Environmental risk assessment of shipwrecks Model...

i

THESISFORTHEDEGREEOFDOCTOROFPHILOSOPHY

Environmentalriskassessmentofshipwrecks–Modeldevelopmentandapplication

HANNALANDQUIST

TheDepartmentofShippingandMarineTechnologyCHALMERSUNIVERSITYOFTECHNOLOGY

Gothenburg,Sweden2016

ii

Environmentalriskassessmentofshipwrecks–ModeldevelopmentandapplicationHANNALANDQUISTISBN978‐91‐7597‐490‐3©HANNALANDQUIST,2016DoktorsavhandlingarvidChalmerstekniskahögskolaNyserienr4171ISSN0346‐718X

TheDepartmentofShippingandMarineTechnologyCHALMERSUNIVERSITYOFTECHNOLOGYSE‐41296GothenburgSwedenPhone+46(0)31‐7721000www.chalmers.seCover:TheshipwreckofSkytteren.CourtesyoftheSwedishMaritimeAdministration.PrintedbyChalmersReproserviceGothenburg,Sweden2016

iii

Environmentalriskassessmentofshipwrecks–ModeldevelopmentandapplicationHANNALANDQUISTDepartmentofShippingandMarineTechnologyChalmersUniversityofTechnologyABSTRACTPotentially polluting shipwrecks containing oil or other hazardous substances pose athreattothemarineenvironment.ThisisaglobalproblemandmanyshipwrecksstemfromtheSecondWorldWarhavingbeendeterioratingontheseafloorsincethen. OnlyinSwedishwatersmorethan300wrecksareestimatedtoposeanenvironmentalthreat.Together, thesewrecks are estimated to contain between1,000 and15,000 tonnes ofbunkeroil.Everyshipwreckposesauniquecasedependingon,forexample,thetypeofvessel,causeofsinking,andenvironmentalpreconditions.Thisimpliesthattheproblemiscomplexandalsothattherearemanyuncertaintiesinvolved.

It isnot feasible toremediateall shipwrecksdue to largecosts,butaproactiveapproachwould reduce the need for and high costs of reactive response in case of adischarge.Untilnow,therewerenocomprehensiveprobabilisticmethodforassessingtheenvironmental riskposedby shipwrecks inorder toprovidenecessary support todecision‐makers.

In order to prioritise and effectively use resources, proper decision support isneeded. Risk assessments and the overall process of riskmanagement are importantmeanstoprovidesuchdecisionsupport.Thepurposeofthisthesishasthereforebeentodevelop,applyandevaluateamodel forcomprehensiveriskassessmentofpotentiallypollutingshipwrecks.Acomparisonofcurrentmethodsforriskassessmentofshipwreckswas performed in order to identify development needs. Based on the comparison, ageneric framework forriskmanagementof shipwreckswassuggested.Furthermore,amethod for estimating the probability of discharge of hazardous substances wasdeveloped, using a probabilistic fault tree method. The method makes it possible toconsider possible activities that may damage the wreck as well as physical andenvironmentalconditionsaffectingthewreck.Anapproachforconsequenceassessmentof discharges from shipwrecks, consisting of an aggregation of methods was alsodevelopedwithinthisthesiswork.

The generic framework for risk management of shipwrecks clearly shows theimportantstepsrequiredandhowtheyarelinked.Italsoemphasizestheneedofproperassessmentstofacilitateprioritisationofshipwrecksandanefficientresourceallocationfortheseenvironmentalthreats.Theresultisaprobabilisticandcomprehensivemodelfor risk assessment of shipwrecks including possibilities to cope with the vastuncertaintiesinvolvedinshipwreckriskassessment.

Keywords: environmental risk assessment, shipwreck, fault tree analysis, decisionsupport,Bayesianupdating

v

Listofpapers

Thethesisisbasedontheworkpresentedinthefollowingpapers,referredtointhetextbyRomannumerals:

I. Landquist,H.,Hassellöv,I.‐M.,Rosén,L.,Lindgren,J.F.andDahllöf,I.(2013).Evaluatingtheneedsofriskassessmentmethodsofpotentiallypollutingshipwrecks.JournalofEnvironmentalManagement,119,85‐92.

II. Landquist,H.,Rosén,L.,Norberg,T.,Lindhe,A.,Hassellöv,I.‐M.andLindgren,J.F.(2014).Afaulttreemodeltoassessprobabilityofcontaminantdischargefromshipwrecks.MarinePollutionBulletin,88,239‐248.

III. Landquist,H.,Rosén,L.,Lindhe,A.,Norrman,J.,Norberg,T.,Hassellöv,I.M.&Lindgren,J.F.(2016).Expertelicitationforderivinginputdataforprobabilisticriskassessmentofshipwrecks.ManuscriptforsubmissiontoJournalofEnvironmentalManagement.

IV. Landquist,H.,Rosén,L.,Lindhe,A.,Norberg,T.,&Hassellöv,I.M.(2016).Bayesianupdatinginafaulttreemodelforshipwreckriskassessment.ManuscriptforsubmissiontoScienceoftheTotalEnvironment.

V. Landquist,H.,Rosén,L.,Lindhe,A.,Hassellöv,I.M.(2016).VRAKA–Aprobabilisticriskassessmentmethodforpotentiallypollutingshipwrecks.FrontiersinEnvironmentalScience,4.

Divisionofworkbetweentheauthors

InPaperI,alltheauthorsparticipatedinstatingtheaimandscopeandtookactivepartinperformingthework.Rosénoutlinedtheframeworkforriskmanagementofshipwrecks.Landquistperformedthecomparisonandevaluationofmethodsforriskassessmentofshipwrecksandwasthemainauthorofthepaper.AlltheauthorscontributedwhentheaimandscopeofPaperIIandIVweredefined.Landquist,Rosén,Lindhe,Norberg,andHassellövdevelopedthefaulttreemodel.Norbergwasthemaindeveloperofthemathematicalfoundationwhichwasalsobasedondiscussionsinthegroup.LindheandRoséncontributedlargelytothetheoreticaldescriptionofthefaulttreemodel.Landquistperformedthesimulationsandwasthemainauthor.

vi

ThescopeandaimofPaperIIIwasdevelopedbyallauthors.Landquist,Rosén,Norberg,LindheandHassellövdevelopedtheelicitationset‐up.Allauthorshavebeeninvolvedinwriting.AllauthorscontributedtoaimandscopeofPaperV.Landquistwasthemaindeveloperoftheconsequenceassessmentapproachandallauthorswereinvolvedinwriting.Otherworkandpublicationsnotappended

Larssonetal.,(2011).Environmentalrisksposedbyshipwrecks.InSwedish,MiljöriskerfrånFartygsvrak,pp.48‐53.TheSwedishMaritimeAdministration.

Landquist,H.(2013).Methoddevelopmentforenvironmentalriskassessmentofshipwrecks.LicentiateThesisNoR13:144,ChalmersUniversityofTechnology.

Landquist,H.(2013).EnvironmentalRiskAssessmentofShipwrecks:AFault‐treeModelforAssessingtheProbabilityofContaminantRelease.IntegratedEnvironmentalAssessmentandManagement(IEAM)bytheSocietyofEnvironmentalToxicologyandChemistry(SETAC).

Hassellöv.,I.M.,Olsson,U.,Ekberg,G.,Östin,A.,Simonsson,F.,Larsson,C.,Sender,U.,Landquist,H.,Lindgren.,J.F.,Lindhe,A.,Rosén,L.&Tengberg,A.(2014).Environmentalrisksposedbyshipwrecks.InSwedish,Miljöriskersjunknavrak.Undersökningsmetoderochmiljöaspekter.Dnr:1399‐14‐01942‐6.Sjöfartsverket.

Hassellöv.,I.M.,Olsson,U.,Ekberg,G.,Östin,A.,Simonsson,F.,Larsson,C.,Landquist,H.,Lindgren.,J.F.,Lindhe,A.,Rosén,L.&Tengberg,A.(2015).EnvironmentalrisksposedbyshipwrecksII.InSwedish,MiljöriskersjunknavrakII.Undersökningsmetoderochmiljöaspekter.Dnr:1399‐14‐01942‐15.Sjöfartsverket.

Landquist,H.,Lindhe,A.,Rosén,L.,&Lindgren,J.F.(2015).SWERADeliverable2.2.EU.BONUS.

Lindgren,F.J,Hassellöv,I.M.,Landquist,H.&Dahllöf,I.(2015).LowconcentrationsofPAHsinducetoleranceinnitrifyingbacteria.FrontiersinMarineScience,2.

Andersson,K.,Brynolf,S.,Landquist,H.&Svensson,E.(2016).MethodsandToolsforEnvironmentalAssessment.In:Andersson,K.,Brynolf,S.,Lindgren,F.J.&Wilewska‐Bien(eds.).ShippingandtheEnvironment:ImprovingEnvironmentalPerformanceinMarineTransportation.Berlin,Heidelberg:SpringerBerlinHeidelberg.

vii

Lindgren,J.F.,Andersson,K.,&Landquist,H.(2016).MarineSpatialPlanningandShipwrecks.In:Andersson,K.,Brynolf,S.,Lindgren,F.J.&Wilewska‐Bien(eds.).ShippingandtheEnvironment:ImprovingEnvironmentalPerformanceinMarineTransportation.Berlin,Heidelberg:SpringerBerlinHeidelberg.

Etkin,D.,FrenchMcCay,D.,Horn,M.,Landquist,H.,Hassellöv,I.M.&Wolford,A.J.

(Inprep).QuantificationofOilSpillRisk.In:FINGAS,M.(ed.)OilSpillScienceandTechnology.Burlington,USA:Elsevier.

Landquist,H.,Rosén,L.,Lindhe,A.&Hassellöv,I.M.Estimatingconsequencesofdischargefrompotentiallypollutingshipwrecks.(2015).Abstract,presentation.Inproceedingsof,Science,Policy,Society–bridgingthegapbetweenriskandscience.SRAEurope,June15‐17.

Landquist,H.,Rosén,L.,Hassellöv,I‐M.,Lindgren,J.F.,Norberg,T.&Lindhe,A.

(2012).Environmentalriskassessmentofshipwrecks:afault‐treemodelforassessingtheprobabilityofcontaminantrelease.Poster,theSETACNorthAmerica33rdAnnualMeeting,LongBeach,November11‐15.

Johansson,A.M.,Hassellöv,I.M.,Eriksson,L.B.,Lindgren,F.J.,Berg,A.,Carvajal,G.,Landquist,H.(2014).RemotesensingforriskanalysisofoilspillsintheArcticOcean.PosteratEGU,Wienna,Austria.

Johansson,A.M.,Hassellöv,I.M.,Eriksson,L.B.,Lindgren,F.J.,Berg,A.,Carvajal,G.,Landquist,H.(2014).RiskanalysisofoilspillintheArcticOcean.PosteratArcticFrontiers,Tromsø,Norway.

Landquist,H.,Rosén,L.,Lindhe,A.,Norberg,T.,Hassellöv,I.M.,Lindgren,J.F&

Dahllöf,I.(2014).Afaulttreemodeltoassessprobabilityofcontaminantdischargefromshipwrecks.PosteratGeoArena–Mötesplatsgeologi,Uppsala,Sweden.

ix

AcknowledgmentsFirstofall,thisprojectwouldnothavebeenpossiblewithoutfundsfromTheSwedishResearchCouncilFORMAS,SWERA,DAIMON,ChalmersInnovationOffice,LighthouseandtheSwedishMaritimeAdministration.Thankyou!Iwouldliketothankmysupervisors.Ida‐MajaHassellöv,yourenthusiasmandabilitytoencourageisadmirable.AndreasLinde,foralwaysbeingpreparedtodiscussthetiniestofdetailsinthelatestpaperorsolveamodellingproblem,youbroughtmuchappreciatedflowtomywork.LarsRosén,forinspiringlysharingyourknowledgeformorethantenyears.BeingaPhDstudentinvolvesagreatdealofworkbutitalsorepresentsfiveyearsofhavingtheutterluxuryofanindividuallydesignededucation.Iamverygratefulforhavinghadtheopportunitytobeyourcolleagueandlearntheskillsofaresearcherfromyouall.IwouldalsoliketotakethisopportunitytosaythankyoutoallthepeopleinvolvedinSWERA,Miljöriskersjunknavrakandreferencegroupsforfruitfulcollaboration.Furthermore,thisthesiswouldnothavebeenpossiblewithoutalltheexpertswhocontributedwiththeirknowledgeatlastyears’workshop.ThankyouTommyNorbergforalwaysdiscussingstatisticalpuzzleswithgreatcalmnessandcuriosity.IwouldalsoliketoextendmygratitudetoJennyNorrmanforyourfruitfulcollaborationinexpertelicitationandwriting.MythanksalsotoGöranEkbergandBertWestenbergforcontributingtotheprojectwithyourknowledgeandever‐cheerfulmood.SpecialthankstoDagmarEtkinandDeborahFrench‐McCayforinspiringcollaboration.IwouldalsoliketothankPerWedel.Alwaysopentootherinterpretationsandwithapassionforgeology,youareapedagogicalrolemodel.ThankyouFredrikLindgrenformostenjoyablecooperationand,togetherwithIngelaDahllöf,forallowingmetoassistinyourlaboratorywork.ThankyoutoallmycolleaguesattheDepartmentofShippingandMarineTechnologyandtheDivisionofGeologyandGeotechnicsforthemanyenjoyablediscussionsaroundthecoffeetable.Thankyoualsoforallowingmetouseyourspareofficechairsfordiscussions,crayfishpartiesandfriendship.SpecialthankstothegroupformerlyknownastheEnvironmentGroupforyourscientificdiscussions,wallpainting,windowgardeningandlotsoffun.Thankyoufamilyandfriendsforcheeringmeonandmakingmelaugh!Andofcourse,thankyouEskilforallyourdevotionandsupport.Iloveyou!

HannaLandquist Gothenburg,2016

x

Thispageisintentionallyleftalmostblank(Wrightetal.,2014).

xi

TableofContents

1. Introduction.............................................................................................................................................................1 1.1. Aimandobjectives.........................................................................................................................................2

1.2. Scopeofwork...................................................................................................................................................3

1.3. Limitations.........................................................................................................................................................5

2. Conceptsofrisk..........................................................................................................................................7 2.1. Riskandshipwrecks...................................................................................................................................10

2.1.1. Hazardousactivities..........................................................................................................................10

2.1.2. Site‐specificandwreck‐specificindicators.............................................................................11

2.2. Theriskmanagementprocess...............................................................................................................11

2.3. Riskassessment...........................................................................................................................................13

2.4. Decision‐making...........................................................................................................................................15

2.5. Uncertaintyandprobability....................................................................................................................16

3. Riskassessmentapproachesforshipwrecks....................................................................................19 3.1. Identifiedmethodsandapproachesforenvironmentalriskassessment

ofshipwrecks................................................................................................................................................19

3.2. Comparisonwithaninternationalstandardforriskmanagement........................................21

4. Ariskmanagementframeworkforpotentiallypollutingshipwrecks...............................25

5. Methods...................................................................................................................................................................27 5.1. Faulttreeanalysis........................................................................................................................................27

5.1. MonteCarlosimulation.............................................................................................................................28

5.2. Expertelicitation..........................................................................................................................................29

5.2.1. Experts,heuristicsandbiases.......................................................................................................29

5.2.2. Stakeholders.........................................................................................................................................30

5.2.3. Elicitationmethods...........................................................................................................................31

5.2.4. SHELF......................................................................................................................................................32

5.3. OilspilltrajectorymodellingandSeatrackWeb............................................................................33

5.4. OilspillimpactandtheDigitalEnvironmentalAtlas....................................................................33

xii

6. VRAKA–acomprehensivemodelforriskassessmentofshipwrecks..........................................35 6.1. Part1–Estimationoftheprobabilityofdischarge.......................................................................36

6.1.1. Afirstversionofthefaulttreemodel–PaperII...................................................................37

6.1.2. Thenovelfaulttree–PaperIV.....................................................................................................39

6.1.3. Expertelicitation–PaperIII..........................................................................................................41

6.2. Part2–Consequenceassessment–PaperV....................................................................................43

6.3. Part3–RiskEvaluationinVRAKA.......................................................................................................44

6.4. AtoolforapplyingVRAKA.......................................................................................................................44

7. Discussion...............................................................................................................................................................47 7.1. Agenericramework...................................................................................................................................47

7.2. VRAKA–acomprehensivemodel.........................................................................................................47

7.2.1. Afaulttreemethod............................................................................................................................47

7.2.2. Consequenceassessment................................................................................................................48

7.2.3. Riskestimation...................................................................................................................................48

7.2.4. Riskevaluation....................................................................................................................................49

7.3. Expertelicitation..........................................................................................................................................50

7.4. ValidationofVRAKA...................................................................................................................................50

7.5. Outlook.............................................................................................................................................................50

7.6. Suggestionsforfurtherresearch...........................................................................................................51

8. Conclusions............................................................................................................................................................53

9. References..............................................................................................................................................................55

PapersI‐V

AppendixA1–VRAKA,Probabilisticriskassessmentofshipwrecks,Handbook.VersionOct.2016.

