MURAL - Maynooth University Research Archive Library PUBL … · 2021. 7. 23. ·...

Investigating therecentapparenthiatusin surfacetem peratureincreases:1.Construction oftwo30-m em berEarth System M odelensem blesStephen Outten1,PeterThorne1,2,Ingo Bethke3,and Øyvind Seland4

1Nansen Environm entaland Rem oteSensing Centre,BjerknesCentreforClim ateResearch,Bergen,Norway,2Departm ent

ofGeography,M aynooth University,M aynooth,Ireland,3UniResearch Clim ate,BjerknesCentreforClim ateResearch,

Bergen,Norway,4Norwegian M eteorologicalInstitute,Oslo,Norway

AbstractTherecentIntergovernm entalPanelonClim ateChangereport,alongwithnum erousstudiessince,hassuggestedthattheapparentglobalwarm inghiatusresultsfrom som ecom binationofnaturalvariabilityand

changesto externalforcings.Herein theexternalforcingsforgreenhousegases(GHGs),long-lived tracegases,

volcanicand troposphericaerosols,and solarirradiancehavebeen replaced in theNorwegian Earth System

M odelusing recentobservationalestim ates.Thepotentialim pactofthesealternativeforcings,and byresidual

theinternallygenerated variability,isexam ined through two30-m em berensem blescovering theperiod 1980

to 2012.TheReferenceensem bleusestheCoupled M odelIntercom parison Projectphase5historicalforcings

extended with theRepresentativeConcentrationPathway8.5(RCP8.5)scenario,whiletheSensitivityensem ble

usesthealternativeforcings.Overthehiatusperiod de ned herein as1998–2012,alloftheforcingsshow

som echangebetween theSensitivityand Referenceexperim entsand haveacom bined netforcingchangeof

0.03W m 2.TheGHG forcing is0.012W m 2 higherin theSensitivityforcings.Thealternativesolarforcing

differsfrom theReferenceforcingby 0.08W m 2,thesam easthealternativevolcanicforcingthatwasbased

on thelatestestim atesfrom NASA Goddard InstituteforSpaceStudies.Anthropogenicaerosolem issions

werereplaced using theEU-EclipseV4adatasetand produceam ean forcing changeof0.11W m 2 overthe

period.Part1detailsthecreationofthetwo30-m em berensem blesandtheircharacterizationforparam etersof

particularrelevanceto theexplanation ofthehiatus.A detailed investigation ofthetworesulting ensem bles

globalsurfacetem peraturebehaviorisgiven in Part2,along with com parisonsto observationaldatasets.

1.Introduction

The globalm ean surface tem perature hasincreased m ore slowly overthe past15yearsthan overthe last

50years.The trend has fallen from 0.11Kdecade 1 over1951–2012 to 0.04Kdecade 1 over1998–2012

[Flato etal.,2013],although updatesto observationalestim atessince [Cowtan and W ay,2014;Karletal.,

2015]change these num berssom ewhatforsom e observationalestim ates.Even so,in allcurrentobserva-

tionaldatasets,the period 1998–2012 exhibitsaslowerrate ofglobalm ean surface tem perature warm ing

thanthepreceding15years.W hilethisapparentpauseorhiatusintheglobalwarm ingispresentinthem ean

globalannualsurfacetem peraturetrends,ithasbeenshowntohavebothaseasonalandageographicaldis-

tribution,with thelargestdecreasein trendsfound overthecontinentsoftheNorthern Hem isphereduringthewinter[Cohen etal.,2012].Theglobalwarm ing hiatushasdrawn greatattention,in partbecauseithasnotbeen reproduced bythe vastm ajorityofglobalclim ate m odelrunsupon which future clim ate projec-

tions are based [Fyfe etal.,2013].The fth assessm entreport(AR5)ofthe Intergovernm entalPanelon

Clim ateChange(IPCC)statedthatthehiatusislikelyduetoacom binationofinternalvariabilityintheclim ate

system and areduced trend intheexternalforcings(m edium con denceand expertjudgm ent).W hilethese

ndingshavebeen m ostlycon rm ed byrecentstudies[e.g.,Hubberand Knutti,2014;Kosaka and Xie,2013;

Meehletal.,2014;Schm idtetal.,2014;MarotzkeandForster,2015],therenow existnum eroushypothesesin

the literature regarding the differentfactorscontributing to the globalwarm ing hiatus.Itisim portantto

stress thatseveralm odelruns by severaldifferent,independentm odelscapture notjustthe hiatusbut

potentially im portantaspectsofinternalvariability such asPacicOcean heatcontent[Meehletal.,2014]orElNiño–Southern Oscillation (ENSO)[Risbeyetal.,2014]which m aybeim portant.

One ofthe prevailing hypothesesregarding internalvariability in the clim ate system wasputforward by

KosakaandXie[2013].Theyperform ed coupled m odelexperim entsinwhichtheyprescribed theseasurface

OUTTEN ETAL. INVESTIGATING RECENTHIATUS,1 8575

PUBLICATIONSJournalofGeophysicalResearch:Atm ospheres

RESEARCH ARTICLE10.1002/2015JD023859

Thisarticleisacom panion to Thorneetal.[2015]doi:10.1002/2014JD022805.

Key Points:•Two setsofalternativeforcingshavebeen applied in NorESM overthehiatus

•Som eforcingshad largechangesbutwith littlenetchange(m axof0.1W m

2)

•Two 30-m em berensem bleswerecreated to coverperiod ofthehiatus

Supporting Inform ation:•TableS1

Correspondence to:S.Outten,[email protected]

Citation:Outten,S.,P.Thorne,I.Bethke,andØ.Seland (2015),Investigating therecentapparenthiatusin surfacetem peratureincreases:1.Constructionoftwo30-m em berEarthSystem M odelensem bles,J.Geophys.Res.Atm os.,120,8575–8596,doi:10.1002/2015JD023859.

Received 26 JUN 2015Accepted 28JUL2015Accepted articleonline30JUL2015Published online4 SEP 2015

©2015.TheAuthors.Thisisan open accessarticleundertheterm softheCreativeCom m onsAttribution-NonCom m ercial-NoDerivsLicense,which perm itsuseand distri-bution in anym edium ,provided theoriginalworkisproperlycited,theuseisnon-com m ercialand no m odicationsoradaptationsarem ade.

brought to you by COREView metadata, citation and similar papers at core.ac.uk

provided by MURAL - Maynooth University Research Archive Library

https://core.ac.uk/display/297019495?utm_source=pdf&utm_medium=banner&utm_campaign=pdf-decoration-v1

tem perature(SST)in theEastern Pacicfrom historicalanom alies.Theresultshighlighted theroleofPacic

variabilityandENSO,suchasthestrongElNiñoeventin1998andtheLaNiña-likedecadalcooling,inexplain-

ing m uch ofthecooling thatisthecauseofthehiatus.Otherworkshavebuilton the ndingsofKosakaand

Xie[2013]byexam ining the clim ate m odelsbased on theirrepresentation ofENSO [Risbeyetal.,2014],by

exam ining changesin the windsthatdrive the cooling forPacic variability [England etal.,2014],and by

adjusting windsin a globalclim ate m odel(GCM )to reproduce the cooling effect[W atanabe etal.,2014].

Onecaveatto the ndingsofKosaka and Xie[2013]isthattheirm odelproduced strong warm ing overthe

wintertim e in Eurasia in contradiction to the strong cooling thathasbeen observed [Cohen etal.,2012]

and thatisapartofthehiatus.

TheobservedEurasianwintertim ecoolinghasbeenlinkedtothereductionofseaiceintheEurasiansectorof

theArctic,which causesstrong localwarm ing and theform ation ofatherm allow.Thisresultsin adecrease

ofthem eridionaltem peraturegradientbetweentheArcticandm idlatitudesandaweakeningofthewesterly

ow thattransportsheatandm oisturefrom theAtlantictothecontinent[OuttenandEsau,2012].Atthesam e

tim e,thelarge-scale ow ischanged,facilitating thetransportation ofcold airfrom theSiberian Arcticdown

to the m idlatitudesofEurasia [Petoukhovand Sem enov,2010;Morietal.,2014].Itrem ainsunclearto what

extentthisEurasian cooling isapredictablesecondaryresponseto anthropogenicforcing orisattributable

to naturalvariability.Norisitobviouswhethersuch cooling would continueinto thefutureundertransient

clim atechange.

W hile internalvariabilityislikelypartiallyresponsible fordeterm ining the globalwarm ing hiatus,the other

potentialpartoftheexplanation liesin changesto theexternalforcings,including greenhousegases(GHG

— including ozone-depleting substances),volcanic and anthropogenic aerosols,solarforcing,ozone,and

the effectofstratospheric watervapor.The inclusion ofthese forcings into the globalclim ate m odels

(GCM s)has,forthe past20years,been coordinated through the Coupled M odelIntercom parison Project

(CM IP),the fth phase ofwhich,CM IP5,hasform ed the basisofthe m odelrunsassessed within IPCC AR5.

In CM IP5,theobserved forcingsareapplied onlyuntiltheend ofthehistoricalexperim entsperiod in 2005.

Beyond this, the forcings are based on the Representative Concentration Pathway (RCP) scenarios.

Furtherm ore,duetom odelrun tim esand delaysinassim ilatingthelatestobservations,thefew yearsim m e-

diatelypreceding2005m ayalsohavequestionableforcings.Hence,theCM IP5m odelrunsonlyincludeaccu-

rate forcing estim atesforthe rst5 to 7yearsofthe hiatusperiod,de ned here as1998–2012.Num erous

individualstudieshave investigated how aslowdown in globalwarm ing could be the resultofchangesto

a single setofforcings,e.g.,non-CO2 GHGs [Hansen etal.,2000],ozone-depleting substances [Estrada

etal.,2013],volcanic[Ridleyetal.,2014;Solom onetal.,2011;Santeretal.,2014],and anthropogenicaerosols

[Neelyetal.,2013].

Atleasttwo preceding studieshave considered the sensitivityto forcing m isspecication in clim ate m odel

sim ulationsspecically[Schm idtetal.,2014;Santeretal.,2014].Santeretal.[2014]showed thatbetterspeci-

cationofvolcaniceruptionsproduced a15% decreaseinthedifferencesbetweenobservationsand sim ula-

tionsofsurfacetem perature.W hatism issingtodateisanaggregatedapproachinwhichtheim pactofm ost

ifnotalloftheindividualexternalforcingsareincludedsim ultaneouslyinfullycoupledruns.Aninvestigation

wasm adeinthisveinbyutilizingapproxim ationsofhow theforcingshavechanged,i.e.,linearinterpolations

forGHGsand ozone-depleting substances,estim ated aerosolim pacts,estim atesofvolcanicaerosols,etc.,

and adjusting the CM IP5 ensem ble with acorrection based on asim ple im pulse/response m odel[Schm idt

etal.,2014].

Aspartofan assessm entofhow changesin externalforcingshavecontributed totheobserved warm ing

hiatus,thisstudydetailstheunderlying experim entaldesign,and thealternativeforcingsused to drive

the Norwegian Earth System M odel(NorESM )to bring them in line with in each case one realization of

recentobservations.These forcingsare then used to drive two 30-m em berensem bles— a Reference

ensem ble using CM IP5-prescribed forcings and a Sensitivity ensem ble using the alternative forcings

discussed here. Section 2 outlines the broad experim ental design and its underlying rationale.

Section 3 describesthe alternative forcingsand com paresthem to the originalforcingsused.Section 4

describesthe NorESM m odeland how the two 30-m em berensem bleswere created.Section 5 outlines

som e principalcharacteristicsand salientfeaturesofthe ensem bles.Section 6 providesa discussion of

caveats,and section 7sum m arizeswith ahand-offto thesecond paper.Theaccom panying Part2paper

JournalofGeophysicalResearch:Atm ospheres 10.1002/2015JD023859


[Thorneetal.,2015]then com paresthe resulting ensem blesto the available observationalestim atesof

surface tem peratures.

