storylines: an alternative approach to representing ...sobel/papers/storylines_revisionv2.pdf67 as...

20
1 Storylines: An alternative approach to representing uncertainty in physical 1 aspects of climate change 2 Theodore G. Shepherd a , Emily Boyd b , Raphael A. Calel c,d , Sandra C. Chapman e,f , Suraje 3 Dessai g , Ioana M. Dima-West h , Hayley J. Fowler i , Rachel James j,k , Douglas Maraun l , 4 Olivia Martius m , Catherine A. Senior n , Adam H. Sobel o , David A. Stainforth d , Simon F. 5 B. Tett p , Kevin E. Trenberth q , Bart J. J. M. van den Hurk r,s , Nicholas W. Watkins d,e,f , 6 Robert L. Wilby t , Dimitri A. Zenghelis d 7 a Department of Meteorology, University of Reading, Reading RG6 6BB, UK 8 b Lund University Centre for Sustainability Studies, 221 00 Lund, Sweden 9 c McCourt School of Public Policy, Georgetown University, Washington, DC 20057 10 d London School of Economics, London WC2A 2AE, UK 11 e Centre for Fusion, Space and Astrophysics, Department of Physics, University of 12 Warwick, Coventry CV4 7AL, UK 13 f Center for Space Physics, Department of Astronomy, Boston University, Boston, MA 14 02215 15 g Sustainability Research Institute, School of Earth & Environment, University of 16 Leeds, Leeds LS2 9JT, UK 17 h Willis Re, London EC3M 7DQ, UK 18 i School of Engineering, Newcastle University, Newcastle upon Tyne NE1 7RU, UK 19 j Environmental Change Institute, University of Oxford, Oxford OX1 3QY, UK 20 k Department of Oceanography, University of Cape Town, Rondebosch 7701, South 21 Africa 22 l Wegener Center for Climate and Global Change, University of Graz, 8010 Graz, 23 Austria 24 m Institute of Geography, Oeschger Centre for Climate Change Research, University 25 of Bern, 3012 Bern, Switzerland 26 n Met Office, Exeter EX1 3PB, UK 27 o Department of Applied Physics and Applied Mathematics and Department of Earth 28 and Environmental Sciences, Columbia University, New York, NY 10027 p School 29 of Geosciences, University of Edinburgh, Edinburgh EH9 3FF, UK 30 q National Center for Atmospheric Research, Boulder, CO 80307 31 r Royal Netherlands Meteorological Institute (KNMI), 3730 AE De Bilt, Netherlands 32 s Institute for Environmental Studies, VU University Amsterdam, 1081 HV 33 Amsterdam, Netherlands 34 t Department of Geography, Loughborough University, Loughborough LE11 3TU, UK 35 36

Upload: others

Post on 24-May-2020

4 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Storylines: An alternative approach to representing ...sobel/Papers/Storylines_revisionV2.pdf67 As climate-change research becomes increasingly applied, the need for actionable 68

1

Storylines:Analternativeapproachtorepresentinguncertaintyinphysical1aspectsofclimatechange2

TheodoreG.Shepherda,EmilyBoydb,RaphaelA.Calelc,d,SandraC.Chapmane,f,Suraje3Dessaig,IoanaM.Dima-Westh,HayleyJ.Fowleri,RachelJamesj,k,DouglasMaraunl,4OliviaMartiusm,CatherineA.Seniorn,AdamH.Sobelo,DavidA.Stainforthd,SimonF.5B.Tettp,KevinE.Trenberthq,BartJ.J.M.vandenHurkr,s,NicholasW.Watkinsd,e,f,6RobertL.Wilbyt,DimitriA.Zenghelisd7

aDepartmentofMeteorology,UniversityofReading,ReadingRG66BB,UK8bLundUniversityCentreforSustainabilityStudies,22100Lund,Sweden9

cMcCourtSchoolofPublicPolicy,GeorgetownUniversity,Washington,DC2005710

dLondonSchoolofEconomics,LondonWC2A2AE,UK11eCentreforFusion,SpaceandAstrophysics,DepartmentofPhysics,Universityof12Warwick,CoventryCV47AL,UK13

fCenterforSpacePhysics,DepartmentofAstronomy,BostonUniversity,Boston,MA140221515

gSustainabilityResearchInstitute,SchoolofEarth&Environment,Universityof16Leeds,LeedsLS29JT,UK17

hWillisRe,LondonEC3M7DQ,UK18iSchoolofEngineering,NewcastleUniversity,NewcastleuponTyneNE17RU,UK19

jEnvironmentalChangeInstitute,UniversityofOxford,OxfordOX13QY,UK20

kDepartmentofOceanography,UniversityofCapeTown,Rondebosch7701,South21Africa22

lWegenerCenterforClimateandGlobalChange,UniversityofGraz,8010Graz,23Austria24

mInstituteofGeography,OeschgerCentreforClimateChangeResearch,University25ofBern,3012Bern,Switzerland26

nMetOffice,ExeterEX13PB,UK27

oDepartmentofAppliedPhysicsandAppliedMathematicsandDepartmentofEarth28andEnvironmentalSciences,ColumbiaUniversity,NewYork,NY10027pSchool29ofGeosciences,UniversityofEdinburgh,EdinburghEH93FF,UK30

qNationalCenterforAtmosphericResearch,Boulder,CO8030731rRoyalNetherlandsMeteorologicalInstitute(KNMI),3730AEDeBilt,Netherlands32

sInstituteforEnvironmentalStudies,VUUniversityAmsterdam,1081HV33Amsterdam,Netherlands34

tDepartmentofGeography,LoughboroughUniversity,LoughboroughLE113TU,UK35

36

Page 2: Storylines: An alternative approach to representing ...sobel/Papers/Storylines_revisionV2.pdf67 As climate-change research becomes increasingly applied, the need for actionable 68

2

Authorfullnamesandemailaddresses37

TheodoreG.Shepherd,[email protected](correspondingauthor;38Phone:+441183788957;ORCID0000-0002-6631-9968)39

EmilyBoyd,[email protected]

RaphaelA.Calel,[email protected]

SandraC.Chapman,[email protected]

SurajeDessai,[email protected]

IoanaM.Dima-West,[email protected]

HayleyJ.Fowler,[email protected]

RachelJames,[email protected]

DouglasMaraun,[email protected]

OliviaMartius,[email protected]

CatherineA.Senior,[email protected]

AdamH.Sobel,[email protected]

DavidA.Stainforth,[email protected]

SimonF.B.Tett,[email protected]

KevinE.Trenberth,[email protected]

BartJ.J.M.vandenHurk,[email protected]

NicholasW.Watkins,[email protected]

RobertL.Wilby,[email protected]

DimitriA.Zenghelis,[email protected]

58

Acknowledgements:ThisworkdevelopedfromaworkshopsponsoredbytheRoyal59Society,withadditionalfundingprovidedbytheEuropeanResearchCouncil60AdvancedGrantACRCC(grant339390).Usefulcommentswerereceivedfrom61MichaelaHegglin,LisaLloyd,andthreeanonymousreviewers.62

63

Theauthorsdeclarenoconflictofinterest.64

65

Page 3: Storylines: An alternative approach to representing ...sobel/Papers/Storylines_revisionV2.pdf67 As climate-change research becomes increasingly applied, the need for actionable 68

