Climate Change & the Uncertainty Monster · ! no basis for developing a PDF (no mean, weakly defended upper bound) Statistically-manufactured ‘fat tails,’ with arguably impossible

Climate Change & the Uncertainty Monster

Judith Curry Climate Forecast Applications Network

Uncertaintyandthescience–policyinterface

observations

research assessment

policyRiskRuinBlackswansDragonkings

KnowledgetransmissionPoliticization

ExpertjudgmentConsensusbuilding

KnowledgefrontierModelinadequacyUnknowns

Socialinvestment

Meas.ErrorsHistoricallimitationsSpatialrepresentativeness

bias

Uncertainty Monster The “monster” is a metaphor used in analysis of the response of the scientific community to uncertainties at the science-policy interface. Confusion and ambiguity associated with: §  knowledge versus ignorance §  objectivity versus subjectivity §  facts versus values §  prediction versus speculation §  science versus policy

Genealogy of the Uncertainty Monster Monster theory: monster as symbolic expressions of cultural unease that pervade a society and shape its collective behavior Monster metaphor of Dutch philosopher Martintje Smits: co-existence of public fascination and discomfort with new technologies Uncertainty monster of Dutch social scientist Jeroen van der Sluijs: ways in which the scientific community responds to the monstrous uncertainties associated with environmental problems

Uncertainty monster coping strategies Monsterhiding.Neveradmiterrorstrategymotivatedbyapoliticalagendaorfearofbeingjudgedaspoorscience.

Monsterexorcism.Reducing uncertainty through more research.

Monstersimplification.Subjectivelyquantifyingandsimplifyingtheassessmentofuncertainty.

Monsterdetection.Scientists,auditors,merchantsofdoubt.

Monsterassimilation.Givinguncertaintyanexplicitplaceinthecontemplationandmanagementofenvironmentalrisks.

VanderSluijs(2005)

UnderstandingtheuncertaintymonsterReasoning about climate uncertainty. Climatic Change (2011)

Climate science and the uncertainty monster. Bull Amer Meteorol. Soc. (2011)

Nullifying the climate null hypothesis. WIRES Climate Change (2011)

Climate change: No consensus on consensus. CAB Reviews (2013)

Climate uncertainty and risk. risk. CLIVAR Variations. (2018)

Sea level rise: what’s the worst case? (in prep)

Levelofu

ncertainty Deep

Scenario

Statistical

Controllability

Controllable Uncontrollable

optimalcontrol

cost/benefitanalysishedginginsurance

adaptivemanagement

precautionscenarioplanning

robustnessresilienceanti-fragility

Decision–analyticframeworks

Shouldwehaveconfidenceinfutureprojectionsfromclimatemodels?

TheIPCCAR4providedthefollowingconclusion:

“Thereisconsiderableconfidencethatclimatemodelsprovidecrediblequantitativeestimatesoffutureclimatechange,particularlyatcontinentalscalesandabove.”

Isthislevelofconfidenceinclimatemodelprojectionsjustified?

Epistemicstatusofclimatemodels

ClimatemodelsimulationsrepresentpossibleclimatefuturesClimatemodelscannotfalsifyorverifypossibleclimatefuturesAreclimatemodelsconsistentwithbackgroundknowledge?•  Climatemodelsmakeassumptionsthatweknowtobefalse•  Climatemodelshavebeenknowntoproducerealistic

simulationsofsomevariablesVerifiedpossibilitiesrequiresthatclimatemodelsareimperfectcredibleworldswithregardtosomeoftheirprojections

Howmuchconfidenceshouldwehaveinclimatemodels?

Mauritsenetal.(2013)

Scenariosofclimatefutures

Existingclimatemodelsdonotallowexplorationofallpossibilitiesthatarecompatiblewithourknowledgeofthebasicwaytheclimatesystemactuallybehaves.Someoftheseunexploredpossibilitiesmayturnouttoberealones.Scientificspeculationonplausible,high-impactscenarios.Theworst-casescenarioisthemostextremescenariothatcannotbefalsifiedasimpossiblebaseduponourbackgroundknowledge.