2.Experim entalDesign

Asoutlined in section 1,there existm ultiple published hypothesesforthe underlying causesofthe recent

warm ing hiatusthatinvolveforcingsand/orvariability.Them ostam enabletoolsto investigatetheseissues

furtherareclim atem odels,whichcanberunwithdifferentexternalforcingsandbeused toexplorepossible

internalclim ate system variabilityasitisdiagnosed bythe m odel.Butclim ate m odelsare com putationally

expensiveto run,m eaning thatcarefulforethoughtin experim entaldesign isnecessaryand thereareinevi-

tabletrade-offsrequired.Sexton etal.[2003]discussed thisforatm osphereonlyruns,butto ourknowledge

no sim ilaranalysishasbeen applied to such an estim ation approach in Earth System M odels(ESM s)such as

NorESM .Furtherm ore,itisnotclearhow experim entaldesign considerationsdepend upon the spatiotem -

poralscaleofthequestionbeingassessed.Certainly,lookingatdecadalversusm ultidecadalorglobalversus

regionalm ay require very distinctm odeling strategies.Forthe hiatus,ourinterestisin a relatively short

period and understanding theregionalsignatures.

Ouroverriding concern wasto attem ptascom prehensivelyaspossibleto testthevariousexisting hypoth-

eses.Thereisno singlerightwayto dothis,especiallygiven niteavailablecom puterresources.Theconun-

drum wasto ndabalancebetweenexploringforcinguncertaintyeffectsandinternalvariabilityeffectsinan

ESM with no obviousaprioribasisupon which to decide,in an inform ed m anner,how to weightthesetwo

aspects.W eknow them odeltreatstheprescribed forcingsasentirelydeterm inistic.Given thenotionalcom -

putationalresourceavailablefor60runs,wecould haverun 60runseach with som ewhatdiffering forcings,

twosetsof30runswithsolelytwodifferentforcingscenarios,oranythingbetween.Theonlywayinwhichto

exploreforcinguncertaintyistorunthem odelm ultipletim eswithdistinctforcingancillaries.Onthe ipside,

cleanlyassessing the role ofvariability alone requiresconsideration ofa seriesofrunswith the sam e pre-

scribed forcings,i.e.,an ensem ble.

Running 60 distinctly forced runswould m ake a clean distinction between internalvariability and forced

response im possible,while running two 30-m em berensem blesyieldsonly 2 degreesoffreedom in the

forced response dim ension.Given (i)readily apparentdecadaltim e scale intraensem ble spread in global

m eansurfacetem peraturesinthoseCM IP5subm issionsthatundertookm ultipleensem blesand(ii)precursor

analysesofe.g.,HawkinsandSutton[2009]showing thelikelydom inantroleofinternalvariabilityin decadal

predictability[seeThorneetal.,2015,Figure1],weconsciouslychoseto undertaketwo ensem blesofequal

size.W ereasoned thatitm ayrequireafarlargernum berofrunstoelucidatevariabilityonhiatustim escales

than theeffectsofanydistinction in forcings.

In sum m ary,theuseoftwo 30-m em berensem bleswasfeltattheoutsetoftheprojectto bepreferablein

elucidatingthepotentialrolesofforcingand internalvariabilitycleanly.Caveatsandlim itationsassociated

with thischoice,m any ofwhich naturally only becam e apparentatprojectcom pletion,are returned to

in section 6.

3.Clim ate Forcings

This section describes which forcings were considered in creating a new setofancillaries which takes

advantage ofthe latest(attim e ofprojectinception in early 2014)available observationalestim ates of

changesthrough atleast2012.Herein,the term forcing refersto “externalforcing”asde ned in the IPCC

Assessm entReport5glossary[Planton,2013],thus,

Externalforcing refersto a forcing agentoutside the clim ate system causing a change in the clim ate

system .Volcanic eruptions,solar variations and anthropogenic changes in the com position of the

atm osphereand land usechangeareexternalforcings.Orbitalforcing isalso an externalforcing asthe

insolation changeswith orbitalparam eterseccentricity,tiltand precession oftheequinox.

In each case,the revised forcingsare com pared to those used in our30-m em berReference ensem ble,

which were identicalto the forcingsused in the version ofNorESM extended historicalrunssubm itted



to the CM IP5 archive.ForNorESM ,the extended historicalruns used RCP8.5 forcings post-2005 and

the sam e choice is m ade in the Reference ensem ble.The changes to the forcings include new

m easurem ents taken since 2005,as wellas m odications m ade to the forcings priorto 2005.These

m odi cationsm aycom efrom inclusion ofnew dataordueto changesin theinterpretation ofpreexist-

ing m easurem ents.

Asoutlined in section 2,the forcingsare treated entirelydeterm inistically within the m odel.Therefore,by

de nition,thereisabsolutelyno uncertaintyin theapplied forcings.Rather,theuncertaintyarisesin which

ofthe (form any forcing factors)severalpossible forcing realizationsto selectand presentto the m odel.

Uncertainty in the selection offorcing historiesto use and potentialim plicationsforthe analysisin Part2

isreturned to in depth in section 6.Forthe rem ainderofthissection,we concentrate upon describing the

applied Sensitivityforcings,discussing where we sourced them to provide fulldata provenance,justifying

where forcingswere notchanged between the two ensem bles,and, nally,characterizing the differences

in applied RadiativeForcing (RF)between theensem bles.

3.1.Greenhouse Gases

Theprim arygreenhousegas(GHG)concentrationsusedinNorESM consistofcarbondioxide(CO2),m ethane

(CH4),nitrous oxide (N2O),CFC-11 (including additionalCFCs and hydrochloro uorocarbons (HCFCs)via

nudging),and CFC-12.Forthe historicalrunsin the CM IP5 archive,observationsofprim aryGHGsare used

where available forthe period of1850 to 2005,although such observationsonlybegan during the lastsix

decadesand forseveralofthesegases,reliableobservationsonlystarted m uch m orerecently.Theconcen-

trationsused in the Reference runsofthisstudywere an average ofthe annualsurface globalm ean m ole

fractionsheld bythe NationalOceanicand Atm osphericAdm inistration (NOAA)and the Advanced Global

Atm ospheric Gases Experim ent(AGAGE)[Prinn etal.,2000].In the CM IP5 archive,the concentrationsin

the earlierpartofthe historicalrunscom e from reconstructionsbased on airtrapped in polarice cores

[Myhreetal.,2013].ForCO2,forexam ple,observationsatM auna Loa,Hawaii,and atthe South Pole start

in the late 1950s,with earliervaluesbased on airextracted from ice coresand rn.From 2005 until2012,

the concentrationsin the Reference ensem ble are taken from the Representative Concentration Pathways

8.5 scenario (RCP8.5).Thisisthe sam e choice asin the subm itted NorESM historicalextended runsin the

CM IP5 archive.Forthe Sensitivity runs,observations are used where available from 1980 until2012 as

detailed below.

3.1.1.Carbon Dioxide

Thealternativecarbondioxideconcentrationswereobtained from theNOAA dataset(NOAA.2014,http://www.

esrl.noaa.gov/gm d/ccgg/trends/global.htm l.)andcoveredtheperiodof1980to2012.Figure1showstheabsolute

concentrationsofCO2 and resultantclim ateforcing relativeto preindustrialforcing asused in theReference

andSensitivityensem bles.Them ainfeatureoftheCO2concentrationsisasteadyincreasebetween1980and

2012thattotalsapproxim ately16% com pared to the1980leveloran increasein forcing of0.81W m 2.The

concentrationsare alm ostidenticalbetween the Reference and Sensitivity ensem bles,with only a sm all

differenceseen after2005.Thelargestdifferenceisseen in 2012and equatesto adecreaseof 0.02W m 2.

3.1.2.M ethane

The m ethane concentrationsforthe alternative forcingswere from the W orld M eteorologicalOrganization

(W M O)data set[Tsutsum ietal.,2009].Thisonly covered the period of1984 until2012,so concentrations

between 1980and 1983werem aintained attheirCM IP5values(Figure1).Thereisam arked system aticbias

betweentheCM IP5concentrationsusedintheReferencerunsandthealternativeconcentrationsusedinthe

Sensitivityruns.Thisresultsin an increase ofaround 2.5% in the forcing relative to preindustrialcaused by

m ethane oraround 0.01W m 2.The offsetisrelativelyinvariantwith tim e afterthe initialshockin 1983/1984.

Thisisinthespin-upperiodfortheensem bles(section4)andpredatesthehiatus,whichisthefocusofthepresent

studies,byapproxim ately15years.

3.1.3.NitrousOxide

Fornitrousoxide,thealternativeconcentrationswerealso obtained from theW M O dataset[Tsutsum ietal.,

2009].Since N2O hasbeen system atically observed since the late 1970s,these coverthe fullperiod ofthe

ensem blerunsfrom 1980until2012.Thealternativeforcingsshow am eandifferenceoverthehiatusperiod

of 0.0003W m 2 from the originalconcentrationsused in CM IP5 (Figure 1).Nitrousoxide concentrations

haveincreased steadilyoverthisperiod resulting in an increasein forcing of0.07W m 2.



3.1.4.CFC-11

The m ostcom plete record ofCFC-11 concentration wasfound in the AGAGE data set(NASA-AGAGE,2014,

http://agage.eas.gatech.edu/data.htm .),and this was used to replace the concentrations ofCFC-11 forthe

Sensitivityruns.ItshouldbenotedthatinNorESM ,theCFC-11 eldisadjustedtoincorporatetheeffectoftrace

gasesincluding long-lived GHGsand ozone-depleting substances.Figure 1 com paresthe adjusted CFC-11

concentrationsandresultantclim ateforcingintheReferenceandSensitivityforcings.Thelargestchangesoccur

from 2009onward and resultin an increasein forcing relativeto preindustrialofapproxim ately0.004W m 2.

To incorporate the radiativeeffectofthetracegases,theyareconverted to an equivalentconcentration of

CFC-11and added tothat eld,givingrisetotheadjusted CFC-11.Thisisdoneusingthefollowingequation,

which utilizestheratio ofradiativeef cienciesofthetracegasescom pared to thatofCFC-11.

CFC-11equivalentofX pptð Þ¼ X pptð Þradiative efficiency ofX

radiative efficiency ofCFC-11

NorESM includesforcingsfrom num eroustracesgasesin thisway,asshown in Table1.Figure1 showsthe

adjustedCFC-11alongwiththetruem easureofCFC-11andtheequivalentconcentrationofalloftherem ain-

ing contributorytrace gasescom bined.Untilthe late 1980s,the com bined contribution ofthe trace gases

Figure1.(topleft)Concentrationsofcarbondioxide,(topright)m ethane,(m iddleleft)nitrousoxide,(m iddleright)CFC-11adjusted,and (bottom right)CFC-12in NorESM fortheReference(blue)and Sensitivity(red)ensem bles.Theseparation of(bottom left)CFC-11adjusted intoitscom ponentsofCFC-11(blacksolid)and tracegases(blackdashed)isalsogiven.Theconcentrationsofthetracegaseshavebeen converted to CFC-11equivalentconcentrationsusing therelativeradiativeef cienciesoftheindividualtracegases.Therightaxisineachplotshowstheforcingrelativetopreindustriallevels.Despiteappearancesoflinearityoversuch asm allrange,therightaxesforCO2,CH4,and N2O arenonlinear.



wascom parable to thatofCFC-11;however,the successofthe M ontrealprotocol[UNEP OzoneSecretariat,

1987]anditssubsequentam endm entshasm eantthatCFC-11concentrationshavebeensteadilydecreasing

sincearound 1990,whilethetracegaseshavesteadilyincreased.Theneteffectisthatsince2008,thecom -

bined forcing ofthetracegasesism orethan doublethatofCFC-11 alone.

Sincethetracegasesaregenerallynotaswellm onitored asGHGs,itwasnotpossibleto nd alternativecon-

centrationsform anyofthem ,especiallyextending asfarbackas1980.Theirconcentrationsweretherefore

replaced using whateverobservationscould befound.CF4 and C2F6 wereboth replaced for2004until2012

from the AGAGE data set(NASA-AGAGE,2014,http://agage.eas.gatech.edu/data.htm .).CFC-113 was also

taken from the AGAGE dataset(NASA-AGAGE,2014,http://agage.eas.gatech.edu/data.htm .)forthe period

of1985 until2012.NOAA’s Halocarbons and otherAtm ospheric Trace Species Group (HATS)provided

the alternative concentrationsofHFC-134a,HCFC-141,CH3Br,CH3Cl,SF6,HCFC-22,HCFC-142,Halon-1211,

and Halon-1301 from 1994,1992,1993,1999,1995,1991,1992,1992,and 2004 respectively (NOAA HATS,

2014,http://www.esrl.noaa.gov/gm d/hats/data.htm l.).Finally,CFC-114 and CFC-115 were taken from the

AGAGEdataset(NASA-AGAGE,2014,http://agage.eas.gatech.edu/data.htm .),startingfrom 2006.Thism eans

that27% oftheannualconcentrationsofthetracegaseswerereplacedduringtheperiod of1980until2012.