3

Abstract66

Asclimate-changeresearchbecomesincreasinglyapplied,theneedforactionable67informationisgrowingrapidly.Akeyaspectofthisrequirementisthe68representationofuncertainties.Theconventionalapproachtorepresenting69uncertaintyinphysicalaspectsofclimatechangeisprobabilistic,basedon70ensemblesofclimatemodelsimulations.Inthefaceofdeepuncertainties,the71knownlimitationsofthisapproacharebecomingincreasinglyapparent.An72alternativeisthusemergingwhichmaybecalleda‘storyline’approach.Wedefinea73storylineasaphysicallyself-consistentunfoldingofpastevents,orofplausible74futureeventsorpathways.Noaprioriprobabilityofthestorylineisassessed;75emphasisisplacedinsteadonunderstandingthedrivingfactorsinvolved,andthe76plausibilityofthosefactors.Weintroduceatypologyoffourreasonsforusing77storylinestorepresentuncertaintyinphysicalaspectsofclimatechange:(i)78improvingriskawarenessbyframingriskinanevent-orientedratherthana79probabilisticmanner,whichcorrespondsmoredirectlytohowpeopleperceiveand80respondtorisk;(ii)strengtheningdecision-makingbyallowingonetowork81backwardfromaparticularvulnerabilityordecisionpoint,combiningclimate-82changeinformationwithotherrelevantfactorstoaddresscompoundriskand83developappropriatestresstests;(iii)providingaphysicalbasisforpartitioning84uncertainty,therebyallowingtheuseofmorecredibleregionalmodelsina85conditionedmanner;and(iv)exploringtheboundariesofplausibility,thereby86guardingagainstfalseprecisionandsurprise.Storylinesalsoofferapowerfulwayof87linkingphysicalwithhumanaspectsofclimatechange.88

89

Keywords:Climatechange,uncertainty,risk,vulnerability90

91

RevisedforClimaticChange,2May201892

93

Page 4: Storylines: An alternative approach to representing ...sobel/Papers/Storylines_revisionV2.pdf67 As climate-change research becomes increasingly applied, the need for actionable 68

4

1.Introduction94

Whatwillthefutureclimatelooklike?Conventionalresponsestothisquestion95offeredbytheclimatesciencecommunity(asintheIntergovernmentalPanelon96ClimateChangeassessmentreports)involvecreatinglargeensemblesofsimulations97offutureclimateusingavarietyofglobalclimatemodels(GCMs).Theseareusedin98turntoderivepredictionsofmeteorologicalfields(e.g.,rainfall),expressedinterms99ofstatisticalquantitiesorprobabilitiesofchangewithinthemodelledworld.The100predictionsaremadeunderanassumedscenariooffutureclimateforcings(e.g.,101greenhousegasemissions),andarecalledprojections.Theimplicationsofthese102projectionsforparticularimpacts(suchasagriculture)aretheninvestigatedby103propagatingthedatathroughachainofimpactmodels.Althoughtheforcing104scenariosareunderstoodtobenon-probabilistic,beingsubjecttohumanactions,it105isstandardpracticetorepresentthephysicalaspectsofclimatechange106probabilistically(Swartetal.2009).107

Asthescientificandpolicydiscussionshiftsfromwhetheranthropogenicclimate108changeisrealtotheevaluationofpotentialregionalimpactsandresponseoptions,109thelimitationsoftheseprobabilisticapproachestothephysicalaspectsofclimate110changearebecomingincreasinglyapparent(Kenneletal.2016).Climatemodels111havestructuralerrors,manyofwhichareshared,whichchallengesaprobabilistic112interpretationofmulti-modelensembles(Knuttietal.2013),andthereisnowayto113directlyassessthereliabilityoffutureclimateprojections,asthereiswithweather114forecasting(Parker2010).Asregionalclimatephenomenasuchasstormtracks115responddifferentlytoclimatechangeindifferentmodels(Shepherd2014),amulti-116modelmeancanleadtoawashed-outresponsethatdoesnotcorrespondtoany117modelsimulation.Effectivebias-correctionofsystematicerrors—especiallyof118multivariaterelationships,suchasthoseinvolvedincompoundevents—requires119vastamountsofdatathatgenerallydonotexist.Inanycase,itisnotknownhowto120correctmodelbiasesinsimulatingclimatechanges(asopposedtosimulationsofthe121presentclimatestate)(Maraunetal.2017).Estimatesofuncertaintiesatthe122regionalscalecanquicklyaccumulatetoapointwherethisknowledgehinders123ratherthansupportsscenario-ledclimateadaptationdecision-making(Wilbyand124Dessai2010).125126Alternativeapproachesarethusemergingthatdonotseektoquantifyprobabilities,127butinsteadtodevelopdescriptive‘storylines’,‘narratives’,or‘tales’ofplausible128futureclimates.Webroadlyrefertotheseas‘storyline’approaches.Whilsttherehas129beenvariationintheuseoftheseterms,therearesomecommoncharacteristics:in130particular,theemphasisonqualitativeunderstandingratherthanquantitative131precision,andtheacceptancethatstorylinesarenotprobabilistic.Herewedefinea132storylineasaphysicallyself-consistentunfoldingofpastevents,orofplausible133futureeventsorpathways.(Atrendcanbeconsideredalong-lastingevent.)Asnoa134prioriprobabilityofthestorylineisassessed,itisnotaprediction.Emphasisis135placedinsteadonunderstandingthedrivingfactorsinvolved,andtheplausibilityof136thosefactors(orofchangesinthosefactors).Typicallymorethanonestorylineis137

Page 5: Storylines: An alternative approach to representing ...sobel/Papers/Storylines_revisionV2.pdf67 As climate-change research becomes increasingly applied, the need for actionable 68

5

considered,toexploremultipleplausiblefutures.Howeverwealsoincludepast138events,becausehistoricaleventsarenotsimplysingledatapointsbutinvolve139detailedstorieswhichcanbeunpacked(Marchetal.1991;Wooetal.2017).140

Anexampleofsuchastorylineisprovidedbytherain-on-snoweventintheSwiss141Alpsthatoccurredon10October2011,whichledtoseverefloodingandmudflows142(Figure1).On8October2011acoldfrontreachedSwitzerland,andinthefollowing143twodayssustainednortherlywindsresultedinsubstantialsnowaccumulationsin144theAlps.Thiswasfollowedbythepassageofawarmfronton10October2011,145combinedwithanatmosphericriver—anelongatedandnarrowareaofveryhigh146atmosphericmoisturetransport—flowingdirectlyintoSwitzerland.Thecombined147effectofthewarmingandthelargemoisturetransportagainsttheAlpsresultedin148snowmelttogetherwithlargeamountsofrainfallinthewesternSwissAlps(Rössler149etal.2014,Piagetetal.2015).Thecombinedwaterfromthesnowmeltandthe150rainfallledtofloodingandtheformationofmudflowsintheKanderandLötschen151valleys.Theeffectswereextremelylocal,asthesoutheastwardfacingslopes152receivedmorethanthreetimestheprecipitationamountsthatfellonthe153northwestwardfacingslopes,becauseoforographiceffects(Rössleretal.2014).154Themudflowsonthesoutheastwardfacingslopesandthefloodshadsignificant155impactsonthelocaltransportationinfrastructure,blockingroadsandatunnel.A156mudflowdykewassubsequentlybuilttoprotectthetunneldamagedintheevent157fromfuturesimilarevents.158

Therearemanyreferencestostorylines,narratives,andscenariosinclimatechange159literature,andthetermsaresometimesusedinterchangeably.Theterm‘narrative’160isoftenusedbysocialscientiststocharacterisepeoples’views,understandings,or161perspectives.Narrativeanalysisisusedtoinvestigateclimatechangediscoursesand162theframingofclimatechangebythemedia,policy-makers,orotherstakeholders.163Whilsttheseareallimportantaspectsofclimate-changeresearch(Fløttumand164Gjerstad2017),ourfocushereisontheratherdifferenttaskofconstructing165storylinestorepresentuncertainty.Theterm‘scenario’isoftenusedindecision-166makingtorepresentanimaginedfuture,withreferencetothescenariotechniqueof167HermanKahnandcolleagues(KahnandWiener1967).Theemphasisonself-168consistencyandplausibility,andthefactitisnotaprediction,hasmuchincommon169withstorylines,asdoesthebenefitforriskawareness(Lempert2013).Indeed,the170termstorylineisoftenusedinthiscontext.Withinclimate-changescience,theterm171‘scenario’isgenerallyassociatedwithsharedsocio-economicpathwaysandtheir172associatedclimateforcings,wheretheuncertaintyliesinthedomainofhuman173choices,andconcernsthefuture.Ourfocushereisinsteadonuncertaintyinthe174physicalaspectsofclimatechange,includingconsiderationofpastevents.175

