0279)

(θ)

N (0°, 40.5°)

(Conti and Demer, 2006)

(38, 120 kHz)

(L)

TS

Fig. 7

.

별하기위한주파수차범위를나타낸것으로

, 2010

가

(SC-CCAMLR-XXIX, Annex 5, CCAMLR 2010; Field- ing et al., 2011)

SDWBA

TS

.

38 kHz

,

120 kHz

.

120 kHz

38 kHz

.

25-60 mm

,

0.36-13.17 dB

.

과학어군탐지기로부터음향산란층내에존재하고있는남극 크릴과다른생물간의종식별을위해

, 2

38 kHz

120 kHz

,

의해추정된남극크릴의

TS

.

(L￣)

TS

란강도의비율과동일하다는것으로

,

(3)

.

△MVBS =TS(120kHz, L￣)-TS(38kHz, L￣)

=SV(120kHz, L￣)-SV(38kHz, L￣) ···(3)

크릴을식별하기위한주파수차범위는

2010

사용한

SDWBA (Stochastic distorted-wave Born approxima- tion)

TS-

(Fig. 6)

S

(min-max)

(Table. 3, SC- CCAMLR-XXIX, Annex 5, CCAMLR 2010; Fieding et al.,

Target strength (dB)

Length (mm)

dB difference (dB)

y = 0.0435x + 318.35 R² = 0.43

0 200 400 600 800 1000 1200

0 2000 4000 6000 8000 10000 12000

Catch (kg) NASC (m

/nmi

)

SV raw data

NOISE SV removed

noise Data range bitmap 1 Mask 1

Erosion filter 3×3

Dilation filter 5×5 Dilation filter 7×7 Data range

bitmap 3

Mask 2

Select Median filter 7×7

Data range bitmap 2

Wang et al. (2015) De Robertis and Higginbottom

(2007)

38 kHz raw data Noise 38 kHz

Cleaned 38 kHz Resample by ping

38 kHz

SV difference 120-38 kHz (dB)

120 kHz raw data

Noise 120 kHz Cleaned 120 kHz

Resample by ping 120 kHz

Data range bitmap

Mask (Krill) Resample vertically

120 kHz Match ping times

Elephant Island Sou th S hetla nd I slan d

ANTARTICA Subarea 48.1

Longitude

Latitude

1.0

0.8

0.6 0.4

0.2

0 -20 -10 0 10 20

MVBS

(dB) 10 ping×2 m

T1 T2 T3 T4 T5 T6 T7

Cumulative Distribution Function

Fig. 6. Relationship between TS and body length calculated by SDWBA model (SC-CCAMLR-XXIX, Annex 5, CCAMLR 2010; Fieding et al., 2011).

(a) 120 kHz raw

(b) TVT

(c) Minimum SV

(d) Erosion filter 3×3

(e) Dilation filter 5×5

(f) Dilation filter 7×7

(g) Median filter 7×7

(i) Select (dB)

38kHZ 120kHZ

SV raw data

De Robertis and Higginbottom

(2007)

Wang et al.

(2015)

(a) 5 ping x 2 m (b) 10 ping x 2 m

(c) 20 ping x 2 m (d) 30 ping x 2 m

(e) 40 ping x 2 m (f) 50 ping x 2 m (dB)

Fig. 7. Comparison of De Robertis and Higginbottom’s method and Wang’s method using the noise removal of 38 kHz and 120 kHz.

최석관

ㆍ

ㆍ

ㆍ

ㆍ

ㆍ

794

2011).

를적용하였다

.

결 과

잡음 제거 결과

연구조사선은높은운영비용많은인력으로인해조업선에 서음향조사진행후획득한음향자료분석하는방법이중요시 되고있다

.

.

로기본적인잡음을제거하는방법을제안한

De Robertis and

Higginbottom (2007)

Wang et al. (2015)

(Fig. 7).

S

raw data

38

120 kHz

,

.

De Robertis and Higginbottom (2007)

.

, Wang et al. (2015)

erosion filter

dilation filter

. erosion filter

,

,

dilation filter

.

,

.

주파수 특성

Table 3. The recommended ranges (min-max) of S_V values (in dB) used to classify different size distributions of Antarctic krill Eu- phausia superba during the 2010 B0 assessment (CCAMLR 2010 (SC-CCAMLR-XXIX, Annex5), Fielding et al, 2011). The values shown on the upper and lower lines of each cell represent the S_V ranges for 120-38 kHz and 200-120 kHz, respectively

Minumum krill length (mm)

Maximum krill length (mm)

20 30 40 50 60

10 14.3-16.9 12.0-16.9 8.7-16.9 4.5-16.9 0.4-16.9 3.9-7.0 2.2-7.0 5.3-7.0 5.3-7.0 5.3-7.0 20 - 12.0-14.3 8.7-14.3 4.5-14.3 0.4-14.3

- 2.2-3.9 5.3-3.9 5.3-3.9 5.3-3.9

30 - - 8.7-12.0 4.5-12.0 0.4-12.0

- - 5.3- 2.2 5.3-0.7 5.3-1.4

40 - - - 4.5-8.7 0.4-8.7

- - - 3.9-0.7 3.9-1.4

50 - - - - 0.4-4.5

- - - - 0.7-1.4

(a) 120 kHz raw

(b) TVT

(c) Minimum SV

(d) Erosion filter 3×3

(e) Dilation filter 5×5

(f) Dilation filter 7×7

(g) Median filter 7×7

(i) Select (dB)

38kHZ 120kHZ

SV raw data

De Robertis and Higginbottom

(2007)

Wang et al.

(2015)

(a) 5 ping x 2 m (b) 10 ping x 2 m

(c) 20 ping x 2 m (d) 30 ping x 2 m

(e) 40 ping x 2 m (f) 50 ping x 2 m (dB)

Fig. 8. Example of echogram as to the volume of cell according to the characteristics of the frequency of Antarctic krill Euphausia superba.