Thisincreasestoaround40% fortheperiodof1990until2012andtoaround70% for2000until2012.Thenet

changein forcing bythesetracegasesoverthehiatusperiod is0.003W m 2.

3.1.5.CFC-12

The AGAGE data setprovided the alternative CFC-12 concentrations (NASA-AGAGE,2014,http://agage.eas.

gatech.edu/data.htm .).Thesecoverthefullperiod of1980until2012and areshown in Figure1along with the

CFC-12 concentrationsfrom the Reference ensem ble.The alternative forcingsrelative to preindustriallevels

areconsistentlyhigherthanthosefoundinCM IP5runs,witham eandifferenceofaround12pptor0.004W m 2.

3.1.6.Sum m ary ofGHGs

M ostoftheGHGsshow littlechangefrom theCM IP5concentrationsusedintheReferenceruns,withthelargest

changearisingfrom thealternativem ethaneconcentrations.Thecom binedeffectofalloftheGHGsisgivenin

Figure2.Theaveragedifferencebetween theseforcingsin the two ensem blesrelativeto preindustriallevels

Table 1. Long-Lived Trace Gases and Ozone-Depleting Substances Included in NorESM as Partofthe AdjustedCFC-11 Concentration

TraceGas RadiativeEf ciencyM ean Concentration Between

1980 and 2012(ppt)EquivalentConcentration

ofCFC-11(ppt) M odied From

CFC-11 0.25 239 239 1980CF4 0.1 71.2 28.5 2004C2F6 0.26 2.6 2.7 2004C6F14 0.49 0.02 0.03 –HFC-23 0.19 12.6 9.6 –HFC-32 0.11 1.3 0.6 –HFC-43-10 0.4 0.1 0.2 –HFC-125 0.23 2.0 1.8 –HFC-134a 0.16 17.2 11.0 1994HFC-143a 0.13 3.3 1.7 –HFC-227ea 0.26 0.3 0.4 –HFC-245fa 0.28 2.0 2.3 –SF6 0.52 3.8 8.0 1995CFC-113 0.3 66.7 80.0 1985CFC-114 0.31 15.3 18.9 2006CFC-115 0.18 6.4 4.6 2006Carb-Tet 0.13 97.5 50.7 –M CF 0.06 73.2 17.6 –HCFC-22 0.2 124.9 99.9 1991HCFC-141 0.14 8.9 5.0 1992HCFC-142 0.2 8.7 7.0 1992Halon 1211 0.3 3.0 3.6 1992Halon 1301 0.32 2.0 2.6 2004Halon 2402 0.33 0.3 0.4 –CH3Br 0.01 8.7 0.3 1993CH3Cl 0.01 520.6 20.8 1999



overtheperiodof1980to2012is0.014W m 2;approxim ately0.01W m 2ofwhichcom esfrom thealternative

m ethane,while around 0.004W m 2 ofwhich com esfrom the alternative CFC-12.However,the com bined

effectofthe alternative greenhouse gasforcings,including the long-lived GHGsand ozone-depleting sub-

stances,isan increase ofonlyaround 0.6% oftotalGHG forcing relative to preindustrialand correspondsto

arelativeerrorofaround1.2% onthechangeofforcingsbetween1980and2012.Them eannetchangeinfor-

cing from GHGsincluding long-lived trace gases and ozone-depleting substancesforthe hiatusperiod is

0.012W m 2.

3.2.AnthropogenicAerosols

NorESM includesthreetypesofaerosolsotherthan volcanic:blackcarbon (BC),sulfurdioxide(SO2),and pri-

m aryorganicm atter(POM )[Kirkevågetal.,2013;Iversen etal.,2013].Theseareread into them odelasem is-

sions(kgm 2s 1)from both fossilfueland biom assburning sources.Blackcarbon isalso read in based on

em issionsfrom airtraf c.Thesearede nedateverylocationandoneverylevelasm onthlyaverages;however,

theyareonlyreadinforasingleannualcycleonceperdecade.Forexam ple,inthehistoricalrunsinCM IP5,12

m onthlyaverageswereread in for1850,1860,etc.until1990and 2000,with a nalannualcycleincluded for

2005,the end ofthe sim ulation.However,in the Reference runsdiscussed in thiswork,the 2005 pointis

ignoredandinterpolationisperform edbetweenthe2000historicalem issionsandthe2010RCP8.5em issions.

NorESM perform salinearweightingtodeterm ineaerosolem issionsfortheyearsbetweentheinputdecades.

Alternative aerosolem issionswere obtained from the Norwegian M eteorologicalInstitute and were taken

from the EU-Eclipse V4adata set[Klim ontetal.,2013a,2013b].These included new annualcyclesforeach

decadebetween1970and2000,andforeachyearbetween2005and2010.Theyalsoincludeda nalannual

cycleestim ated for2020to ensurethatNorESM could interpolateem issionsfor2011and 2012.Thealterna-

tiveem issionsdonotincludenew valuesforblackcarbonem itted byairtraf casthisdatawereunavailable.

TheoriginalReferenceandnew Sensitivityem issionsareshowninFigure3.Inordertoascertaintheradiative

im pactoftheperturbationsto theaerosolsem issions,weperform ed asetofauxiliarysim ulations.Because

the aim isto estim ate the effectrelative to Preindustrial(PI)atm ospheric burdens,three setsofrunsare

required.Theserunsareundertaken utilizing,respectively,thefollowing:Preindustrial(m ainlynatural)em is-

sionswithnotim evariationbeyond aseasonalcycle(PI),tim e-varyingCM IP5em issionsasused inReference

(Ref),and tim e-varying Eclipse V4aem issionsasused in Sensitivity(Sens).In allcases,the radiation (direct)

andcloud nucleation( rstindirecteffect)im pactsarecom putedin“online”m odeusingtheprescribed aero-

solconcentrationpathwaysprovidedbyNationalCenterforAtm osphericResearchaspartoftheCom m unity

Atm osphereM odel.Thereafter,theyarerecom puted in “of ine”m odeusing theaerosolconcentrationsand

aerosol-cloud-radiation calculations thatresultfrom the distinctem ission scenarios in PI,Ref,and Sens.

[Kirkevåg etal.,2013].The nalaerosolclim ate forcing wasestim ated asthe difference between Ref-PIand

Sens-PIin thecalculated of inetop ofatm osphere(TOA)netradiation.

The alternative em issionsshow higherlevelsofblack carbon after2000,which provide positive clim ate

forcing bydecreasing thealbedo and allowing theatm osphereto absorb m oreheat.Theyalso show higher

levelsofSO2,which serveto stronglycooltheatm osphere.Finally,Prim aryOrganicM atter(POM )em issions

Figure2.Com bined forcingsrelativetopreindustriallevelsforCO2,CH4,N2O,CFC-12,and theadjusted CFC-11,includingtracegasesforthe(left)Referenceensem ble(blue)and theSensitivityensem ble(red).(right)ThedifferenceofSensitivitym inusReferencein forcingsisalso shown.



are lowerwith am arked relative m inim um in 2009.POM em issionsare “brown”and so have both positive

and negative directeffects.However,theirindirecteffects are undoubtedly a negative RF [Myhre etal.,

2013,Table 8.4].By design,the Sensitivity aerosolem issions exhibit substantively greater interannual

variabilitythan thoseused in Referenceoverthehiatus.Thedifferencesbetween theaerosolancillariesare

substantialoverthisperiod.Allofthe aerosolsundoubtedly contribute to the difference in RF in Figure 3

(bottom right).However,Figure 3 suggeststhatthe changesin POM are dom inantasthe peak in 2009

coincideswith arelativem inim um in BC and thestrong m inim um in POM .M axim um relativechanges(note

thatFigure3yaxisrangesdonotstartfrom zero)aregreaterinPOM (~25% )thaninSO2(~5% )orBC (~10% ).

IfBC werethedom inantfactor,thepeakinthedifferencebetweenReferenceand SensitivityRFwouldoccur

in 2010.Furtheranalysisbeyond thecurrentstudywould beneeded to diagnosetheim pactsoftheindivi-

dualaerosolsin them odel.

Regardless,them ean netdifferencein radiativeforcing between theReferenceand theSensitivityaerosols

overthehiatusperiod is0.11W m 2,with thelargestdifferenceoccurring in 2009and reaching 0.28W m 2.

Changesinallthreesetsofem issionsarelikelytobeim portant,butthechangeinPOM em issionsdom inates

thechangein overallRFbetween theReferenceand theSensitivityforcings.

3.3.Solar

Solarforcing isincluded inNorESM asanannualm ean valueoftotalsolarirradiance,with diurnaland seaso-

nalcyclesinsolarforcingappliedthroughthem odelcode.ThevaluesusedintheNorESM CM IP5sim ulations

werebasedonLean[2000].Thesewereadjustedbyafactorof0.9965basedontheworkofW angetal.[2005]

to bring them in line with observationsfrom the TotalIrradiance M onitor(TIM ),a space-based instrum ent

launched in 2003 as part ofNASA’s Earth Observing System SOlar Radiation and Clim ate Experim ent

(SORCE).From 2009 onward,the solarforcing in CM IP5 wasarepeatofthelastfourcyclesofthe historical

Figure 3.Totalaerosolem issionsfor(top left)blackcarbon,(top right)prim ary organicm atter,and (bottom left)sulfurdioxidefrom biom assburning and fossilfuelsfortheReference(blue)and Sensitivity(red)em issionsused in thiswork.Theem issionsfortheReferencerunareprescribedateachdecadefrom 1970until2020;whiletheRCPscenariosincludeavalueat2005.In theSensitivityruns,theem issionsareprescribed ateach decadefrom 1970to 2000and ateach yearfrom 2005to2010,with a nalvalueprescribed at2020.(bottom right)Thenettop ofatm osphereforcingassociated withthechangein allaerosolsisalso shown.



record. Som e concerns have been

raised over this approach since the

cyclefrom 1996until2008wasinreality

only12.2yearslong,not13yearsasitis

in the m odel[Lean and Rind,2009].

Furtherm ore,the two cycles prior to

1996 were also atypical, being two

of the shortest on record.The solar

forcing in our30-m em berensem bleof

Reference runs was identicalto those

of the CMIP5 sim ulations, including

therepeated cyclefrom 2009onward.

The Sensitivity ensem ble uses the

solarforcings from the Physikalisch-

MeteorologischesObservatorium Davos

(PM OD)thatare based on a com bina-

tion ofdata from severalinstrum ents:

Active Cavity Radiom eterIrradiance Monitor-I(ACRIM -I),ACRIM-II,ACRIM -III,Variability ofsolarIRradiance and

GravityOscillations(VIRGO),TotalIrradianceM onitor(TIM),and Earth Radiation BudgetExperim ent(ERBE).The

PM OD data are generally considered the m ostaccurate setofsolarforcings,given how itwasconstructed

[FröhlichandLean,1998;Fröhlich,2006].Uncertaintyin solarforcingsisdiscussed furtherin section 6.1.1.

Itshould benoted thatwhilethedifferencebetween thePM OD solarforcing used in Sensitivityand thatof

the originalNorESM CM IP runsused in Reference reaches 0.5W m 2 in 2009,thisisthe totalsolarirradi-

ance.TheEarth receivesthisradiation effectivelyasatwo-dim ensionaldiskratherthan athree-dim ensional

sphere,andhencewithaquarterofthesurfaceareathatittrulyhas.Therefore,theeffectiveforcingisshown

on the leftaxisin Figure 4,and thisisone quarterofthe actualTOA forcing.The difference between the

PM OD forcingsandtheNorESM originalforcingisthereforeuptoaround 0.125W m 2,althoughthisvaries

with tim e,with them ean overthe hiatusperiod being 0.08W m 2.Thisisan orderofm agnitude greater

than thechangein GHGsasdiscussed in section 3.1(Figure2)and potentiallycom parable,butofopposite

sign,to thehighlyuncertain effectofaerosolchanges(section 3.2).

3.4.Volcanic

Theforcing effectofvolcanicaerosolsisincluded in NorESM through azonalm assm ixing ratio specied for

each latitudeand on each verticallevel.In practice,them odelusesthezonalm ean colum n m assofaerosols

distributed overtheverticallevelsbyasim pleshapefunction.Thecolum n m assofvolcanicaerosolsused in

NorESM wasbased on the work ofSato etal.[1993].The lastvolcaniceruption included in thisdata setis

Pinatubo in 1991.Tracesofthiseruption existuntil2001,afterwhich,there isno volcanicaerosolforcing

in NorESM .SincePinatubo,therehavebeen aseriesofsm alleruptionseach adding sm allquantitiesofaero-

solsto theatm osphere[Santeretal.,2014;Ridleyetal.,2014].

