531
1 INTRODUCTION
Ships were significantly damaged by the Great East
JapanEarthquakeandtheensuingtsunamionMarch
11, 2011. Because Nankai Trough Earthquake and
Tokai Earthquake occurrence are highly probable in
thenearfuture,theimportanceofevacuationmeasure
for ships has become increasing. While various
measureshavebeendiscussed,ithasnotbeendoneto
clarify dangerous waters and safe place from actual
marine traffic analysis when the tsunami predicted.
To let the Tsunami go past safely, the navigators
normallynavigateshipstothedeep‐waterareathatis
sometimes said more than 200 meters (Kamaishi
Japan Coast Guard 2004). If it is not possible, they
instructabandontheshipandevacuateonthelandas
one choice. Actually, several ships are evacuated to
thesafedepthareaattheGreatEastJapanEarthquake
(Hidenari M. et al. 2011). Preliminary measures in
consideration ofthe marine traffic situation leads to
quicklyandsafelyevacuateshipstosafetydepthsea
area.
The authors introduced marine traffic analytical
methodsbasedontheGasmodel(GenF.etal.2013)
andOZT (GenF. &RuriS.2016).In this paper,one
year AIS data has been studied with both models.
Then results are compared with estimated fishery
tracksby hearing observation (Gen F. 2016). Finally,
shipsthatareactuallyheadingtoOZTareanalyzed.
2 CALCULATIONMODELS
2.1 Gasmodelbasedriskanalysis
The Gas model based calculation for the purpose of
developing Tsunami measure is introduced and
detailsareshowninGenF.&KyokoT.2015.Therisk
is become higher when more vessels exist and take
different courses in the area during the sampling
time. In this study, 1 hour average rate (TR) are
calculatedasfollows:
Development Of Analytical Method for Finding the
High Risk Collision Areas
G.Fukuda
TokaiUniversity,Tokyo,Japan
R
.Shoji
TokyoUniversityofMarineScienceandTechnology,Tokyo,Japan
ABSTRACT:ThetrafficconditionhasbeenanalyzedbymodelsthatarebasedonGasmodelandObstacleZone
byTarget(OZT)byusing1yearAISdataforTsunamimeasure.ByapplyingGasmodelbasedmethod,itis
possibleto analyzethe areainterms ofships’relative angle, size,speedand density.TheOZT isoriginally
developedtoshowthecollisionzoneonthetargetship’scourse. Byusing thismethodtothemarinetraffic
analysis,highpossibilityofcollisionareasareestimated.Moreover,shipsthatcoursesareheadingtoOZTare
identifiedandstudiedtheirpositionandhowlongtheyproceedcourseheadingtoOZT.
http://www.transnav.eu
the International Journal
on Marine Navigation
and Safety of Sea Transportation
Volume 11
Number 3
September 2017
DOI:10.12716/1001.11.03.20
532
1
24
ij ij
iij
DVt
TR
dS
(1)
D
ij and Vij are the geometrical collision diameter
and relative velocity of ships i and j. The S is a cell
area(1/4mile×1/4mile).Thetanddarethesampling
time and days respectively. Pedersen’s model has
been used to calculate the geometrical collision
diameter(PedersenP.T.1995).
2.2 Estimatingcollisioncourse
areaswithOZT
The OZT model is introduced by Hayama I. et al.
2002. This method is able to show the estimated
collision areas called Obstacle Zone by Target.
Therefore navigators can sassily understand the
collisioncurseasshowninFigure1.The calculation
model for OZT is used Hayama I.
2014, because of
lesscomputationtimecomparedwithHayamaI.etal.
2002. Firstly, the zone is decided with a circle of
radiusraroundown‐ship.Inthisstudy,theradiusr
iscalculatedbyusingthecollisiondiameter(Pedersen
P.T.1995)dividedbytwo.Thenitdraws
tangentsto
thecirclefromthetargetship.ThecourseCowhichis
DCPA=r is calculated with angleαin the tangential
directionisgivenby:
1
0
sin sin( )
T
T
V
Co Az Az C
V
(2)
TheAz,V
TandCTaretargetshipʹsdirectionfrom
own ship, speed and course respectively. The V
0 is
ownshipʹsspeed.Therearecalculatedupto4TCPAs
fromtherelativespeedV
RandcourseCRasfollows:
cos( 180)
R
R
dCAz
TCPA
V
(3)
TheOZTisfoundbyusingcalculatedTCPAtimes
V
TasshowninFigure1.
