445

1 INTRODUCTION

The Japan Coast Guard consolidated the Tokyo Bay

Maritime Traffic Center and the four port traffic

control offices (Tokyo, Yokohama, Kawasaki, and

Chiba) into the Tokyo Wan Vessel Traffic Service

Centre on 31 January 2018 [5]. This is expected to

improve both safety and efficiency in Tokyo Bay,

which is one of the most congested sea areas in Japan,

by eliminating poor traffic service areas where ships

are not provided with accurate information, allowing

the ships under control to navigate on time without

waiting for traffic lights or congestion and allowing

ships not under control to navigate seamlessly with

the information provided, such that they are not

affected by the ships under control. The bay entrance,

in the vicinity of the Tsurugisaki-Sunosaki Line and

the entrance to the Uraga Suido Traffic Route (Uraga

Suido), is known to be accident-prone; collisions have

occurred because of the intermingling of vessel traffic.

Accordingly, a study on marine traffic control

measures at the entrance of Tokyo Bay was conducted

[13]. In this study, a questionnaire survey was

conducted among the concerned parties regarding the

areas where they experienced discomfort. To rectify

marine traffic based on virtual automatic

identification system (AIS) route signs in this area, the

marine traffic flow was simulated, and the status of

vessel traffic in the area was determined.

Consequently, the necessity of establishing a virtual

AIS route sign in the southern part of Uraga Suido

and separating the traffic flow of vessels entering and

leaving the channel by a line connecting the sign and

the lighted Buoy 1 at the centre of Uraga Suido, was

realized. The location of the virtual AIS was studied,

and the changes in marine traffic flow due to the

installation of the virtual AIS were investigated by

simulation. Consequently, it was confirmed that the

Estimation of Changes in Marine Traffic Flow Due to

Installation of Virtual Buoys Based on the OZT Method

G. Fukuda, H. Tamaru & R. Shoji

Tokyo University of Marine Science and Technology, Tokyo, Japan

ABSTRACT: In this study, changes in maritime traffic flow due to the installation of virtual buoys were

analysed from the perspective of collision, based on the obstacle zone by target (OZT) method. The existence of

an OZT within ±2° in the bow direction is considered to be dangerous; all the OZTs that a vessel encountered at

the time when this orientation prevailed were subjected to analysis. For the calculation of the OZT model, an

SD3 model was used. The estimated OZT was used to calculate the OZT density, and the positions of the ships

that were encountering the OZT were used for the analysis. It was found that the OZT density at the locations

where collisions had occurred in the past was approximately 0.01 [times/km

]. After the installation of virtual

buoys, most places around there were reduced to around approximately 0.005 [times/km

]. The results imply

that the installation of virtual buoys has reduced the risk of collisions. In contrast, the OZT density in the

southern region of Buoy 1 increased from approximately 0.005 [times/km

] to approximately 0.01 [times/km

suggesting that a more detailed analysis of this area is required.

http://www.transnav.eu

the International Journal

on Marine Navigation

and Safety of Sea Transportation

Volume 15

Number 2

June 2021

DOI: 10.12716/1001.15.02.24

446

rectification effect of the installation of the virtual AIS

improved the safety. Since 1 March 2019, a new

maritime traffic route designation has been

established in accordance with the Maritime Traffic

Safety Act for the Tokyo Bay entrance sea area [6].

Accordingly, the vessels entering and exiting Uraga

Suido are expected to navigate directly between the

recommended passage established off the west coast

of Izu Oshima and the sea area where the rectification

is expected to occur, and the traffic flow of vessels

entering and exiting Tokyo Bay is expected to be

organised in the sea area between the west coast of the

island and Uraga Suido [13]. This is believed to

reduce dangerous encounters between vessels, such as

the one that actually led to the collision described

later in this paper. However, no study has yet been

conducted to analyse changes in traffic flow after the

installation of virtual buoys, from the perspective of

collisions. Therefore, in this study, we investigated the

changes in traffic flow by applying the obstacle zone

by target (OZT) analysis method, which the authors

have been studying for some time now [1–3]. In this

study, under the condition that the OZT exists in the

range of ±2°, based on the direction of a ship’s course,

the OZT density and the position of the ship at the

time when this orientation prevails, are analysed. By

analysing the changes in traffic flow at the past

collision sites based on the proposed analysis method,

we investigated the trend of the change in marine

traffic flow in terms of collision zones.