Alternativevolcanicaerosolconcentrationswereobtainedfrom theNASA GoddardInstituteofSpaceScience

(NASA-GISS,2014,http://data.giss.nasa.gov/m odelforce/strataer/.)and arebased on theupdateofthetabulation

ofstratosphericaerosolopticalthicknessgiveninSatoetal.[1993].Figure5com paresthecolum nm assofvolcanic

aerosolsforthe originalNorESM concentrationsand the alternative concentrations,along with the associated

latitudinallyintegrated forcings.The forcingswere calculated using the sam e form ulaaswasused in IPCC AR5

(Table 8.SM .9),where RF= 25*Atm ospheric OpticalDepth in W m 2 [Hansen etal.,2005].The weaker

volcanoesare clearlym issing from the Reference concentrations.Furtherm ore,the aerosolsfrom the Pinatubo

eruption have decreased in peak m agnitude in the alternative concentrationsbuthave a greaterpresence at

m iddle and high latitudes.Assum ing the presence ofa long-term (m ultidecadal)response to episodic large

volcanoes[Gregory,2010],sucham isspecicationm ayhaveim plicationsform odelbehaviorwithinthehiatus

period even though thiseruption occurred priorto this.Specically,m odelsim ulationsusing the Reference

forcingsm ay show lowerinitialocean heatcontent(OHC)valuesthan sim ilarm odelsim ulationsusing the

alternativeforcingsasaresultofthestrongerspecied volcanicaerosolsin theReferenceforcing volcanic

seriesin theearly1990s.Thiscom plicatesaclean com parison ofthelikelyeffectoftheforcing differences.

Figure 4.Solarforcing in W m 2 ofthe Reference forcing from NorESM(blue)and theSensitivityforcingsfrom Physikalisch M eteorologischesObservatorium Davos(red).



Them ean changein forcing for1980

to 2012 between the Reference and

Sensitivity forcings is 0.17W m 2.

Overthe hiatus period,the alterna-

tive forcings exhibit system atically

greaterloadingsthatwould im parta

cooling in uence.The m ean change

in forcings overthe hiatus period is

0.08W m 2,which com pares well

to the estim ate from Box 9.2 ofthe

IPCC reportofaround 0.09W m 2.

3.5.ForcingsNotAltered

3.5.1.StratosphericW aterVapor

Between 2000 and 2005,the strato-

spheric water vapor shows a sharp

decline;however,the concentrations

have since risen steadily until2012

[Dessleretal.,2013].A recentstudy

byUrban etal.[2014]hasshown that

anothersharp drop in stratospheric watervaporconcentrations has occurred overthe last18m onths.

Decreasesin stratospheric watervaporare expected to warm the stratosphere butcoolthe troposphere

[Solom on etal.,2010].Although thisideawasdisputed byKaufm ann etal.[2011],who found no statistically

signicantrelationship between changing stratosphericwatervaporand surfacetem peratures,Dessleretal.

[2013]quantied thefeedbackofstratosphericwatervaporto beapproxim ately0.3W m 2K 1.

Unfortunately,thestratosphericwatervaporconcentrationsarenotread into NorESM and cannottherefore

be changed directly.This,to som e extent,re ectsthe realitythatstratosphericwatervaporalthough de -

nitelyradiativelyim portantisstrictlyspeakingam echanism ofinternalfeedback/variabilityratherthanwhat

m ayclassicallybeconsidered asaforcing which isan externaldrivertothesystem .In an attem ptto include

theeffectofchanging stratosphericwatervapor,theeffectofm oistureon variouslevelswasadjusted inthe

longwaveand shortwavecom ponentsofthecode.Thiswould havetheeffectofchanging thewatervapor

on those levelsforradiative purposes.However,the resulting tem perature response to thischange was

contraryto existing theoryand literature on thissubjectand no explanation could be determ ined forthis.

Sincewecurrentlyhaveno otherm ethod foradjusting thestratosphericwatervaporin theNorESM m odel,

and theonlym ethod wedohavepro-

duces results that are inconsistent

withthecurrentliterature,stratospheric

water vapor was not altered in the

Sensitivityrunsin thisstudy.

3.5.2.Ozone

NorESM includes ozone concentra-

tions as a forcing ancillary eld.

These concentrations are spatially

com plete,being read in foreverygrid

point and on every m odel level.

However,the ozone concentrations

are included in the m odelon a tem -

porally interm ittentbasis.Forexam -

ple,in the historicalm odelruns,the

m onthly ozone concentrations fora

single yearare read into the m odel

every 10yearsfrom 1850 until1990,

and every 5years from 1995 until

2005 (Figure6).

1840 1860 1880 1900 1920 1940 1960 1980 2000 20207.5

8

8.5

9

9.5

10

year

ozon

e co

ncen

trat

ion

[1e−

6 m

ol/m

ol]

Figure 6.M onthly ozone concentrations input to NorESM once every10yearsuntil1990 and every5yearsuntil2005.

Figure 5.Colum n m assofvolcanicaerosolsin kgm 2 bylatitude with them eridionallyintegrated forcingsunderneath,forthe(top)Referenceforcingsfrom theNorESM CMIP5runsand the(bottom )Sensitivityforcings.Theseareboth based on theworkofSato etal.



The input frequency of ozone into

NorESM is insuf cient to resolve

its decadal scale im pacts which

would require forcing estim atesthat

varied continuously in space and

tim e.W hile com plete observationsof

vertically resolved ozone across the

globeforthewholeperiod ofinterest

are available, since no hypothesis

has been proposed in the literature

professing the role ofozone in caus-

ing the currently observed warm ing

hiatus,the ozone driving eld has

rem ained unchanged in ourcurrent

experim ent.A m ore com plete study

would be able to incorporate m ore

accurate ozone elds into NorESM .

This would not only allow for the

investigation ofthe im pactofchan-

ging ozoneconcentration on decadal

tim escalesbutwould alsohelp quantifytheim portanceofm oretem porallycom pleteozonerepresentation

in futureruns.

3.5.3.Land Surface Changes

NorESM includestheCom m unityLandSurfaceM odelversion4asitslandsurfacecom ponent.Thelandusedata

wereunchanged betweentheReferenceandSensitivityrunsand isthesam easthelandusedataspeciedfor

CM IP5 sim ulations.These consistofland use estim atesfrom theHistoryDatabaseofthe GlobalEnvironm ent

sm oothedtocom bineseam lesslywiththefuturepredictionsundertheRCP8.5scenario.Thelandusedatawere

notm odied in thisstudyprim arilybecause atthe tim e ofthe projectsexecution,land use changeshad not

beenpositedasam ajorfactorincausingorexplainingtheglobalwarm inghiatus.Furtherm ore,wedidnot nd

an observationaldatasetofland usethatwassignicantlydifferentfrom thestandard inputofNorESM.

3.6.Sum m ary ofCom bined Radiative EffectsofChanged Forcings

In thissection,we have docum ented changesto anum berofforcing ancillariesre ecting the observational

understanding attim eofprojectinception in early2014.Changeswerem adeto GHGsand long-lived green-

house gases (LLGHGs)(section 3.1),anthropogenic aerosols (section 3.2),solarradiation (section 3.3),and

volcanicaerosols(section3.4).A num berofadditionalforcingswerenotaltered(section3.5).Alloftheforcings

thatwererevisitedwerealtered withsom ebecom ingm orepositive/lessnegative(LLGHGsandanthropogenic

aerosols)and otherslesspositive/m orenegative(solarandvolcanoes).On average,theSensitivityTOA forcing

isslightlylesspositiveovertheperiodofthehiatus.However,changestoallfourforcingsaretim evariantsuch

thatthe totaldifference in netforcing between the two ensem blesvariesconsiderablyovertim e (Figure 7).

During theend ofthetwentieth century,theforcing differencesarestronglydom inated byvolcanicaerosols,

with theElChichióneruption in1982and M ountPinatuboin1992;however,itshould benoted thatthealter-

nativeforcingsincludenonew anthropogenicaerosoldataduringthisperiod.Overthehiatusperiodof1998to

2012,thedifferencein solarforcing becom esapparentand ofcom parablem agnitudeto thevolcanicforcing.

Thecom bined effectofthevolcanicand solarforcingisneverthelessm orethancom pensatedforbythesharp

risein anthropogenicaerosolforcing,which resultsin anetpositivedifferencein forcingsbetween 2008and

2010.Them eannetforcingdifferencebetweentheReferenceandSensitivityensem blesoverthehiatusperiod

is 0.03W m 2,although itvariesbetween 0.10W m 2 and 0.07W m 2.Thelineartrend in forcing ism ore

positiveforSensitivitythan itisforReferenceoverthehiatusby0.08W m 2dec 1.

4.M odelConfiguration and Ensem ble Creation4.1.M odelSystem

This study uses the m edium resolution con guration ofthe Norwegian Earth System M odelversion 1

(NorESM 1-M )which has provided output to the fth Coupled M odelIntercom parison Project (CM IP5)

1980 1985 1990 1995 2000 2005 2010−0.5

−0.4

−0.3

−0.2

−0.1

0

0.1

0.2

0.3

year

Figure 7.Netdifference ofSensitivity m inus Reference in radiative forcing(black).Thiscom binesthedifferencesin forcingsforGHGsand LLGHGs(green),anthropogenicaerosols(red),volcanoes(blue),and solarirradiance(m agenta).



[Tayloretal.,2012].The m odelisbased on the Com m unityClim ate System M odelversion 4 (CCSM 4)[Gent

etal.,2011]withim portantchangesbeingthereplacem entofthezleveloceancom ponentwithanisopycnic

coordinateocean m odel,im proving therepresentation ofwaterm asses,and theem ploym entofadvanced

chem istry-aerosol-cloud-radiation interaction schem es,enabling the m odelto account for the indirect

aerosolfeedback.The sea ice and land com ponents— the LosAlam osSea Ice M odeland the Com m unity

Land M odel— have been adopted from CCSM 4 withoutchanges.The atm ospheric and land com ponents

arecon gured on aregularhorizontalgrid with a1.9°×2.5°resolution.In thevertical,theatm osphericcom -

ponentcom prises26hybridsigm a-pressurelevelsextendingupto3hPa.Theoceanandseaicecom ponents

arecon guredonacurvilinearhorizontalgridwith1°resolutionalongtheequatorandthenortherngridsin-

gularityshiftedoverGreenland.Inthevertical,theoceancom ponentcom prisesatwo-layerbulkm ixed-layer

representation with 51 isopycnic layers beneath.A detailed description ofthe m odeland evaluation of

standard sim ulationsaregiven in Bentsen etal.[2013]and Iversen etal.[2013].

4.2.Experim entalSetup

W ehaveperform ed twosetsofsim ulations,oneReferenceensem blewhichusesthedefaultCM IP5forcings,

and one Sensitivity ensem ble which uses the alternative forcings outlined in section 3.The sim ulations

covered theperiod of1980until2012.Thisperiod waschosen forthreereasons.

1. Itincludestheperiod ofinterestforstudying theobserved hiatusofglobalwarm ing.

2. Sincethehiatusstarted in the1990s,thischoiceallowsthem odeloveradecadein ordertoadjusttothe

m odied forcings,thusavoiding sharp changesin theforcingsduring theperiod ofinterest.

3. Starting overa decade priorto the period ofinterestallowsthe individualrunsto spin-up and diverge

such thattheyexhibitabroad rangeofbasicstatesofthe m odeled Earth system (including cryosphere

and ocean)consistentwith them odelphysicsbythestartoftheperiod ofstudy.Theend dateisdriven

bytheavailabilityofsom eoftheSensitivityrun-based forcingsbeing solelythrough 2012.

To allow fora robustdetection ofany differencesin forced response and an elucidation ofthe effectsof

m odelinternalvariability on decadaltim e scales,each ensem ble com prisesofa totalof30 sim ulations.