Possibilityverification

Stronglyverifiedpossibility–basictheoreticalconsiderations,empiricalevidence

Corroboratedpossibility-ithashappenedbeforeVerifiedpossibility–consistentwithrelevantbackground

knowledge(Un)verifiedpossibility-climatemodelsimulationBorderlineimpossible–consistencywithbackground

knowledgeisdisputed(‘worstcase’territory)Impossible–inconsistentwithrelevantbackgroundknowledge

GregorBetz

unverifiedpossibilities

borderlineimpossible

verifiedpossibilities

strong

Classifyingpossibilities

Allpossibilities(withappropriateclassification)areusefulforpolicymaking

EquilibriumClimateSensitivitytoDoubledCO2

011.52.02.53.03.54.04.561020T(oC)

IPCCAR5likelyrange

CMIP5modelrange

Lewis/Curry2015likelyrange

corroboratedpossibilities

(un)verifiedpossibilities

borderlineimpossible

IPCCAR5SPM:“Nobestestimateforequilibriumclimatesensitivitycannowbegivenbecauseofalackofagreementonvaluesacrossassessedlinesofevidenceandstudies.“

IPCCAR5perspective

EquilibriumClimateSensitivitytoDoubledCO2

011.52.02.53.03.54.04.561020T(oC)

theo

retical

no-fe

edback

sensitivity

Lewis/Curry2018v.likelyrange1.15-2.7C

stronglyverified

corroboratedpossibilities

What’stheworstcase?

JC’sperspective

upperlimitv.likelyrange IMPOSSIBLE

EquilibriumClimateSensitivitytoDoubledCO2

011.52.02.53.03.54.04.57.161120T(oC)

CMIP5modelrange

Weitzmann(2008)

0.01%

0.05%

Perspectivefromeconomistsandthesocialcostofcarbon

valueusedinIPCCAR5WGIII

Socialcostofcarbonisdrivenbyextremevaluesofclimatesensitivity

5%

USIWGSCC

Worstcasescenarioversus‘fattail’

Uncertaintystatusofequilibriumclimatesensitivity:scenario/deepuncertainty;Level4uncertainty!nobasisfordevelopingaPDF(nomean,weakly

defendedupperbound)

Statistically-manufactured‘fattails,’witharguablyimpossiblevaluesofclimatesensitivity,aredrivingcalculationsofthesocialcostofcarbonà MonstercreationThebestthatwecandoisboundtherange:bestcase,worstcasescenarios

loppingoffthe‘fattail’

On trial in MN: Social Cost of Carbon

A 1990s law requires the MN Public Utilities Commission (PUC) to establish externality values for CO2 and other power plant pollutants to help guide utility planning decisions. The PUC adopted the federal Social Cost of Carbon (SCC) PUC/SCC was challenged by energy companies and industry groups in a recent trial

Pindyck:Use&MisuseofModelsforClimatePolicy

“Buildingandusingelaboratemodelsmightallowustothinkthatweareapproachingtheclimatepolicyproblemmorescientifically,butintheend,liketheWizardofOz,wewouldonlybedrawingacurtainaroundourlackofknowledge.”Instead:• Focusoncatastrophicoutcomes• Avoidthepretensethatweknowthedamagefunction,climatesensitivity,etc.

• Useexpertopiniontodeterminetheinputstoasimplemodel

“Paynoattentiontothemanbehindthecurtain!”–WizardofOz

IsCO2the‘controlknob’forsealevelrise?

“Inlookingforcauses,Ihaveappliedthe‘SherlockHolmesprocedure’ofeliminatingonesuspectafteranother.Theprocedurehasleftuswithoutanygoodsuspect.ThermalexpansionwasthecandidateofchoiceatthetimeofthefirstIPCCreview.Thecomputedstericrise[fromwarming]istoolittle,toolate,andtoolinear.”– WalterMunk

Predictingsealevelrise:2100“Weareintheuncomfortablepositionofextrapolatingintothenextcenturywithoutunderstandingthelast.”–WalterMunkProjectionsoffuturesealevelrise:àSemi-empiricalapproachesbasedonpastrelationshipsofsealevelrisewithtemperatureàProcess-basedmethodsusingmodels

•  Thermalexpansionfromoceanwarming:deriveddirectlyfromtheglobalclimatemodelsimulations

•  Changesinglacierandsurfacemassbalance:regionalmodels;empiricalrelations

•  Contributionsfromicesheetdynamics:icesheetmodels,expertjudgment,and/orstatisticalprojections.