2 ESTIMATION OF OZT DENSITY AND VESSEL

POSITION UNDER THE CONDITION THAT OZT

EXISTS IN THE DIRECTION OF HEADING

OZT was estimated based on the safety passing

distance (SD) according to the matching relationship

with the target vessel [1, 4]. In the OZT density

analysis, all OZTs within a set distance range were

included in the calculation [1]. However, a problem

was encountered because few OZTs did not pose any

danger. Therefore, to avoid calculating non-dangerous

OZTs, the authors only considered the OZTs

appearing in the direction of the ship path for

calculation [12]. While calculating the OZTs, it was

difficult to estimate which OZTs increased the overall

OZT density because the OZTs lengthened,

particularly, in the situations where ships were

navigating courses that were close and almost similar

to that of the test ship. To solve this problem, the

authors divided the diameter of the SD into three

parts and employed the SD3 method, which was first

introduced by Prof. Imazu Hayama in our OZT study

group, to calculate the OZT. In addition, a vertical line

was drawn from the proximate side of the calculated

OZT to the line connecting the farther side of the OZT

and the ship, and the intersection point was used as

the new OZT. All the methods are described in the

following sections.

2.1 OZT calculation based on SD3 method

The basic method to determine the OZT is to set the

SD in the vicinity of the ship and establish a course

that does not allow other ships to enter that area. In

this study, we set the SD such that, it is divided into

three circles within the conventionally used SD. The

method of determining the OZT course by setting the

SD in this manner is described as follows.

1. A normal SD is drawn that divides the

conventional SD into three equal parts in the

direction of the velocity vector of a nearby target

ship (hereinafter referred to as “target”).

2. Tangent lines are drawn from the position of the

target ship to the SD (blue, purple, green, and

yellow dashed lines in Figure 1).

3. Each tangent line is parallelised such that it touches

the tip of the velocity vector of the target ship,

extended from the position of the test ship.

4. The intersections of the parallel tangent lines and

the speed circle of the test ship are determined.

5. The range of the line connecting the outermost

point of each intersection with the position of the

test ship is the range of the OZT course (OZT Co1,

OZT Co2, OZT Co3, and OZT Co4 in Figure 1).

The courses on both sides that represent each OZT

range are defined as OZT Co1 and OZT Co2 and OZT

Co3 and OZT Co4. The vector connecting each OZT

Co1, Co2, Co3, and Co4 to the end of the velocity

vector of the target ship is the direction and

magnitude of relative velocities of the ship as it

traverses each course. For example, the direction from

the intersection of OZT Co1 and the test ship’s

velocity circle to the tip of the target ship's velocity

vector is the direction of the relative vector when the

test ship travels along Course OZT Co1, and its

magnitude is the relative velocity. The Time to the

Closest Position of Approach (TCPA) when

navigating each OZT Co can be obtained by dividing

the length of the tangent line drawn from the target

ship to each SD by the respective relative velocity.

Figure 1. Procedure to determine the OZT Course (Co) by

drawing.

It is considered that OZT1 is the endpoint of the

OZT closest to the test ship, OZT2 is the endpoint of

the OZT on the farther side, the angle of inclination is

θ, and the distance from the test ship to OZT1 is D1.

Dropping the vertical line from OZT1, the intersection

of the line connecting the test ship and OZT2 is OZT2'

as shown Figure 2. Hereinafter, the zone connecting

OZT1 and OZT2' is referred to as SD3C OZT.

447

Figure 2. Relationship between OZT1, OZT2, and OZT2’.