The initialconditions are generated from the three m em bers ofNorESM ’s CM IP5 historicalexperim ent

[Bentsen etal.,2013]asfollows:Both setsaresplitinto threesubsetsof10sim ulationseach.Thesim ulations

ofeach subsetare allinitialized with the1980-01-01 stateofthesam e CM IP5 NorESM historicalsim ulation,

i.e.,thesim ulationsofsubset1areinitializedwiththestateofthe rstCM IP5historicalrealization,thesim ula-

tionsofsubset2areinitializedwiththestateofthesecondCM IP5historicalrealization,andthesim ulationsof

subset3 are initialized with the state ofthe third CM IP5 historicalrealization.The three historicalrunsthat

werethem selvesspun offthem odelcontrolrun separated byseveraldecadesand by1980haveeach been

running for130years.W ithin each subset,the initialspread isgenerated by adding m icroscopic(O(1 6K))

noise to the ocean m ixed-layertem peratures.The internalspread then growswith tim e and (atleastfor

theatm osphere)reachessaturation afteradecadeorso,i.e.,beforeourperiod ofprim aryinterest.

5.Characterization ofthe M odel

Asdiscussed previously,variousalternative hypothesesotherthan forcingsinvolving internalvariabilityor

feedbacks within the system have been proposed to explain the observed hiatus in globalwarm ing.

Beforedetailed analysisofthetwo ensem blescan beundertaken,itisim portanttoplaceNorESM in context

ofthewiderrangeofCM IP5m odelsand to establish whetherornottheintroduction ofthealternativefor-

cingshasim pacted these alternative m echanism sin the m odel.Figure 8 showsthe Equilibrium Clim ate

Sensitivity(ECS)againsttheTransientClim ateResponse(TCR)forarangeofCM IP5m odels(Thisisam odied

version ofFigure9.42 from Chapter9 ofIPCC AR5)[Flato etal.,2013].NorESM hasan ECSofapproxim ately

2.8°C and aTCR ofapproxim ately1.4°C and thuslieswellwithin thespread ofCM IP5m odels.W hileitdoes

lie furtherfrom the CM IP5 ensem ble m ean than the CCSM 4 m odelused in the workofSanteretal.[2014],

NorESM isapparentlynotunusualam ong theCM IP5m odelsforitsresponseto changesin clim ateforcings.

The powerspectrum density ofglobalm ean tem perature variance in the historicalsim ulationsofCM IP5

m odelsisgiven in Figure9(Thisisam odied version ofFigure9.33from Chapter9ofIPCC AR5)[Flatoetal.,

2013].NorESM isgenerally consistentwith observationsand showsreasonable powerathiatus-like tim e

scalesthatisin broad with both otherm odelsand theobservations.



5.1.Changesin Ocean HeatContent

Variousstudieshavediscussed thepossi-

bilityofincreased heatsequestrationinto

the world oceans[e.g.,Meehletal.,2011,

2013, 2014; Chen and Tung, 2014;

Allan et al.,2014;England et al.,2014].

Figure 10 showsthe anom aliesofglobal

OHC forthetwoensem blesintheshallow

ocean and the deep ocean,de ned as

the surface to 325m and below 325m ,

respectively. Both ensem bles show a

sim ilarincrease occurring between 1980

and 2012 ofapproxim ately 100ZJin the

shallow ocean and 200ZJ occurring in

thedeeperocean.ThetotalOHC increase

of approxim ately 300ZJ is sim ilar to

the observational estim ate reported in

the IPCC report [Rhein et al., 2013].

Both ensem bles show with a sim ilar

m agnitudelocalm inim aaround 1992resulting from thePinatubo eruption thataregreatestin theupper-

m ost325m .

Intheshallow ocean,theSensitivityensem bleshowsincreasedvariabilityoverthehiatusperiod,especiallyafter

2006 when the RCP8.5 forcingswere replaced with observationalestim ates.The Sensitivityensem ble in the

deep ocean hasapproxim atelythe sam e variabilityasthe Reference ensem ble butwith aslightlydecreased

m ean.Thisistobeexpected sincethetotaloceanheatcontentshouldbeverysim ilartothetotalEarthsystemheatcontentwhich iscontrolled bytheforcings.Sincein thenettheSensitivityforcingsshow adecrease,wewould expectthe totalOHC anom alyto also decreasesom ewhat.Figure 10 suggeststhatthetrendsin OHC

during thehiatusperiod of1998to 2012areslightlyhigherin theSensitivityensem blethan in theReference

ensem bleforboth theshallow and deep oceans.Figure10furthersuggeststhatthereisgreaterheatcontent

increaseinthedeepoceanandlessheatcontentincreaseintheshallow oceanduringthehiatusperiodthanin

thewarm ing periodoftheearly1990s.Toelucidatethis,acom parison ofthetrendsinOHC fordeep and shal-

low oceansoverthehiatusperiodisgivenforall30m em bersofboththeReferenceandSensitivityensem bles

(Figure11).Thetrendswerecalculatedfrom thegloballyaveragedOHC annualseriesbyordinaryleastsquares

(OLS)regression overthe period 1998–2012.As expected [Meehletal.,2013],there is an anticorrelation

wherebywhen near-surfaceuptakeisrelativelyrapid,deep ocean uptakeisrelativelyslow and viceversa.

ThetrendsofthetotalOHC wereexam ined forthehiatusperiod and found to be13.4ZJyr 1and 13.9ZJyr 1

fortheReferenceand Sensitivityensem bles,respectively.A student’sttestwasappliedtoshow thatthetrends

inOHCbetweenthetwoensem blesweredifferentatthe5% signicancelevel.Overall,theinclusionofthealter-

nativeforcingsappearstohavehadm inim alim pactonthebehaviorofthem odeledOHCintheshallow ocean.

In the deep ocean,the alterna-

tive forcings reduce the m ean

OHCintheSensitivityensem ble

com pared to the Reference

ensem ble,butincreased trends

over the last 15years act to

decrease this difference by

the end ofthe run in 2012.A

recent paper by M ori et al.

suggested the need fora high

num berofensem ble m em bers

(atleast80)in orderto detect

asignicantresponseinsurface

airtem perature to changes in

Figure 9.Globalclim ate variability as represented by power spectraldensity for1901–2010globalm ean surfacetem peratureforboth historicalCM IP5sim ulationsandtheobservations.Thegreyshadingprovidesthe5to95% rangeofthesim ulations.Thisisam odied version ofFigure9.33b from Chapter9oftheIPCC AR% [Flatoetal.,2013].

Figure 8.Com parison ofthe Equilibrium Clim ate Sensitivity(ECS)andTransientClim ateResponse(TCR)ofNorESM to otherCM IP5m odels.Thisisam odied versionofFigure9.42from Chapter9oftheIPCC AR5[Flato etal.,2013].



boundaryconditions.AswillbediscussedinPart2,m orerunsintheSensitivityensem blebeganthehiatusper-

iodwithaLaNiñasituationandendwithanElNiñosituationthanintheReferenceensem ble.Thiscouldinpart

explain thedifferencein OHC observed here.

5.2.ENSO Variability and Trends

Theocean tem peraturesoftheEastern Pacichavebeenofparticularinterestduetotheirrolein drivingthe

El-Niño–SouthernOscillation(ENSO).Com paringthetem peratureanom aliesovertheENSO 3.4regionshows

thatthefrequency,m agnitude,and seasonalityofthevariationsareallcom parableto thosein theobserva-

tionsforboth theReferenceand Sensitivityensem bles(Figure12).Theperiodogram highlightsthatabroad

peak in variability occursbetween 2 and 7years,in agreem entwith the observed variability ofENSO.The

introductionoftheSensitivityforcingsappearstohavehadnosignicantim pactonENSO asitisreproduced

byNorESM ,based on astudent’sttestofthepowerspectrum .

5.3.TOA Radiation

Thenetim balanceofradiation atthetop ofatm osphereprovidesinsightinto both thechangesin radiative

forcing and the changesin clim ate response.Thisnetim balance hasbeen shown to capture interannual

variability in the clim ate system from both volcanoesand ENSO [Allan etal.,2014]and hasbeen shown

to change in response to lower

frequency unforced m odes such as

the Interdecadal Pacic Oscillation

[Brown etal.,2014].Figure 13 shows

this netim balance ofTOA radiation

from the Reference and Sensitivity

ensem bles,com paredtotheobserva-

tionsfrom theCloudsand theEarth’s

Radiation EnergySystem (CERES)and

the Earth Radiation BudgetSatellite

(ERBS) as prepared by Allan et al.

[2014].The observations only begin

in 1985 atthe startofthe ERBS data.

The volcaniceruptionsofElChichón

and Pinatubo are clearly visible in

the 1980s and 1990s, respectively,

although the observations do not

cover the period ofthe ElChichón

eruption.Thereisreasonableconcur-

rence between the observational

record and the m odel ensem bles

overtheperiod thattheyoverlap.

20 25 30 35 40 45 50 55 60 65 7070

75

80

85

90

95

100

105

110

Figure 11.Crossdependency ofOHC changesbetween shallow and deepoceans(boundaryde ned at325m )over1998–2012in theReference(blue)and Sensitivity(Red)ensem bles,witheachpointdescribingthevaluesforasingle-ensem blem em ber.TrendsinOHChavebeencalculatedfrom thegloballyaveraged OHC annualseriesbyOLSregression overtheperiod1998–2012.Thenum bersin theupperrightcornerdenoteR2 values.

Figure10.Anom aliesofglobalm ean ocean heatcontentfortheReferenceensem ble(blue)and theSensitivityensem ble(red)in Zeta(1021)Joulesfor(left)shallow and (right)deep oceans.Shallow and deep oceansarede ned asfrom thesurfaceto 325m and below 325m respectively.



5.4.ArcticSea Ice Extent

Finally,the Arctic sea ice extent,previously linked to cooling overm idlatitude Eurasia [Petoukhov and

Sem enov,2010;Outten and Esau,2012;Honda etal.,2009],isshown in Figure 14 forthe two ensem bles

and com pared to observationalestim ates from the NationalSnow and Ice Data Centre [Fettereretal.,

2002].TheNorESM m odelhasclearde cienciesinitsseasonalcycleofArcticregionseaice,havingaseasonal

cycle thatistoo sm all,even though the m ean extentoverthe calendaryearm ay be broadly correct.The

wintertim em axim um istoo sm all,and thesum m ertim em inim um too large.Thism isspecication m aylim it

thevalueofNorESM forconsidering atleastsom easpectsofseaice-atm ospherefeedbacks.

Differencesin thetrendsand variabilityofseaiceextentbetween thetwo ensem blesarenotsignicantat

the5% levelaccording toastudent’sttest;hence,theintroduction oftheSensitivityforcingshasnotsigni-

cantlyalteredthispotentialsourceofclim atefeedback.Inbothensem bles,them odelshowsaslightlyslower

decreaseinthewintertim eseaiceextentthanisseeninobservations,anditfailstoreproducethesteepdrop

in sum m ertim eseaiceobserved overthepast10–15years.

1981 1986 1991 1996 2001 2006 2011−4

−3

−2

−1

0

1

2

3

year

tem

pera

ture

ano

mol

y [K

]

101 10210−2

100

102

Period (months)

Pow

er

Figure12.Tem peratureanom aliesinthe(top)ENSO 3.4regionfrom HadCRUT4(black)[Moriceetal.,2012],theReferenceensem ble(blue),and Sensitivityensem ble(red),with theaccom panying (bottom )periodogram .Thesolid verticallinesontheperiodogram indicateperiodsof2yearsand 7years.

Figure 13.Netim balance in radiation atthe top ofthe atm osphere from the Reference ensem ble (blue)and Sensitivityensem ble(red),com pared to observationsfrom CERESand ERBS[Allan etal.,2014].Plotted arethe95% bootstrapcon dencebounds(shading),togetherwith ensem blem eans(solid line).



6.Discussion

TherecentIPCCreport[Flatoetal.,2013]proposedthatthereducedtrendintheexternalforcings,largelyasa

resultofsolarand volcanicfactorswith apoorlyquantied contribution from anthropogenicaerosols,wasa

factorroughlyequalin im portto internalvariabilityin explaining thecurrentlyobserved warm ing hiatus.To

investigate the role offorcings,the forcingsin the Norwegian Earth System M odeldue to prim ary green-

housegases,solarirradiance,volcanicaerosols,anthropogenicand biom assburningaerosols,and anum ber

oflong-lived GHGsand ozone-depleting substanceshavebeen m odied based upon m ultiplestate-of-the-

artobservationalsources.Thesem odicationsnotonlyincluded new valuesfortheperiod of2006to 2012,

i.e.,beyond theCM IP5historicalrun,butalso m odicationsfortheperiod priorto 2006based on im proved

understandingand associated datasetinnovationsand updates.Toassessthepossibleroleofinternalvaria-

bility,eachsetofforcingswasrun30tim esstartingfrom distinctinitialconditionswith18yearsofspin-upto

ensure divergence.There are a num berofcaveatsrequired in both the forcingsand ensem ble creation

choiceswhich m ayim pactthesubsequentanalysisand thatweoutlinein thissection.

