Midlatitudestormsinamoisterworld:lessonsfromidealizedbarocliniclifecyc - PDF document

Midlatitudestormsinamoisterworld:lessonsfromidealizedbarocliniclifecycleexperimentsJamesF.BoothShuguangWangLorenzoPolvaniReceived:23April2012/Accepted:23July2012/Publishedonline:7August2012Springer-Verlag2012AbstractTheresponseofmidlatitudestormstoglobal J.F.Booth(NASAGoddardInstituteforSpaceStudies,2880Broadway,NewYork,NY10025,USAe-mail:jbooth.atmos@gmail.comS.Wang lowertroposphere,weakeninganimportantsourceofavailablepotentialenergyformidlatitudestorms(e.g.,Holton,Ch.8).Ontheotherhand,anincreaseinlow-levelmoisturewillincreasetheenergyassociatedwithlatentheatreleaseduringstormdevelopment(e.g.,Carlson,pp.216Ð217,orMartin2006,pp.290Ð295).Thepresenceoftwoopposingmechanismsmakestheassess-mentofprojectedstormstrengthdifÞcult,aswenowbrießyreview.First,recentanalysesofmidlatitudestormsusingmodeloutputfromGCMsrununderglobalwarmingscenarios,havereportedcontradictoryresultsastothechangesinextremestormeventsinthefutureclimate.IndependentstudiesusingtheMax-PlankGCM(Bengtssonetal.andtheHadleyCentreGCM(Cattoetal.),foundnoincreaseinthefrequencyorstrengthofextremestorms(basedontherelativevorticityat850hPa).Incontrast,ananalysisthatexaminedalloftheGCMsfromtheWorldClimateResearchProgrammeÕs(WCRPÕs)CoupledModelIntercomparisonProjectphase3(CMIP3)arguedthatthenumberofextrememidlatitudestormswillincreasewithglobalwarming(LambertandFyfe).However,thestormstrengthmetricusedinthelatterstudy,sealevelpressure(SLP)minimum,mayhaveintroducedabias.ThisisbecausetheclimatologicalSLPislowernearthepoles,andtherefore,apolewardshiftinstormtracklocationcanincreasethefrequencyofdeepstorms,evenifthestormsÕcirculationsarenotactuallystronger(Ulbrichetal.Inaddition,otherstudiesoftheCMIP3modelsfoundthatthenumberofextremewindevents(GastineauandSoden)andtheextremeprecipitation(OÕGormanandSchneider)inthemidlatitudesincreaseswithglobalwarming.Bothtypesofeventstypicallyoccurwithinmidlatitudestorms,althoughadirectattributioncannotbemadefromthesestudies.Inadditiontothesedifferences,onemightbeconcernedastotheabilityofGCMstofaithfullycapturethemoistevolutionofmidlatitudestorms,intermsofboththedynamicsandthermodynamics,andtheseconcernsraisesomedoubtsaboutGCMprojectionsoftheresponseofmidlatitudestormstomoisturechanges.AcomparisonofoneCMIP3GCMtocloudsatelliteobservationsfoundthattheGCMhadalowbiasinitsupper-levelcloudswithinmidlatitudestorms(Naudetal.),whichimpliesaweakverticalmoisturetransportinthemodel.Inadditiontomodelphysics,thetypicalresolutionofcurrentGCMsisalsoacommonconcern.Inthestudiesmentionedabove,Bengtssonetal.()andCattoetal.()usedmod-iÞedGCMswithaspatialresolutionof63km,whichisÞnerthanthatoftheCMIP3models.However,typicalweatherforecastingmodelsareusuallyintegratedwithaÞnerhorizontalgridspacing,soastoproperlyresolvemidlatitudestormÕsfrontaldynamics.Therefore,theabilityofGCMstoproperlycapturethemoistevolutionofmid-latitudestormsremainsanopenquestion.Indeed,Cham-pionetal.()andLietal.()foundthatthestrengthofthemostextremestorms,intermsofsurfacewindsandprecipitation,increaseswhenthehorizontalresolutionisincreasedfrom100to25km.Second,adifferentlineofresearchhasbeenprobingthedynamicsofbarocliniclifecycles,goingbacktothelate1970Õs.MostoftheseearlystudieswerecarriedoutinaverysimpliÞedcontext,notablyintheabsenceofmoisture(e.g.,SimmonsandHoskins;Rotunnoetal.Thorncroftetal.).Laterworkhasincludedmoistphysics,andithasbecomeclearthatmoistprocessescangreatlyinßuencestormdynamics.Usingatheoreticalframework,Emanueletal.