The distance, D2, between the test ship and OZT2'

is obtained using the following formula:

2 1cosDD



(1)

The parameters for the test ship and target ships

were set as shown in Table 1. Figure 3 shows the

relationship between the conventional OZT, the OZT

when SD is divided into three parts (SD3 OZT), and

the OZT obtained from the angle of elevation, θ

(SD3C OZT), when the course of the target ship is

turned every 30° from 0° to 360°. The diameter of the

conventional SD was set to 290 m. As can be seen

from Figure 3, the SD3C OZT used in this study

clearly shows the horizontal range where the target

ship cannot proceed when viewed from the

perspective of the test ship. However, it should be

noted that the SD3C OZT used in this study cannot be

used in the evaluation of a case where both ships

avoid each other over a short distance.

Table 1. Parameters for test ship and target ship.

_______________________________________________

SOG [knot] Length [m] Width [m]

_______________________________________________

Test Ship 10 200 36

Target Ship 7 90 12

_______________________________________________

Figure 3. Relationship between Normal OZT, SD3 OZT, and

SD3C OZT.

2.2 Calculation conditions for OZT

For the existence of an OZT with one of the target

ships within ±2° on the test ship’s course [12], all the

OZTs within ±45° of the test ship’s course and within

4 miles, were included in the analysis, as shown in

Figure 4. The distance between the ships and the

distance to the OZTs were also set within 4 miles. If

one endpoint of an OZT was within 4 miles and the

other endpoint was set at a point beyond 4 miles, the

time to reach the farther OZT was calculated as 30

min. TS stands for “Target Ship”. The ship positions

in the estimation results (e.g. Figure 9) that appear

later show the positional relationship between the test

ship and target ships.

Figure 4. Example of OZT calculation conditions.

2.3 OZT Density

The OZT density is the sum of the existence time of

the OZTs in the grid divided by the observation time.

However, because this study uses data that are

interpolated over 30 s, it presents an error of ±30 s. For

the navigational density of the test ship, the time that

the test ship has existed in the grid can be estimated

from the test ship’s speed; however, it is difficult to

estimate the existence time of the OZT, whose state

changes depending on the paths of the target and the

test ships. A feasible solution is to use the AIS data

interpolated to 1 s [1]; however, it is extremely time-

intensive to estimate the existence time, owing to the

heavy marine traffic near the entrance of the Tokyo

Bay. The OZT density at the time of the OZT

encounter is defined as the number of OZTs per unit

area, per unit time, in each mesh and is calculated

using the following formula:

OZT

eOZT





(2)

where:

DOZT: OZT density [times/km

eOZT: Existence time in the OZT mesh for each

matchmaking relationship [s],

a: Mesh area [km

T: Total Time [s], and

n: Number of target ships that generated the OZT in

the mesh.

448

2.4 Ship position when OZT calculation conditions are

satisfied

When the conditions for calculating the OZT

described in Section 2.2 are met, the positions of all

the ships that generate OZT in relation to the test ship

are examined, and the average number of ships per

month is shown on the mesh, where the ships are

present. The examination was conducted to determine

the locations of the test and target ships, when the

conditions were met.

3 ANALYSIS OF SEA AREA AND DATA USED

3.1 Analysis of sea area

For the analysed area shown in Figure 5, the

estimation has been performed based on the method

proposed in this study. The locations of the collisions

shown in the figure correspond to the collision cases

IV and V which are similar to the collision cases I–III

that have been used as references when installing the

virtual buoy.

Figure 5. Analysed sea area, positions of the collision, the

tracks, and the virtual buoys.

3.2 Collision cases in the analysis

Three collision cases (I, II and III) have been discussed

and used for verification in the virtual buoy-

installation study group [13]. In this study, the

locations of two additional collisions (IV and V),

involving a north-bound ship and a south-bound ship,

are described.

Case I (Collision between south-bound and north-

bound ships)

Cargo Ship A was heading south-southwest, and

Container Ship B was heading northeast towards

Keihin Port, Tokyo, when they collided at 03:10 on 18

March 2014 at the mouth of Tokyo Bay, off the

southeast coast of Tsurugisaki, Miura City, Kanagawa

Prefecture. When the two ships approached each

other, Cargo Ship A turned to the right and Container

Ship B turned to the left and continued straight ahead,

resulting in a collision between the two ships [7].