6.1.Forcings

The alternative forcings discussed in section 3 show differences between the Reference and Sensitivity

forcings;however,thenew forcingsshow onlyasm allchange(relativeto thecom m on signalofincreasing

netforcing)from those applied in the Reference runs,which were the CM IP5 forcings extended with

RCP8.5.Forthehiatusperiod of1998–2012,theSensitivityforcingsshowed differencesfrom theReference

forcingsofapproxim ately 0.02W m 2 due to the GHGs and long-lived trace gases, 0.08W m 2 due to

decreased solar forcing,0.11W m 2 due to anthropogenic aerosolchanges,and 0.08W m 2 due to

increased volcanic activity.Thus,the largestdifferenceswere found in the solarirradiance,volcanic,and

troposphericaerosols.

Thereisanim plicitassum ptionthatthedatasetsusedtocreatetheSensitivityforcingsinthisworkwerenot

them selvesm odied orextended unlesstheproducerofthatdatasetbelieved itwasanim provem enttodo

so.W hilethisisself-evidentfordatasetsthathavesolelybeen extended with new observationsthatdid not

previouslyexist,aquestion could beraised abouttheim provem entobtained bym odifying preexisting data

(e.g.,volcanic aerosolsfrom Pinatubo).Forthis reason,the authorsm ake no claim in thiswork thatthe

Sensitivityforcingsarebetterthan thosein theReferenceruns,onlythattheyhavebeen m odied to bring

them inlinewithcurrentunderstandingandaredifferentfrom thoseintheReferenceruns.Further,itshould

be noted thatthe black carbon em issionsfrom airtraf c,stratosphericwatervapor,land use albedo,and

ozone concentrationswere notm odied due to eithera lack ofsuitable observationaldata ora lack ofa

suitable m ethod form odifying the forcing in NorESM .The im plications ofthese om issionsby de nition

rem ainunknown.However,based uponMyhreetal.[2013],theforcingeffectsofozoneorblackcarbonfrom

airtraf careexpected to besm allcom pared with theim pactfrom otheraerosolsand thesolarforcing.

Asdetailed in section 2,there wasan inevitable trade-offin the experim entaldesign wherebywe allowed

solely 2 degreesoffreedom in the forcings.Yetitisbeyond dispute thatthere isuncertainty in m any of

theapplied forcings.Thoseforcingswhich changed m ostsubstantivelybetween Reference and Sensitivity,

Figure14.Arcticseaiceextentforwintertim e(Decem ber-January-February,topblacktraceandcolortraces)andsum m ertim e(June-July-August,lowerblacktraceand colortraces)from theReferenceensem ble(blue)and Sensitivityensem ble(red),com pared to theobservationalestim atesfrom theNationalSnow and IceDataCentre(black).Plotted arethe95% bootstrapcon dencebounds(shading),togetherwith ensem blem eans(solid line).



solar, volcanic, and anthropogenic

aerosols are exactly those forwhich

there is greatest uncertainty [Myhre

etal.,2013].Hence,ifwe had chosen

differentrealizationsoftheseforcings

to apply,then the resulting RFwould

have differed.Here we lim it further

specic discussion to issues around

each ofthesethreeforcingsand how

theywereapplied.

6.1.1.Solar

W hile m odifying the solar forcing,

severalpossible alternative estim ates

of the observed evolution were

found, based on different sources

and different algorithm s.Figure 15

shows the three m ain alternatives

com pared to the original forcings

used in NorESM for CM IP5 including the nalchoice outlined in section 3.The Institut Pierre Sim on

LaplacesolarforcingisidenticaltothatofNorESM until2009,whenthepreviouscyclesbegintoberepeated

inNorESM .Thesolarforcingfrom SORCEisfrom theTM Iinstrum entfor2003onward.Priortothis,theSORCE

forcings are based on previous reconstructions [Krivova et al.,2010]but include a sm alloffset based

on the TM Im easurem ents forthe period from 2003 onward.The solarforcings from the Physikalisch-

M eteorologischesObservatorium Davos(PM OD)are based on a com bination ofdata from severalinstru-

m ents:ACRIM -I,ACRIM -II,ACRIM -III,VIRGO,TIM ,andERBE.ThePM OD dataaregenerallyconsideredthem ost

accuratesetofsolarforcings,given how itwasconstructed [Fröhlich andLean,1998;Fröhlich,2006].Forthis

reason,and in orderto stretch ourSensitivitytestssince PM OD hasthelargestdifferencefrom theoriginal

NorESM runs,thisestim ateofsolarforcing wasused.However,driving lesforNorESM werecreated forall

threeofthesolarforcingsshown in Figure15,and futureworkcould exploresuch sensitivity.

6.1.2.VolcanicForcing

RecentworkbyRidleyetal.[2014]thatpostdatestheproduction oftheensem blesand initialsubm ission of

thispapersuggeststhatthevolcanicforcingsfrom Sato etal.used in theReferenceforcingsand Sensitivity

forcingsare biased low.Thisisbased upon acom parison to otherdatasetsbased upon observationsfrom

severallidarsand otherinstrum ents.Specically,an assum ption regarding the lowerm ostlevelin which

stratosphericvolcanicaerosolscan reside leadsto an underestim ate ofthe stratosphericburden in m iddle

tohighlatitudes,particularlyinwinter.Thisrequiresareassessm entofhow toapplythisforcingsothestrato-

sphericaerosolloading basepressurelevelvariesseasonallyand latitudinally.Thisisnotthecaseineitherof

thecurrentancillariesin NorESM ,and forthatm atterm ostothercontem poraryGCM sand ESM s,which load

above a xed pressure levelwith no globalorseasonalredistribution in the verticalextentoverwhich the

aerosolloading isbeing spread.

Forthesm allvolcanoessince2000,observationssuggestthattheeffectsofthisunderestim ation m aybeas

greatashalfoftheforcing im plied byestim atessuch asSato etal.atsom elatitudes.Therealm agnitudeof

post-2000 volcaniceffectsisreported byRidleyetal.to be ofthe order 0.19±0.09W m 2.Therefore,the

Sensitivityrunsm aybe high biased relative to realitybythe orderof0.1W m 2.Asdiscussed in section 3,

theReferencerunshaveeffectivelyno loading throughoutthehiatusperiod and arethereforehigh biased

throughoutto an even greaterdegree,lacking even an episodicinjection clearly presentin eitherSato or

Ridleyetal.series.A subsequentstudyfocusing furtheron volcanicforcingsiscurrentlybeing undertaken.

6.1.3.Aerosols

First,theestim ateofforcing dueto theapplied anthropogenicaerosolsin section 3isincom pleteasitdoes

notaccountforthe directeffectofaerosolson longwave radiation,which isassum ed to be sm all.Further,

whileitaccountsforchangesin thecloud liquid waterpath,itdoesnotaccountforeitherchangesin cloud

fractionortheindirecteffectofaerosolsincoldorm ixed-phaseclouds(thelatterhavingalow levelofunder-

standing)[e.g.,seeKirkevåg etal.,2013].

Figure 15.Solarforcing in W m 2 ofthe Reference forcing from NorESM(blue),theInstitutPierreSim on Laplace(black),theSOlarRadiation andClim ateExperim ent(green),and theSensitivityforcingsfrom PhysikalischM eteorologischesObservatorium Davos(red).



The aerosolissuesastheypertain to the presentanalysiswere highlighted in m anyareasofthe recentFifth

Assessm entReportwith,in particular,a whole chapterconcerned with aerosolsand clouds[Boucheretal.,

2013].Theradiativeeffectsofaerosolsrem ain an areaofactiveresearch ashighlighted byrecenthigh pro le

workon theissueby,forexam ple,Shindel[2014]and Stevens[2015].Relevantprincipalaspectsto thepresent

studyare asfollows.Hartm ann etal.[2013]concluded thatwhile regionalchangeswere apparentin regions

withsuf cientinsiturecords,thesearegrosslyincom pleteandthatcon denceinsatellite-basedaerosoloptical

depth recordsislow.Myhreetal.[2013]highlighted large uncertaintyin historicalaerosolforcingsand their

radiativeeffects,and thatthesecontinued to bethepredom inantsourceofuncertaintyin historicalforcings.

Kirtman etal.[2013,section 11.3.6.1]cautioned thatthe aerosolem ission-RCP scenariosin the shortterm

m ay notbe realistic,rem oving the aerosolburden too quickly.They furthernote thatwhen and where the

aerosolsareem itted m aybeim portant,so aconsideration oftheeffectofaerosolsshould notnecessarilybe

consideredinterm ssolelyoftheglobalm eanRF.Postdatingthereport,Shindel[2014]suggeststhattheef cacy

ofaerosolforcingsisunlikelytobeinvariantinspaceandtim e.Collinsetal.[2013]highlightaroughlyequalsplit

betweenm odelsthatprescribeaerosolburdensdirectlyorcalculatethem interactivelybaseduponprescribed

em issions(seetheirTable2.1).TheNorESM m odelconsidered herein usesprescribed em issions.

Given thatNorESM iscon gured such asto derive aerosolconcentrationsinteractively from em ission-based

driverancillaries,the use ofdirectobservationally based aerosolburdens was precluded from the outset.

Historicalem ission-based scenariosdepend upon the veracityofnationalem issionsinventorieswhich cannot

betakenasagiven.Inparticular,nationalem issionsinventoriestendtoconcernindustrialscalepollutionrather

thandom esticsourcepollution.Thism ayleadtosignicantunderreportingofem issionsincountrieswithheavy

dom esticem issionssuch asChinaandIndia.Thererem ainsconsiderableuncertaintyonwhatthecorrectem is-

sionswereandthelackofadequateobservationalcapabilitiesofspeciatedaerosolsgloballycom binedwiththeir

shortlifetim eprecludeseitheratop-downorabottom -upclosureofthebudgetasispossibleform anyLLGHGs.

Duringtheexperim entalsetup,anadditionalpossiblesetofem issionsdriverswasconsidered.However,this

yielded burdensofthe speciated aerosolsthatwere substantially and system atically differentto the CM IP

runs in allcases.Use ofsuch ancillaries,while they m ay wellbe closerto the truth,would have led to

large-scaleshockterm sbeing applied to them odelin term softheRF.W ecarefullyconsidered thepossible

realism oftheaerosoldriversweused in Sensitivity.Such an assessm entislim ited bythequestionablevera-

cityoftheRCPscenarios’short-term aerosolbehavior[Kirtm anetal.,2013].Figure16com parestheReference

and Sensitivity burdensto those forallfourRCP scenarios.Underthe assum ption thatthe RCP scenarios

describeplausibledecadalscalevariation,theappliedSensitivityforcingsdonotappeartobegrosslyunrea-

listic,although thePOM concentrationsaresom ewhatoutsidetheRCP bounds.

W econcludethatofthevariousforcingsaltered in thiswork,both theem issionsand theconcentrationsof

tropospheric aerosols have the greatestuncertainties associated with them ,which com e from both the

uncertaintiesin observations/em issionsofaerosolsand from the RCP scenariosdue to uncertaintiesin the

prediction offuture em issions.One studyhasshown thatincreased sulfurem issionsin Asia are im portant

at least regionally,since the em itted sulfur counteracts the anthropogenically driven warm ing [Neely

etal.,2013];however,recentwork by Murphy [2013]suggeststhatthe shiftofpollution from Europe to

South EastAsiahashad little effecton the clear-skyradiative forcing.The uncertaintiesin aerosolsm aybe

com parableto thecom bined effectsofchangesin thesolarand volcanicforcingsoverrecentyears,asseen

in thisstudy.In theSensitivityensem ble,theeffectstend to cancelwhich m ay,orm aynot,re ectthereal-

world situation.Only im proved aerosolem issionsancillariesbased upon m ore rigorousunderstanding of

the historicalrecord could close this issue.One aspect of the aerosolconcentrations that is m issing

from the CM IP5 sim ulationsentirely isthe interannualvariability which can be quite large and im portant,

given thesubannualresidencetim escaleofaerosols.Thisaspectcould beresolved iftheaerosolconcentra-

tionsorem issionswere prescribed in the m odelsforeach yearinstead offoreach decade.Innovationsin

reanalysesofaerosolsunderthe M onitoring Atm osphericCom position and Clim ate (M ACC)projectand its

follow-onsm ayperm itsuch im proved ancillariesin thefuture,atleastsincesom epointin the1980s.