Storylinescanbeperceivedasanecdotalandthusunscientific.Itis,therefore,176importanttounderstandtheirbasisandhowtheycontributetorepresentingand177communicatinguncertaintyinthephysicalaspectsofclimatechange.Weidentify178someofthemainwaysinwhichthestorylineconceptisbeingused,structuringour179evaluationaroundatypologyoffourreasonsfortakingastorylineapproach.180

Page 6: Storylines: An alternative approach to representing ...sobel/Papers/Storylines_revisionV2.pdf67 As climate-change research becomes increasingly applied, the need for actionable 68

6

Figure1showshowallfourreasonsaremanifestintheconsiderationoftheSwiss181rain-on-snoweventdiscussedabove.Thefloodingandmudflows(panelc)improve182riskawarenessbyprovidingasalientevent,moreoveronethatwasnotforecasted.183Strengtheneddecision-makingisevidentinthedevelopmentofdefense184infrastructure(paneld).Thesynoptic-scaleweathersetting(panela)providesa185physicalbasisforpartitioninguncertaintytoassessfuturerisk,includingthe186distinctionbetweenthermodynamic(warmthandmoisture)anddynamic(wind)187factors.Finally,thestrongenhancementoftheprecipitationonthesoutheastward188facingslopesofthevalleywasduetoaverylocalcavitycirculationandassociated189‘seeder-feeder’effect(paneld),ameteorologicalphenomenonnotcapturedbythe190weatherforecastmodels—whichiswhytheeventwasnotforecasted—andwhich191therebyexplorestheboundariesofplausibilityrelativetowhatiscurrently192representedinthemodels.193

1942.Improvingriskawareness195

Twohumantendenciesareintuitivelyevidentinhowwethinkaboutrisk.Itmaybe196usefultoclassifythemusingadistinctionmadefirstbyTulving(1972,2002)about197memory:knowingfacts(semantic)versusrelivingevents(episodic).Sincethen,it198hasbecomeclearthatepisodicmemoryhasaroleinanticipatingthefuture199(Schacteretal.2007),andnewneuroscientificdiscoveriesaregivingapictureof200constructivememoryandepisodicfuturesimulation,whichactaswarningbells,201andhelpustoconceiveofpossibleextremephenomena.Consideredwithinthis202broadercontext,theconventionalapproachtoclimate-changeriskissemantic(e.g.,203whatisa1in1000yearevent?),whereasstorylineapproachesareepisodic(e.g.,204haveweseenthisbefore;andifso,whatmightthenexteventbelike?).205

Behaviouralpsychologyshowsthathumanshavedifficultyrespondingrationallyto206risksfromeventsthatareoutsidetheirexperience,evenwhenaccuratequantitative207informationontheserisks,andthebenefitsofrationalmitigationactions,is208available.Evenwhengivensuchinformation,weactasthoughtheprobabilityofa209badoutcomeislessthanitreallyisifaneventofthattypehasnothappenedtous210recently(orever),andmoreprobablethanitreallyisifithas.Thisasymmetrical211responseisknownasthe‘availabilitybias’(Kahneman2011).Essentially,we—212eventhosewithquantitativescientifictraining—aremorelikelytorespondto213episodicthantosemanticinformation.214

Theavailabilitybiasisapparentinthehistoryofphysicalinfrastructuremeasuresto215mitigatenaturaldisasterrisk,whichshowsthatsuchmeasuresareusuallytaken216onlyindirectresponsetodisastersthathavejustoccurred.Thiswasthecasewith217theNorthSeafloodof1953,whichledtheNetherlandstodeveloptheDeltaWorks218andtheUKtheThamesbarrier,andmorerecentlyHurricaneSandyinNewYork219City(Sobel2014).Threatsfromnaturaleventssufficientlyrareastobeoutsidethe220livingmemoryoflocalpopulationsanddecision-makersdonotmotivate221investmentsinphysicalinfrastructurethatwouldreducerisk,despitegood222semanticunderstandingofthecaseforsuchinvestments,becausethenecessary223

Page 7: Storylines: An alternative approach to representing ...sobel/Papers/Storylines_revisionV2.pdf67 As climate-change research becomes increasingly applied, the need for actionable 68

7

episodicunderstandingisnotpresent.Inthebestcases,scientificinformationwell224inadvanceofaneventhasledto(lesscostly)effortstoimproveemergency225managementprocedures.Thisappearsmostlikelytohappenwhenthescientific226informationismadeavailableinsufficientlyepisodicform,i.e.,asastoryline.This227arguablyoccurredinthecaseofapossibleearthquakeandtsunamiinthePacific228NorthwestoftheUnitedStates,whereapopularmagazinearticle(Schulz2015)—229augmented,perhaps,bywideawarenessofthe2011FukushimaeventinJapan—230ledtoorganizedpreparationeffortsbythefederalgovernment(FEMA2016).231

Climatechangeisessentiallysimilar.Thebestscientificinformationpredictsa232futurefundamentallydifferentfromthepast,buteventhosewhoareawareofthe233sciencehavedifficultyprioritizingactionspreciselybecauseoftheunfamiliarityof234thatfuture(Weber2006).Here,scientificallyconstructedstorylineshelpasa235complementaryapproachtoraiseriskawareness,byincorporatingepisodic236informationandmakingthepredictedfuturemoretangible(e.g.Matthewsetal.2372016,2017).238

ThisistheapproachadvocatedbyHazelegeretal.(2015),whoconstructstorylines239(called‘tales’)offutureweatherthatillustratetheimplicationsofclimatechangefor240real-lifehigh-impactweatherevents.Thisisdonebymappinghistoricalor241hypotheticaleventsontofutureclimateconditions.Examplesoftheapproach242includererunningweatherepisodesinalimited-areaweather-predictionmodel243withelevatedtemperaturesasboundaryconditions(Attemaetal.2014;Preinetal.2442016),diagnosingunprecedentedstormsinfutureclimateintegrations(Haarsmaet245al.2013),andexploringmultivariatedriversoflocalextremewaterlevels(Vanden246Hurketal.2015).Thatsuch‘simulatedexperience’ismoreeffectiveatconveying247riskthanstatisticalcharacterizationshasbeenrecognizedmorewidelyinthe248decision-makingliterature(HogarthandSoyer2015),buildingoninsightsfrom249experimentalpsychology(GigerenzerandHoffrage1995).250

Thereisalsotheissueofhowtocommunicatethestoryline.Well-designedboard251andcardgamescanofferanimmersiveexperienceandtellstories,andhavebeen252usedtoaiddevelopmentdecisions(Tintetal.2015),kick-startdiscussionsonthe253effectsofclimatechange(Chuang2017),andinformparticipantsaboutclimate254change(ReckienandEisenack,2013),althoughmostgamesfocusonnationallevel255policymakers(e.g.MatznerandHerrenbrück,2017).TheRedCrossClimateCentre256(http://climatecentre.org/resources-games/)hasdevelopedmanysimplegamesto257aidstakeholders’understandingofdifferenthumanitarianissues.Onewayto258engageinfrastructureownersandmanagerswithpotentialfutureclimatechange259couldbetoproduceadeckofcardswitheachcardcontainingasimpletextual260descriptionofaweatherorclimateevent.Thedeckofeventsisastorylineofa261potentialfutureclimate.Theparticipantsthendiscussthepotentialimpactofeach262eventontheirinfrastructure,howtheimpactofthateventonotherinfrastructure263mightaffectthem,andhowtheymightmitigatethedamagefromtheevent.This264allowsdecision-makerstounderstandhowpotentiallycascadingeventsmightaffect265theirinfrastructure,includingthecumulativeimpactofmultipleevents.The266