Case II (Collision between south-bound and north-

bound ships)

Cargo Ship A was proceeding south-southwest,

and Cargo Ship B was proceeding northeast, when at

approximately 05:18 on 13 April 2006, the bow of

Cargo Ship A collided with the port bow of Cargo

Ship B, off the east coast of the Miura Peninsula in

Kanagawa Prefecture. During the night, visibility was

restricted off the east coast of the Miura Peninsula,

and owing to inadequate radar coverage of both ships,

they inevitably approached each other. Because the

speeds of the ships were not reduced to the minimum

necessary to maintain their respective courses, the

ships could not stop when necessary, and the two

vessels collided. The ships were flooded and they

eventually sank [10].

Case III (Collision between two north-bound ships)

Cargo Ship A was travelling north-northwest, and

Cargo Ship B was travelling north-northeast, when the

starboard centre of Cargo Ship A collided with the

port bow of Cargo Ship B at approximately 03:49:40,

on 16 July 2014, off the northwest coast of Susaki,

Tateyama City, Chiba Prefecture. Cargo Ship A

suffered a dent with a crack on its starboard hull, and

Cargo Ship B suffered a dent with a crack on its port

hull. Nonetheless, there were no casualties on either

ship. The collision occurred during the night, when

the visibility was restricted off the northwest coast of

Susaki, despite the fact that Cargo Ship B tried to turn

the bow of the ship to 039°. The instructions of the

navigator were unclear and Cargo Ship B continued to

turn to the right, towards the front of Cargo Ship A.

Moreover, the captain of Cargo Ship A might have

assumed that Cargo Ship B would eventually turn to

the left, and therefore, Cargo Ship A continued to

make a small change in direction, and the two vessels

collided [8].

Case IV (Collision between south-bound and north-

bound ships)

Cargo Ship A was proceeding south and Cargo

Ship B was proceeding northward, when at 18:03 on 7

October 1998, at a location 137.5° and 4.8 NM from the

Tsurenzaki Lighthouse, the starboard bow of the

Cargo Ship A collided with the port bow of the Cargo

Ship B at an angle of 21° from astern, while continuing

its original course at the initial speed [9].

Case V (Collision between south-bound and north-

bound ships)

Car Carrier A was heading north, and Container

Ship B was heading south, when at 6:10 p.m. on 24

January 2000 at a location 159° and 3.2 NM from

Tsurugizaki Lighthouse, they collided. Car Carrier A

was heading 336° when the port bow of the Container

Ship B collided with the starboard centre of the Car

Carrier A at an angle of 40° from astern [11].

449

3.3 AIS Data

AIS data of three months (from October to December

2018) before the installation of the virtual buoy, and

three months (from July to September 2019) after the

installation of the virtual buoy, were used. In both

cases, before use, the AIS data were interpolated to

every 30 s. Figures 6 and 7 show the positions of the

north-bound and south-bound ships before and after

the installation, respectively. The arrows in the figure

indicate the general direction of the ship traffic. It can

be seen that before the installation of the virtual buoy,

both north-bound and south-bound ships were

observed in the vicinity of sites I, II, IV and V, where

the collision between the north-bound and south-

bound ships occurred as detailed in Section 3.2. After

the installation of the virtual buoy, the number of

north-bound ships in the vicinity of sites I and V,

decreased.

Figure 6. (a) South-bound and (b) north-bound ship

positions where one or more ships were observed before the

virtual buoy was installed.

Figure 7. (a) South-bound and (b) north-bound ship

positions where one or more ships were observed after the

virtual buoy was installed.

4 RESULTS

An absolute evaluation index for the OZT density has

not been established yet. Therefore, a comparative

evaluation was conducted for the following ship-

encounter situations: (1) north-bound test ship and

north-bound target ship; (2) north-bound test ship and

south-bound target ship; (3) south-bound test ship

and north-bound target ship; and, (4) south-bound

test ship and south-bound target ship. The evaluation

was conducted considering the situation where an

OZT exists with one of the ships within ±2° in the

direction of the main ship (test ship). Ships with

speeds of 6 knots or more and an inter-ship distance

of 4 miles or less were analysed using the proposed

analysis method. However, tugboats and pilot boats

were excluded from the analysis. The maximum value

of the colour bar was set such that the sea area where

the collision occurred in each matching relationship

was close to the middle value and was evaluated

comparatively. In this study, ships emerging from the

inner part of Tokyo Bay will be referred to as

emerging from Uraga Suido, even if they are not

navigating through Uraga Suido. Similarly, ships

entering Tokyo Bay will be referred to as entering

Uraga Suido, even if they are not navigating through

Uraga Suido.

