Port 
Starboard 
  
performing such analysis, it must be assumed that the 
acoustical data obtained by scientific echo-sounders 
are representative samples of a population. 
However the effect of an approaching ship on a fish 
school must be considered. Misund and Aglen 
(1992a) observed with sonar that herring schools 
escape from purse seine fishing boats. They 
suggested that it is necessary to grasp the three- 
dimensional directivity pattern of underwater noise 
radiated from a ship and determine how noise affects 
fish behavior. To determine how a fish school reacts to 
a moving ship, horizontal projections were sliced into 
10m-deep layers, and counted the fish schools 
appeared by the distance from the ship's course 
(Figure 9). In the figure, the horizontal projection (B) is 
made by the specified layer from 20m to 30m in depth 
in the side projection (A), and the histogram of fish 
appearance against to the distance from the ship (C). 
: TS Frequenc 
Horizontal projection o of mney 
OI 
    
   
souereadde Jooyog 
Vertical projection 
Extracted layer 
20~30m 
  
  
Figure 9 Fish school distribution obtained by 
horizontal projection of specified layer. 
Figure 10 displays the frequency distribution of fish 
schools that appeared in each layer. The vertical axis 
indicates the number of fish, and the horizontal axis 
indicates the distance from the ship track. The central 
vertical broken line denotes the area just under the 
ship. If fish did not avoid the ship, each chart would 
show a flat distribution despite of distance from the 
ship. However it is clear in the top three charts that the 
shallower fish schools showed a large bias from the 
center, while the bottom two charts show nearly flat 
distributions. These figures suggest that fish schools 
possibly avoid ships, especially in shallow waters 
(Misund, 1990, 1993a). 
Misund (1994) observed the swimming direction of 
herring schools during trawl fishing and found that the 
schools escaped in the same direction as the ship was 
moving, not to the sides. He suggested that the 
732 
directivity of underwater noise emitted by ships is 
strong in the broadside direction and weak forward 
and to the rear of the ships. By determining the three- 
dimensional directivity pattern of underwater noise 
emitted by ships, the bias of fish school distribution 
patterns just under ships will be elucidated. 
  
40 Fr 
30 b 10—20m 
20 D, Dun ——L 
10 f a | 
O A a A 
40 
  
  
Ww 
o 
pop 
Ww 
o 
n 
© 
8 
  
NO 
oo 
TT TU 
  
50 — 60m 
Port Center Starboa 
100m 100m 
30 F 
  
  
  
Figure 10 Frequency distribution of fish school 
appearance in relation to fish avoidance. 
4.2 Species classification by characteristics of fish 
school shape 
If a fish school is observed, it is useful to identify the 
species and to predict the behavior of the school (Weill 
et al., 1993, Lu and Lee, 1995, Reid and Simmonds, 
1993). Unfortunately, optical observation using 
underwater cameras is limited to close range because 
of the rapid attenuation of light in seawater (Pitcher 
and Partridge, 1979). 
Hara (1985) observed from an airplane that moving 
sardine schools are crescent shaped, with the convex 
side facing forwards the direction of movement. In this 
study, fish schools observed by sonar were classified 
into several shape types. In order to test the 
relationship between the shape and species or 
behavior, the volume, lengthwise-to-crosswise ratio 
(elongation), circularity, fractal dimension, and fish 
school depth were extracted as characteristic 
parameters of each fish school shape. 
Figure 11 shows the scattergram matrix of three 
representative parameters: school depth, elongation, 
and school volume  (logarithmic value). The 
relationship between these parameters show that 
small fish schools often occurred in deep water, and 
St 
a “no az 
— 
m 
«" "r3 T3 —1 CO) «€ t3 CY) a (MN. DO: —* 
DO -T— 
ar