1

1. Introduction

Theabilitytoidentifyadverseeventsthatmightpossiblyoccurinthefuture,andtomakegood and relevant choices from among alternative actions that could be employed tomanagesuchevents,isakeyfactorinmodernsociety.Understandingriskthusallowsustomakerationaldecisions(Bernstein,1996).Limitedsocietalresourcescombinedwithnumerouspotentialareasofneedfortheseresourcesimplythatprioritisationdecisionsarerequired.Oneoftheseresource‐demandingareasisenvironmentalprotectionandaspecificareathathasattractedgrowingattentionistheproblemofpotentiallypollutingshipwrecks.Polluting shipwrecks are a global and multi‐faceted problem. Estimates indicate thatseveralthousandwreckslittertheoceanfloors(Micheletal.,2005).InSwedishwatersalone,2,700wrecks(≥100grosstonnage)warrantfurtherinvestigationandmorethan300ofthesewereassumedtoposeanenvironmentalthreatastheymaycontain1,000‐15,000 tonnes of bunker oil (Larsson et al., 2011). Many shipwrecks have beendeterioratingontheseabedforseveralyearsandmanyoriginatefromtheSecondWorldWar containing unknown amounts of oil and other hazardous substances. Oil can bepresentbothasbunkerandcargo,makingitthemostcommonhazardoussubstanceinwrecks. Assessing risks posed by shipwrecks requires the adoption of a holisticperspective.Theentirechainofevents,fromwreckconditionandcargotoenvironmentalconditionsatthewrecksiteandexternalfactors,mustbetakenintoaccounttoobtainarelevantgraspoftheproblem.Thethreatfromshipwrecksrequiresattention,andthereis a need for guidance, decision‐making support and methods to assess and handleefficiently the environmental risks and facilitate resource allocation for mitigationmeasures(PaperI;Micheletal.,2005).Theenvironmentalriskposedbyeachwreckisuniquewithregardtotheprobabilityofleakageanditspotentialimpact.Eachvesselwillresponddifferentlytoexternalforcesandenvironmentalconditionsarespecifictoeachwrecklocation(Etkinetal.,2009).Theeffectsofanoildischargefromashipwreckwilldependonwhentheoilisreleased,wherethe releaseoccurs, thespreadingrateof theoil slickand the typeofoil (Micheletal.,2005). Oil discharged into amarine environment is harmful to living organisms (e.g.Kingston,2002;Lindgren,2015).Oilspillscanalsocauseeconomicdamagetopropertyandmarket goods such as tourismand fisheries aswell asnon‐market goods such asrecreation(Fejesetal.,2011).Thewidevarietyoffactorsinfluencingapotentialdischargeimpliesthattherearealsoanumberofuncertaintiesassociatedwithriskassessmentsof

H.Landquist

2

shipwrecks,andhencealsowithpossiblemeasuresaimedatmitigatingtherisksposedbyshipwrecks.Etkin(2009)arguesthatareactiveapproachtooilleakagefromashipwreckwouldleadtohigherresponseanddamagecostscomparedtoawell‐planned,proactiveoilremovaloperation.Nevertheless, it is very costly to remediate shipwrecks. It is estimated thatremediation of onewreck could costUSD1‐100million (Michel et al., 2005) and it isthereforenot economically feasible to remediate every sunken shipwreck.The300ormorewrecksonlyinSwedenmentionedpreviouslywouldcosthundredsofmillionsofdollarstoremediate.Todealwiththeproblemofpotentiallypollutingshipwrecks,thereis a vital need for decision supportmethods to facilitate prioritisation ofwrecks andidentificationofhowavailableresourcescanbeusedmostefficiently.Shipwreck risk assessment requires knowledge from many fields and collaborationbetween several bodies to make relevant decisions regarding potential mitigationmeasures. A key aspect of utilising available resources efficiently is to facilitatecooperation between these bodies. The responsible authorities, such as the SwedishAgencyforMarineandWaterManagement,theMaritimeAdministration,theCoastGuard,shipwreckremediationcompaniesandresearchersinSwedenandabroad,arejustsomeexamplesofpotentialcollaborativeorganisations.Animportantmeansofprovidingdecisionsupportistoapplyriskassessmentpracticesthat are widely employed in several disciplines, including engineering, ecotoxicology,public health and economics (Burgman, 2005). Risk assessment is part of riskmanagementand includesRisk identification,Riskanalysis andRiskevaluation. Severalapproaches to risk assessment specific to shipwrecks can be found in the literature(Alcaroetal.,2007;Etkinetal.,2009;Idaas,2005;Louzisetal.,2009;Micheletal.,2005;NOAA,2009;NOAA,2013;SPREPandSOPAC,2002;Ventikosetal.,2013;Ventikosetal.,2016). However, studies presented in this thesis show that current methods fail toprovideaholisticand/orfullyprobabilisticriskassessmentofshipwrecks.Consequently,itcouldresult in inefficientuseofsocietalresourcesand inadequateriskmitigationofpotentiallypollutingshipwrecksduetoalackofwell‐foundeddecisioninput.

1.1. Aimandobjectives

Theoverallaimofthisthesiswasto:

Develop,applyandevaluateamodelforcomprehensiveriskassessmentofpotentiallypollutingshipwrecks.

Tomeettheoverallaim,thefollowingspecificobjectivesweredefined:

1. Introduction

3

Compareandanalyseidentifiedcurrentriskassessmentmethodsforpotentiallypolluting shipwrecks with regard to how these methods comply with aninternationalstandardforriskmanagement.

Suggestagenericframeworkforriskmanagementandassessmentofshipwrecks,includingriskidentification,riskanalysisandriskevaluation.

Develop a method to quantitatively estimate the probability of discharge ofhazardoussubstancesfromshipwrecks.

Developanapproachforconsequenceassessmentofthedischargeofhazardoussubstancesfromshipwrecksbycombiningexistingmethodsforoilspillmodellingandreceptorsensitivity.

Presentacomprehensivemodelforconductingariskassessmentofshipwrecks,including a method for estimating the probability of discharge of hazardoussubstances and an approach for making a consequence assessment of such adischarge. The model should also provide a means for risk evaluation bycomparisonwithacceptablerisklevelsormakingananalysisoftherelationshipbetweenriskmitigationandcost.

A structured framework for managing risks associated with shipwrecks providesnecessaryguidanceontheimportantstepsneededtohandlethisenvironmentalproblem.Understandingtheriskarisingfrompotentiallypollutingshipwrecksalsoentailshandlinguncertainties.Tocopewithuncertainty,aquantitativeprobabilisticapproachisappliedby using a logic tree model for estimating the probability of discharge of hazardoussubstances.Anexpertpanelwasalsoinvolvedtogatherinformationonuncertainaspectsof the probability of discharge. Release of hazardous substances into the marineenvironment will give rise to environmental consequences and implications insocioeconomic terms.Theseeffectsmustbeaddressedandaconsequenceassessmentapproachisthereforedeveloped.Acomprehensivemodelformakingariskassessmentofpotentiallypollutingshipwrecksisbasedontheprobabilityestimationmethodandtheapproachtoconsequenceassessment.

1.2. Scopeofwork

Theappendedpapersareinterlinked,asshowninFigure1.PaperIpresentsaliteraturereviewofexistingmethodsforriskassessmentofshipwrecksandsuggestsagenericriskassessmentframeworkforwrecks.Thepaperwaspublishedin2013andsincethenthreenewapproacheshaveemergedandarediscussedinChapter3.However,themainfindingofthepaperstillholds:existingapproacheslackkeyelementsofaholisticriskassessmentforshipwrecks.PapersIIandIVdescribeafaulttreemethodforprobabilisticestimationofthedischargeof hazardous substances fromwrecks. This quantitativemethod facilitates an explicituncertaintyanalysis.Agenericestimationoftheprobabilityofanopeninginthewreckisupdatedbysite‐specificandwreck‐specificindicators,e.g.timesincesinking,salinity,and

H.Landquist

4

water temperature. A total probability of discharge is estimated by combining theprobability of opening with the rate of hazardous activities and the probability thathazardoussubstancesarestillcontainedinthewreck.Thefaulttreemethodprovidesthefoundation for the VRAKA tool, an Excel‐based tool for estimating the probability ofdischarge from shipwrecks. Paper III presents the structure and outcomes of acomprehensive expert elicitation process that provided further input to the fault treemethod.Paper IVmakesuseof theseresultsandnovelmethodology for incorporatingsite‐specificandwreck‐specificindicatorsintheriskmodelispresented.InPaperV,acomprehensivemodelformakingariskassessmentofpotentiallypollutingshipwrecksispresented.ThemodeliscalledVRAKA(theSwedishwordforwrecking)andcomprisestheVRAKAmethodforprobabilityestimationofdischargeandathree‐tieredapproachforaconsequenceassessmentofsuchadischarge.VRAKAisdevelopedandoperationalisedinthespreadsheetprogramMicrosoftExcel©

withtheadd‐insoftwareCrystalBall©forMonteCarlosimulations.TheVRAKAHandbookisappendedtothethesis(A1).

Figure1.Schematicsummaryofthescopeofthethesisinrelationtotheappendedpapers.

1. Introduction

5

1.3. Limitations

Thepotentialriskposedbyshipwrecksisacomplexprobleminvolvingriskassessmentandmanagement,decision‐makingandvalue‐basedjudgement.Italsoinvolvesanumberofsubstancesinwrecksonoceanfloorsallovertheworld.Encompassingthefullcomplexityofshipwrecksisavastendeavourandthefollowinglimitationsofthisworkneedtobehighlighted:

Thesuggestedframeworkforriskmanagementandassessmentofshipwrecksdescribes the entire risk management process. However, the comparisonbetweentheidentifiedmethodsfocusesonriskassessment,asdoestheVRAKAmethod.

Theframeworkandmodelpresentedinthisthesisareintendedtofacilitateaproactiveapproachandspecificallytheriskassessmentofpotentiallypollutingshipwrecksandtheprovisionofdecisionsupporttodealwiththeserisks.Theassessment results aim to support prioritisation of wrecks and decisionsregardingmeasuresforpreventingleakageofoil,althoughtheactualprocessof selecting and performing preventive options or monitoring appliedmeasuresisnotaddressedinthisthesis.

Thereareshipwrecksallovertheworldcontainingawiderangeofdifferent

hazardous substances. This thesis is focused on shipwrecks located inScandinavian waters that contain oil. A wider geographical scope andexpansiontoothersubstancesarepossiblefuturedevelopments.ThemethodforestimatingtheprobabilityofdischargeisnotrestrictedtoaspecificareabutitshouldbepointedoutthattheexpertelicitationprocessthatprovidesinputismadeusingexpertsfromSwedenandisbasedontheSwedishwreckpopulation.TheconsequenceassessmentcanforthelessdetailedtiersoftheVRAKAmethodbeperformedinotherpartsof theworldalthoughthemostelaborate tier isonlyapplicable inSweden.Thesemethodsprovideasoundbasisforfurtherdevelopmentforotherlocations.

H.Landquist

6

7

2. Conceptsofrisk

TheconceptsofriskasweseethemtodayoriginatefromtheHindu‐ArabicnumberingsystemanddeeperstudiesthattookplaceduringtheRenaissance.In1645,BlaisePascalandPierredeFermatdiscoveredthetheoryofprobability,whichisthecoreoftheconceptofrisk,whensolvingagamblingproblem(Bernstein,1996).Etymologically,riskstemsfromtheearlyItalianwordrisicare,meaningtodare.Whenviewingriskfromthatlatterperspectivethereisachoiceinvolved,asopposedtodependenceonfaithwhichwastheprevailingviewoflifeinthepast(Bernstein,1996).Todaymanydefinitionsexist(e.g.Avenetal.,2015)althoughageneralviewofriskisthatit is the chance of an adverse event occurring within a specific time‐frame and withspecificconsequences(Burgman,2005).Anotherdefinitionisthatriskisacombinationoftheseverityandprobabilityofeffectsfromacertainaction,whereseverityisthenatureandmagnitudeoftheeffects(Suter,2007).Acommondefinitionthatcan,however,bedeceptivetosomeextent is thatofriskexpressedasanexpectedconsequence, i.e. theprobabilityofaneventmultipliedbyitsconsequence.Withthisdefinitionaneventwithlowprobabilityandhighconsequencewouldnotbeseparablefromaneventwithhighprobabilityandlowconsequence(KaplanandGarrick,1981).Aven (2010) argues that uncertainty should be included in the definition of risk andsuggeststhatprobabilityisonlyatooltoexpressorrepresentuncertaintyandthatriskisnotlimitedtoaninitiatingevent,itsconsequencesandtheassociatedprobabilities. Aven(2010) further discusses that probabilities assigned are based on backgroundinformation and assumptions that could hide uncertainties and prevent them fromreceivingproperattention.AvenandRenn(2009)defineriskasthe“uncertaintyaboutandseverityoftheconsequences(oroutcomes)ofanactivitywithrespecttosomethingthathumansvalue”.ISO(2009)alsoincludesuncertaintyanddefinesriskasthe“effectofuncertaintyonobjectives”.KaplanandGarrick(1981)prefertostatethataquantitativeriskanalysisshouldspecificallyanswerthefollowingthreequestions:

‐ Whatcanhappen?‐ Howlikelyisit?‐ Whataretheconsequences?

Theriskapproachchosenshouldbeappropriatefortheapplicationathand.InthisthesistheapproachbyKaplanandGarrick(1981)isapplied,takinguncertaintiesintoaccountwhenanswering the last twoquestionsabove.This isdescribedby IEC(1995)as “the

H.Landquist

8

combination of the frequency, or probability, of occurrence and the consequence of aspecified hazardous event”. However, it should be noted that this thesis examinescombinationsofseveralhazardousevents.FurtherdiscussionregardingthedefinitionsofriskappliedinthisthesisispresentedinChapter7.Thereisanambiguitysurroundingrisk,andtheterminologydiffersgreatly(Aven,2012).SomeofthecommonlyuseddefinitionsareshowninTable1anddefinethemeaningofthedifferenttermsasusedinthisthesis.Anumberofexamplesforashipwreckcontextarealsogiven.

2. Conceptsofrisk

9

Table1.Definitionsoftermsbasedon(Aven,2012), ISO31000(2009)and IEC(1995)withsomeexamplescoupledtoshipwreckriskassessment.

Aleatoryuncertainty Thevariationofquantitiesinapopulation.

E.g.Oxygenlevelsofthewateraroundashipwreck.

Consequence Theoutcomeofaneventaffectingobjectives.

E.g.Effectsoildischargeinthemarineenvironment.

Epistemicuncertainty Alackofknowledgeaboutanunknownquantity.

E.g.Amountofoilcontainedinashipwreck.

Event Theoccurrenceofaparticularsetofcircumstances.

E.g.Openinginashipwreckthoughwhichoilcanbedischarged.

Hazardousevent Aneventthatcancauseharm.

E.g.Constructionwork,TrawlingorShippingtrafficpossiblycausingdamagetoashipwreck.

Hazardoussubstance Asubstancewiththepotentialtocauseharm.