4.1 North-bound test ship and south-bound target ship

encounter situation

This section describes the results of estimating the

OZT density and the position of the ship at that time,

in relation to the ship emerging from Uraga Suido

(north-bound) when the test ship enters Uraga Suido

(south-bound). First, Figure 8 (a) shows that before the

installation of the virtual buoy, the value of OZT

density was relatively high, at approximately 0.015

[times/km

] or above, in the vicinity of the virtual

Buoy 3 (the northernmost buoy), and the value of OZT

density decreased as it further moved north. The

value of OZT density is approximately 0.01

[times/km

] in the vicinity of collision sites I, II and IV

in this relation. It was also confirmed that the area

where the collision occurred did not have the highest

OZT density in the comparative evaluation. Figure 8

(b) shows that the spread of the OZT density is

smaller than that of Figure 8 (a) before the installation

owing to the rectification of the flow of ships

emerging from Uraga Suido. From Buoy 3 onwards,

the density of OZT in the vicinity of Collision II

increases. Figure 9 shows that when the ship entered

the Uraga Suido, it intersected with the target ships

that generated OZT over a wide area. In the vicinity of

the past collision sites I, II, IV and V, it was confirmed

that the positions of the test ship and target ships

overlapped when the test ship encountered the OZT.

Figure 10 (a) shows that the position of the ship, when

it encounters the OZT, is concentrated, owing to the

rectification of the traffic flow. Between Buoys 1 and

3, it can be confirmed that there are numerous ships

experiencing OZTs; however, Figure 10 (b) shows that

the positions of the test ship encountering OZTs do

not overlap with the positions of target ships that may

be causing them. It can also be confirmed that few test

ships have encountered the OZT in the vicinity of

collision sites I and V in the past. Similarly, in the

450

vicinity of site IV, no target ships were present. Figure

10 (a) shows the ship encountering OZT, and Figure

10 (b) shows the presence of the target ships in the

vicinity of the increased density of OZT on the north

side of virtual Buoy 1.

Figure 8. OZT densities (a) before and (b) after the

installation of virtual buoy for the south-bound (test) and

north-bound (target) ships.

Figure 9. Positions of the (a) test and (b) target ships, when

the OZT is calculated before the installation of the virtual

buoy.

Figure 10. Positions of the (a) test and (b) target ships, when

the OZT is calculated after installation of the virtual buoy.

4.2 South-bound test ship and north-bound target ship

encounter situation

Next, the OZT density and position of the ship that

generates it when the test ship leaves the Uraga Suido

(south-bound) and the target ship enters the Uraga

Suido (north-bound), are discussed. Figure 11 (a)

shows that the OZT density is highest at

approximately 0.015 [times/km

] in the vicinity of

virtual Buoy 3, and extends to the sea area near Buoy

1 with value of approximately 0.01 [times/km

]. This

trend is similar to that shown in Figure 8. Relatively

high values of approximately 0.01 [times/km

] can

also be observed in the vicinity of the past collision

sites I, IV and II. Figure 11 (b) shows that the

installation of the virtual buoy congests the areas with

relatively high OZT density in the path of the ships

entering Uraga Suido, and the density is lower than

that before the installation. However, the value of

OZT density increased to approximately 0.01

[times/km

] on the south side of Buoy 1, where

collision II occurred in the past. Figure 12 (a) and (b)

show that when an OZT occurs, there is a possibility

that the target ship is approaching from the starboard

side of the ship. In particular, in the vicinity of the

collision sites I, II and IV, there is a possibility that the

test ship is encountering an OZT and that the target

ship is approaching from the starboard side. Figure 13

shows that after the installation of the virtual buoy,

there are no (or particularly few) ships approaching

from the starboard side of the ship, when the ship

encounters an OZT from Buoy 1 to 3. However, the

number of ships encountering OZT near Buoy 1

increased by approximately 3–5 ships, and the

presence of the target ships that may generate OZT

south of Buoy 1 was confirmed.