6.2.Ensem ble Experim entalDesign

Inthisstudy,wehavecreatedtwo30-m em berensem blestoinvestigatetheim pactoftheSensitivityforcings

and the internalvariabilityasdiagnosed bythe NorESM m odel.W hile thisisalarge num berofensem bles

com pared to the num berofhistoricalruns subm itted to the CM IP5 archive forany single m odel,these



ensem blesare stillnotsuf cientlylarge to sam ple the entire solution space ofpossible realizationsforthis

m odelorto fully explore the effectsofim perfectknowledge ofthe forcingsoverthe period (section 6.1).

Given thechanceto repeattheexperim ent,therearecertain thingsthatwewould do differently.

IftheOHC initialstateisim portantforcorrectlysam pling thesolution spaceoftheinternalvariability[Meehl

etal.,2014],then starting from justthreestatesofOHC m ightnotbesuf cientunlessoneofthesestatesby

chance wasreasonablyproxim alto the 1980sOHC state.Furtherm ore,itm aytake longerthan the 18year

spin-up used here to attain reasonable spread in OHC statesin the abyssaloceans.Itwould be preferable

to start from a broaderrange ofinitialOHC states to betterassess whether the ocean initialstate is

potentiallyim portant.

W ith regardsto ensem ble design,when the runswere started,we had notdiagnosed the RF ofthe two

ensem bles.Given the large changesin the ancillariesform anyofthe forcings,itwasexpected thatthe RF

would differsubstantiallygloballyand certainlyregionally;and hence,two distinctpathwayswould suf ce

to answeratleastto rstorderwhatroleforcing m ayplay.In hindsight,theresulting globalRFissom ewhat

sim ilar(section 3.6).Itm ay wellbe the case thatallofthe applied ancillariesin Sensitivity are reasonably

proxim alto the unknown truth and thatthe CM IP-5 runsthrough fortuitouscalculation oferrorsgotclose

to this.Butitwould be incredibly naïve to place any faith in such an inference,given the uncertaintiesin

m anyoftheforcings(section6.1).Ifweweretorepeattheanalysis,wewouldrunasom ewhatlargernum ber

1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 20202000

2200

2400

2600

2800

3000

3200

3400

year

1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 20201

1.1

1.2

1.3

1.4

1.5

year

1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 20202

2.5

3

3.5

4

4.5

5

year

Figure 16.Com parison ofthetwo applied setsofaerosolem ission forcingsto thosefortheentiresuiteofRCP scenariosthrough 2020.



ofensem bleswithsom ewhatsm allerpopulationssothatwecould betterexploreforcinguncertaintyeffects

such thatwehad three orfourdifferentSensitivityensem blesand hence 4 or5 degreesoffreedom in the

appliedforcings.Effortstoexploreuncertaintiesinsolarirradiance,addressinadequaciesinvolcanicforcings

thathaveonlyveryrecentlybecom eapparent,and m orefullyconsideraerosolem issionsuncertaintywould

beprioritized.

Perhaps,m ostim portantly,theuseofasinglem odelinthisstudyensuresthattheresultsdonotsam plethe

entireparam eterspaceofrealizationsofthehiatusobtainablebyclim atem odels,ingeneral,undertherange

offorcingswe considered.The Sensitivityforcingsused in thisprojecthave been m ade available through

supplem entary m aterialboth in the interestsoftransparency in ourwork and in the hopesthatthey m ay

beofusetootherm odelinggroupsinterestedinperform ingcom parablestudiesoftheirown.W ewouldalso

encouragegroupsto im proveupon ourexperim entaldesign.

7.Sum m ary

Variousforcingsin theNorwegian Earth System m odelwerem odied to bring them in linewith new obser-

vationalorem ission-based estim ates.Thesewereappliedtoa30-m em berSensitivityensem blecoveringthe

period of1980until2012.A sim ilar30-m em berReferenceensem blewascreated using theCM IP5historical

forcingsextended with RCP8.5.Exploratoryanalysishasrevealed thattheinclusion ofthese alternative for-

cings did not greatly alter the ENSO variability or Arctic sea ice extent between the Reference and

Sensitivityruns.Ocean HeatContentdid show an increased positivetrend in responseto theSensitivityfor-

cings.ENSO behaviorissim ilarto thatobserved,whileseaiceextentshowsm arked errorsin them ean state

and trendsin allseasons.

These ensem blesprovide aunique,to ourknowledge,toolsetthatenablesan investigation ofthe relative

rolesofpositedforcingm isspecications(totheextentdescribedbytwoofabroadrangeofpossibleforcing

histories)and internalm odelvariabilitym echanism sin explaining the hiatusin surface tem peraturesfora

single clim ate m odel.Asan initial“ rstlook”atthe im pactofthese Sensitivity forcings,the annualm ean

globaltem peratureforthetwo ensem blesisgiven in Figure17.Thechangesto theforcings,including solar

and aerosols,appearto havelittleim pacton theglobalm ean tem peraturetrends.A thorough investigation

ispresented in Part2 ofthisstudy[Thorneetal.,2015],where the ensem blesare com pared to various

observationaldata sets,and both ensem ble m ean behaviorand individualm em bersofthe ensem bles

are discussed.

ReferencesAllan,R.P.,C.Lui,N.G.Loeb,M .D.Palm er,M .Roberts,D.Sm ith,andP.L.Vidale(2014),Changesinglobalnetradiativeim balance1985–2012,Geophys.Res.Lett.,41,5588–5597,doi:10.1002/2014GL060962.

Bentsen,M .,etal.(2013),TheNorwegian Earth System M odel,NorESM 1-M — Part1:Description and basicevaluation,Geosci.ModelDev.Discuss.,5,2843–2931,doi:10.5194/gm dd-5-2843-2012.

Boucher,O.,etal.(2013),Cloudsand aerosols,inClim ateChange2013:ThePhysicalScienceBasis.ContributionofW orkingGroupItotheFifthAssessm entReportoftheIntergovernm entalPanelon Clim ateChange,edited byT.F.Stockeretal.,pp.571–658,Cam bridgeUniv.Press,Cam bridge,U.K.,and New York,doi:10.1017/CBO9781107415324.016.

Brown,P.T.,W .Li,L.Li,and Y.M ing (2014),Top-of-atm osphereradiativecontribution tounforced decadalglobaltem peraturevariabilityinclim atem odels,Geophys.Res.Lett.,41,5175–5183,doi:10.1002/2014GL060625.

Chen,X.,and K.K.Tung (2014),Varying planetaryheatsinksled to global-warm ing slowdown and acceleration,Science,345,897–903.

Figure 17.Annualglobalm ean surface airtem perature from the 30-m em berReference (blue)and Sensitivity (red)ensem bles.Plotted arethe95% bootstrap con dencebounds(shading),togetherwith ensem blem eans(solid line).

Acknowledgm entsTheforcingsdataused in thisworkcam efrom theAdvanced GlobalAtm osphericGasesExperim ent(http://agage.eas.gatech.edu/index.htm ),theEarth Research System LaboratoryoftheNationalOceanicand Atm osphericAdm inistration (http://www.esrl.noaa.gov/gm d/ccgg/),theW orldM eteorologicalOf ce’sW orld DataCentreforGreenhouse Gases(http://ds.data.jm a.go.jp/gm d/wdcgg/),theGoddard InstituteofSpaceStudiesattheNationalAeronauticsand SpaceAdm inistration (http://data.giss.nasa.gov/m odelforce/strataer/),and theW orld Radiation CentreatthePhysikalisch-M eteorologischesObservatorium Davos,Switzerland(http://www.pm odwrc.ch/pm od.php?topic=tsi/com posite/SolarConstant),including unpublished datafrom theVIRGO Experim enton thecooperativeESA/NASA M ission SoHO.Theaerosoland aerosol-precursorem issionswereprovided bytheEU projectsEclipseandPegasos(http://eclipse.nilu.no/).ThisworkwaspartoftheEXPLAIN projectwhich wasfunded bytheFastTrackInitiativeschem eattheCentreforClim ateDynam icsattheBjerknesCentreand hasbeen supported bytheResearch CouncilofNorwaythroughprojectEVA (grant229771).Com putationaland storageresourceshavebeen provided byNORTUR(nn2345k)and NorStore(ns2345k).Theauthorswould liketo thankHongm eiLiforproviding codeand adviceforFigures8and 9,putting NorESM intocontextwith otherCM IP5m odels,andIselin M edhaug fordiscussionsregard-ing solarforcings.PaulStackhouse,Richard Allan,and Norm an Loeb areacknowledged fortheirexpertiseonTOA observations.Gavin Schm idtprovided adviceon theSchm idtetal.analysis.Drew Shindellprovided usefulinputon how to calculateradiativeequivalentforcingsforvolcanicandaerosolforcings.AlfKirkevåg helped setup and provided adviceon theinter-pretation ofthem odelrunsto derivetheradiativeim pactsoftheaerosolchanges.Jochem M arotzkeprovidedadviceon NorESM in thecontextoftheCM IP-5 ensem ble.Sim on Good pro-vided adviceon observed OHC recordssuitability.Finally,wethankGeraldM eehland two anonym ousreviewersforthorough and constructive reviews,which served to im provethepaper.



Cohen,J.L.,J.C.Furtado,M .Barlow,V.A.Alexeev,and J.E.Cherry(2012),Asym m etricseasonaltem peraturetrends,Geophys.Res.Lett.,39,L04705,doi:10.1029/2011GL050582.

Collins,M .,etal.(2013),Long-term clim atechange:Projections,com m itm entsand irreversibility,in ClimateChange2013:ThePhysicalScienceBasis.ContributionofW orkingGroupItotheFifthAssessm entReportoftheIntergovernmentalPanelonClimateChange,editedbyT.F.Stockeretal.,pp.1029–1136,Cam bridgeUniv.Press,Cam bridge,U.K.,and New York,doi:10.1017/CBO9781107415324.024.

Cowtan,K.,and R.W ay(2014),CoveragebiasintheHadCRUT4tem peratureseriesand itsim pacton recenttem peraturetrends,Q.J.R.Met.Soc.,140,1935–1944,doi:10.1002/qj.2297.

Dessler,A.E.,M .R.Schoeberl,T.W ang,S.M .Davis,and K.H.Rosenlof(2013),Stratosphericwatervaporfeedback,Proc.Natl.Acad.Sci.U.S.A.,110,18,087–18,091.

England,M .H.,S.M cGregor,P.Spence,G.A.M eehl,A.Tim m erm ann,W .Cai,A.S.Gupta,M .J.M cPhaden,A.Purich,and A.Santoso (2014),Recentintensication ofwind-driven circulation in thePacicand theongoing warm ing hiatus,Nat.Clim .Change,4,222–227.

Estrada,F.,P.Perron,and B.M artínez-López(2013),Statisticallyderived contributionsofdiversehum an in uencesto twentieth-centurytem peraturechanges,Nat.Geosci.,6,1050–1055,doi:10.1038/ngeo1999.

Fetterer,F.,K.Knowles,W .M eier,and M .Savoie(2002),updated daily.Sea IceIndex.1980to 2012,NationalSnow and IceDataCenter,Boulder,Colo.,doi:10.7265/n5qj7f7w.

Flato,G.,etal.(2013),EvaluationofClim ateM odels,inClim ateChange2013:ThePhysicalScienceBasis.ContributionofW orkingGroupItotheFifth Assessm entReportoftheIntergovernm entalPanelon Clim ateChange,edited byT.F.Stockeretal.,pp.741–866,Cam bridgeUniv.Press,Cam bridge,U.K.,and New York,doi:10.1017/CB09781107415324.020.

Fröhlich,C.(2006),Solarirradiancevariabilitysince1978:Revisionofthe{PM OD}com positeduringsolarcycle21,SpaceSci.Rev.,125,53–65,doi:10.1007/s11214-006-9046-5.

Fröhlich,C.,andJ.Lean(1998),Cyclesandtrendsinthepasttwodecadesandassociatedclim atechangeuncertainties,Geophys.Res.Lett.,25,4377–4380,doi:10.1029/1998GL900157.