Figure1.RelationshipwithOZTandotherparameter
Shipsarechosenbycertaintimerange.Then,each
shipʹs OZT is calculated with all other ships. If the
OZTispassing through the cell (cell number: cn) at
timet,the1isaddedtothecellrateTC.
1
cn cn
tt
TC Cell
TheddaystotalcellrateTC
iaresummedupand
thendividedtotalhours.
1
24
n
i
i
n
TC
ATC
d
(4)
3 RESULTOF1YEARANALYSIS
3.1 1yearresultswithGasmodelandOZTmodelbased
calculation
The Shimizu Port area has been found higher risk
thanotherareainSurugaBay(GenF.2015,GenF.&
Kyoko T. 2015, Gen F. & Ruri S. 2016). Therefore,
around the Shimizu Port area is studied with Gas
model and OZT based analysis by using 1‐year AIS
datafromJuly2014toJune2016.
Cell sizes are decided 1/4 miles square for Gas
modelbasedanalysisand1/100milessquareforOZT
model based analysis. The sampling time is
set 0.5
hoursforGasmodelbasedanalysis.
The OZT is calculated if it is satisfied following
parameters:
DCPA:lessthan0.5miles,
distancetoothership:lessthan3miles,
relative angles to the OZT: less than ±90 degrees
fromownshipcourse,
ownship
speed:morethan8knots,
othershipʹsspeed:morethan0.5knots.
Figure2. One year analysis data with Gas model based
calculation
The1‐yearanalysisdatabasedon theGasmodel
calculation is shown in Figure 2. The Shimizu Port
entranceismarkedashigh‐risk rate.Sinceshipsare
533
navigatednormallysamecourseorpassingthorough
about 180 degrees relative angles, the north part is
foundlessriskratecomparedwithsouthsideofport
entrance.Since thesouthside ishigh‐riskrate,extra
careisnecessaryforthesafetyevacuation.
If there are more OZT found, there
are more
collisionriskisexisted.Thehighestratethatisabout
0.05OZTexistencesper hour is found near the port
entranceNo.1inFigure3.Itshouldbeavoidedtaking
coursetowardthisarea.Intheport,thereisfoundthe
lineofhighrateareathatshows
2.Thislineisalong
to the container ship course from the container
terminal. Because the containership is using similar
course when the evacuation is carried out, this area
might be dangerous for other ships. High OZT area
indicatedby3isthecourse,whichnormallyusedby
outboundvessels.Inbound
vesselsshouldavoidthis
areaforevacuation.Thehighrateareasarealsofound
outsideofthe port.Eventhougharound 4 areas are
morethan200‐meterdepthandclosetotheport,itis
betternottousefortheevacuation.The5isnormally
the course for
out bound vessels. The 6 is normally
thecourseforinboundvessels.Sinceitisdifficultto
see with other analysis method such as using ships
density,thisinformationmightbeveryusefulforthe
navigatorsformakingtheevacuationplan.
Figure3. 1‐year analysis data with OZT model based
calculation
3.2 Comparingwithestimatedfisheryboattrajectory
It is known that the white baits fishing boats and
Sakura shrimp fishing boat are crossing near the
entranceoftheShimizuPort(GenF.2016).Thewhite
baits fishing starts from 21st of March until 15th
Januaryinnextyear.Thestartingtime
isfromsunrise
until about 9 oʹclock in the morning. Sakura shrimp
fishingstartsfromMarchuntil5thJuneandOctober
untilDecember. Theybeginfishingfromthe sunset
untilaround21oʹclock.
Figure4. From March to May based on Gas model based
calculation
Figure 4 and Figure 5 are shown results from
MarchtoMayandOctobertoDecemberbasedonthe
Gas model calculation. The Gas model based
calculation shows very similar risk trend compare
withFigure2.