Figure 11. OZT densities (a) before and (b) after the

installation of the virtual buoy for the north-bound (test)

and south-bound (target) ships.

Figure 12. Positions of the (a) test and (b) target ships, when

the OZT is calculated before the installation of the virtual

buoy.

451

Figure 13. Positions of the (a) test and (b) target ships, when

the OZT is calculated after installation of the virtual buoy.

4.3 North-bound test ship and north-bound target ship

encounter situation

Figure 14 (a) and (b) show that there is no significant

change in the OZT density between the test ship and

the target ship before and after the installation of the

buoy when the ships enter Uraga Suido. The same is

true in the vicinity of collision site III, where there is

no change in the OZT density of approximately 0.01

[times/km

]. From the south-side of Buoy 1, the values

of 0.015 [times/km

] or above, are sustained which

may be due to the influence of the continuous

existence of the OZT owing to the ships navigating in

a row for a long time. Figure 15 shows that before the

buoy was installed, OZT may have occurred over a

wide area when approaching a ship on the starboard

side. Furthermore, Figure 16 shows that the buoy has

improved the traffic flow for ships that navigate in the

same direction and that, when the ship encounters the

OZT from Buoy 1 to Uraga Suido, the matching

relationship, where the ship approaches the position

of the target ship that may be on the right side of the

ship, is significantly reduced.

Figure 14. OZT densities (a) before and (b) after the

installation of the virtual buoy for the north-bound (test)

and north-bound (target) ships.

Figure 15. Positions of the (a) test and (b) target ships, when

the OZT is calculated before the installation of the virtual

buoy.

Figure 16. Positions of the (a) test and (b) target ships, when

the OZT is calculated after the installation of the virtual

buoy.

4.4 South-bound test ship and south-bound target ship

encounter situation

Figure 17 (a) shows that the OZT density of south-

bound ships spreads in a triangular shape from the

area where the ships leave Uraga Suido. It can be seen

that the density of OZT among the ships leaving

Tokyo Bay to Eastern Japan was approximately 0.015

[times/km

] until Buoy 3. After the installation of the

virtual buoy, it can be confirmed that the density of

OZT among the ships leaving Tokyo Bay to Eastern

Japan increased to approximately 0.015 [times/km

]

along the buoys where approximately 0.01

[times/km

] before the installation; this trend

disappears and the density of OZT decreases after

leaving Buoy 1 (the northernmost buoy). In addition

to the area along the buoy, the area with a relatively

high OZT density decreases. Comparing Figures 18

and 19, the number of ships encountering the OZT

along the buoy increased after the installation of the

virtual buoy. Comparing Figures 18 (a) and 19 (a), it

can be observed that the routes of the ships that are

heading for western Japan and those that are heading

south are becoming increasingly separated. In case of

the OZT density of ships that are not along the buoys

heading towards western Japan, the relatively high

area decreases.

452

Figure 17. OZT densities (a) before and (b) after the

installation of the virtual buoy for the south-bound (test)

and south-bound (target) ships.

Figure 18. Positions of the (a) test and (b) target ships, when

OZT is calculated before the installation of the virtual buoy.

Figure 19. Positions of the (a) test and (b) target ships, when

OZT is calculated after the installation of the virtual buoy.

5 DISCUSSION

Considering the relationship between the test ship

going towards Uraga Suido and the target ships

emerging from Uraga Suido, it can be inferred that at

this time, the test ship is the course-keeping vessel

and the target ship is obligated to avoid the test ship.

Before the installation of the virtual buoy, the test and

target ships could not be differentiated in an area with

a high OZT density. However, after the installation of

the virtual buoy, the test ship could be differentiated

from the target ships along the buoy; it is certain that

the ships can navigate more safely than before the

installation of the virtual buoy, by following the buoy

even when OZT exists. However, in the vicinity of the

past collision site II, the OZT density increased from

approximately 0.005 [times/km

] to approximately

0.01 [times/km

], and the positions of the test ship

encountering the OZT and its possible target ship

overlapped. Therefore, it is difficult to judge whether

the installation of virtual buoys improves safety,

based on this analysis method. In Figure 10 (b), the

position of the south-bound ships shows that the

traffic flow of ships leaving the Uraga Suido and

continuing southward, is concentrated along the

virtual buoy. This makes it easier for the north-bound

test ship to predict the position of the target ship,

which potentially leads to improved safety.