Fyfe,J.,N.P.Gillett,and F.W .Zwiers(2013),Overestim ated globalwarm ing overthepast20 years,Nat.Clim .Change,3,767–769.Gent,P.R.,etal.(2011),TheCom m unityClim ateSystem M odelversion 4,J.Clim .,24,4973–4991,doi:10.1175/2011JCLI4083.1.Gregory,J.M .(2010),Long-term effectofvolcanicforcingonoceanheatcontent,Geophys.Res.Lett.,37,L22701,doi:10.1029/2010GL045507.Hansen,J.,M .Sato,R.Ruedy,A.Lacis,and V.Oinas(2000),Globalwarm ing in thetwenty- rstcentury:An alternative scenario,Proc.Natl.Acad.Sci.U.S.A.,97,9875–9880.

Hansen,J.,etal.(2005),Ef cacyofclim ateforcings,J.Geophys.Res.,110,D18104,doi:10.1029/2005JD005776.Hartm ann,D.L.,etal.(2013),Observations:Atm osphereand surface,in Clim ateChange2013:ThePhysicalScienceBasis.Contribution ofW orkingGroupItotheFifthAssessm entReportoftheIntergovernm entalPanelonClim ateChange,editedbyT.F.Stockeretal.,pp.159–254,Cam bridgeUniv.Press,Cam bridge,U.K.,and New York,doi:10.1017/CBO9781107415324.008.

Hawkins,E.,andR.Sutton(2009),Thepotentialtonarrow uncertaintyinregionalclim atepredictions,Bull.Am .Meteorol.Soc.,90,1095–1107,doi:10.1175/2009BAM S2607.1.

Honda,M .,J.Inoue,and S.Yam ane(2009),In uenceoflow Arcticsea-icem inim aon anom alouslycold Eurasian winters,Geophys.Res.Lett.,36,L08707,doi:10.1029/2008GL037079.

Hubber,M .,and R.Knutti(2014),Naturalvariability,radiativeforcing and clim ateresponsein therecenthiatusreconciled,Nat.Geosci,7,651–656.

Iversen,T.,etal.(2013),TheNorwegian Earth System M odel,NorESM 1-M — Part2:Clim ateresponseand scenario projections,Geosci.ModelDev.,6,389–415.

Karl,T.,A.Arguez,B.Huang,J.Lawim ore,J.M cM ahon,M .M enne,T.Peterson,R.Vose,andH.Zhang(2015),Possibleartifactsofdatabiasesintherecentglobalsurfacewarm ing hiatus,Science,doi:10.1126/science.aaa5632.

Kaufm ann,R.,H.Kauppi,M .M ann,and J.Stock(2011),Reconciling anthropogenicclim atechangewith observed tem perature1998–2008,Proc.Natl.Acad.Sci.U.S.A.,108,11,790–11,793.

Kirkevåg,A.,etal.(2013),Aerosol-clim ateinteractionsin theNorwegian Earth System M odel— NorESM 1-M ,Geosci.ModelDev.,6,207–244.Kirtm an,B.,etal.(2013),Near-term clim atechange:Projectionsand predictability,in Clim ateChange2013:ThePhysicalScienceBasis.Contribution ofW orking Group Ito theFifth Assessm entReportoftheIntergovernm entalPanelon Clim ateChange,edited byT.F.Stockeretal.,pp.953–1028,Cam bridgeUniv.Press,Cam bridge,U.K.,and New York,doi:10.1017/CBO9781107415324.023.

Klim ont,Z.,S.J.Sm ith,andJ.Cofala(2013a),Thelastdecadeofglobalanthropogenicsulfurdioxide:2000–2011em issions,Environ.Res.Lett.,8,014003,doi:10.1088/1748-9326/8/1/014003.

Klim ont,Z.,etal.(2013b),ECLIPSE V4a:Globalem ission data setdeveloped with the GAINS m odelforthe period 2005 to 2050 Keyfeaturesand principaldata sources,IIASA -Eclipse Technicalnote.[Available athttp://eccad.sedoo.fr/eccad_extract_interface/doc/pdf/ECLIPSE-V4a_info.pdf.]

Kosaka,Y.,and S.Xie(2013),Recentglobal-warm ing hiatustied to equatorialPacicsurfacecooling,Nature,501,403–407,doi:10.1038/nature12534.

Krivova,N.A.,L.E.A.Vieira,and S.K.Solanki(2010),ReconstructionofsolarspectralirradiancesincetheM aunderm inim um ,J.Geophys.Res.,115,A12112,doi:10.1029/2010JA015431.

Lean,J.(2000),Evolution oftheSun’sspectralirradiancesincetheM aunderm inim um ,Geophys.Res.Lett.,27,2425–2428,doi:10.1029/2000GL000043.

Lean,J.L.,andD.H.Rind(2009),How willEarth’ssurfacetem peraturechangeinfuturedecades?,Geophys.Res.Lett.,36,L15708,doi:10.1029/2009GL038932.

M arotzke,J.,and P.M .Forster(2015),Forcing,feedbackand internalvariabilityin globaltem peraturetrends,Nature,517,565–570,doi:10.1038/nature14117.

M eehl,G.A.,J.Arblaster,J.Fasullo,A.Hu,and K.Trenberth (2011),M odel-based evidenceofdeep-ocean heatuptakeduring surface-tem peraturehiatusperiods,Nat.Clim .Change,1,360364.

M eehl,G.A.,A.Hu,J.N.Arblaster,J.Fasullo,and K.Trenberth (2013),Externallyforced and internallygenerated decadalclim atevariabilityassociated with theInterdecadalPacicOscillation,J.Clim .,26,7298–7310,doi:10.1175/JCLI-D-1200548.1.

M eehl,G.A.,H.Teng,and J.M .Arblaster(2014),Clim atem odelsim ulationsoftheobserved early-2000shiatusofglobalwarm ing,Nat.Clim .Change,4,898–902.

M ori,M .,M .W atanabe,H.Shiogam a,J.Inoue,and M .Kim oto(2014),RobustArcticsea-icein uenceonthefrequentEurasian cold wintersinpastdecades,Nat.Geosci.,7,869–873.

M orice,C.P.,J.J.Kennedy,N.A.Rayner,andP.D.Jones(2012),Quantifyinguncertaintiesinglobalandregionaltem peraturechangeusinganensem bleofobservationalestim ates:TheHadCRUT4 dataset,J.Geophys.Res.,117,D08101,doi:10.1029/2011JD017187.



M urphy,D.M .(2013),Littlenetclear-skyradiativeforcing from recentregionalredistribution ofaerosols,Nat.Geosci.,6,258–262.M yhre,G.,etal.(2013),Anthropogenicand naturalradiativeforcing,in Clim ateChange2013:ThePhysicalScienceBasis.Contribution ofW orking Group Ito the Fifth Assessm entReportofthe Intergovernm entalPanelon Clim ate Change,edited by T.F.Stockeretal.,pp.659–740,Cam bridge Univ.Press,Cam bridge,U.K.,and New York,doi:10.1017/CBO9781107415324.018.

Neely,R.R.,etal.(2013),RecentanthropogenicincreasesinSO2from Asiahavem inim alim pactonstratosphericaerosol,Geophys.Res.Lett.,40,999–1004,doi:10.1002/grl.50263.

Outten,S.,and I.Esau (2012),A linkbetween Arcticseaiceand recentcooling trendsoverEurasia,Clim .Change,110,1069–1075.Petoukhov,V.,and V.Sem enov(2010),A linkbetween reduced Barents-Karaseaiceand cold winterextrem esovernorthern continents,J.Geophys.Res.,115,D21111,doi:10.1029/2009JD013568.

Planton,S.(Ed.)(2013),IPCC,2013:AnnexIII:Glossary,in Clim ateChange2013:ThePhysicalScienceBasis.Contribution ofW orkingGroupItotheFifth Assessm entReportoftheIntergovernm entalPanelon Clim ateChange,edited byT.F.Stockeretal.,pp.1447–1466,Cam bridgeUniv.Press,Cam bridge,U.K.,and New York,doi:10.1017/CBO9781107415324.031.

Prinn,R.G.,etal.(2000),A historyofchem icallyandradiativelyim portantgasesinairdeducedfrom ALE/GAGE/AGAGE,J.Geophys.Res.,105,17,751–17,792,doi:10.1029/2000JD900141.

Rhein,M .,etal.(2013),Observations:Ocean,in Clim ateChange2013:ThePhysicalScienceBasis.Contribution ofW orking Group Ito theFifthAssessm entReportoftheIntergovernm entalPanelon Clim ateChange,edited byT.F.Stockeretal.,pp.255–316,Cam bridgeUniv.Press,Cam bridge,U.K.,and New York,doi:10.1017/CBO9781107415324.010.

Ridley,D.A.,etal.(2014),Totalvolcanicstratosphericaerosolopticaldepthsand im plicationsforglobalclim atechange,Geophys.Res.Lett.,41,7763–7769,doi:10.1002/2014GL061541.

Risbey,J.S.,S.Lawandowsky,C.Langlais,D.P.M onselesan,T.J.O’Kane,and N.Oreskes(2014),W ell-established globalsurfacewarm ing inclim atem odelprojectionsselected forENSO phase,Nat.Clim .Change,4,835–840.

Santer,B.D.,etal.(2014),Volcaniccontribution to decadalchangesin tropospherictem perature,Nat.Geosci.,7,185–189.Sato,M .,J.E.Hansen,M .P.M cCorm ick,and J.B.Pollack(1993),Stratosphericaerosolopticaldepth,1850–1990,J.Geophys.Res.,98,22,987–22,994,doi:10.1029/93JD02553.

Schm idt,G.A.,D.T.Shindell,and K.Tsigaridis(2014),Reconciling warm ing trends,Nat.Geosci.,7,158–160.Sexton,D.M .H.,H.Grubb,K.P.Shine,and C.K.Follard (2003),Designand analysisofclim atem odelexperim entsfortheef cientestim ationofanthropogenicsignals,J.Clim .,16,1320–1336.

Shindel,D.T.(2014),Inhom ogeneousforcing and transientclim atesensitivity,Nat.Clim .Change,4,274–277,doi:10.1038/nclim ate2136.Solom on,S.,K.H.Rosenlof,R.W .Portm ann,J.S.Daniel,S.M .Davis,T.J.Sanford,and G.Plattner(2010),Contributionsofstratosphericwatervaporto decadalchangesin therateofglobalwarm ing,Science,327,1219–1223.

Solom on,S.,J.S.Daniel,R.R.NeelyIII,J.-P.Vernier,E.G.Dutton,and L.W .Thom ason (2011),Thepersistentlyvariable“Background”stra-tosphericaerosollayerand globalclim atechange,Science,333,866–870.

Stevens,B.(2015),Rethinking thelowerbound on aerosolradiativeforcing,J.Clim .,28,4794–4819,doi:10.1175/JCLI-D-14-00656.1.Taylor,K.E.,R.J.Stouffer,and G.A.M eehl(2012),An overview ofCM IP5 and theexperim entdesign,Bull.Am .Meteorol.Soc.,93,485–498,doi:10.1175/BAM S-D-11-00094.1.

Thorne,P.,S.Outten,I.Bethke,andØ.Seland(2015),Investigatingtherecentapparenthiatusinsurfacetem peratureincrease:2.Com parisonofm odelensem blesto observationalestim ates,J.Geophys.Res.Atm os.,120,doi:10.1002/2014JD022805.

Tsutsum i,Y.,K.M ori,T.Hirahara,M .Ikegam i,andT.J.Conway(2009),Technicalreportofglobalanalysism ethodform ajorgreenhousegasesbytheW orld DataCenterforgreenhousegases,GAW Rep.,184.

UNEP OzoneSecretariat(1987),TheM ontrealprotocolon substancesthatdepletetheozonelayer.[Availableathttp://ozone.unep.org/Publications/m p_ nal_act/M ontreal_Protocol_Final_Act_1987-E.pdf.](lastaccessed October2014).

Urban,J.,S.Lossow,G.Stiller,and W .Read (2014),Anotherdrop in watervapor,EosTrans.AGU,95,245–246,doi:10.1002/2014EO270001.W ang,Y.M .,J.Lean,and N.R.Sheeley(2005),M odeling theSun’sm agnetic eld and irradiancesince1713,Astrophys.J.,625,522–538.W atanabe,M .,H.Shiogam a,H.Tatebe,M .Hayashi,M .Ishii,and M .Kim oto (2014),Contribution ofnaturaldecadalvariabilityto globalwarm ing acceleration and hiatus,Nat.Clim .Change,4,893–897.



MURAL - Maynooth University Research Archive Library PUBL … · 2021. 7. 23. ·...

Documents

Transcript of MURAL - Maynooth University Research Archive Library PUBL … · 2021. 7. 23. ·...