E.g.Bunkeroil.

Likelihood Thechanceofsomethinghappening.

Risk Acombinationofthefrequency,orprobability,ofoccurrenceandtheconsequenceofaspecifiedhazardousevent.

Riskanalysis Aprocessemployedtocomprehendthenatureofriskandtodeterminethelevelofrisk.

Riskassessment Theoverallprocessofriskidentification,riskanalysisandriskevaluation.

Riskcriteria Thetermsofreferenceagainstwhichthesignificanceoftheriskisassessed.

Riskevaluation Theprocessofcomparingtheresultsofriskanalysiswithriskcriteriatodeterminewhethertheriskand/oritsmagnitudeisacceptableortolerable.

Riskidentification Theprocessoffinding,recognisinganddescribingrisks.

Riskmanagement Thecoordinationofactivitiestodirectandcontrolanorganisationwithregardtorisk.Includesriskassessmentandriskreductionandcontrol.

Riskmanagementframework

Asetofcomponentsthatprovidethefoundationsandorganisationalarrangementsfordesigning,implementing,monitoring,reviewingandcontinuallyimprovingriskmanagementthroughouttheorganisation.

Stakeholder Apersonororganisationthatcanbeaffectedby,orperceivethemselvestobeaffectedby,adecisionoractivity.

E.g.theSwedishMaritimeAdministration,theSwedishAgencyforMarineandWaterManagement,theSwedishCoastGuard,theSwedishMeteorologicalandHydrologicalInstitute.TheSwedishGeologicalSurvey,MaritimeNationalMuseumsandtheSwedishDefenceResearchAgency.

H.Landquist

10

2.1. Riskandshipwrecks

The largenumberofwrecks, theassociatedpotentialremediationcosts,andthemanyfactorsaffectingtheconsequencesofadischargefromwrecksjustifyariskassessmentapproach (Michel et al., 2005; Etkin et al., 2009). There are alsomajor uncertaintiesassociatedwithassessingtheriskofhazardousdischargesfromwrecks,mainlybecausethere are fewmeasurements of howdifferent activities, such as constructionwork ortrawling,affectwrecks(Micheletal.,2005).Thelackofmeasurementsandavailabledataimplies that assessments including expert judgements are necessary and that theirestimationsarealsouncertain.Moreover,site‐specificconditionsmakegeneralisationsconcerningtheeffectsofhazardousdischargedifficult.Thismeansthatthehandlingofuncertaintiesiscrucialinordertoobtainusefulresultsfromariskassessmentofwrecks.

2.1.1. Hazardousactivities

As mentioned in the introductory chapter, there are a number of factors that causeshipwrecks to deteriorate and ultimately this can lead to a discharge of hazardoussubstancessuchasoil.Anobviousprerequisiteisthatthewreckstillcontainshazardouscargo or bunker fuel for propulsion. A number of hazardous activities that carry thepossibility of inducing discharge of oil from wrecks have been identified throughliterature reviews (Louzis et al., 2009; Michel et al., 2005; Etkin et al., 2009) andbrainstormingeventswithgroupsofexperts inwhichSwedishwatersareusedas thefocalarea.Someactivitiesareinducedbyhumanswhilstotheractivitieshavenaturalorenvironmentalcauses.Below isadescriptionofeachof theeight identifiedhazardousactivities presented to experts when eliciting information for VRAKA. The elicitationprocedureispresentedinPaperIII.ConstructionworkConstructionwork can vary in extent and can include preliminary investigations andworkrangingfromaquaculturetopipelines,dredging,portconstructionandmulti‐useplatforms.DeteriorationDeteriorationincludesthewreck’sgeneralconditionandhowitisaffectedbycorrosionandothertypesofdegradationsincethetimethevesselsank.DivingDiving isdescribedasoneormorediversapproachingawreck theyknowofbyboat.Divers anchor and dive around or inside the wreck. Material from the wreck can bebroughttothesurface.

2. Conceptsofrisk

11

MilitaryactivityThisincludesactivitiessuchasbasicseamanshipexercisesastowing,tominesweepinganduseofmilitaryforce.ShippingtrafficShippingtrafficinvolveseverythingfrompleasureboats,ferrytraffic,coastguardvesselsandicebreakerstocruiseships,bulkcarriersandcontainerships.Storms/ExtremeweatherThisisdefinedaswindsofatleast98‐102km/h(Storm).TrawlingTrawlingmeansfishingwitheverythingfromsmalltrawlsmadeofsteel‐reinforcedwoodclosetothecoastandinmoreshallowwaters,tofishingwithlargesteeltrawlsfurtheroutfromthecoastatconsiderabledepth.UnstableseabedAvaryingamountofmaterialontheseabedismovedoveradistancethatcanalsovary.

2.1.2. Site‐specificandwreck‐specificindicators

Theprobabilityandextentofthedamagethespecifichazardousactivitiescanleadtoareassumedtobedependentonanumberofsite‐specificandwreck‐specificindicators.Theseindicatorswerearrivedatduringbrainstormingsessionswithexpertsandthroughliteraturereviews(Louzisetal.,2009;Micheletal.,2005;Etkinetal.,2009;Sender,2011).Theindicatorsare(PaperIII,IV):

Averagebottomwateroxygenconcentration Averagebottomwatersalinity Averagebottomwatertemperature Averagebottomwatercurrentspeed Averagehullthickness Seabedcharacter Shipuse Timesincesinking Waterdepth Wreckpositiononseabed

2.2. Theriskmanagementprocess

Thereareseveralframeworksthatdescribetheriskmanagementprocess(e.g.IEC,1995;ISO,2009).Theyaretypicallygenericinthesensethattheyarenotlimitedtoonespecific

H.Landquist

12

field of application and they share certain basic steps (Figure 2): risk assessment,consistingofriskanalysisandriskevaluation,andriskreductionandcontrolcollectivelycompleteafullriskmanagementprocess.

Figure2.Asimplifiedrepresentationoftheriskmanagementprocess,basedonIEC(1995).

An expanded framework is provided by ISO (2009) (Figure 3) where a broaderpresentationoftheriskmanagementprocessisgiven.Inestablishmentofthecontext,theorganisationsetsthescopeandriskcriteriafortheprocess.Withinriskidentification,theaimistoproduceacomprehensivelistofwhatmightaffectachievingtheobjectives.Riskanalysisistheprocessofunderstandingthenatureanddeterminingthelevelofriskandshouldprovideinputfortheriskevaluationstep,whichcouldactassupportwhenmakingdecisionsrelatedtothehandlingofrisks.Riskevaluationinvolvescomparingtheresultsoftheriskanalysiswithriskcriteriaandevaluatingpossibleactions.Thisshowswhethertherisksfoundareacceptableornot.Duringthefinalstep,risktreatment,oneormoreactionsarechosenandimplementedtomitigatetherisk.

In order to enter into a dialogue with stakeholders and to ensure an exchange ofinformationwithrelevantparties,communicationandconsultationareimportantduringall the riskmanagement phases.Monitoring and review should also be part of a riskmanagement process, including providing input to improve the process, detectingchanges thatshouldbereflected inearlierstages,and identifyingemergingrisks (ISO,2009).

Riskmanagementcanbeahelptodecision‐makerswhenmakinginformedactionsandwhen prioritising between options (ISO, 2009). More specifically, it is through riskassessmentsandanalysesinformationtosupportgooddecisionscanbeelicited(Aven,2012).

2. Conceptsofrisk

13

Figure3.TheriskmanagementprocessafterISO(2009).

2.3. Riskassessment

As noted above, the purpose of risk assessment is to provide input and support to adecision‐making process (Burgman, 2005). It is performed within many disciplinesincludingmedicine,engineeringandenvironmentalregulation(Suter,2007).Themainsteps in risk assessment are risk identification, risk analysis and risk evaluation asdescribedabove(ISO,2009).Numerous tools are available for hazard and consequence identification, such aschecklistsandbrainstorming,hazardmatrices,hazardandoperabilityanalysis(HAZOP),failuremodesandeffectsanalysis(FMEA)andhierarchicalholographicmodelling(HHM).Several tools could be applied for the same hazard identification process in order toincreasethepotentialtodetectasmanyhazardsaspossible(Burgman,2005),whichisalsothepurposeofthisstep(ISO,2009).Riskanalysisisperformedtodevelopanunderstandingoftherisk.Theconsequencesandtheir likelihoods are determined. The analysis can be qualitative, semi‐qualitative orquantitativeandshouldincludeasensitivityanalysis.Thetypeofanalysisdependsonthelevelofdetailsoughtandothercircumstances,suchasthepurposeoftheriskassessment(ISO,2009).Intheriskevaluation,identifiedlevelsofriskcanbecomparedtopre‐setcriteriaforriskacceptanceinordertoanalysewhethertheriskisacceptable.OnemeansofevaluatingriskagainstsetcriteriainpracticaltermsistheuseoftheALARPprinciple,i.e.“aslowasreasonablypracticable”,whereitisacknowledgedthatevenifinmostcircumstancesrisk

H.Landquist

14

canbemitigated,theeffortsinvolvedtomitigatetherisktothatextentareincreasinglycostly.(Jones‐LeeandAven,2011;Burgman,2005)(Figure4).Thereisanumberofdecisioncriteriaavailableforstatingtheacceptedrisk.Preventingadverse effects, mitigating risk, and balancing costs and benefits, are some examples(Suter,2007).Criteriaofthisnaturecanbetermedutility‐basedandrights‐basedcriteriawheretheformerarebasedonavaluationofoutcomeswhilethelatterarefocusedonprocess and permitted action. An example of a utility‐based criterion is benefit‐cost(probabilisticordeterministic)whichisaimingatidentifyingthealternativethatproducesthehighestnetbenefit.Rights‐basedcriteriaareinsteadexpressedase.g.zerorisk,whereriskiseliminatedindependentlyofbenefitsandcosts.Anotherexampleofarights‐basedcriterionisboundedorconstrainedrisk,wherethelevelofriskisconstrainedtoaspecificlevelormeetsaspecificcriterion.Combinationsofthesetypesofcriteriaarealsopossible(MorganandHenrion,1990).Arangeofmethodsareavailabletosupportthedecision‐makingprocessandtoevaluateifthesetcriteriahavebeenmet.Riskrankingisasimpleformofriskcomparisonwhererisks are ordered. Risk classification involves assigning risks as e.g. acceptable orunacceptable (Suter, 2007), see also ALARP above. Examples of more elaboratetechniquesareMultiCriteriaDecisionAnalysis(MCDA)andCost‐BenefitAnalysis(CBA),whereMCDAisameansofguidingthedecision‐makerbydefiningtheallrelevantcriteriaforevaluatingtherisk(Burgman,2005).Cost‐BenefitAnalysisisanexampleofamethodwithaneconomicfocus,measuringwhetherthebenefitsaregreaterthanthecostsforaparticularactionfromasocietalpointofview(HanleyandBarbier,2009).Furthermore,Cost‐EffectivenessAnalysis(CEA)isameansoffindingtheoutcomewiththeleastcostinrelationtootheroutcomesratherthaninabsolutenumbers(Munier,2004).

2. Conceptsofrisk

15

Figure4.TheALARP‐principlebasedonBurgman(2005).

2.4. Decision‐making

Abasicstructureforthedecision‐makingprocessisshowninFigure5(Aven,2012).Thestructureisfoundedontheassumptionthatdecision‐makingisaprocesswhereriskanddecisionanalyses,e.g.CBAorMCDA,provideinput.Informalmanagerialjudgementandreview should follow, to result in a decision. The risk management process and thedecision‐makingstructuredonotconflict;rather,theformerisanimportantcomponentofthelatter.

Figure5.Basicstructureofthedecision‐makingprocess(Aven,2012).

H.Landquist

16

2.5. Uncertaintyandprobability

Twocommondenominatorsforproblemsconcerningriskarethatadecisionneedstobemade and that uncertainties are involved (Suter, 2007). Uncertainty is a far‐reachingconcept and can emerge from statistical variation, linguistic indistinctness, variability,inherent randomness, and so on (Morgan and Henrion, 1990). Potential sources ofuncertaintyaremany,althoughuncertaintyisingeneralcategorisedasoneoftwotypes:epistemicandaleatory. Epistemicuncertaintyisassumedtobederivedfromalackofdataandaleatory(stochastic)uncertaintyreferstointrinsicrandomnessanddescribesnaturalvariabilityinpopulations(KiureghianandDitlevsen,2009;Aven,2012).However,purelyaleatoryuncertaintiesseemraresincetheyoftenhaveanepistemiccomponent(O'Hagan and Oakley, 2004). Uncertainty can be defined as “Imperfect or incompleteinformation/knowledgeaboutahypothesis,aquantity,ortheoccurrenceofanevent”(Avenetal.,2015).

Thequalityoftheriskassessmentwilllargelydependonhowuncertaintiesaredealtwith.Uncertaintiesmustbetreatedinacomprehensive,repeatableandtransparentmanner(Burgman,2005).Thealeatorytypeofuncertaintyisdeterminedbythecircumstancesandcannotbereduced,onlyacknowledged.However,theepistemicuncertaintiescanbereducedthroughincreasedinformation.Thereareseveraltechniquesforevaluatingandhandling uncertainties in risk analysis. One approach, applied in this thesis, isprobabilisticriskanalysis,alsoknownasquantitativeriskanalysis,whichcanbeusedtoestimate or cope with uncertainties regarding our knowledge about probability andassociatedconsequences(e.g.BedfordandCooke,2001).

An important means of evaluating and handling uncertainty is to apply a Bayesianapproach,facilitatingaformalhandlingofsubjectiveinformation.AroundonehundredyearsafterthetheoryofprobabilitywasformulatedbyFermatandPascal,asmentionedinthebeginningofthechapter,ThomasBayesdiscoveredhownewinformationcouldbemathematically combinedwithold information toprovidebetterdecisions (Bernstein,1996).Bayesianprobabilityisabelief‐typeapproachanassessorcanapplytorationallychange a belief when new evidence is found (Hacking, 2001). When applying Bayes’theory, a prior distribution can be updated to a posterior distribution. The priordistributionisbasedonknowledge,includingmeasurementdata,andexpertise,whiletheposterior distribution also acknowledges additional information, e.g. from a newlyperformed sampling campaign. The extensive use of the Bayesian approach can beattributed to the possibility of combining expert judgements with scientific evidence(BedfordandCooke,2001).

Furthermore, it is crucial to accept the fact that there is uncertainty in unknownunknowns,i.e.aspectsthatwerenotpossibletoknowbeforehand.Itissimplyimpossibletoincludeallrelationshipsandvariables(Metzetal.,2007).Probabilitycanbeinterpretedandexpressedindifferentwaysaseitherfrequentistorsubjectivist.Afrequentistapproach“definestheprobabilityofanevent´soccurringina

2. Conceptsofrisk

17

particulartrialasthefrequencywithwhichitoccursinalongsequenceofsimilartrials”(Morgan and Henrion, 1990, 48). The subjectivist, or Bayesian, definition views theprobabilityofaneventratheras“thedegreeofbeliefthatapersonhasthatitwilloccur,givenalltherelevantinformationcurrentlyknowntothatperson”(MorganandHenrion,1990,49).Althoughsubjective,aBayesianprobabilitymust,tobelegitimate,followtheaxioms of probability. For example, if a probability of an event occurring is pi, thecomplement, i.e. the event not occurring, is 1‐pi (Morgan and Henrion, 1990). In thisthesis,theBayesianorsubjectivistapproachisappliedsincethereisalackofbackgroundinformationandobservationaldatathatcouldserveas input for themodel.Moreover,everyshipwreckisuniqueandaneventintheformofamoreextensivedischargeisnotrepeatable.Amain focusof this thesiswas todevelop amethod for calculating theprobability ofdischarge of oil from shipwrecks. A Bayesian approach was applied to estimateprobabilities as degrees of belief. An assessment of site‐specific and wreck‐specificparameterswasappliedtoupdateearlierestimationsoftheprobabilityofopeninginawreck.Sincethebackgroundinformationandknowledgefortheprobabilityestimationisincomplete, theprobability estimation itself isuncertain. Statisticaldistributionswerethusassignedtorepresenttheuncertaintyoftheprobabilityestimations.Theapproachtakenhereisinaccordancewiththeapproachtoprobabilisticriskanalysisdescribedby,forexample,BedfordandCooke(2001).