In the case of a ship leaving Uraga Suido, it is

highly probable that the ship will be obligated to take

evasive action in relation to a ship entering the

channel. Before the installation of the buoy, the

position of the ship encountering the OZT overlapped

with the position of the target ship at that time,

indicating that the ship may have been navigating in a

manner that avoided the course of the target ship, to

avoid the risk of collision. After the installation of the

buoy, ships entering Uraga Suido began to pass to the

south of Buoy 1. Therefore, it can be inferred from the

decrease in OZT density and the positions of the test

and target ships encountering OZTs that the ships are

capable of adjusting their course to align with the

destination more safely because of the decrease in the

number of evasive manoeuvres. However, the OZT

density increases from approximately 0.005

[times/km

] to approximately 0.01 [times/km

], south

of Buoy 1. As mentioned earlier, further analysis is

required in this area.

In the case where both the test ship and the target

ship were heading towards Uraga Suido, the

installation of the virtual buoy significantly reduced

the number of ships that are normally from western

part of Japan approaching from their starboard side.

No significant change in the OZT density or the

number of ships was observed in the area of Collision

III. However, with the installation of the buoy, it is

assumed that ships from the eastern part of Japan

navigate along the buoy at an early stage. Therefore, it

is implausible that the vessels will navigate through

the same course as that in Collision III.

As a result of the installation of the virtual buoy, it

is now less probable that ships leaving Uraga Suido

and heading southwest and south will encounter

ships heading south of Buoy 1 into Uraga Suido. It can

be inferred from the decrease in the OZT density that

the ships heading southwest can now safely navigate

a wider area. It is inferred from the increase in OZT

density that the ships sail along the buoy for a long

time in the same direction, similar to the ships sailing

in the north direction.

Finally, the past collisions did not occur in the area

with the highest OZT density; however, in the area

with the OZT density of approximately 0.01

[times/km

], after the installation of the virtual buoy,

the area south of Buoy 1 and the area of collision III

are the most probable locations for the occurrence of

453

collision. As mentioned above, the installation of

virtual buoys may have improved safety in these

waters; however, given the fact that collisions have

occurred in waters with an OZT density of

approximately 0.01 [times/km

], it is necessary to

navigate these waters more carefully and prevent

collisions caused by human error.

6 CONCLUSION

In this study, a virtual buoy installed in Tokyo Bay (by

a study committee [13]), was analysed in terms of

collisions based on the OZT before and after

installation. The existence of OZTs within ±2° in the

bow direction was considered a hazard, and all

occurrences of OZTs encountered by the ship at that

time were included in the analysis. The OZT model

was changed from an SD model to an SD3 model, and

the OZT was converted to be viewed horizontally

from the perspective of the ship. The estimated OZT

was used to calculate the OZT density. The positional

relationship between the test ship and the target ship

in the OZT was used in the analysis. The estimated

OZT was used to calculate the OZT density, and the

positional relationship between the test ship and the

target ship, during the calculation of the OZT, was

used in the analysis. It was found that the OZT

density was approximately 0.01 [times/km

] in the

places where collisions had occurred in the past. It

was confirmed that the installation of virtual buoys

reduced the risk of collisions in the locations where

collisions had occurred in the past. It was also found

that the buoys facilitated in organising the flow of

marine traffic, making it easier to predict the

movement of ships. Three new passages were created

on the south side of virtual Buoy 1, two from south-

east and south direction to Uraga Suido, and a third

from Buoy 1 to the south. After the installation of the

buoy, the OZT density in this area increased from

approximately 0.005 [times/km

] to approximately

0.01 [times/km

], which is the density observed

around past collision sites before the installation of

virtual buoys, suggesting that a more detailed analysis

is needed in this area. Since there are few analyses

that use the location of ship collisions, the results of

the OZT density estimation were limited to analysing

the changes using the locations where collisions

occurred in the past as indicators. However, we were

able to demonstrate that the places where ships

collide are not necessarily the most congested areas

and not the places with the highest OZT density. In

the future, if we can develop an absolute index using

OZT density, we will be able to make quantitative

evaluations and contribute to the reduction of ship

collisions.