Globalsealevelrise:2100(m)

00.20.30.60.81.01.52.02.5359

IPCCAR4likelyIPCCAR5likely

expectedSLRexperttestimony

worstcaseestimates

expectedSLRHansenetal

CCSR2017likely

NOAA2017H++

Horton2014xperteliclikely

ΔT=4.5C

Expertassessments

Globalsealevelrise:2100(m)

00.20.30.60.81.01.52.02.5359

IPCCAR4likely

corroboratedpossibilities IMPOSSIBLEverified

possibilitiescond.onΔT

borderlineimpossible

NOAA2017H++

Horton2014xperteliclikely

ΔT=4.5C

JCperspective

worstcaseunverifiedpossibility

unverifiedpossibilitiescond.onΔT

Sealevelrise2100:What’stheworstcase?

Estimatesofworstcaserangefrom1.6mto3.0mNOAAH++scenario2.5m:probabilityof0.05%Probabilityforworstcaseismisleading:unknownprobability‘Worstcase’scenarioisdrivinginfrastructureplanningandlawsuits

WestAntarcticIceSheet(WAIS)

•  Completecollapsewouldraisesealevelby10feet•  IPCCAR5:mediumconfidencewouldnotexceedseveraltenthsof

ameterofsealevelriseduringthe21stcentury(~8inches)

Confoundingfactors:•  138volcanoes•  WAisrising41cm/yr

Chengetal.2017

Whatiscausingrecentsealevelrise?

GreenlandmeltingandtheAMO

Fettweisetal.2008

Framingerror:‘unknownknowns’

Unknownknowns:known‘neglecteds’.Knownprocessesoreffectsthatareneglectedforsomereason.Climatechangeproblemframedtoonarrowly:drivenbyCO2Predictionsof21stCsealevelchange-knownneglecteds:

•  Solarvariabilityandsolarindirecteffects•  Volcaniceruptions•  Naturalinternalvariability(large-scaleoceancirculations):

impactonglacier/icesheetmassbalanceGreenland•  Geothermalheatsources:Greenland,Antarctica•  Geologicuplift:Greenland,Antarctica

Blackversuswhiteswans

Human-causedclimatechange

• Highlycomplexdynamicalsystem• Nosimplecauseandeffect• Climateshiftsnaturallyinunexpectedways.

•  CO2istheprimaryclimate‘controlknob’

Naturalclimatevariability

50 YEARS, 100 YEARS

Emissions

Scenarios of future climate

Solar effects

Volcanic eruptions

Regional change Extreme events Black swans

Climatemodels

Historical and paleo observations

Statistical models

Climate dynamics

Unknowns

Natural internal variability Long range

processes

Are GCMs the best tool for developing scenarios of future climate change?

Challenges:

• GCMs do not simulate the full likely range of climate sensitivity, nor full range of climate-relevant processes

•  current GCMs inadequate for simulating natural internal variability on multidecadal time scales

• computational expense precludes adequate ensemble size

•  GCMs currently have no skill in simulating regional climate variations

Alternativescenariogenerationmethods

•  Hybridmodels:integrateGCMs,processmodels,empiricalmodels,expertjudgment(un)verifiedpossibilities

•  Climatology(historical,paleoclimate)(verifiedpossibilities)

•  Extrapolationofrecenttrend(corroboratedpossibilities)•  Dynamicclimatologyempiricalmodel(Suckling/Smith)(verifiedpossibilities)

•  Network-baseddynamicclimatology(Wyatt&Curry)(verifiedpossibilities)

•  Secularglobalwarmingasamultipliereffect(conditionallyverified)

•  “Whatif”scenariosrelativetovulnerabilitythreshold

•  Sensitivityanalyses(unverifiedpossibilities)

Levelofu

ncertainty Deep

Scenario

Statistical

Controllability

Controllable Uncontrollable

optimalcontrol

cost/benefitanalysishedginginsurance

adaptivemanagement

precautionscenarioplanning

robustnessresilienceanti-fragility

Decision–analyticframeworks

Oversimplification!alarmism

PROBLEM

SOLUTION

positivefeedback

oversimplifyinguncertainty

monstercreation

framingerror

policycartb4scientifichorse

enragedmonsters

A tamed uncertainty monster

“Beingopenaboutuncertaintyshouldbecelebrated:inilluminatingwhereourexplanationsandpredictionscanbetrustedandinproceeding,then,inthecycleofthings,toamendingtheirflawsandblemishes.”-BruceBeck

http://judithcurry.com

• Climate science •  Uncertainty •  Communications •  Social psychology •  Philosophy of science •  Policy and politics •  Ethics •  Energy tech & policy