Page 8: Storylines: An alternative approach to representing ...sobel/Papers/Storylines_revisionV2.pdf67 As climate-change research becomes increasingly applied, the need for actionable 68

8

participantswouldworkthroughthecarddeckdiscussingeachevent,observedand267supportedbyafacilitator.Importantly,theparticipants,likeinreality,wouldnot268knowwhateventsarestilltocome.Attheendofthedeckthefacilitatorwouldthen269discuswiththeparticipantswhichevents,orcombinationofevents,matteredto270themandforwhichonesdamagemitigationwouldrequiresubstantialinvestment.271Itisforthoseeventsthatsubsequentriskassessmentcouldbemade.272

Thedialecticbetweensemanticvsepisodicknowledgeseemsrelatedtothat273betweentheinvisibilityvsvisibilityofclimatechange(Rudiak-Gould2013).The274orthodoxclimate-scienceviewwouldbethatclimatechangeisinvisibletothenaked275eye,asitisinherentlyastatisticalconcept(semanticknowledge).However,many276wouldarguethatsomeaspectsofclimatechangearealsovisible,asinthecaseof277manyoftheobservedcryosphericchanges(episodicknowledge).Ifanobserved278changeissufficientlylargethatevenasingleoccurrenceisattributable(Shepherd2792016),andifitisvisibletothenakedeye(asopposedtosomethingthatcanonlybe280measured),thenvisibilityofclimatechangecanbereconciledwiththestatistical281perspective.Inasimilarway,storylinesallowepisodicknowledgeofclimatechange282tobesetwithinthecontextofsemanticknowledge.283

3.Strengtheningdecision-making284

Fewsocietaldecisionsaredrivensolelyby,orareframedonlyby,concernsabout285climatechange,butratherbysustainabledevelopmentmoregenerally.Storyline286approachesacknowledgethiscontextbycommunicatingthepotentialconsequences287ofclimatechangeinwaysthatarerelevanttothespecificdecisionandthespecific288decision-maker(Hazelegeretal.2015).Theyarealsonaturallysuitedtonon-289probabilisticdecision-makingframeworks,whichseekadaptationoptionsthatare290robustwithinacontextofdeepuncertainty(Kalraetal.2014;Simpsonetal.2016).291Storylinesallowawaytoexploresuchuncertaintiesinatraceableandphysically292plausiblemanner(seealsothediscussioninSection5concerningsea-levelriskin293theNetherlands).Itisimportanttoundertakesuchanalysisusinganiterative294processinwhichdecision-makerconcernsandinterestsshapethescientific295researchandmodellinginaprocessofco-production(Kalraetal.2014;Bhaveetal.2962018).297

Anexampleofsuchaniterativeprocessisprovidedbyananalysisofthephysical298andlegaldimensionsofwaterdiversionsfromtheUpperColoradoRiverBasin299understorylinesofclimatechange(Yatesetal.2015).Inthiscase,regionalclimate300simulationswereusedtostresstestamodelofthewatersupplysystemwithand301withoutanadaptationoption.Stakeholderswereengagedfromtheoutsetinthe302developmentofplausiblestorylinesthatcaptureboththedirectandindirect303consequencesofclimatechangeontheheadwaterareas.Forinstance,ifhotterand304drierconditionswereimagined,thenwatermodelparameterswereadjustedto305reflectpossiblehydrologicalchangesassociatedwithdustonsnowpackorwildfires306destroyingforestedareas.Stakeholdersalsospecifiedmetricsofwatersystem307behaviourandstaterelevanttotheirriskmanagementcontext.Inthisway,both308watermanagersandresearchersarrivedatasharedunderstandingofthesystem309

Page 9: Storylines: An alternative approach to representing ...sobel/Papers/Storylines_revisionV2.pdf67 As climate-change research becomes increasingly applied, the need for actionable 68

9

vulnerabilitiesandtheextenttowhichclimateriskscanbemanagedusingalegal310adaptationinstrument.311

Storylinesareespeciallyeffectiveforconsideringrisksfromcompoundextreme312events,namelythosewheresevereimpactsaretriggeredbytheinteractionofmore313thanonevariable.Examplesincludedroughtwithheatwaves(Ciaisetal.2005),314windstormswithheavyprecipitation(Martiusetal.2016),andfluvialfloodingwith315stormsurges(Kewetal.2013).Figure1providesanotherexample.Neithervariable316itselfneedstobeveryextremebutthecombinationresultsinsevereimpacts317(Leonardetal.2014).Tostudycompoundextremesstatistically,longtimeseries318andrefinedstatisticalmethodssuchascopulasareneededtocaptureandquantify319theinter-dependenceofthevariables.Astorylineapproachisaveryattractive320complementarymethodtoaddresscompoundextremesasitallowsonetodefine321verycomplexeventsstartingfromtheimpactsinaveryflexibleway,including322spatialandtemporaldependence(clustering)ofextremes,whichcangeneratethe323impacts(e.g.,Muchanetal.2015).Storylinesbasedonimpactscanthenbeassigned324acategoricalplausibilityusingstatementslike“worldwideatleastonesimilarevent325occurredbefore”.TheSwissFederalOfficeforCivilProtectionusessuchan326approachtoprepareforabroadrangeofperils,includingatmospherichazards,but327alsotechnicalandsocialthreats(BABS2013).Foreachperil,theydefinethree328storylineswithincreasingseverityandimpacts.329

Anotheremergingapplicationofthestorylineapproach,althoughverymuchinits330infancy,isintheinsurancesector.Theinsuranceandreinsuranceindustriesare331slowlybeginningtoacknowledgetheimpactsofclimatechangeonworldwide332insuredlosses(Lloyd’sofLondon2014).Havingarobustunderstandingofpossible333catastrophelosses,particularlyoftheextremecatastrophes,isessentialforany334insurance/reinsurancecompany’sinternalcapitalmanagement,definitionofown335riskappetite,andsolvency(EIOPA2016).Themaintoolsforassessingsuchrisk336fromnaturalperilsarecatastrophemodels.Bynature,catastrophemodelsarebuilt337toestimateprobabilisticrisk,throughtheextrapolationofthelimitedobserved338record,inordertoestimatearangeofpotentialcatastrophes,theirassociated339probabilitiesandcorrespondinglosses(GrossiandKunreuther2005).However,the340samemodelscanbeandfrequentlyareusedtoestimatedeterministicriskfrom341variousstormscenarios.Assuch,modelsareruntocomputepossiblelossesfroma342relevanthistoricaleventoramodifiedversionofthehistoricalevent,foragiven343portfolio(Wooetal.2017).Thisstorylineapproachofassessingriskforinsured344propertiesisanextremelyusefulmethodofstress-testingtheclient’sexposureto345variousweatherandclimateconditions.3464.Partitioninguncertainty347

Manystudieshaveshownthattheresponsetoclimatechangecanbeusefully348understoodasacombinationofwhatmightbecalledathermodynamiccomponent349(surfacewarming,moistening,meltingofice)andadynamiccomponent(changesin350circulation)(e.g.,Deseretal.2014).Thedistinctionisnotprecisebecauseboth351componentsoftheclimatesystemarecoupled;inpractice,eitherthedynamic352

Page 10: Storylines: An alternative approach to representing ...sobel/Papers/Storylines_revisionV2.pdf67 As climate-change research becomes increasingly applied, the need for actionable 68