Figure 6 and Figure7 are the OZT based
calculationresultfromMarchtoMayand
Octoberto
Decemberrespectively.Theredarrowsinthefigures
are estimated fishing boats tracks by hearing
observation (Gen F. 2016). The route, which is
indicated as 1 in the both figures, is used mainly
Sakurashrimpfisheryboats.Therefore,thisestimated
routeisfoundatthesunsettime
goingtothefishery
areaandafter21oʹclockgoingbacktotheport.The
Sakurashrimp routeestimatedroute iscrossing and
along areas where the OZT existence rate is 0.04 to
0.06 in the Figure 6. When planning the evacuation
plan at nighttime in these seasons, it needs
to be
considered these fishery boats. White baits fishery
boatsandrecreationalfishingboatfromtheShimizu
port mainly use the route, which is indicated as 2.
Theseshipsarenormallyfoundinthemorning.The
routeis seemedto benotdangerous comparedwith
routeindicatedas1.However,
theseshipsaremight
be drifted by the current caused by the Tsunami. In
this case, the entrance of the port could be more
dangerous especially around the areas that OZT
existencerateisabout0.16to0.18.
534
Figure5. From October to December based on Gas model
basedcalculation
Figure6.FromMarchtoMaydatawithOZTmodelbased
calculationandestimatedfisheryboattrack
Figure7.FromOctobertoDecemberdatawithOZTmodel
basedcalculationandestimatedfisheryboattrack
4 ANALYSINGSHIPSTHATAREHEADINGTO
OZT
ThereisoneactualcollisionaroundtheShimizuPort
EntranceshowninFigure8.ByanalyzingwithOZT,
theshiptowardstheOZTfrom5:36:12forabout160
seconds until the collision. The S1 ship stopped
engine(ʹʹS1ship(StopEngine)ʹʹas
showninFigure8)
atabout5:36A.M.andproceedhercourseheadingto
quarantine anchorage. But at this time, the ship has
already been heading to OZT as shown in Figure 8.
PilotisonboardonS2ship.TheycontactbyVHFat
5:38A.M.withS1ship
totelltopasstheirstern.Then,
theS2shipstartschanginghercoursetoportat5:41
A.M.TheS2shipcontinuallychanginghercourseto
theportandcollisionoccurred.Thereisnocasefound
headingtoOZTsuchalongtimeheadingtoOZTor
changinghercourse
rapidlyafterheadingOZTatthis
time. However, AIS big data can be analyzed with
thismethodforfindingthenearmisscase.Fordoing
this,itneedsmorecollisionstudiesbyusingOZT.
Figure8.ShowingOZTandshipsʹpositionbyusingactual
collisiondata
Thereare3timesthatisfacingOZTcontinuously
OZTfor6secondsinthepreviouscollisionexample.
Therefore, it has conducted a survey of vessels
continuously facing to OZT in the survey area. By
using AIS data that is converted to every 1second
fromJuly2014 toDecember2014,OZT
iscalculated
andextractedvesselsthatareproceedingtowardsto
OZT.Afterextractionofships,thedatathatvirtually
nocollisionoccurred,suchasownshipcourse,target
shipcourseandOZTintersects breakwaters and the
lands are deleted. Continuous data at 1‐second
intervals are counted and shown with
ship‐ship
distanceintheFigure9.
535
Figure9.ThenumbersofshipsareheadingtoOZT
Table1.HeadingtoOZTsecondsandship‐shipdistance
_______________________________________________
Seconds/Distance(m) 5 6 7 8 9 10
_______________________________________________
d<10006 7 1 1 1 0
d<200022 5 1 1 0 0
d<300036 10 2 2 2 1
d<400040 10 7 2 0 1
d<500031 9 7 2 0 0
d<600019 4 4 1 0 0
_______________________________________________
Total154 45 22 9 3 2
_______________________________________________
Fromthedata,mostofvesselsthatarefacingOZT
in succession are within 1 seconds to 4 seconds.
During this period, the maximum time towards to
OZTis10seconds.Thereare3,9,24,45and154cases
foundbetween5to9secondsasshownin
Table1.
Analysisisthencarriedoutusingthefigure.From
8to10secondsand 7 seconds heading to OZT data
are shown in Figure 10 and Figure 11 respectively.
Thewhitelineandgreenlineshowown‐shipcourse
and target ship course respectively. The red lines
represent OZT.
In the figure, 9sec_1 means that the
ship is heading to OZT for 9 seconds continuously.