H.Landquist

18

19

3. Riskassessmentapproachesforshipwrecks

In Paper I, Evaluating the needs of risk assessment methods of potentially pollutingshipwrecks, six methods and approaches for risk assessment of shipwrecks wereanalysedinordertoevaluateiftheycouldproviderelevantdecisionsupportinthisarea.Theidentifiedmethodswerecomparedto,inthecaseforshipwrecks,relevantpartsofthe ISO standard for risk assessment, ISO 31000, Risk management – Principles andguidelines(ISO,2009),seeSection2.1forfurtherdetailsofthestandard.Moreexplicitly,parts assumed to be relevant correspond to the necessary steps for a comprehensiveenvironmental risk assessment of shipwrecks and are presented in Table 2. Sincepublication,threefurtherworkspresentapproachesforriskassessmentofshipwrecks,NOAA(2013),Ventikosetal.(2013)andVentikosetal.(2016).SomeoftheauthorsofthefirstpublicationauthoredthepublicationoftheWORPproject(NOAA,2009)presentedbelowasapproachA.ThelatterpublicationsshareauthorswiththepublicationbyLouzisetal.(2009),presentedbelowasmethodF,butthereisnoclearreferencebetweentheearlierandlatterworks.ThesemorerecentmethodsareaddedtotheoriginalcomparisonandasummaryispresentedinTable2.ThefullcomparisonisappendedtoPaperIassupplementaryinformation.

3.1. Identifiedmethodsandapproachesforenvironmentalriskassessmentofshipwrecks

The methods and approaches for risk assessment of wrecks were found in scientificpapers and other official reports. Other material, such as the Nairobi InternationalConventionontheRemovalofWrecks(InternationalMaritimeOrganization,2007),wasnotincludedsinceitwasnotintendedasaframeworkforriskassessmentofshipwrecks.Furthermore, the IMO Guidelines for Formal Safety Assessment (FSA) (InternationalMaritimeOrganisation,2002)isexcludedsinceitisnotawreck‐specificguideline.

Theevaluatedmethodsandapproacheswere:

A. TheWreck Oil Removal Program (WORP), aimed to use a scientifically‐basedapproachtoremoveoilandfocusedspecificallyontryingtominimisecostsandtheriskofpollutionfromsunkencommercialvessels(NOAA,2009).

B. Michel et al. (2005) presented an oil release risk assessment guide for

shipwrecks. The goal was stated as being to objectively analyse shipwrecksaccording to their potential threat of petroleum discharge and to provideguidanceonaddressingthisissue.

H.Landquist

20

Etkinetal.(2009)presentedasimilarguide,althoughtakingintoaccountboth

theprobabilityofleakageandtheimpactofsuchaleakageoccurring,inordertoassess the environmental risk of shipwrecks. It was a strategic modellingapproachtoprioritiseshipwrecksandtheaimwastoprovidedecisionsupportforauthorities.

C. DEvelopment of European Guidelines for Potentially Polluting Shipwrecks

(DEEPP)aimedtoprovideEuropeancoastalstatesandnationaladministrationswithguidelinesandcriteria tohandle theenvironmental threatof shipwrecks(Alcaroetal.,2007).

D. TheNorwegianPollutionControlAgency(NPCA)describedawreckproject in

threephases:registration,priorityrankingandrequiredaction.Theaimwastohave a complete overview of shipwrecks along the Norwegian coast (Idaas,2005).

E. The Pacific Ocean Pollution Programme (PACPOL) within the South Pacific

RegionalEnvironmentProgramme(SPREP)wasaimedatpollutionofthemarineenvironment fromship‐basedsources.This specific strategywasaimedate.g.preventingormitigatingthenegativeimpactofshipwrecks(SPREPandSOPAC,2002).

F. AriskanalysisstrategyforshipwrecksinGreekwaterswaspresentedbyLouzis

etal.(2009).ItwasaimedatoilleakageandwasbasedontheIMOFormalSafetyAssessment(2002).

G. NOAA (2013) presented a comprehensive method for risk assessment ofpotentiallypollutingwrecksinU.S.waters.Pollutionpotentialwasestimatedaswellastheconsequencesofarangeofspills.Possibleresponsealternativeswerealsoexplored.

H. Ventikos et al. (2013) presented the development of an application for riskanalysisofshipwrecksinGreekwaters.Adatabasewasdeveloped,anapproachfor risk assessment was presented and finally GIS tools were used in thevisualisationofdataandresults.

I. In“Enhanceddecisionmakingthroughprobabilisticriskassessment:Focusingon the situation in Greece”, Ventikos et al. (2016) presented a method forestimating the probability of release using dynamic fault tree modelling.Consequences were estimated qualitatively and then combined with theprobabilityusingariskmatrix.


21

3.2. Comparisonwithaninternationalstandardforriskmanagement

Eachofthemethodsfoundweregradedonafour‐pointscalewithrespecttoeachoftheselectedparameters fromthe ISOstandard.Theparameterswereevaluated from“notconsideredatall”,“considered”,“fulfilledtosomeextent”,to“fulfilledtoalargeextent”.AsummaryispresentedinTable2.

H.Landquist

22

Table2.Theresultsofthecomparisonofriskassessmentmethodsforshipwrecks.A=TheWreckOilRemovalProgramme(NOAA,2009),B=Potentiallypollutingwrecksinmarinewaters(Micheletal.,2005),C=TheDEEPPproject,DevelopmentofEuropeanguidelinesforPotentiallyPollutingshipwrecks(Alcaroetal.,2007),D=TheNorwegianPollutionControlAuthority(Idaas,2005),E=TheSouthPacificRegionalEnvironmentProgramme(SPREPandSOPAC,2002),F=Louzisetal.(2009),G=ThemethodbyNOAA(2013),H=TheapproachbyVentikosetal.(2013)andI=Ventikosetal.(2016).

The comparison inPaper I showed that therewasno comprehensivemethod for riskassessmentofshipwrecks.Themainweaknessesfoundwerethatmanyofthemethodswerequalitativeandingeneraldidnotincludesensitivityoruncertaintyanalysis.Overall,riskevaluationdidnotcorrespondtothelevel“fulfilledtoalargeextent”.Themethodsfound provide some guidance but none provide comprehensive support for decision‐making with regard to shipwrecks. To some extent, the new methods compriseduncertainty analysis and NOAA (2013) also incorporate a quantitative probabilisticmethodforconsequenceassessment,althoughnotregardingtheprobabilityofdischarge.

Method

A B C D E F G H IEstablishmentofthecontext

Scope

Definingriskcriteria ‐ Shipselection ‐ Geographiccoverage

Riskassessment

Riskidentification Sourcesofrisks

Areasofimpact

Events

Causesofevents

Riskanalysis

Estimationofconsequences

Estimationoflikelihood/probabilitiesofconsequences

Riskestimation ‐ Qualitative ‐ Semi‐quantitative ‐ Quantitative

Uncertaintyanalysis

Sensitivityanalysis

Riskevaluation

Comparisonofrisklevelswithriskcriteria

Considerationofrisk‐reductionalternatives

Legend;Leveloffulfilment

Fulfilledtoalargeextent Fulfilledtosomeextent Considered Notconsidered


23

Ventikosetal.(2013)andVentikosetal.(2016)providepartlyqualitativeestimationsand Ventikos et al. (2013) includes uncertainty analysis while Ventikos et al. (2016)includesbothsensitivityanalysisandaddressestheissueofuncertainty.Itshouldalsobepointedoutthatsomeoftheapproachescomparedareatthepreparatorystagewhilee.g.NOAA (2013) and Idaas (2005) are actually applied to a larger number ofwrecks. Insummary,NOAA(2013)andVentikosetal.(2016)arethemostdevelopedapproachesaccordingtothecomparisonpresentedinTable2butthereisstillnofullyprobabilisticcomprehensiveriskassessmentmethodforshipwrecksthatprovidesdecisionsupportinrelation to this threat. Lack of holistic risk assessment and decision support forshipwreckscouldpotentiallyleadtoinefficientallocationofresourcesforriskmitigation.ThecomparisonpresentedhereandinPaperIformsthebasisforthedevelopmentoftheframeworkandmodelforprobabilisticriskassessmentofshipwreckspresentedinthisthesis.Theframeworkclearlystatesthepartsthatshouldbeincludedandstressesthelink between risk management and decision‐making. The suggested framework ispresented in more detail in in Chapter 4 and in Chapter 6 the developed model isdescribed.

H.Landquist

24

25

4. Ariskmanagementframeworkforpotentiallypollutingshipwrecks

Environmental risk management of shipwrecks is complex and should therefore beperformedinastructuredway.AgenericframeworkforriskmanagementofshipwreckswassuggestedinPaperI(Figure6Figure 1),basedonISO(2009),whichisaninternationalstandard. The purpose was to emphasise the link between decision‐making and riskmanagementandstress thatriskassessment isanessentialbasis fordecision‐making.Certainadditionalpartswereaddedtotheoriginalguidelinestostress,forexample,thecommunicationofresultstostakeholdersandtheneedtohandleuncertainties.Exchangeofinformationandcommunicationareimportantinthistypeofproceduresinceitisessentialthatthoseresponsibleforimplementingtheresultsarewellinformedaboutthebasisforpossibledecisions.Riskmanagementisaniterativeprocessandstakeholdersshould be continuously involved in the different steps. A review should be made toincorporatenewinformationandupdatepartsoftheriskmanagementworkduringtheprocess if necessary (ISO, 2009). Furthermore, the complexity implies that a team ofexpertsshouldbe involvedtoensure thatknowledgeofmarineactivities, thephysicalpropertiesofaship,theriskassessmentprocessandotherfactorsaredulyconsidered.The framework presents a holistic risk management structure for shipwrecks and isbasedonwell‐establishedviewsofriskmanagement.Onlybyapplyingacomprehensiveframework and adopting a holistic view of the problem can a full risk managementprocess and assessment be made. When a framework is in place, methods can bedevelopedthatarewellsuitedtothepurposeofeachspecificpartandtotheprocessasawhole.The frameworkadvocatesproactive riskmanagement tomitigate the riskofpollutionfrom shipwrecks. However, risk assessment cannot be the sole source of informationinput behind a decision, aspects not included in the risk assessment should also beconsideredinthedecision‐makingprocess(Aven,2003).Inordertoimplementtheframeworkinpractice,itisnecessarytouseanadequatesetofmethodsandtools.Largelygeneralisedassessmentsareduetothecomplexcause‐and‐effectchainsofrisksinvolvedwithshipwrecks,notlikelytoprovideusefulresultsandtoolsshouldthereforebechosenwithcare.Furthermore,uncertaintiesareassumedtobeconsiderable and shouldbehandledproperlybyadoptingaquantitative, probabilistic

H.Landquist

26

approach.DevelopmentofmethodsandtoolsformakingariskassessmentofshipwrecksisinitiatedandpresentedinChapter6andinPapersI‐Vinthisthesis.

Figure6.Genericframeworkforriskmanagementandassessmentofpotentiallypollutingshipwrecks(PaperI).

27

5. Methods

Thischapterdescribesthemethodsappliedintheappendedpapers(Figure7).ThefaulttreeapproachisusedlargelyinPapersIIandIV,andlikewiseMonteCarloanalysis,asmeans for estimating the probability of discharge of hazardous substances. ExpertelicitationandtheSHELFpackage(SheffieldElicitationFramework)areappliedinPaperIIIforobtaininginputdataforVRAKA.TheDigitalEnvironmentalAtlasandtheoilspilltrajectory model Seatrack Web are used in Paper V. The theory for Paper I, riskassessmentandmanagement,isdescribedinChapter2.

Figure7.OverviewofmethodapplicationinPapersI‐V.

5.1. Faulttreeanalysis

Logictrees,suchasfaultandeventtreeslinkinghazardouseventsandprocesses,haveawiderangeofapplications(Burgman,2005).Aneventtreedescribessubsequenteventsfollowingastartingeventandthefinaloutcomesfordifferentscenarioscanbespecified.Fault trees on the other hand seek to find the actions or events that cause such aninitiatingevent.The fault treeemerges frombasiceventsandpassesvia intermediateeventstotheanalysedtopevent(BedfordandCooke,2001).AnexampleofafaulttreeisshowninFigure8.In a fault tree, several logic gates describe how the underlying events are connected(BedfordandCooke,2001;Burgman,2005;Robertsetal.,1981).Inthisthesis,theANDandORgatesareapplied.TheANDgatedefinesanoutcomewherealltheinputevents

H.Landquist

28

_

_ 1 1

mustoccursimultaneouslywhiletheORgatedescribesanoutcomewhereonlyoneoftheinputeventsmustoccur(Robertsetal.,1981).MathematicaldescriptionsofthetwogatesappliedareprovidedinEquations1and2,wherePrepresentsprobabilityandindex irepresentsaneventinthefaulttree.

Figure8.Exampleofafaulttree.Simplifieddescriptionofapotentialoildischargefromashipintraffic.

Equation1

Equation2

5.1. MonteCarlosimulation

MonteCarlosimulationisatechniqueappliedinuncertaintyanalysis(e.g.BedfordandCooke, 2001). It is based on random sampling and provides a means to cope withuncertaintyininputinformationandresults(Burgman,2005).InVRAKA,inputdataisprovided as ranges or probability distributions. ThroughMonte Carlo simulation, theoutcomeestimatesoughtiscalculatedusingrandomsamplingoftheinputdistributionsforasetnumberofiterations(seeFigure9foraschematicvisualisation).Aprobabilitydistributionisobtainedthatrepresentsthewholespectrumofpossibleoutcomesratherthan a point value. This technique facilitates sensitivity and uncertainty analysis (e.g.BedfordandCooke,2001;Burgman,2005).

5. Methods

29

Figure9.TheconceptofMonteCarlosimulationaccordingtoLindhe(2010).Inputinformationisgivenasdistributions,andresultsarepresentedincludinguncertainties.

5.2. Expertelicitation

Shipwreckriskassessmentisassociatedwithuncertainties.Lackofmeasurabledataordifficultyobtainingdataduetocostorotherconditions,suchasthedepthofthewreck,complicate the risk assessment. One means of gathering or estimating this type ofinformationandreducingtheuncertaintiesoftheoutcomeistoaskforanexpertopinionandsubjectiveprobabilities(BedfordandCooke,2001).