10

componentisdefinedasthecomponentcongruentwithinternalvariability(Deser353etal.2004)withthethermodynamiccomponentobtainedasaresidual,or354(especiallyforprecipitation)thethermodynamiccomponentisdefinedasthe355componentobtainedintheabsenceofchangesincirculation,withthedynamic356componentobtainedasaresidual(Pfahletal.2017).357

Thereasonthedistinctionisusefulis,first,thatthereisastrikingcontrastbetween358thedegreeofconfidencewehaveinthetwocomponents(Shepherd2014;359Trenberthetal.2015).Thermodynamicaspectsofclimatechangearegenerally360robustintheory,observations,andmodels(althoughthereissubstantial361quantitativeuncertaintyassociatedwithclimatesensitivity).Incontrast,dynamic362aspectsarenotanchoredinacceptedtheories,havenotyetemergedinobservations,363anddivergebetweenmodels(Shepherd2014).Second,modeluncertaintiesin364climatesensitivityandindynamicaspectsofclimatechange(afterscalingbyglobal-365meanwarming)appeartobeuncorrelated(e.g.,ZappaandShepherd2017),which366makessenseastheyareassociatedwithdifferentaspectsofmodelerror.The367distinctionbetweenthermodynamicanddynamicaspectsofclimatechange368providesanalternativetotheusualapproachofconsideringensemblesofmodel369simulations,whichmixtogetheruncertaintiesinthetwoaspects,andmayhelpto370provideinformationonwherethelargestuncertaintieslie(e.g.,Pfahletal.2017).In371particular,storylinescanbeconsideredforeachaspect.372

Understandingtheroleofclimatechangeinweatherandclimateextremesisof373considerablesocietalrelevance.Thisisnotstraightforwardtoresolve,sinceextreme374eventsalwaysresultfromanintersectionofnaturalvariabilityofsomesort,riding375onandaugmentingglobalwarmingeffects,andeveryextremeeventisunique(NAS3762016).Theconventionalapproachtothisquestionfocusesonchangesin377probability,butdoesnotdealsatisfactorilywiththeuniquenatureofeachevent378(Shepherd2016).Analternativestorylineapproachmightaskinsteadhowmuch379worsetheeventoutcomeswerebecauseoftheknownthermodynamicaspectsof380climatechange,suchasoceanwarming(Trenberthetal.2015).381

Thedistinctionbetweenthermodynamicanddynamicaspectsofclimatechangeis382alsousefulwhenitcomestorepresentingtheuncertaintyinprojectionsoffuture383change.ApracticalexampleofthisistheKNMIClimateChangeScenarios(see384http://www.climatescenarios.nl/),whichprovidefourdiscretesetsofweatherand385sea-levelvariables,assumingagivenglobaltemperatureincreaseandaregional386amplificationduetocirculationresponses(VandenHurketal.2014).Inessence,387theseprovidestorylinesofregionalclimategivenlarge-scalechangesinphysical388climate.389

Onacontinentalscale,itispossibletomakestatementsaboutfutureclimate(such390asextremeprecipitationinthemid-latitudeswillincrease)withahighdegreeof391confidencesincethespatiallyaggregatedsignalisdrivenprimarilybythe392thermodynamicresponsetoradiativeforcingandglobaltemperatureincrease393(Fischeretal.2013).However,ataregionalscale,mid-latitudeprecipitation394extremesarestronglyinfluencedbythedynamicresponse,andtheycouldeither395

Page 11: Storylines: An alternative approach to representing ...sobel/Papers/Storylines_revisionV2.pdf67 As climate-change research becomes increasingly applied, the need for actionable 68

11

increaseordecreasedependingonthechangesincirculation(Fischeretal.2013;396Pfahletal.2017).Manzinietal.(2014)showedthatmuchofthemodelspreadin397wintertimecirculationchangesoverthenorthernextratropicsisrelatedtothe398modelspreadinremotedriverssuchastropicalwarming,Arcticwarming,and399stratosphericvortexchange.ZappaandShepherd(2017)usedthisframeworkto400developstorylinesofEuropeancirculationchange,conditionedonagivenlevelof401globalwarming.Thattheinfluenceoftheremotedriversiscausalissupportedby402evidencefromseasonalpredictionandfromsingle-forcingmodelexperiments.It403wasfoundthatforcentralEuropeanwintertimestorminessandwintertime404Mediterraneandrying,twoimportantclimate-changeimpactsforEurope,the405differencebetweenthemostextremestorylineswasequivalenttothatfromseveral406degreesofglobal-meantemperatureincrease.407

Theuncertaintyinlong-termglobaltemperatureincrease(foragivengreenhouse408gasforcing)ismainlyassociatedwiththatinclimatesensitivity.Theobserved409warmingprovidesonlyaveryweakconstraintonlong-termequilibriumwarming410orclimatesensitivity,anditsdirectapplicabilityisamatterofconsiderablecurrent411debate.Astorylineapproachattemptstoarticulatethemuch-richerphysical412understandingofhowclimateprocesseschangeasclimatewarms(orcoolsunder413pastclimatechanges)toprovideamoremechanisticwayofconstrainingthe414problem.Usingphysicallydevelopedstorylines,Stevensetal.(2016)showthat415greaterconfidenceinanoverallnetpositivecloudfeedbackorbetter416reconstructionsoftropicaltemperaturesattheLastGlacialMaximumcouldbe417enoughtoraisethelowerboundonclimatesensitivity.Criticaltestsfortheupper418boundcouldincludetestingofrecentargumentsforamoremodestaerosolforcing.419

5.Exploringtheboundariesofplausibility420

Ifclimateprojectionscouldbereliablyarticulatedasprobabilitypredictions421(conditionedonscenariosoffutureGHGemissions)thenthesocietalconsequences422couldbepresentedasquantifiedrisksoraschangingriskprofiles.Theconventional423approachtoclimateprojectionisfocusedonmakingsuch(conditional)probability424predictionsbyapplyingcomplexpost-processingmethodologies(includingbias425correctionanddownscaling)totheoutputofGCMs.Iftheclimatesystemwaswell426understoodandcouldbesimulatedaccuratelyatallrelevantscales,thenthis427engineeringapproachtoclimateprojectionswouldbeappropriate.However,sucha428propositionhasbeenchallenged(Smith2002;McWilliams2007;VanOldenborghet429al.2013).Furthermore,21stcenturyprojectionswithGCMshaveonlypartial430explorationofmodeluncertainty(Knuttietal.2013)andofinitialcondition431uncertainty(Hawkinsetal.2016).Confidenceintheprocessessimulatedbyclimate432modelsrestsinsteadontherobustnessofthemodelbehaviourwhencompared433withavarietyoftypesofobservations,withinacausalframework(Lloyd2015).As434aconsequence,thereisastrongneedtoincorporatephysicalunderstandingofthe435processesofclimatechangedirectlyintoclimate-relatedstatementsaboutthe436future(Maraunetal.2017).437

Page 12: Storylines: An alternative approach to representing ...sobel/Papers/Storylines_revisionV2.pdf67 As climate-change research becomes increasingly applied, the need for actionable 68

12

Storylineapproachescandothisbyplacingtheemphasisonphysicalplausibility438whenmakingstatementsaboutfutureclimate.Theydosobyusingmodelsastools439tosupportandtesttheoriesregardingtheinteractionsofclimateprocesses,rather440thanasasourceofquantitativeclimatepredictions.Scientificunderstandingisused441topushouttheboundariesofplausiblefuturesbyproposingmechanismsthrough442whichoutcomesnotcurrentlyseeninGCMs,eitherbecauseofmissingphysicsor443becauseofinsufficientsampling,couldarise(Hazelegeretal.2015;Zscheischleret444al.2018).Thisprovidesamorethoroughexplorationoftherangeofuncertainty445reflectedbytoday’slevelofscientificunderstanding,comparedtothelimitedrange446representedbytheextantensemblesofGCMprojections.447