ACKNOWLEDGEMENT

The authors would like to express their sincere gratitude to

the Japan Coast Guard, who provided the AIS data used in

this study. This work was supported by JSPS KAKENHI

(Grant Number: JP18K13960).

REFERENCES

1. Fukuda, G., Shoji, R.: Development of Analytical

Method for Finding the High Risk Collision Areas.

TransNav, the International Journal on Marine

Navigation and Safety of Sea Transportation. 11, 3, 531–

536 (2017). https://doi.org/10.12716/1001.11.03.20.

2. Fukuda, G., Shoji, R.: Estimating Collision Course Area

Using OZT in Offshore Refugee Outside the Port. The

Journal of Japan Institute of Navigation. 135, 129–134

(2016). https://doi.org/10.9749/jin.135.129.

3. Fukuda, G., Takashima, K.: Analyzing the marine traffic

condition for estimating the high risk areas in the

emergency evacuation. In: 2015 International

Association of Institutes of Navigation World Congress

(IAIN). pp. 1–6 (2015).

https://doi.org/10.1109/IAIN.2015.7352245.

4. Imazu, H.: Evaluation Method of Collision Risk by

Using True Motion. TransNav, the International Journal

on Marine Navigation and Safety of Sea Transportation.

11, 1, 65–70 (2017).

https://doi.org/10.12716/1001.11.01.06.

5. Japan Coast Guard: Centralizing Marine Traffic Control

in Tokyo Bay,

https://www.kaiho.mlit.go.jp/e/Centralizing%20Marine

%20Traffic%20Control%20in%20Tokyo%20Bay.html,

last accessed 2021/03/11.

6. Japan Coast Guard: New Shipping Route Designation at

Tokyo Bay Entrance,

https://www.kaiho.mlit.go.jp/03kanku/ichigenka/pdf/ko

kuji_e.pdf, last accessed 2021/03/11.

7. Japan Transport Safety Board: Ship Accident

Investigation Report,

https://www.mlit.go.jp/jtsb/ship/rep-acci/2016/MA2016-

5-1_2014tk0009.pdf, last accessed 2021/03/12.

8. Japan Transport Safety Board: Ship Accident

Investigation Report,

https://www.mlit.go.jp/jtsb/ship/rep-acci/2015/MA2015-

9-15_2014yh0087.pdf, last accessed 2021/03/12.

9. Marine Accident Inquiry and Safety Investigation

Association: Cargo ship Kasho Maru cargo ship Blue

Peak collision incident,

https://nippon.zaidan.info/seikabutsu/2002/00888/conten

ts/0662.htm, last accessed 2021/03/12.

10. Marine Accident Inquiry and Safety Investigation

Association: Cargo ship Tsugaru Maru cargo ship

Eastern Challenger collision incident,

https://maia.or.jp/wp-content/uploads/pdf/accidents/ ㉜

heisei%20zenki.pdf, last accessed 2021/03/12.

11. Marine Accident Inquiry and Safety Investigation

Association: Collision of cargo ship Yongchai with cargo

ship Toyo Maru No. 12,

https://nippon.zaidan.info/seikabutsu/2002/00889/conten

ts/0032.htm, last accessed 2021/03/12.

12. Nishizaki, C., Kayano, J., Shoji, R., Imazu, H.: Encounter

Features of Collision Risk in the West Side Sea Area of

Izu Oshima using OZT. The Journal of Japan Institute of

Navigation. 139, 48–54 (2018).

https://doi.org/10.9749/jin.139.48.

13. Tokyo Wan Association for Marine Safety: Study of

marine traffic control measures in the waters at the

mouth of Tokyo Bay in 2017,

http://www.toukaibou.or.jp/img/file18.pdf, last accessed

2021/03/11.