5.2.1. Experts,heuristicsandbiases

Expert judgement is almost unavoidable in environmental risk assessment. Whenempirical evidence or extrapolations are not available, experts can step in. They canestimate intervals and make point estimates or statistical distributions by usingfrequency concepts of probability without any measurement information (Burgman,2005).Anexpertcould,forexample,besomeonewiththebestavailableknowledgeaboutthequantitiesofinterest(Garthwaiteetal.,2005),orapersonwithagreatdealoftrainingandknowledgeinaspecificfieldwhoprovidesanopinion(Ayyub,2001).Peoplemakejudgementsunderuncertaintybasedonanumberofheuristicprinciples.Thesearestrategiesorinstinctiveprocessesthatreduceacomplexquestiontosimplerjudgements.A couple ofheuristicsmightbeusefulbut canalso lead to serious errors(TverskyandKahneman,1974).HeuristicsmentionedbyTverskyandKahneman(1974)include:(1)representativeness,(2),availabilityand(3)adjustmentandanchoring.TverskyandKahneman(1983)describerepresentativenessasestimatedcorrespondencebetweenasampleandapopulationoranoutcomeandamodel.Alternatively,itcouldbe

H.Landquist

30

thedegree towhichA is representativeofBor resemblesB (TverskyandKahneman,1974).RepresentativenessisspecificallyrelevantwhenelicitingconditionalprobabilitieswheretheexpertmightassessprobabilitybyhowwelloutcomeArepresentsmodelBratherthantheprobabilityofobtainingoutcomeBgivenmodelA(O'Haganetal.,2006).The conjunction fallacy is one type of cognitive bias that can be attributed torepresentativeness.Theprobabilityoftheconjunction,P(A&B)cannotbelargerthantheconstitutingprobabilitiesP(A)andP(B),astheformerisdescribedusingpartsofthesesubsets(orthewholesubsetsifP(A)andP(B)areequal).However,theconjunctioncanbemorerepresentativethaneachofP(A)orP(B),whichiswhyanerrorcouldbemade.Another difficulty is to separate P(A|B) from P(B|A), known as the inverse fallacy(VillejoubertandMandel,2002).Availabilitycanbeappliedwhenassessingafrequencyorprobability.Frequenteventsaremoreeasilyrecalledthanlessfrequenteventsandthejudgementisbasedontheeasetheoccurrencecanbebrought tomind(TverskyandKahneman,1974).Theeasier it is torecallanevent,thehighertheassessmentofprobability.Aneventwithlowfrequencybuthighmediacoveragewouldthusalsobeassignedahighprobability(O'Haganetal.,2006).If, however, the assessor or expert has greatpersonal experience of the problem thatneedstobeassessed,thisheuristicmightperformquitewell(MorganandHenrion,1990).Anchoringandadjustmentrefertowhenpeoplemakejudgementsbasedonaninitialvalueand then adjust that judgement to reach a final assessment (Tversky and Kahneman,1974).However,thereisatendencytowardsnotadjustingthefinalvaluesufficientlyandtheestimateisbiasedwithregardtotheinitialvalue.Theanchoringandadjustmenteffectcanapplytoanykindofquantitativejudgement(O'Haganetal.,2006).Both experts and laymen provide assessments influenced by heuristics even if someelementaryerrorscanbeavoidediftheexpertistrainedintheuseofstatistics.Heuristicscanbeaveryeffectivemeansofmakingdecisionsalthoughunderstandingtheheuristicscanleadtobetterdecisionsunderuncertainconditions(TverskyandKahneman,1974).Theprocessneedstobeconstructedinawaythattheinfluenceofheuristicsandbiasintheelicitationareminimised.Anchoringeffectscould,forexample,behandledinpartbycarefully choosing the order of the questions posed to the expert. The process couldencourageanalyticalthinkingby,forexample,applyingaidsandtheexperts’experienceshouldbeconsideredinrelationtopossibleanchoringandavailabilityeffects(O'Haganetal.,2006).

5.2.2. StakeholdersPublicconcernandawarenessofthemanyaspectsofriskhasbecomestrongerduetothegrowth in communications, availability of information and scientific advances (Aven,2012).Apersonororganisationthatisparticularlyaffectedbytheconsequencesofanenvironmentalmanagementdecisioniscalledastakeholder(Suter,2007).Therearealso

5. Methods

31

definitions containing non‐human entities, such as endangered animals and plants(Burgman,2005).Thepublicasawholecanhaveaninterestinenvironmentaldecisionsand decisions not accepted by those affected would be politically unacceptable. Thegeneralpublicandthestakeholdersthatareaffectedspecificallybythedecisionshouldbeseparatedandconsequentlythelattergroupissmaller.Itcouldbenecessarytoinvolvealargergroupfromthegeneralpublicinadditiontothestakeholdersinordertobalancetheinputsincethestakeholderstendtobebiased(Suter,2007).

Whenusingstakeholdersasasourceofinformationandinputfordecision‐making,itisimportantthattheirknowledgeisusedeffectively.Thisinvolves,forexample,choosingthe right stakeholders for the problem at hand, eliciting information rigorously andapplying appropriate analysis techniques. There should be a clear statement andcommunicationofthepurposeoftheassessmenttothestakeholders.Aclearandhonestcommunication process throughout the assessment is necessary for effective dialogue(Glicken,2000).

Anexpertcouldbeastakeholderandviceversa.Inthisthesiswork,numerousexpertshavebeeninvolvedintheexpertelicitationprocessesandalargenumberofstakeholdershavebeeninvolvede.g.inreferencegroups.However,someofthestakeholderswerealsobroughtinasexpertsandviceversa.

5.2.3. ElicitationmethodsSeveral approaches, general suggestions, methods and guidelines for eliciting expertopinionandsubjectiveprobabilitiesaredescribedinthe literature(WallsandQuigley,2001; Cooke, 1991; Morgan and Henrion, 1990; Meyer and Booker, 2001; Jenkinson,2005; Phillips, 1999; Cooke and Goossens, 2000; Kuhnert et al., 2010; Ayyub, 2001).However, consensus prevails regarding the broader concept and Jenkinson (2005)suggeststhefollowingprocess:

1. Backgroundandpreparation2. Identifyingandrecruitingexperts3. Motivatingtheexperts4. Structuringanddecomposition5. Probabilitytraining6. Elicitation

Thefirststepinvolvesidentifyingthevariablestobeassessed.Herethefacilitatoralsohastheopportunitytoacquireknowledgeofthespecificfieldinordertocommunicateeffectively with the experts. Preparation of the elicitation session and arranging therequireddocumentsarealsopartofthefirststep(O'Haganetal.,2006).The second step, identifying and recruiting experts for elicitation, is a very importantstage.CertaincriteriathatcouldbeappliedwhenselectingexpertsareprovidedbyHora

H.Landquist

32

and Von Winterfeldt (1997): tangible evidence of expertise, reputation, willingness toparticipateandavailability,understandingofthegeneralproblemarea,lackofaneconomicorpersonalstakeinthepotentialfindingsandimpartiality.O'Haganetal.(2006)alsopointout that it might be worth the effort to investigate the expert’s level of statisticalunderstanding.Ifthisknowledgeispoor,trainingcouldbeprovidedatalaterstage.Thethirdstepinvolvesmotivatingtheexperts.O'Haganetal.(2006)suggestthatexpertsshould be informed about why their judgement is sought and how the results of theelicitationwill beused. Furthermore, it is suggested that a record is kept and that itspurpose is explainedclearly to theexperts. Jenkinson (2005)alsopointsout that it isimportanttoestablishagoodbondwiththeexpertsandprovidethemwithsupportandencouragement in the elicitation process. This is followed by structuring anddecomposition,whichisastageofdefiningpreciselythequantitiestobeelicitedandtheformatoftheelicitation(CookeandGoossens,2000).Expert knowledge can be divided into substantive andnormative. The first applies toknowledgeandexperienceofthefieldinquestionwhilethelatterisrelatedtotheformofresponsedesiredintheelicitation.Theresponsemodecouldbeprobabilities,ranksorpairwisecomparisons(MeyerandBooker,2001).Someexpertsmightbefamiliarwiththeresponsemodechosenandsomemaynot,thushighlightingtheneedforprobabilitytraining.Thefiveinitiatingstepsaretakeninpreparationfortheactualelicitation.Garthwaiteetal.(2005)describetheprocessofelicitingprobabilitydistributionsasfollows:

1. Thesetup.Incorporatethefiveinitiatingstepsalreadydescribed.2. Elicitthesummariesoftheexpert’sdistributionsregardingtheareasofinterest.

Thisisthemainstageoftheexpertelicitationprocess.3. Fitaprobabilitydistributiontotheelicitedsummariesinstep2.4. Assesstheadequacyoftheelicitation.Heretheelicitationcanalsogobacktostep

2toelicitmoresummariesinaniterativeprocess.

5.2.4. SHELFSHELFisanexpertelicitationframeworkdevelopedbyO'HaganandOakley(2010).Itisopenlyavailabletoanyonewhowishestouseitanditisappliedinthisthesis.Itprovidesa formal procedure for an elicitation process and improves quality and defensibility.SHELFisatoolandprovidesadviceandguidanceontheprocessforafacilitator.SHELFcanbeappliedtoelicitdistributionsforoneormoreexpertsasagroupandthepackageprovidesseveralpossibilities forelicitingadistribution.Theseare: theP (Probability)method,whereprobabilitiesarederivedfromtheexperts;theQ(Quartile)method,wheretheexpertisaskedtoestimatethemedianandupperandlowerquartiles;theR(Roulette)method,wherethefacilitatorasksforprobabilitiesofrangesofvalueswithintenbins,

5. Methods

33

andtheT(Tertile)method,wherethemedianandupperandlowertertilesarerequestedfromtheexpert.Initially, the elicitation should be done individually, after which the group makes acommonelicitation.Eachofthesestepsispresentedinformsthefacilitatorandexpertscanusetogothroughtheelicitationprocess.Duringelicitation, theexpertsareshowndistributionscorrespondingtotheiranswers.TheseshouldbecomputedinrealtimeandtheSHELFpackage includesproceduresadapted to the statistical softwarepackageR,Rpanel, for performing this. By visualising the distributions, the experts are given theopportunity to provide feedback and revise their estimations in order to produce thedistributiontheybelievein.

5.3. OilspilltrajectorymodellingandSeatrackWeb

In order to adequately determine the spreading of oil discharge in the marineenvironment,oilspilltrajectorymodellingisrequired.OneexampleofsuchatoolistheSpill Impact Model Application Package (SIMAP), which comprises three‐dimensionalmodelsforbiologicaleffectsandoilfates.Itcanberuninastochasticmodeforanalysisoftheprobabilityofdifferentoutcomes(McCayetal.,2004).Anothertoolthatprovidesthepossibilityofmodellingoilspilltrajectories,andwhichisappliedintheBalticSea,isSeatrackweb.DuetoitsavailabilitytoSwedishauthorities,ithasbeenchosenfortheworkathand.Theuserprovidesthedate,locationandamountofpossibledischargeandSeatrackWebusesavailablemeteorologicalandhydrologicaldatato estimate a trajectory. Results are provided as amounts of hazardous materialtransported to the seabed, water surface and shore (SMHI, 2015; SMHI, 2011). Thetrajectoryestimationisbasedonforcing(forecastsofflowandwindfields)andtheoildriftmodelPADM(ParticleDispersionModel),andispresentedinagraphical,web‐basedinterface.TheforcingfieldsarebasedonoceanandweatherforecastsatSMHI(SwedishMeteorologicalandHydrologicalInstitute),i.e.theHIRLAM(HighResolutionLocalAreaModelling)forweatherforecasts,andHIROM(HighResolutionOperationalModelfortheBaltic)foroceanforecasts.LongerperiodsofforcingweatherfieldsaremodelledbytheEuropeanCentreforMedium‐RangeWeatherForecasts.TheprincipalareascoveredbySeatrackWebaretheBalticSea,Kattegat,SkagerrakandtheNorthSeatoaround3°east.ThePADMhasbeendevelopedbytheDanishMaritimeSafetyAdministration(DAMSA)andSMHIandthegraphicalinterfacebySMHI.Forfurthertechnicaldetails,seeLiungmanandMattson(2011).

5.4. OilspillimpactandtheDigitalEnvironmentalAtlas

Estimationofoilspillimpactshouldcontainecologicalandsocioeconomicaspects.Toolsthat can aid the analysis of the impact could include e.g. sensitivity mapping of the

H.Landquist

34

coastline, mapping of tourist areas, the businesses possibly affected, environmentallyprotectedareas,sensitivityestimationsofspeciesinthearea.The Digital Environmental Atlas is applied in this thesis andmaps the Swedish coastaccordingtosensitivitytooilspillandphysicalcharacteristics.Thesensitivitymappingisbasedonthetypeofshore,itsexposureandtosomeextentonbiologicalprerequisites.(Kulanderetal.,2010).ThetoolisopenlyavailableonthewebandisadministeredbytheCountyAdministrativeBoardofVästraGötaland(CountyAdministrativeBoardofVästraGötaland,2015).

35

6. VRAKA–acomprehensivemodelforriskassessmentofshipwrecks

VRAKA– the comprehensivemodel for riskassessmentof shipwrecks,wasdevelopedbased on the conclusions from Paper I and the framework for risk management ofshipwrecks.ThepurposeofVRAKAistofacilitateriskassessment,includingriskanalysisand risk evaluation (Figure 10). Risk analysis involves scope definition, hazardidentificationandriskestimation.Scopedefinitionisataskfortheassessor,i.e.itinvolvesdecidingwhichwreckswillbeincludedintheassessmentwhilehazardidentificationandrisk estimation are performed in VRAKA. VRAKA also provides a means for riskevaluation.AsshowninFigure10,VRAKAcanbeappliedtoprovidedecision‐supportregardingprioritisationofmitigationmeasuresforwrecks.Aspectslinkedtoimplementationandmonitoringaspartofriskreductionandcontrolarenotincluded.

Figure10.VRAKAinrelationtotheriskmanagementprocessafter(PaperIV).

H.Landquist

36

TheVRAKAmodelcanbedescribedasconsistingofthreeparts:PartI–Estimationoftheprobabilityofdischarge,PartII–ConsequenceassessmentandPartII–Riskevaluation(Figure11).PartIisamethodbasedonfaulttreeanalysisdescribingthecombinationofactivities possibly leading to an opening in the wreck and a possible discharge ofhazardoussubstances.PartIIisanaggregationofmethodsforconsequenceassessmentofapotentialdischarge.Theassessmentcanbeperformedaccording toa three‐tieredstructure.Thechoiceoftier,orlevelofdetail,ismadewithrespecttolevelofambitionand available resources. Part I is described in Papers II‐IV and is summarised in thischapter in Section 6.1. Part II is described in Paper V and is also summarised in thischapterinSection6.2.PartIII,riskevaluation,isdiscussedfurtherinSection6.3.

Figure11.ThethreepartsofVRAKA.

6.1. Part1–Estimationoftheprobabilityofdischarge

Theprobabilityofdischargeismodelledusingafaulttree(Figure12)andtwoeventsaredeemed necessary for discharge; an opening in the wreck occurs and hazardoussubstancesarepresent.Anopeninginthewreckisassumedtobearesultofoneormoreactivities,andhazardoussubstancesareassumedtobepresentifeitherthefuelorbunkerishazardousandisstillcontainedinthewreck.ThedevelopmentoftheVRAKAfaulttreemodel isdescribedinPapersIIandIV.AnimportantdifferencebetweenPapersIIandPaperIVisthat inthelattertherehasbeenanimportantrevisionofthemathematical


37

foundation,byinclusionofBayesianupdating,regardinghowtheopeningprobabilityiscalculated. Paper IV is presented here directly after Paper II to clearly display thesedifferences.ExtensiveeffortshavealsobeenmadetocollectmoreinputinformationforthemodelandthesearedescribedseparatelyinPaperIII.

Figure12.AschematicfaulttreestructureofVRAKAfromPaperIII.

6.1.1. Afirstversionofthefaulttreemodel–PaperIIHazardousactivitiesthatareassumedtopossiblycauseanopeninginashipwreckwereidentifiedthroughaliteraturereview(Etkinetal.,2009;Louzisetal.,2009;Micheletal.,2005;Sender,2011)andbrainstormingsessionswithexperts.TheactivitiesarefurtherdescribedinSection2.1.Aworkshopthatbroughttogetherexpertsinthisfieldwasalsoheld as part of the process of developing the fault treemethod. The output from theworkshop served as input for further work relating to e.g. the choice of hazardousactivities.