Thefactthatclimatemodelsdonotfullysimulatetherangeofresponseuncertainty448isparticularlyevidentforlocal-scaleextremeevents.Forexample,ithasbeenfound449thatshort-duration(hourly)summertimeprecipitationextremesarewell-simulated450onlyinvery-high-resolution(<4km)regionalclimatemodels(RCMs)thatexplicitly451representdeepconvection,andnotinconventionalRCMsthatparameterisedeep452convection,andthatintensityincreaseswithwarmingaremuchlargerinthehigh-453resolutionsimulations(Kendonetal.2017).Thissuggeststhatconventional454nationalclimatechangescenariosmayunderestimatethepotentialchangesto455short-durationprecipitationextremes.However,very-high-resolutionregional456simulationscanbeusedtogetherwithobservationsandtheoreticalunderstanding457toprovideriskestimateswithinastorylineframework.Meredithetal.(2015)458arguedthataconvectiveeventleadingtointensefloodingcouldbeattributedto459BlackSeawarmingusinganRCMwithexplicitconvectionconstrainedbylarge-scale460reanalysis,butthattheresponsesimulatedwithparameterisedconvectionwas461physicallyimplausible.Preinetal.(2016)usedasimilarapproachtomake462statementsaboutfutureriskofextremeprecipitationandfloodingfortheUnited463States.Suchtargetedvery-high-resolutionsimulationsofrelevantphenomena464combineprocessunderstandingwiththesamplingofotherwiseunexplored465uncertaintiesandtherebygreatlycontributetotheestablishmentofrobust466projectionsoflocalisedextremeevents.467

Asanexampleofanapplicationtoclimateimpacts,UKWaterIndustryResearch468addressedthechallengeoflong-termchangeinintensestormsbycommissioninga469studytobuildaplausible‘story’usingthebestscientificevidenceofhowthese470stormsmightevolveunderglobalwarming.Thestudyusedtheoretical471understandingofpotentialchangeinprecipitationextremesfromtheClausius-472Clapeyronrelation(7%/K),andexaminedclimateanalogues(usingsummer473temperatureandprecipitationclimatologyasanalogues)andoutputsfrommodels474withexplicitconvectiontoproduceplausiblefuturescenarios,whichwerethen475translatedintodesignguidanceforstormdrainagenetworks(Daleetal.2017).In476thiscasetheclimateanalogues,high-resolutionmodelpredictionsandtheory477agreedonacommontrajectory,whichgaveadditionalconfidenceintheresults.478

IntheNetherlands,wheresealevelriseconstitutesanexistentialrisk,therehas479beenaparadigmshiftfrompredictingtoexploringthefuture,whichhasincreased480

Page 13: Storylines: An alternative approach to representing ...sobel/Papers/Storylines_revisionV2.pdf67 As climate-change research becomes increasingly applied, the need for actionable 68

13

theopportunityforrobustdecision-making(HaasnootandMiddelkoop2012).The481DeltaCommissionoftheNetherlandsaddressedthechallengeofhigh-end,long-482rangesealevelchangebydeconstructingallthecomponentsofsealevelriseand483buildingplausibleaccountsofhoweachmightevolveunderextremeglobalmean484warming(Katsmanetal.2011).Inthisstudyeachsourceofinformation,whether485climatemodeloutput,palaeoclimaticdata,observationsorexpertelicitation,is486presentedinatransparentandauditableformatthatisopentochallengeand487revisionasscientificunderstandinggrows.Thisfieldofexpertiseisdeveloping488rapidly,andfrequentlynewinsightsonmajorcontributorstosealevelriseare489explored(e.g.DeContoandPollard2016),whichtriggerscientificandpublic490debatesontheseverityoftherisksforextremesealevelrise.Usingasimilar491process-baseddecompositionmethod,LeBarsetal.(2017)presentedprojectionsof492sealevelriseexceedingthoseofKatsmanetal.(2011),whichtriggeredtheDutch493DeltaProgrammetocommissionanewstudyonpotentialconsequencesofthese494projectionsforthesafetyoftheNetherlands.Suchaconceptof“adaptivedelta495management”asappliedintheNetherlandsisbasedontheprinciplethatnew496insightsoreventsmaytriggertheneedtochangethedefensestrategy,andrelies497heavilyonitsabilitytomonitoranddigestnewevidenceandinformationofferedby498consecutivestorylinesofmajorclimaticphenomena.499

6.Conclusion500

Wehavepresentedatypologyoffourreasonsfortakingastorylineapproachin501climate-changeresearchandcommunication.Storylinescanraiseriskawarenessby502framingriskinanevent-orientedratherthanaprobabilisticmanner—where503eventscanbelong-lasting,suchasatrend—whichcorrespondsmoredirectlyto504howpeopleperceiveandrespondtorisk.Storylinescanprovideaneffective505mechanismforstrengtheningdecision-makingbyallowingonetoworkbackward506fromaparticularvulnerabilityordecisionpoint,combiningclimate-change507informationwithotherrelevantfactorstoaddresscompoundriskanddevelop508appropriatestresstests.Storylinescanprovideaphysicalbasisforpartitioning509uncertainty,therebyallowingtheuseofmorecredibleregionalmodelsina510conditionedmanner.Finally,storylinescanexploretheboundariesofplausibility,511beyondwhatmightbeprovidedbyastandardsetofmodellingtools,thereby512guardingagainstfalseprecisionandsurprise(ParkerandRisbey2015).513

Theterm‘post-normalscience’wasintroducedbyFuntowiczandRavetz(1993)to514describeasituationwhereeitherthedecisionstakesorthesystemsuncertainties515arehigh,andhencetheuseofevidenceiscontested.Theynotedthatmany516environmentalissuesfallintothiscategory,includingclimatechange.Insucha517situation,traditionalmethodsofscientificproblem-solving(reductionist,expert-518based)areineffective.Thefailuretorecognizetheselimitationshelpsexplainwhy519therehasbeenapersistentgapbetweentheproductionanduseofclimate520information(Kirchhoffetal.2013).521

AccordingtoFuntowiczandRavetz(1993),post-normalscienceacknowledges522unpredictability,incompletecontrol,andapluralityoflegitimateperspectives.The523

Page 14: Storylines: An alternative approach to representing ...sobel/Papers/Storylines_revisionV2.pdf67 As climate-change research becomes increasingly applied, the need for actionable 68

14

goalisnottobanishuncertainty,buttomanageit.Themodelforscientificargument524isnotaformalizeddeduction,butaninteractivedialogue.Storylinesoffermany525possibilitiesintheserespects,notleastas‘conversationstarters’.Morebroadly,526theyprovideameansofnavigatingthe‘cascadeofuncertainty’(e.g.Wilbyand527Dessai2010)byrepresentinguncertaintyinphysicalaspectsofclimatechange528withoutlosingsightoftherobustaspects,whilstusingconceptsthatrelateto529people’sexperienceandtapintotheirepisodicwayofthinking.Giventheir530prevalenceinconsiderationofthehumandimension,storylinesalsooffera531powerfulwayoflinkingphysicalwithhumanaspectsofclimatechange.532

533

Page 15: Storylines: An alternative approach to representing ...sobel/Papers/Storylines_revisionV2.pdf67 As climate-change research becomes increasingly applied, the need for actionable 68

15

References534

AttemaJJ,LoriauxJM,LenderinkG(2014)Extremeprecipitationresponsetoclimate535perturbationsinanatmosphericmesoscalemodel.EnvironResLett9:14003.536

BABS(BundesamtfürBevölkerungsschutz)(2013)MethodenzurRisikoanalysevon537KatastrophenundNotlagenfü[email protected]

BhaveAG, Conway D, Dessai S, Stainforth DA (2018) Water resource planning under 539future climate and socioeconomic uncertainty in the Cauvery River Basin in Karnataka, 540India. Water Resources Res 54:708–728.541