Thetotalnumbersof235datathatareheadingtothe
OZT time from 5 to 10 seconds are studied. For
example, 9sec_1 ship is changing the heading to
starboardand entering port after
this. 8sec_5shipis
changing her course to starboard. 8sec_6 ship
proceedingsamecourse,buttargetshipchangingher
course to starboard. Looking at the location of the
OZTofashipnavigatingtowardsOZTfor7seconds
to 10 seconds, it is understood tendency for OZT to
appeararoundthe
Shimizuport.Althoughthereisno
nearmissdatafoundatthistime,thismethodcanbe
usedforfindingthenear‐misscasewithAISbigdata.
Fordoingthis,themorecollisioncasewillneedtobe
studiedbyusingthismethodinthefuture.
Figure10.HeadingtoOZTfor8to10secwithshipsʹcourse
andOZTPosition
Figure11.HeadingtoOZTfor7secwithshipsʹcourseand
OZTPosition
5 CONCLUSION
Thetrafficconditionhasbeenanalyzedbyusingthe
GasmodelbasedandOZTmodelbasedmethodsby
usinga1‐yearAISdataforthetsunamimeasure.
The south side of the Shimizu port entrance has
found highest risk place with the Gas model. When
ships are evacuating,
navigators should have extra
cautionaround thisarea.Thedangerouscourses are
known with OZT based analysis. The several high
OZT existing places are found in the port and also
outsideoftheport.Navigatorsshouldbeconsidered
seasonal and time zone of fishery boat movement,
especially near the port
entrance. Ships that courses
are heading to OZT are also identified and studied
their position and how long she proceeds course
heading to OZT. This method can be applied to the
AIS big data for analyzing the near miss ship
collisions.Forthispurpose,itneedstoanalyzeactual
collisions with
the same method and clarify OZT
informationsuchasperiodofheadingtoOZT,ship‐
shipdistance,speedandencounteringsituations.
ACKNOWLEDGMENT
Thisworkwassupported byJSPSKAKENHIGrant‐
in‐AidforYoungScientists(B)Number15K16308.
536
REFERENCES
GenF.&Seung‐GiG.&A‐RaC.&Hye‐RiP.2013,Traffic
RiskAnalysisApplyingtheGasmodelmethodinBusan
Port,AsiaNavigationConference2013Proceedings,Korean
InstituteofNavigationandPortResearch,pp.349‐356
Gen F. 2015, Marine Traffic Condtion Analysis for
Developing Tsunami
Countermeasure in Suruga Bay,
Asia Navigation Conference 2015 Proceedings, Japan
InstituteofNavigation,pp.314‐318
Gen F. & Kyoko T. 2015, Analyzing the marine traffic
condition for estimating the high risk areas in the
emergencyevacuation,2015InternationalAssociationof
InstitutesofNavigationWorldCongressPrague,Czech
Republic
Gen F.
& Ruri S. 2016, Estimating Collision Course Area
Using OZT in Offshore Refugee Outside the Port, The
Journal of Japan Institute of Navigation,Japan Institute of
Navigation,Vol.135,pp.129‐134
Gen F. 2016, The Study of the Non‐SOLAS Vessel
Movementin SurugaBay, AsiaNavigation Conference
2016Proceedings
Hidenari
M.&NovukazuW.&YoshijiY.&ShigekiS.2011,
Research on trend of ship evacuation behavior when
Tsunami warnig, The Journal of Japan Institute of
Navigation, Japan Institute of Navigation, Vol.125,
pp.191‐197
Hayama I., Junji F. & Masashi N. 2002, Obstacle Zone by
TargetanditsExpression,
TheJournalofJapanInstituteof
Navigation,JapanInstituteofNavigation,pp.191‐197
Hayama I. 2014, Computation of OZT by using Collision
Course, NAVIGATION, The Japan Institute of
Navigation,Vol.188,pp.78‐81
Kamaishi Japan Coast Guard 2004, Preparation for
earthquake and tsunami,
http://www.kaiho.mlit.go.jp/02kanku/kamais hi /tsun
ami/tsunami.html
Pedersen,
P. T. 1995, Collision and Grounding Mechanic,
Proceedings of WEMT’95, The Danish Society of Naval
ArchitectsandMarineEngineers,pp.125‐15