H.Landquist

38

Eachactivitywasrepresentedbyanintensity, i.e.thefrequencyofoccurrence,andtheprobabilityofanopeninginthewreckgiventheoccurrenceoftheactivity.Anexceptionwascorrosion,whichwasdescribeddirectlyas theprobabilityofopeningsince itwasassumedtobeacontinuousprocessandnotofthesamenatureastheotheractivities.Agenericprobabilityofopeningwasestimatedforeachhazardousactivity(Popening|acvtivity,i)by a groupof expertsand the risk assessor sets indicatorvalues andestimatesof thefrequencyoftheactivity(λoccurrence,i).Necessaryinputdatawereestimatedbyagroupofexperts,whowereaskedtoassignthehighestreasonabletheprobabilityofeachactivitycausinganopeninginashipwreck.Todescribetheuncertaintiesoftheprobabilityestimations,thevaluesgivenbytheexpertswere interpreted as the 90th percentile of beta distributions. The percentiles and theestimated parameters they are selected to represent are a way of representing thereliability of the expert judgements. A higher percentilemeans that the probability ofopeningismorecertaintobelowerthantheexpertjudgement.Theshapeparameterαofthe beta distributionwas set at a value of 1, implying that themost like value of theprobabilityofanopeningis0.Thiswasestimatedtobereasonable,givingthatthereiscurrentlyno informationavailable inSwedenregardinga largedischargeofoil fromashipwreck.Twelveindicatorswerefoundtoaffectthegenericprobability.Thesewerephysicalandenvironmentalfactorsthatcouldinfluencethedeteriorationrateandthecircumstancesinwhichtheactivitiesoperated.Theindicatorswereidentifiedthroughtheliteratureandbrainstormingsessionswithexperts(Louzisetal.,2009;Micheletal.,2005;Etkinetal.,2009;Sender,2011).The indicatorsarepresented inChapter2.1,although inthe firstversionofthefaulttreemodel,theindicatorsMaintenanceandShipconstructionwerealsoincluded.InPaperII,asemi‐quantitativestructurewassetuptodescribehowthetwelveindicatorsaffectedtheprobabilityofopeningduetothehazardousactivities.Themodelwasinitiallysetatanormalstateandtheassessorcouldchangethisbyeitherdecreasingorincreasingthe robustness of the wreck or environmental prerequisites on a scale from a‘considerablybetter’toa‘considerablyworse’state.Intotal,fiveoptionswereavailable.Themagnitudeof thechangewaspre‐set in themodel.This couldbe repeatedby theassessor forall the indicatorsand theprobabilityofactivities causinganopeningwasalteredwithrespecttothestateoftheindicators.Withthederivedexpertinformation,anexpectedvalue,µprior,wasestimated.Theassessorinformationregardingindicatorswasusedtoupdatethebetadistributionsfromµpriortoµupdated.Duringtheupdating,thesumofalphaandbetawaskeptconstant, indicatingthatnonewinformationregardingtheuncertaintyoftheprobabilityestimationwasprovided.


39

Theprobabilityofanopeningasaresultofaspecificactivitywasalsodescribedbythefrequency or rate of occurrence of the activity. This was modelled using gammadistributionsandtheassessorwasprovidedwithfixedintervals.Furthermore,theassessorestimatedtheprobabilitythattherewasahazardoussubstancepresentinthewreckinthreesteps:

1. Estimationoftheprobabilitythatthereisbunkerremaining.2. Estimationoftheprobabilitythatthereiscargoremaining.3. Estimationoftheprobabilitythatthecargoishazardous.

Theassessmentsregardingtheprobabilityofremainingbunkerandcargoandwhetherthe cargowas hazardouswere performed qualitatively. Possible choices ranged fromImpossibleandVerylowtoVeryhighandCertain.ItwasalsopossibletosetNoinformation.Values corresponding to the different classes were set by the authors enablingquantitativedescriptionofuncertainties.Thecertaintyof theseassessments fromLowuncertainty toHighuncertaintywasalsoestimated.Allassessmentsregardingthevolumeofbunkerandcargoandwhetherthecargowas environmentally hazardous or not,weremade using guidingmatrices. Theuncertaintiesofpossiblechoiceswererepresentedbybetadistributions.Thechoiceofcertaintycategoryalteredtheuppercredibilitylimit.Theassessment,forexample,wouldbeinterpretedasthe85thpercentileiftheassessorsetahighuncertaintyandthe95thpercentileifuncertaintywassetaslow.Theamountofhazardousmaterialcontainedinthewreckwasalsoestimated.Asmallestandhighest reasonable amountwas interpretedas the10thand90thpercentilesof alognormaldistribution.

6.1.2. Thenovelfaulttree–PaperIVInPaperIV,anovelfaulttreebasedonthefaulttreeofPaperIIispresented.Thisfaulttreecontains thesamepotentiallyhazardousactivitiesas theprevious fault tree.Eachactivity is again described in the form of frequency of occurrence and probability ofopening.However,themeansofcalculatingtheprobabilityofopeningdiffers(Figure13).The integration of the effect of site‐ and wreck‐specific indicators on the genericprobabilityofopeningisnowmadebyformalBayesianupdating.Furthermore,thesimpleexpertestimationsregardingagenericprobabilityofopeninginthefirstfaulttreeversionwerereplacedbyinformationacquiredduringanexpertelicitationworkshop(PaperIII).Secondly, the effect of site‐specific andwreck‐specific indicators on the probability ofopeningwasalsoestimatedbyexpertsattheabove‐mentionedworkshop.Thiswas incontrasttoPaperII,wheretheauthorsmadeinitialestimations.Itshouldalsobenotedthattheinitialestimationsweremadeforanothertypeofstructurewheretheeffectofindicatorsweredisplayedasdeviationfroma“normal”wreck‐status.

H.Landquist

40

Figure13.ThenovelfaulttreemodelforVRAKAaccordingtoPaperIV.

Foreachactivity,theeffectoftheindicatorswasevaluatedanddescribedbymeansofuncertaintydistributions.Anassessorcanassignvaluesforindicatorsatthespecificsiteandwreckandonlythepartofthedistributionestimatedbyexpertsthatisvalidforthewreck inquestion is applied in themodel.Thepriorprobability of openingdue to anactivity, ,isthusupdatedbasedonthestatusofthesite‐specificandwreck‐specific

parameters to a posterior estimation , … , . This facilitates amathematically

correctaggregationoftheprobabilityofopeningandindicatorsconcerningthespecificsiteandwreckandisenabledbytheexpertelicitation(PaperIII).Themodelhasthusbeen refined compared to Paper II and nowprovides probabilistic Bayesian updatingincorporatinguncertaintiesthroughtheapplicationofMonteCarlosimulation.


41

Thefrequencyofhazardousactivitiesisassignedonacontinuousscaleratherthanasachoiceofpre‐setsuggestionsaswasthecaseinPaperII.Theassessorestimatesthelowestreasonableandhighestreasonablenumberofoccurrencesperyearfortheactivities,thatcanphysicallyaffectthewreck.Thisisinterpretedasthe10thand90thpercentileofagammadistribution. The probability that hazardous substances are still contained in the wreck and theprobabilitythatthecargoishazardousandisstillcontainedinthewreckisestimatedinthe same way as in Paper II. However, in Paper IV, the approach for modelling theuncertaintyoftheseparametersisaltered.Theprobabilityoftheestimatedparameterismodelled using a beta distribution where the sum of the shape parameters α and βrepresentstheuncertaintyortotal informationavailable.Sumsareassignedtoclasses,whereahighlevelofuncertaintyisrepresentedbyalowersumofαandβandviceversa.ClassesofuncertaintyfromImpossibletoCertainaswellasNoinformationareavailabletotheassessor.

6.1.3. Expertelicitation–PaperIIIAspointedoutintheconclusionsinPaperII,theinputfromexpertstooktheformofroughestimationsandtherewasanapparentneedformorerobustexpertelicitationinordertoobtaininputfortheVRAKAmodel.AworkshopatwhichmorethantwentyexpertswerebroughttogetherwasheldtoretrievenecessaryinputinformationfortheVRAKAmodel.TheworkshopwasbasedontheSheffieldElicitationFramework(SHELF),asdescribedinSection5.2(O'HaganandOakley,2010).Groupsforelicitationwereformedbasedontheexperts’knowledge.Expertswereaskedto estimate the probability of a specific activity causing an opening in a shipwreck.Furthermore, estimations were made regarding the state of site‐specific and wreck‐specificindicators,giventhatopeningandnoopeninghadoccurredduetoaparticularactivity.ThelatterestimationsareappliedinVRAKAtoupdatethegenericprobabilitiesofopeningduetodifferentactivitiesandtoestimatethejointprobabilityofopeningforthewreck in question. Theworkshop also resulted in two indicators being removed,MaintenanceandShipConstruction,andsomecombinationsof indicatorsandactivitieswerefoundtobeirrelevant.InTable3anoverviewisprovidedofwhichindicatorswereassumedtohaveaninfluenceonwhichactivities.

H.Landquist

42

Table3.Summaryofelicitedinfluenceornoinfluenceofindicatorsonactivities.Influenceisshownas“Y”withalightgreybackgroundandnoinfluenceas“N”withadarkgreybackground(PaperIII).

Theworkshopresulted in92probabilitydistributions,representingtheuncertaintyofthestateofindicatorsinrelationtotheprobabilityofopeningornoopeningasaresultofeachofthehazardousactivities.Furthermore,eightdistributionswereelicited,describingthegenericprobabilityofopeningduetoeachoftheactivities.ExamplesofdistributionsarepresentedinFigure14,thegenericprobabilityofopeningduetoConstructionwork,andinFigure15,theuncertaintydistributionforHullthicknessifanopeninghasoccurredduetoshippingtraffic.

Figure14.Theprobabilityofopeningduetoconstructionwork.

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Beta: = 1 = 1.43

Construction

25th percentile50th percentile75th percentile

Summaryofcollecteddataregardingindicators(Datawasonlyelicitedwhereexpertsfoundtherewasaconnection). Construct‐

ionDiving Military

activityDeterior‐ation

Shippingtraffic

Unstableseabed

Storms Trawling

Averagebottomwateroxygenconcentration

N N N Y N N N N

Averagebottomwatersalinity

N N N Y N N N N

Averagebottomwatertemperature

N N N Y N N N N

Averagebottomwatercurrentspeed

Y Y N Y N N N N

Shipuse N N N Y Y Y Y NAveragehullthickness Y Y Y Y Y Y Y YSeabedcharacter Y N Y Y Y Y Y YWreckpositiononseabed

Y N Y Y Y Y Y Y

Depth Y Y N N Y Y Y YTimesincesinking Y Y Y Y Y Y Y Y


43

Figure15.UncertaintydistributionfortheHullthicknessindicator,giventhatanopeninghasoccurredduetoShippingtraffic.

6.2. Part2–Consequenceassessment–PaperV

Estimation of consequences in VRAKA can be performed in three tiers depending onavailableresources,dataanddemandsregardingtheresults.Intier1,riskisdescribedastheexpectedvolumeofoil andpotentialhazardous cargodischarged.Tier2appliesamatrix combining volume, distance to shore and shoreline sensitivity to estimate theconsequence(Table4).Riskisdescribedastheoutcomeofthematrixcombinedwiththeprobabilityofdischarge.Tier3isanaggregationofresultsfromSeaTrackWebforoilspilltrajectory modelling and the Digital Environmental Atlas containing information onshorelinesensitivitytooilspill.Table4.Tier2ConsequenceassessmentinVRAKA.BasedonKulanderetal.(2010)andtheDEEPPproject(Alcaroetal.,2007)accordingtoPaperIII.

Low severity

Moderate severity

High severity

Volume < 10 m3 10 – 500 m3 > 500 m3

Distance to shore > 10 nautical miles 1 – 10 nautical miles < 1 nautical mile

Sensitivity Nearest shore is: Sandy, steep cliffs or rock walls or facilities.

Nearest shore is: Cliff beaches, pebble, boulder or gravel beaches.

Nearest shore is: Reed beds, meadows, fine sediment beaches. or mixed beaches

Having theopportunity to apply consequenceassessment in three tiers allows for thepossibilityofadaptingtheassessmenttothelevelofambitionandresourcesavailable.

0 2 4 6 8 10 12 14

Gamma: k = 0.91 = 0.81

Hull thickness - Shipping traffic - Opening

25th percentile50th percentile75th percentile

H.Landquist

44

However,itshouldbenotedthattheselectionofatierwillhaveanimpactonthetypeofresultsthatcanbeobtainedandhowthesecanorshouldbeused.If,forexample,tier1isapplied,theconsequenceswilldependsolelyontheexpectedvolumeofoildischarged.Applyingaspecifictierthusinvolvesmakingadecisionregardingwhattheresultswillencompass.This reasoning also brings us to a fundamental aspect of risk assessment andmanagement. Developing models and performing risk assessment involves makingdecisionsandchoices.Evenwiththeaimofkeepingthisprocessfreefrombiasandleavingdecisions to the decision‐makers and later steps in the riskmanagement process, thedevelopedmodelsandtoolswill inevitablybecolouredbythedevelopers’backgroundandassumptions.

6.3. Part3–RiskEvaluationinVRAKA

Results from VRAKA can be used to compare wrecks and for deciding which wreckswarrantaction.Theestimatedriskcanbecomparedasameasureitself(tier1)orthetwocomponents – probability of discharge and the consequences – can be comparedseparately(alltiers).Ifseveralwrecksarewithinaclosedistance,suchgroupscanalsobecomparedtoothergroupssincecombinedprobabilitiesandconsequencescanproduceotherresults.Furthermore,remediationcostscouldperhapsbeloweredbycoordinatingoperations.Inthecaseofriskevaluation,i.e.makingrisktolerabilitydecisionsandanalysingoptions,nospecificcomponentofVRAKAisapplicable.Decidingonriskcriteriaandwhatriskistolerableisuptotheassessorordecision‐maker.However,themainbenefitofVRAKAisthat it can be applied to estimate the relative risk between wrecks and prioritiesmitigationmeasurestoevaluatetheriskreductionoptions.Insimpleterms,riskreductioncanbeachievedbyeitherdecreasingtheprobabilityofanactivity or reducing the consequences. VRAKA can for example be used to analyse achangeintheprobabilityofdischargeofhazardoussubstancesbyreducingthefrequencyofahazardousactivityandcomparing theriskrelated to thecurrent frequencyof theactivitywithareducedfrequency.

6.4. AtoolforapplyingVRAKA

VRAKAhas alsobeen realised in the formof anExcel‐based tool. It includes two filescontaininganumberofsheetsthatguidetheassessorthroughtheriskestimation(Figure16). Hereafter followsabriefoverviewof the tool.Furtherdescription isprovided inAppendixA1.


45

Figure16.SchematicviewofthetoolforapplyingVRAKA(VRAKA,Probabilisticriskassessmentofshipwrecks,Handbook.VersionOct.2016).

Figure 17 shows the structure of VRAKA in Excel. The purpose of a more detailedpresentationistoclarifyhowuncertaintiesareincludedintheriskestimation.VRAKAisseparatedintotwofilesduetopracticalitiesrelatingtoCrystalBall.Thecalculationsinfileoneshouldbeperformedbeforethecalculationsinfiletwo.Theformerservesasinputtothelatterandthuscannotbeperformedsimultaneously.File1containsinformationfromtheexpertelicitationregardingsite‐specificandwreck‐specificparameters.ThisisupdatedbasedonthevaluesforthewreckinquestionassignedbytheuserofVRAKA.Inthe second file, all the remaining calculations and simulations are performed. Theconsequenceofthisseparationisthatsensitivityanalysiswillneedtobeperformedintwosteps.

H.Landquist

46

Figure17.ThestructureofVRAKAintheExcel‐basedtool.CBisanabbreviationofCrystalBall,anadd‐insoftwareforMonteCarlosimulations.

47

7. Discussion

Potentially polluting shipwrecks are a growing environmental threat and adequatedecision support is needed to cope with the problem. The risk from shipwrecks isinfluencedbynumerousanduncertainvariablesandtheinteractionofthesevariablesisatrulycomplexmatter.Dataaretypicallylackingandgatheringfurtherinformationcouldbecostlyor,forvariousotherreasons,difficulttoachieve.Handlingoftheseuncertaintiesisnecessaryinordertoobtainproperriskassessmentresultsanddecisionsupport.Thisthesisinitiallyidentifiedaneedforfurtherresearchonriskassessmentofshipwrecks.Aframeworkforstructuringandguidingdevelopmentaswellasriskmanagementwasthensuggested and amodel for risk assessment of wrecks (VRAKA)was developed as anintegralpartoftheframework.VRAKAcanbeappliedtoprioritisemitigationmeasuresandresourceallocationforpotentiallypollutingshipwrecksbasedonscientificallywell‐foundedinputacquiredthroughprobabilisticriskassessment.