ChuangA(2017)Gamechangers.Science355:587.542

CiaisP,etal.(2005)Europe-widereductioninprimaryproductivitycausedbythe543heatanddroughtin2003.Nature437:529–533.544

DaleM,etal.(2017)Newclimatechangerainfallestimatesforsustainabledrainage.545ProcInstCivilEng–EngineeringSustainability170:214–224.546

DeContoRM,PollardD(2016)ContributionofAntarcticatopastandfuturesea-547levelrise.Nature531:591–597.548

DeserC,MagnusdottirG,SaravananR,PhillipsA(2004)TheeffectsofNorthAtlantic549SSTandSeaIceanomaliesonthewintercirculationinCCM3.PartII:Directand550indirectcomponentsoftheresponse.JClim17:877–889.551

DeserC,PhillipsAS,AlexanderMA,SmoliakBV(2014)ProjectingNorthAmerican552climateoverthenext50years:uncertaintyduetointernalvariability.J553Clim27:2271–2296.554

EIOPA(2016)AnoverviewoftherecommendationsregardingCatastropheRiskand555SolvencyII.556https://eiopa.europa.eu/Publications/Stakeholder%20Opinions/IRSG%20own%20557initiative%20paper%20-%20Cat%20Risk.pdf558

FEMA(2016)https://www.fema.gov/media-library/resources-559documents/collections/513560

FischerEM,BeyerleU,KnuttiR(2013)Robustspatiallyaggregatedprojectionsof561climateextremes.NatureClimChange3:1033–1038.562

FløttumK,GjerstadO(2017)Narrativesinclimatechangediscourse.WIREsClim563Change8:e429.564

FuntowiczSO,RavetzJR(1993)Scienceforthepost-normalage.Futures25:739–565755.566

GigerenzerG,HoffrageU(1995)HowtoimproveBayesianreasoningwithout567instructions:frequencyformats.PsycholRev102:684–704.568

Page 16: Storylines: An alternative approach to representing ...sobel/Papers/Storylines_revisionV2.pdf67 As climate-change research becomes increasingly applied, the need for actionable 68

16

GrossiP,KunreutherH(eds.)(2005)CatastropheModeling–ANewApproachto569ManagingRisk.Springer,252pp.570

HaarsmaRJ,etal.(2013)MorehurricanestohitwesternEuropeduetoglobal571warming.GeophysResLett40:1783–1788.572

HaasnootM,MiddelkoopH(2012)Ahistoryoffutures:Areviewofscenariousein573waterpolicystudiesintheNetherlands.EnvironScience&Policy19-20:108–120.574

HawkinsE,SmithRS,GregoryJM,StainforthDA(2016)Irreducibleuncertaintyin575near-termclimateprojections.ClimDyn46:3807.576

HazelegerW,etal.(2015)Talesoffutureweather.NatureClimChange5:107–113.577

HogarthRM,SoyerE(2015)Communicatingforecasts:thesimplicityofsimulated578experience.JBusinessRes68:1800–1809.579

KahnH,WienerAJ(1967)TheYear2000:AFrameworkforSpeculationontheNext580Thirty-threeYears.NewYork:Macmillan.581

KahnemanD(2011)ThinkingFastandSlow.Farrar,Straus,andGiroux.582

KalraN,etal.(2014)Agreeingonrobustdecisions:Newprocessesfordecision583makingunderdeepuncertainty.PolicyResearchWorkingPaperNo.6906,World584Bank,Washington,D.C.,35pp.,585https://openknowledge.worldbank.org/handle/10986/18772.586

KatsmanCA,etal.(2011)Exploringhigh-endscenariosforlocalsealevelriseto587developfloodprotectionstrategiesforalow-lyingdelta–theNetherlandsasan588example.ClimChange109:617–645.589

KendonEJ,etal.(2017)Doconvection-permittingregionalclimatemodelsimprove590projectionsoffutureprecipitationchange?BullAmerMeteorSoc98:79–93.591

KennelCF,BriggsS,VictorDG(2016)Makingclimatesciencemorerelevant.Science592354:421–422.593

KewSF,SeltenFM,LenderinkG,HazelegerW(2013)Thesimultaneousoccurrence594ofsurgeanddischargeextremesfortheRhinedelta.NatHazardEarthSys13:2017–5952029.596

KirchhoffCJ,LemosMC,DessaiS(2013)Actionableknowledgeforenvironmental597decisionmaking:Broadeningtheusabilityofclimatescience.AnnRevEnviron598Resour38:393–414.599

KnuttiR,MassonD,GettelmanA(2013)Climatemodelgenealogy:Generation600CMIP5andhowwegotthere.GeophysResLett40:1194–1199.601

LeBarsD,DrijfhoutS,deVriesH(2017)Ahigh-endsealevelriseprobabilistic602projectionincludingrapidAntarcticicesheetmassloss.EnvironResLett12:044013.603

Page 17: Storylines: An alternative approach to representing ...sobel/Papers/Storylines_revisionV2.pdf67 As climate-change research becomes increasingly applied, the need for actionable 68

17

LempertR(2013)Scenariosthatilluminatevulnerabilitiesandrobustresponses.604ClimChange117:627–646.605

LeonardM,etal.(2014)Acompoundeventframeworkforunderstandingextreme606impacts.WIREsClimChange5:113–128.607

LloydEA(2015)Modelrobustnessasaconfirmatoryvirtue:Thecaseofclimate608science.StudHistPhilosSci49:58–68.609

Lloyd’sofLondon(2014)Catastrophemodellingandclimatechange.610https://www.lloyds.com/~/media/lloyds/reports/emerging-risk-reports/cc-and-611modelling-template-v6.pdf612613ManziniE,etal.(2014)Northernwinterclimatechange:Assessmentofuncertainty614inCMIP5projectionsrelatedtostratosphere-tropospherecoupling.JGeophysRes615119:7979–7998.616

MaraunD,etal.(2017)Towardsprocess-informedbiascorrectionofclimatechange617simulations.NatureClimChange7:764–773.618

MarchJG,SproullLS,TamuzM(1991)Learningfromsamplesofoneorfewer.Org619Sci2:1–13.620

MartiusO,PfahlS,ChevalierC(2016)Aglobalquantificationofcompound621precipitationandwindextremes.GeophysResLett43:7709–7717.622

MatthewsT,MullanD,WilbyRL,BroderickC,Murphy,C(2016)Pastandfuture623climatechangeinthecontextofmemorableseasonalextremes.ClimateRisk624Management11:37-52.625

MatthewsT,WilbyRL,MurphyC(2017)Communicatingthedeadlyconsequences626ofglobalwarmingforhumanheatstress.ProcNatlAcadSciUSA114:3861–3866.627

MatznerN,HerrenbrückR(2017)Simulatingaclimateengineeringcrisis:Climate628politicssimulatedbystudentsinmodelUnitedNations.Simulation&Gaming48:629268–290.630

McWilliamsJC(2007)Irreducibleimprecisioninatmosphericandoceanic631simulations.ProcNatlAcadSciUSA104:8709–8713.632

MeredithEP,MaraunD,SemenovVA,ParkW(2015)Evidenceforaddedvalueof633convection-permittingmodelsforstudyingchangesinextremeprecipitation.J634GeophysRes120:12500–12513.635

MuchanK,LewisM,HannafordJ,ParryS(2015)Thewinterstormsof2013/2014in636theUK:hydrologicalresponsesandimpacts.Weather70:55–61.637

Page 18: Storylines: An alternative approach to representing ...sobel/Papers/Storylines_revisionV2.pdf67 As climate-change research becomes increasingly applied, the need for actionable 68

18

NAS(NationalAcademiesofSciences,EngineeringandMedicine)(2016)Attribution638ofExtremeWeatherEventsintheContextofClimateChange.Washington,DC:The639NationalAcademiesPress,doi:10.17226/21852.640