7.1. Agenericframework

The framework was developed based on a comparison of identified methods andapproachescurrentlyappliedtoshipwrecks(PaperI).Itwasshownthatatthetimeofthestudy therewasno comprehensivemethod for shipwrecks andnone that facilitatedaquantitativeassessment.TheanalysiswasfurtherdevelopedinChapter3ofthisthesis,which include additional, recently developed approaches. It was concluded thatconsequenceassessmentisnowingeneralmoresolidbutthereisnofullyprobabilisticapproach that takesuncertainties intoaccount.The suggested framework stresses thelinkbetweenriskassessmentanddecision‐makingandpresentsaholisticstructurebasedonwell‐establishedviewsofriskmanagement.Hence,theframeworkarguesinfavourofaproactiveapproach.

7.2. VRAKA–acomprehensivemodel

Due to the complex nature of the risks posed by shipwrecks and the inherentuncertainties,itiscrucialtoapplyastructuredandprobabilisticapproach.TheVRAKAmodelthereforecomprisesaprobabilisticmethod,withexplicithandlingofuncertainties,forestimatingtheprobabilityofdischargeandanaggregationofmethodsforestimatingtheconsequences.

7.2.1. AfaulttreemethodSeveral methods for logical modelling exists. The fault three was chosen due to itspossibility to provide structure and transparency to the problem. Furthermore, the

H.Landquist

48

mathematical framework suits the conceptualmodel of VRAKA and probabilistic riskassessmentofshipwreckswell.Applyingaprobabilisticapproachallowsuncertaintiestobetakenintoaccountininputdataandresults.InVRAKA,afaulttreewasdevelopedandappliedtocalculatetheprobabilityofdischargeofhazardoussubstances.Thefaulttreecombinestheprobabilityofanopeningoccurringinthewreckwiththeprobabilitythat there ishazardousmaterial insidethewreck.Statisticaldistributionsdescribetheuncertainties of input data and through the use of Monte Carlo simulation theuncertaintiesoftheresultscanbecalculatedanddisplayed.Aprobabilisticapproachalsoallowsforsensitivityanduncertaintyanalysis.Uncertaintyanalysis facilitates an explicit representation of uncertainties in input data and theireffectsonthemodeloutputs.Sensitivityanalysismakesitpossibletoidentifywhichinputvariables contribute most to output uncertainty, and thus provides guidance ondeterminingwhichadditionalinformationshouldbeprioritisedinordertoachievemorereliableriskcalculations.TheuserofVRAKAcanalsorundifferentscenariosinordertoseehowtheriskisaffectedandthusevaluatepossiblemeasureforreducingtherisk.

7.2.2. ConsequenceassessmentDependingonthepurposeoftheriskassessmentandavailableresources,VRAKAmakespossible theselectionofa relevant tier formakinga consequenceassessment.VRAKAoffersthreetiersofincreasingrefinement:afirsttierwheretheamountofoilservesasaproxyforconsequences,asecondtierwhereasensitivitymatrixisappliedtodescribetheconsequences(thematrixcontainsthevolumeofoil,distancetoshoreandthesensitivityofthespecificshoretype,i.e.asemi‐quantitativeconsequenceassessment.),andathirdtierwhereanoilspilltrajectorytooliscombinedwithsensitivitymappingofthecoast.Uncertainties canbe taken into account to someextent in the latter two since the setvaluesinthesensitivitymatrixareprovidedasanintervalandtheoilspilltrajectorytoolcan be run for several scenarios. Currently, none of the tiers provided offers aconsequenceassessmentfromasocialoreconomicpointofview,whichareofcoursearecrucialareaswhenmakinganassessmenttofindsustainablesolutions.Allthetierscanhoweveratpresentofferanestimationoftheconsequencesoffapotentialdischargeofoilthatcanserveasinputfordecision‐making.TheconsequenceassessmentofVRAKAcanbeappliedtocompareshipwrecksindependentofselectionoftierpossible.

7.2.3. RiskestimationAmajorfocusofthisworkwastodevelopamodeltoestimatetheprobabilityofdischargeof oil and to illustrate how this estimation can be combined with different types ofconsequence assessments. Hence, to estimate the risk posed by potentially pollutingshipwrecks.Inthisthesisriskisdefinedasacombinationofprobabilityandconsequences.However,differentmeansforcombiningprobabilityandconsequencesareappliedwhenestimating

7.Discussion

49

thelevelofrisk.Thechoiceofhowtopresenttheriskwilldependontheaimsoftheriskassessment.Riskcan,asintier1,beestimatedastheexpectedvalueofthedischargedvolumeofhazardoussubstance.Usingthisapproach,theusermustbeawareofthefactthatverysignificantconsequencescanbeobscuredbyasmallprobabilityorviceversaasmentionedinChapter2.Itisthusimportanttobeawareofthelimitationsinthiswayofexpressing the risk. A separate analysis of the probability, the consequences, and thecombinedriskestimatewillprovideamoreholisticpictureoftherisk.Tiers2and3inherentlyencouragetheassessortoevaluatetheprobabilityofdischargeand the consequences separately by not providing actual numbers representing theconsequences.Theassessorthusweighstheestimationsofconsequencesinrelationtotheprobabilityofdischarge.Thechoiceoftierfortheconsequenceassessmentwillalsoinvolveselectingthetypesofconsequencestobeincluded.Anassessormustalwaysbeawareofandtakeintoaccounthowthechoiceofriskdefinitionandthedescriptionoftheconsequences will affect the outcome of the risk estimation when making the finaldecision.

7.2.4. RiskevaluationTherearenoregulationsorcriteriaspecifyingwhenawreck isa threat to themarineenvironment to theextent thatmeasuresshouldbe taken,unlessa largerdischarge isunderway. The role of the decision‐maker is therefore very important and the riskevaluation and comparison of wrecks imperative for supporting decision regardingmitigationmeasures.VRAKAcanbeappliedtoprioritisemitigationmeasuresbyanalysisof potential risk‐mitigation options. For example, by analysing the effect on results ofdecreasingorbanningshippingtrafficinthearea.Riskmitigationmeasuresintermsofreducedrisklevelcouldalsobeevaluatedinrelationto the cost for suchmitigation using e.g. cost‐benefit analysis, thus implying that riskmitigationmustalsobevaluedineconomicterms.Thecostofachievingspecificrisklevelscanalsobeevaluatedbycosteffectivenessanalysis,inordertofindwhichalternativethatwouldreachaspecificriskcriterion,orgoal,tothelowestcost.Anotherexampleismulticriteriadecisionanalysis,whichtakesanumberofcriteriaintoaccount.It is theresponsibilityof thedecision‐makerto,basedone.g. informationretrievedbyVRAKA,decidewhatthemostdesirableoptionis.Forexample,thelargestriskreductiontothelowestcostorsimplyareductionofthelargestrisk?Thereareseveralalternativesthatarerelevantfromdifferentperspectives.Although,VRAKAcanprovidesupportandinputforadecision‐makingprocessindifferentways,itisuptothedecisionmakertosetthegoalsfortheassessmentandchoosetheproperway,e.g.therelevanttier,forthetypeofdecisionathand.

H.Landquist

50

7.3. Expertelicitation

Informationisscarceonshipwreckconditionsanddataderivedbyexpertelicitationisessentialforprobabilisticriskassessmentofshipwrecks.InputforthefaulttreeinVRAKAwasderivedbymeansofarigorousexpertelicitationprocesswhereexpertswereaskedtoprovidetheirsubjectivebeliefsaboutrelevantquantitiessuchastheprobabilityofanopeninginashipwreckdutohazardousactivitiesandtheinfluenceofsite‐specificandwreck‐specificindicatorsonthatprobability.Atotalof23expertsfromvariousareasofexpertiseparticipatedintheelicitation.Anidealsituation would have involved more experts to embody a wider common level ofcompetenceinallthegroups.Certainexpertswereincludedinseveralgroupsforsomeoftheelicitations,butarrangingforallareasofknowledgetoberepresentedforallissueswasnotpossiblewithrespecttothetimeavailable.Althougheffortsweremadetoinvitealargenumberofexperts,itwasnotpossibleatthetimeandforthechosenstructureoftheelicitationtoattractfurtherexpertise.Analternativecouldhavebeentorunseveralworkshopswithfewerexpertsoneachoccasion.Probabilisticriskassessmentofshipwreckswouldbeextremelydifficultifnotimpossibleifexpertelicitationwasnotperformed.Althoughthetaskwasdifficult,itisimportanttobearinmindthatexpertknowledgeisnotexpressedintermsofprecisevaluesandthatuncertaintieswerealsotakenintoaccount.

7.4. ValidationofVRAKA

ValidationinabsolutetermsofamodelsuchasVRAKAisnotfeasiblesincetheprocessmodelledcannotbecontrolledandisnotpossibletofollowup.Suchactionwouldincludefindingalargenumberofwrecksandstudythemforaverylongperiodoftimeinorderto obtain statistically valid results. However, what can be done is to test real andhypotheticalwrecksandtoanalysechangesininputdatainrelationtochangesinresults.VRAKAhasbeentestedonseveralhypotheticalandrealshipwrecks.ResultsshowhowVRAKAcanbeappliedtoestimatetheriskfromshipwrecksintermsoftypesofresultsandhowuncertaintyandsensitivityanalysiscanbeperformed.

7.5. Outlook

The problem of potentially hazardous shipwrecks is complex. Numerous authorities,companies,universitiesandexpertsareinvolved.TheSwedishcoastguardwillactifoilisfound on beaches or on thewater surface, the SwedishMaritime Administrationwillremovewrecks if they pose a threat to shipping trafficwhile the SwedishAgency forMarine and Water Management has overall responsibility. For a risk managementprocess to be efficient, these organisations will have to cooperate. Data for risk

7.Discussion

51

assessmentwillhavetobesharedaswellastherequisiteknowledgeandcompetenceforan effective risk reduction and control. The stakeholders involved might also havedifferentinterestsandviewsaboutwhenariskisunacceptableandhowitshouldbedealtwith.VRAKAhasbeenoperationalisedinanExcel‐basedtool to facilitateriskassessmentofshipwrecksfortheauthorityresponsible.UsingVRAKArequiresknowledgeofwheretofindtheinformationrequiredinthetool,whichiswhyaninterdisciplinaryapproachisimportant. It could be argued that “the assessor” rather should be a team in order tofacilitate assessment since many different aspects need to be considered. It is alsoessentialtounderstandtheconnectionbetweenthequalityandtypeof inputdata,thechoiceoftierforconsequenceassessment,i.e.howriskisexpressed,andwhattheresultswill includeandrepresent.Resultsshouldneverbeconsideredasmerelynumbersbutratheranalysedbearinginmindtheprocessoftheassessment.Resultswillalsoneedtobeweighedagainstother information inorder tobeable to fullyweight costsagainstbenefitsofmitigationmeasuresandtomakeaninformeddecision.Manyassumptionshavebeenmadeduringthecourseofdevelopment.AsstatedinSection6.2, models and tools cannot be entirely free of bias and the influence of the modelconstructors.Tocopewiththisnotion,severalexpertshavebeeninvolvedintheprocessatnumerousoccasions.Areferencegrouphasbeeninplacethroughouttheprocessandexperts groups varying in size have contributed informally and formally to thedevelopment.VRAKAhasbeendeveloped in cooperationwith, among, others the SwedishMaritimeAdministrationandtheSwedishAgencyforMarineandWaterManagement.DiscussionsareabouttogetunderwayregardingtheuseofVRAKAforriskestimationonanationallevel.

7.6. Suggestionsforfurtherresearch.

Basedontheworkcarriedoutwithinthisthesis,thefollowingareidentifiedaspossibleareasforfurtherresearch:

Theestimationofconsequencesintiers2and3issemi‐quantitative.Tofurther

encompassassociateduncertainties,aprobabilisticapproachcouldbeadopted.Anexample of such a method is SIMAP (McCay et al., 2004), which providesprobabilistic estimates of consequences. Initial studies linking VRAKA to theSIMAPmethodhasbeenperformed(Etkinetal.,Inprep).

The Swedish Meteorological and Hydrological Institute is developing SeatrackWebforbroaderanalysiswheretheassessorcanmodelaspillreleasedonseveraloccasionsandobtainprobabilitymapsofwheretheoilmightbedeposited.Thisis

H.Landquist

52

anotherpossibilityforprobabilisticassessmentofconsequencesinfutureversionsofVRAKA.

An interesting development of VRAKA would be to refine the estimation ofenvironmentalconsequencesandalsoprovidemeansforanalysingfurtheraspectssuchastheseasonduringwhichthespilloccurredorspecificendangeredanimalspotentially affected by a spill. It would also be profitable to include furthersocioeconomicaspectsintheconsequenceassessment.

In general, themain focus of oil spills is on large acute spills. Small andmore

continuousspillsareacknowledgedlessfrequently.However,ithasbeenshownbye.g.Lindgrenetal.(2012)thatsmallinputs,resultinginlowconcentrationsofoil, also have a negative effect on microbial and meiofaunal communities.Improving thepossibility of estimating consequencesof small discharges couldthereforebeconsideredinafuturedevelopmentofVRAKA.

AtpresentVRAKAcanbeappliedtoevaluatetherelativerisksfromshipwrecks.ApossiblefurtherdevelopmentcouldbetolinkVRAKAtotoolsformoreadvancedriskevaluation.Forexample,performingacost‐benefitanalysis candisplay thebenefitofmitigatingtheriskbyprovidinginformationonthesocietalprofitability(netbenefit)of riskmitigationmeasures.Multi criteriadecisionanalysiswouldfacilitateamorecomprehensivesustainabilityanalysis,takingintoaccountlocalsocial,environmentalandeconomiceffectsthataredifficultornotpossibletotakeintoaccountincost‐benefitanalysis.

Model development and risk assessment of potentially polluting shipwrecks isongoinginseveralothercountries.Therearemanypossibilitiesforcollaboration.ExpertelicitationwasperformedwithafocusontheSwedishwreckpopulation.However, the structure of theVRAKAmodel is generic and it could be appliedinternationally.

Themethod and tool for probabilistic risk estimation of shipwrecks could, forvalidationpurposes,be further testedandcalibrated.Therearemore than300wrecksinScandinavianwatersthattheSwedishMaritimeAdministration(2011)suggestsmayposeathreattothemarineenvironment.Aselectionofwell‐knownwrecksforwhichin‐situinspectioncanbeperformedwouldbesuitabletouseinordertotestandvalidatethemodel(seee.g.Hassellövetal.,2014;Hassellövetal.,2015).

8.Conclusions

53

8. Conclusions

Thefollowingmainconclusionshavebeenreachedfromthisthesiswork:

Thereisaneedforacomprehensiveriskassessmentapproachforshipwrecks. Thereisalackofmethodsforcomprehensiveandquantitativeriskassessmentofshipwrecks.Specifically,methodsforestimatingtheprobabilityofdischargeandforincludinguncertaintyanalysisoftheriskestimationarelacking.

Aframeworkforriskmanagementofshipwrecksissuggested.ThisthesispresentsagenericframeworkforriskmanagementofshipwrecksinaccordancewiththeISOstandardforriskmanagementthatoffersastructureandsupport for developing risk assessment models, and provides guidance forshipwreck‐relatedriskmanagement.

Novelinputdataisderivedforriskassessmentofshipwrecks. Expert elicitation was used to address the intrinsic uncertainties of riskassessment of shipwrecks. The information obtained has not previously beenavailableandisaprerequisiteforquantitativeriskassessmentofshipwrecks.Theinformationprovidesanessentialfoundationthatcanbeupdatedusingsite‐andwreck‐specific preconditions as a basis. It is now possible to estimate theprobabilityofdischargefrompotentiallypollutingshipwrecksandtoincorporateuncertaintiesthatinfluencetherisk.

VRAKA–amodelforcomprehensiveriskassessmentofshipwrecks.Thedevelopedmodeltakesintoaccountboththeprobabilityofdischargeanditspotential consequences. Risk evaluation can also be performed based on theresultsfromVRAKA.Theriskassessmentmodelisimplementedinaspreadsheettool(MicrosoftExcel)tofacilitateapplication.o Amethodforestimatingtheprobabilityofhazardousdischarge.