ParkerWS(2010)Predictingweatherandclimate:Uncertainty,ensemblesand641probability.StudHistPhilosModPhys41:263–272.642

ParkerWS,RisbeyJS(2015)Falseprecision,surpriseandimproveduncertainty643assessment.PhilTransRoyalSocLondA373:20140453.644

PfahlS,O’GormanPA,FischerEM(2017)Understandingtheregionalpatternof645projectedfuturechangesinextremeprecipitation.NatureClimChange7:423–427.646

PiagetN,etal.(2015)DynamicsofalocalAlpinefloodingeventinOctober2011:647moisturesourceandlarge-scalecirculation.QuartJRoyMeteorSoc141:1922–1937.648

PreinAF,etal.(2016)Thefutureintensificationofhourlyprecipitationextremes.649NatureClimChange7:48–52.650

ReckienD,EisenackK(2013)Climatechangegamingonboardandscreen:Areview.651Simulation&Gaming44:253–271.652

RösslerO,etal.(2014)Retrospectiveanalysisofanonforecastedrain-on-snowflood653intheAlps–amatterofmodellimitationsorunpredictablenature?HydrolEarth654SystSci18:2265–2285.655

Rudiak-GouldP(2013)‘‘Wehaveseenitwithourowneyes’’:Whywedisagree656aboutclimatechangevisibility.WeaClimSoc5:120–132.657

SchacterDL,AddisDR,BucknerRL(2007)Rememberingthepasttoimaginethe658future:theprospectivebrain.NatureReviewsNeuroscience8:657–661.659

SchulzK(2015)Thereallybigone.TheNewYorker,July20,2015issue.660

ShepherdTG(2014)Atmosphericcirculationasasourceofuncertaintyinclimate661changeprojections.NatureGeosci7:703–708.662

ShepherdTG(2016)Acommonframeworkforapproachestoextremeevent663attribution.CurrClimChangeRep2:28–38.664

SimpsonM,etal.(2016)Decisionanalysisformanagementofnaturalhazards.Annu665RevEnvironResour41:489–516.666

SmithLA(2002)Whatmightwelearnfromclimateforecasts?ProcNatlAcadSci667USA99:2487–2492.668

SobelAH(2014)StormSurge:HurricaneSandy,OurChangingClimate,andExtreme669WeatherofthePastandFuture.Harper-Collins.670

Page 19: Storylines: An alternative approach to representing ...sobel/Papers/Storylines_revisionV2.pdf67 As climate-change research becomes increasingly applied, the need for actionable 68

19

StevensB,SherwoodS,BonyS,WebbM(2016)Prospectsfornarrowingboundson671earth’sequilibriumclimatesensitivity.Earth'sFuture4:512–522.672

SwartR,BernsteinL,Ha-DuongM,PetersenA(2009)Agreeingtodisagree:673uncertaintymanagementinassessingclimatechange,impactsandresponsesbythe674IPCC.ClimChange92:1–29.675

TintBS,McWatersV,vanDrielR(2015)Appliedimprovisationtrainingfordisaster676readinessandresponse:Preparinghumanitarianworkersandcommunitiesforthe677unexpected.JHumanitarianLogisticsandSupplyChainManagement5:73–94.678

TrenberthKE,FasulloJT,ShepherdTG(2015)Attributionofclimateextremeevents.679NatureClimChange5:725–730.680

TulvingE(1972)Episodicandsemanticmemory.InETulvingandWDonaldson681(Eds),OrganizationofMemory,pp.381–402.NewYork:AcademicPress.682

TulvingE(2002)Episodicmemory:frommindtobrain.AnnRevPsychol53:1–25.683

VandenHurkB,vanMeijgaardE,deValkP,vanHeeringenKJ,GooijerJ(2015)684AnalysisofacompoundingsurgeandprecipitationeventintheNetherlands.685EnvironResLett10:035001.686

VandenHurkB,etal.(2014)Driversofmeanclimatechangearoundthe687NetherlandsderivedfromCMIP5.ClimDyn42:1683–1697.688

VanOldenborghGJ,DoblasReyesFJ,DrijfhoutSS,HawkinsE(2013)Reliabilityof689regionalclimatemodeltrends.EnvironResLett8:014055.690

WeberEU(2006)Experience-basedanddescription-basedperceptionsoflong-term691risk:Whyglobalwarmingdoesnotscareus(yet).ClimChange77:103–120.692

WilbyRL,DessaiS(2010)Robustadaptationtoclimatechange.Weather65:180–693185.694

WooG,MaynardT,Seria,J(2017)Reimagininghistory:counterfactualrisk695analysis.Lloyd’semergingriskreport,London.696

YatesD,MillerKA,WilbyRL,KaatzL(2015)Decision-centricadaptationappraisal697forwatermanagementacrossColorado’sContinentalDivide.ClimateRisk698Management10:35–50.699

ZappaG,ShepherdTG(2017)Storylinesofatmosphericcirculationchangefor700Europeanregionalclimateimpactassessment.JClim30:6561–6577.701

ZscheischlerJ,etal.(2018)Futureclimateriskfromcompoundevents.NatureClim702Change,inpress,doi:10.1038/s41558-018-0156-3.703

704

Page 20: Storylines: An alternative approach to representing ...sobel/Papers/Storylines_revisionV2.pdf67 As climate-change research becomes increasingly applied, the need for actionable 68

20

705

706

Figure1.a)Schematicdepictionofthesynoptic-scaleweathersituationover707Europeon10October2011at00UTC.Awarmfront(WF)islocatedovertheAlps;708behindtheWF,anatmosphericriver(AR)reachesthewesternAlpsfromthe709northwest,providingasupplyofmoisturetotheAlps.Thecoldfront(CF1)over710GreecehadcrossedtheAlpstwodaysearlier,providingheavysnowfall.TheL711symbolsindicatethelocationsofthelow-pressurecentresassociatedwiththetwo712fronts.b)Schematiccross-section(northwesttosoutheast)acrosstheLötschen713Valley,Switzerland,illustratingthestrongcontrastinprecipitationbetweenthetwo714sidesofthevalley;theprecipitationvalues[mm/12hours]indicatethe715accumulationatseverallocationsbetween00UTCand12UTCon10October2011.716Thesolidbluearrowindicatestheatmosphericriver(AR)andthedashedlinethe717cavitycirculation,resultinginafeedercloud.c)Amudflowblockinganddamaging718theonlyroadintotheLötschenValley,locatedattheredstarinpanelb).Figure719fromBerncantonalpolice,usedwithpermission.d)Mobilefloodprotection720measure.FigurefromBeaverCompany,usedwithpermission.721

722

Figureownership:c)cantonalpoliced)Beavercompany

L

Adaptationmeasures

Alps

Impacts

Synoptic-scaleflow

NW SE4km

45

95110

170AR

Localprecipitationandcirculation

L

CF1WF

[mm/12hours]

a) b)

c) d)

Caption:a)Schematicdepictionofthesynoptic-scaleflowsituationoverEuropeon10October201100UTC.Awarmfront(WF)islocatedovertheAlpsandbehindtheWFanatmosphericriver(AR)reachesthewesternAlpsfromthenorthwest.Thecoldfront(CF1)overGreecehadcrossedtheAlpstwodaysearlier.b)Schematiccross-section(north-westtosouth-east)acrosstheLötschenValley,Switzerlandillustratingthestrongprecipitationgradientsacrossthevalley,precipitationvalues[mm/12hours]indicateprecipitationaccumulationbetween00UTCand12UTC10October2011.Thesolidbluearrowindicatestheatmosphericriver(AR)andthedashedlinethecavitycirculationresultinginafeedercloud.Theredstarindicatesthelocationofpanelc.c)AmudflowblockinganddamagingthemainroadintheLötschenValley.D)Mobilefloodprotectionmeasure.

seederfeeder