The method enables estimation of the annual probability of discharge ofhazardous substances from shipwrecks. Bayesian updating of a genericprobabilityofopeninginawreckisperformedbasedoninfluenceofwreck‐and site‐specific indicators. The Bayesian approach enables a formalintegration of expert judgement regarding potential hazards of differentactivitieswithavailabledataontheoccurrenceandintensity(frequency)ofsuch activities. The logical structure of the method and the type of input

H.Landquist

54

informationcombinedwithMonteCarlosimulationsenablesuncertaintyandsensitivityanalysestobemadeoftheresults.

o Anapproachforestimatingtheconsequencesofapotentialdischarge. In VRAKA, consequences can be estimated in three tiers depending on availableresourcesand,dataandtheaimoftheriskassessment.

ApplicationandvalidationofVRAKA.VRAKAhasbeenappliedtoanumberofhypotheticalandrealshipwrecks.Thisstudyprovidesimportantknowledgeconcerningchangesofresultsinrelationtoinput data also coupled to the uncertainty distributions of input data. Theapplicationshowedthatresultsoftheprobabilityofopeningspanningoveratleasttwoordersofmagnitudeispossible.

VRAKA is a comprehensive, structured, quantitative risk assessment model forshipwrecksthattakesuncertaintiesintoaccount.Fortheveryfirsttime,theprobabilityofdischargecanbequantifiedandcombinedwiththeconsequencesofsuchadischargeto formaquantitative risk estimation for aparticular shipwreck.Anumberof factorsaffecting the risk posed by shipwrecks have been identified and are included in theassessment.Thestructuredquantitativemodelalsoimpliesthatriskassessmentcanbeperformedconsistentlyforallthewrecksstudied.

The framework and model developed facilitates adequate and transparent riskassessmentforshipwrecks,thusprovidingsupportforreachingwell‐informeddecisionsregardingprioritisationandmitigationmeasuresforpotentiallypollutingwrecks.Thisinturn facilitates more cost efficient riskmitigation and efficient resource allocation tocounterthisenvironmentalthreat.

55

9. References

ALCARO,L.,AMATO,E.,CABIOCH,F.,FARCHI,C.&GOURIOU,V.2007.DEEPPProject,DEvelopmentofEuropeanguidelinesforPotentiallyPollutingshipwrecks.ICRAM,InstitutoCentraleperlaRicercascientificaetecnologicaApplicataalMare,CEDRE,CEntredeDocumentationdeRecherchéetd'Epérimentationssurlespollutionsaccidentellesdeseaux.

AVEN,T.2003.Foundationsofriskanalysis,aknowledgeanddecision‐orientedperspective,WestSussex,England,JohnWiley&SonsLtd.

AVEN,T.2010.Onhowtodefine,understandanddescriberisk.ReliabilityEngineering&SystemSafety,95,623‐631.

AVEN,T.2012.FoundationsofRiskAnalysis.Wiley.ProQuestEbookCentral[2september2016].

AVEN,T.,BEN‐HAIM,Y.,ANDERSENBOJE,H.,COX,T.,LÓPEZDROGUETT,E.,GUIKEMA,S.,KROEGER,W.,RENN,O.,THOMPSON,K.M.&ZIO,E.2015.SRAglossary.TheSocietyForRiskAnalysis.

AVEN,T.&RENN,O.2009.Onriskdefinedasaneventwheretheoutcomeisuncertain.JournalofRiskResearch,12,1‐11.

AYYUB,B.2001.ElicitationofExpertOpinionsforUncertaintyandRisks,TheUnitedStatesofAmerica,CRCPressLCC.

BEDFORD,T.&COOKE,R.2001.Probabilisticriskanalysis.Foundationsandmethods,UnitedStatesofAmerica,CambridgeUniversityPress.

BERNSTEIN,P.L.1996.Againstthegods.Theremarkablestoryofrisk.,USA,JohnWileyandsons,INC.

BURGMAN,M.2005.Riskanddecisionsforconservationandenvironmentalmanagement,NewYork,Cambridgeuniversitypress.

COOKE,R.M.1991.ExpertsinUncertainty:OpinionandSubjectiveProbabilityinScience,OxfordUniversityPress.

COOKE,R.M.&GOOSSENS,L.H.J.2000.Proceduresguideforstructuredexpertjudgementinaccidentconsequencemodelling.RadiationProtectionDosimetry,90,303‐309.

COUNTYADMINISTRATIVEBOARDOFVÄSTRAGÖTALAND.2015.DigitalMiljöatlas[Online].Available:http://www.lansstyrelsen.se/vastragotaland/Sv/manniska‐och‐samhalle/krisberedskap/skydd‐mot‐olyckor/Pages/digital‐miljoatals.aspx[Accessed30October2015].

56

ETKIN,D.,FRENCHMCCAY,D.,HORN,M.,LANDQUIST,H.,HASSELLÖV,I.M.&WOLFORD,A.J.Inprep.QuantificationofOilSpillRisk.In:FINGAS,M.(ed.)OilSpillScienceandTechnology.Burlington,USA:Elsevier.

ETKIN,D.,VANROOIJ,H.&FRENCH‐MCCAY,D.2009.Riskassessmentmodelingapproachfortheprioritizationofoilremovaloperationsfromsunkenwrecks.Effectsofoilonwildlife.Tallin,Estonia.

FEJES,J.,COLE,S.&HASSELSTRÖM,L.2011.TheREMEDEProject:Ausefulframeworkforassessingnon‐marketdamagesfromoilspills?:DepartmentofEconomics,UmeåUniversitet.

GARTHWAITE,P.H.,KADANE,J.B.&O'HAGAN,A.2005.Statisticalmethodsforelicitingprobabilitydistributions.JournaloftheAmericanStatisticalAssociation,100,680‐700.

GLICKEN,J.2000.Gettingstakeholderparticipation‘right’:adiscussionofparticipatoryprocessesandpossiblepitfalls.EnvironmentalScience&Policy,3,305‐310.

HACKING,I.2001.Anintroductiontoprobabilityandinductivelogic.,TheUnitedStatesofAmerica,CambridgeUniversityPress.

HANLEY,N.&BARBIER,E.B.2009.Pricingnature.Cost‐BenefitAnalysisandEnvironmentalPolicy.,UK,EdwardElgarPublishingLimited.

HASSELLÖV,I.M.,OLSSON,U.,EKBERG,G.,ÖSTIN,A.,SIMONSSON,F.,LARSSON,C.,LANDQUIST,H.,LINDGREN,J.F.,LINDHE,A.,ROSÉN,L.&TENGBERG,A.2015.EnvironmentalrisksposedbyshipwrecksII.InSwedish:MiljöriskersjunknavrakII.Undersökningsmetoderochmiljöaspekter..Sjöfartsverket.

HASSELLÖV,I.M.,OLSSON,U.,EKBERG,G.,ÖSTIN,A.,SIMONSSON,F.,LARSSON,C.,SENDER,U.,LANDQUIST,H.,LINDGREN,J.F.,LINDHE,A.&ROSÉN,L.2014.Environmentalrisksposedbyshipwrecks.InSwedish:Miljöriskersjunknavrak.Undersökningsmetoderochmiljöaspekter..In:ADMINISTRATION).,S.T.S.M.(ed.).

HORA,S.C.&VONWINTERFELDT,D.1997.Nuclearwasteandfuturesocieties:Alookintothedeepfuture.TechnologicalForecastingandSocialChange,56,155‐170.

IDAAS,K.2005.NorwegianPollutioncontrolauthorityworkonshipwreck,.2005InternationalOilSpillConference,IOSC2005.MiamiBeach,FL,.

IEC1995.DependabilityManagement‐Part3:Applicationguide‐Section9:Riskanalysisoftechnologicalsystems.InternationalElectrotechnicalCommission.

INTERNATIONALMARITIMEORGANISATION2002.GuidelinesforFormalSadetyAssessment(FSA)foruseintheIMOrule‐makingprocess.London.

INTERNATIONALMARITIMEORGANIZATION2007.NairobiInternationalConventionontheRemvalofWrecks.In:INTERNATIONALMARITIMEORGANIZATION(ed.)LEG/CONF.16/19.

ISO2009.Riskmanagement‐Principlesandguidelines,ISO31000:2009(E).Geneva:InternationalOrganizationforStandardization.

57

JENKINSON,D.2005.Theelicitationofprobabilities:Areviewofthestatisticalliterature.Citeseer.

JONES‐LEE,M.&AVEN,T.2011.ALARP—Whatdoesitreallymean?ReliabilityEngineering&SystemSafety,96,877‐882.

KAPLAN,S.&GARRICK,J.1981.Onthequantitativedefinitionofrisk.RiskAnalysis,Vol.1.

KINGSTON,P.F.2002.Long‐termenvironmentalimpactofoilspills.Spillscience&Technologybulletin,7,53‐61.

KIUREGHIAN,A.D.&DITLEVSEN,O.2009.Aleatoryorepistemic?Doesitmatter?StructuralSafety,31,105‐112.

KUHNERT,P.M.,MARTIN,T.G.&GRIFFITHS,S.P.2010.AguidetoelicitingandusingexpertknowledgeinBayesianecologicalmodels.EcologyLetters,13,900‐914.

KULANDER,K.E.,ERICSSON,M.,TEGENACK,A.,FEJES,J.&EVANS,S.2010.Saneringsmanualföroljapåsvenskastränder.

LARSSON,B.‐Å.,EKBERG,G.,SÖDERBERG,M.&JESÄTER,M.2011.Miljöriskerfrånfartygsvrak.TheSwedishMaritimeAdministration,.

LINDGREN,J.F.2015.Evaluatingeffectsoflowconcentrationsofoilinmarinebenthiccommunities.Doctoralthesis,ChalmersUniversityofTechnology.

LINDGREN,J.F.,HASSELLÖV,I.M.&DAHLLÖF,I.2012.Meiofaunalandbacterialcommunityresponsetodieseladditionsinamicrocosmstudy.MarPollBull.,595‐601.

LIUNGMAN,O.&MATTSON,J.2011.ScientificDocumentationofSeatrackWeb;physicalprocesses,algorithmsandreferences.In:SMHI(ed.).

LOUZIS,K.,KOIMTZOGLOU,A.&VENTIKOS,N.P.TheRiskofWrecksortheWreckofRisk?TheGreekParadigm.2ndInternationalConferenceonRiskAnalysisandCrisisReponse,2009Beijing,China.AtlantisPress.

MCCAY,D.F.,ROWE,J.J.,WHITTIER,N.,SANKARANARAYANAN,S.&ETKIN,D.S.2004.Estimationofpotentialimpactsandnaturalresourcedamagesofoil.JournalofHazardousMaterials,107,11‐25.

METZ,B.,DAVIDSON,O.,BOSCH,P.,DAVE,R.&MEYER,L.2007.Mitigationofclimatechange.IPCC.

MEYER,M.&BOOKER,J.2001.ElicitingandAnalyzingExpertJudgment,SocietyforIndustrialandAppliedMathematics.

MICHEL,J.,GILBERT,T.,WALDRON,J.,BLOCKSIDGE,C.T.,SCHMIDTETKIN,D.&URBAN,R.2005.Potentiallypollutingwrecksinmarinewaters.Internationaloilspillconference.

MORGAN,G.M.&HENRION,M.1990.Uncertainty.Aguidetodealingwithuncertaintyinquantiativeriskandpolicyanalysis.,TheUnitedStatesofAmerica,Cambridgeuniversitypress.

58

MUNIER,N.2004.MulticriteriaEnvironmentalAssessmen.Apracticalguide.,Dordrecht,KluwerAcademicPublishers.

NOAA2009.WreckOilRemovalProgram(WORP)2009DemonstrationProjectOverview.NationalOceanicandAthmosphericAdministration,.

NOAA2013.RiskAssessmentforPotentiallyPollutingWrecksinU.S.Waters.SilverSpring,MD:NationalOceanicandAtmosphericAdministration.

O'HAGAN,A.,BUCK,C.E.,DANESHKHAH,A.,EISER,R.J.,GARTHWAITE,P.H.,JENKINSON,D.J.,OAKLEY,J.F.&RAKOW,T.2006.UncertainJudgements:ElicitingExperts'Probabilities,Chichester,England.,JohanWiley&SonsLtd.

O'HAGAN,A.&OAKLEY,J.E.2004.Probabilityisperfect,butwecan'telicititperfectly.ReliabilityEngineering&SystemSafety,85,239‐248.

O'HAGAN,A.&OAKLEY,J.F.2010.TheSheffieldElicitationFramework(SHELF,v2.0).An“OfftheShelf”packageforelicitingprobabilitydistributions.DepartmentofProbabilityandStatistics.TheUniversityofSheffield.UK.(http://www.tonyohagan.co.uk/shelf/).

PHILLIPS,L.D.1999.GroupElicitationofProbabilityDistributions:AreManyHeadsBetterThanOne?In:SHANTEAU,J.,MELLERS,B.A.&SCHUM,D.A.(eds.)DecisionScienceandTechnology:ReflectionsontheContributionsofWardEdwards.Boston,MA:SpringerUS.

ROBERTS,N.H.,VESLEY,W.E.,HAASL,D.F.&GOLDBERG,F.F.1981.Faulttreehandbook,Washington,D.C.,SystemsandReliabilityResearch,OfficeofNuclearRegulatoryResearch,U.S.NuclearRegulatoryCommission.

SENDER,U.2011.Korrosionpåskeppsvrakisvenskavatten.(CorrosionofshipwrecksinSwedishwaters).SwereaKIMAB.

SMHI.2011.SeatrackWeb[Online].TheSwedishMeteorologicalandHydrologicalInstitute.Available:https://stw.smhi.se/[Accessed2016‐02‐242016].

SMHI.2015.Seatrackweb‐Spridningsprognoserföroljeutsläppochandraföroreningar[Online].TheSwedishMeteorologicalandHydrologicalInstitute.Available:http://www.smhi.se/professionella‐tjanster/professionella‐tjanster/sakerhet‐och‐beredskap/seatrack‐web‐spridningsprognoser‐for‐oljeutslapp‐och‐andra‐fororeningar‐1.1646[Accessed2016‐02‐292016].

SPREP&SOPAC.AregionalstrategytoaddressmarinepollutionfromworldwarIIwrecks.13thSPREPMeeting,2002Majuro,MarhsallIslands.ThePacificOceanPollutionPreventionProgramme,PACPOL.

SUTER,G.W.2007.Ecologicalriskassessment,Florida,CRCPress.TaylorandFrancisGroup.

TVERSKY,A.&KAHNEMAN,D.1974.Judgmentunderuncertainty:heuristicsandbiases.Biasesinjudgmentsrevealsomeheuristicsofthinkingunderuncertainty.Science,185,1124‐1131.

TVERSKY,A.&KAHNEMAN,D.1983.Extensionalversusintuitivereasoning:Theconjunctionfallacyinprobabilityjudgment.PsychologicalReview,90,293‐315.

59

WALLS,L.&QUIGLEY,J.2001.BuildingpriordistributionstosupportBayesianreliabilitygrowthmodellingusingexpertjudgement.ReliabilityEngineering&SystemSafety,74,117‐128.

VENTIKOS,N.P.,LOUZIS,K.&KOIMTZOGLOU,A.2013.TheShipwrecksinGreeceareGoingFuzzy:AStudyforthePotentialofOilPollutionfromShipwrecksinGreekWaters.HumanandEcologicalRiskAssessment,19,462‐491.

VENTIKOS,N.P.,LOUZIS,K.,KOIMTZOGLOU,A.&DELIKANIDIS,P.2016.EnhancedDecisionMakingthroughProbabilisticShipwreckRiskAssessment:FocusingontheSituationinGreece.FrontiersinMarineScience,3.

VILLEJOUBERT,G.&MANDEL,D.R.2002.Theinversefallacy:AnaccountofdeviationsfromBayes’stheoremandtheadditivityprinciple.Memory&Cognition,30,171‐178.

WRIGHT,G.,COUDERT,F.,BENTLEY,M.,DEVILLE,S.&STEEL,G.2014.ThisStudyisIntentionallyLeftBlank‐Asystematicliteraturereviewofblankpagesinacademicpublishing.

Environmental risk assessment of shipwrecks Model...

Documents

Transcript of Environmental risk assessment of shipwrecks Model...