()foundthatmoisturereducesgrossstaticstabilityandstrengthensbarocliniceddies.Includingmoisturealsoleadstoacontractionoftheregionofverticalmotion(Emanueletal.;Fantini).Numericalmodelingstudieshaveshownthatthepresenceofmoistureleadstostrongerstormsandfasterdevelopment(Gutowskietal.;Mak1994;WhitakerandDavis).Onlyonestudyoflifecycleresponsetochangesfromdrytofullmoistconditions(Pavanetal.)reportedthatmoisturehaslittleimpactoneddykineticenergy(EKE).Incontrast,morerecentstudieshaveshownalargedifferenceinstrengthbetweendryandmoistlifecycles(Boutleetal.2011,hereafterB2011).MorespeciÞcally,BoutleandcollaboratorsfoundthatstormEKEgrowsastheinitialmoisturecontentisincreased.Inexperimentsinwhichtheinitialconditionswerevariedfromzerorelativehumidity(RH)toRHclosetoobservations,thestormEKEincreasedby100%(Boutleetal.).B2011alsoconÞrmedtheimportanceofthewarmconveyorbelt(WCB,e.g.,WernliandDavies;Carlson,pp.302Ð319)inventilatingmoistureoutoftheplanetaryboundarylayer(PBL).B2011alsocorroboratedthenotion,putforthinearliercase-studies,thatlatentheatingassociatedwithrisingmoistairinßu-encesstormstrength(e.g.,,Davisetal.;Stoelinga;Boothetal.;CampaandWernli),andshowedthattheamountofheatingincreasesastheinitialRHisincreased.Finally,theyshowedthatatightrela-tionshipexistsbetweentheventilationofmoistureoutoftheboundarylayerandtheprecipitationthatfallsundertheWCB.Theobjectiveofthispaperistoindependentlyrepro-duceandextendtheB2011study:herewefocus,speciÞ-cally,ontheimpactofmoistureonmidlatitudestorms.Ourgoalistosystematicallyexplorehowstormschangewithincreasingmoisturecontent,asprojectedunderglobalwarming,usingtheidealizedbarocliniclifecycleexperi-ments.Onewaytoaccomplishthisistosimplymodifytheinitialrelativehumidity.Therefore,inaÞrstexperiment, J.F.Boothetal. weexamineasetofintegrationsinwhichtheinitialRHisprogressivelyincreasedfromdrytomoist,asinB2011.Theseexperimentsaremeanttohelpclarifytheimpactofmoisture,whichappearedtobedifferentinthestudiesbyPavanetal.()andB2011.Asanalternativeapproach,wealsoemploythemethodusedinFriersonetal.(),wherebythemoisturecontentoftheatmosphereischangedbyartiÞciallyalteringthedeÞnitionofthesaturationvaporpressure.Thisisaccom-plishedbyintroducinganartiÞcialcoefÞcientintheClausius-Clapeyronequation,andchangingthecoefÞcientsoastoalterthemoisturecontentoftheatmosphereinaprogressiveway,whiletheinitialRHisheldconstant.Anexperimentofthistypemightbemorerelevanttoglobalwarmingprojections,becauseGCMsandtheorypredictthatthechangeinRHwithglobalwarmingwillbesmall(assummarizedinSherwoodetal.).Wepresenttheresultsofthisexperimentbelow,andshowthattheycon-ÞrmtheconclusionsreachedbyincreasinginitialRH.Needlesstosay,itisalsopossibletochangethemois-turecontentoftheatmospherebyvaryingtheinitialtem-perature.However,thatapproachrequiresanumberofchoiceswithrespecttotreatmentofthemeridionalandverticaltemperaturegradients,aswellastheheightofthetropopause,allofwhichmayimpactthestormÕsresponseincomplexways(e.g.,LorenzandDeWeaverOÕGorman).Therefore,wearesettingthoseexperi-mentsasideforaseparatestudy,andheredealwiththesimplerquestionsofmoistureincreasewithouttemperaturechangesPerhapsthekeynoveltyofthispaperisthatwehereuseawidevarietyofmetricstoexaminethestormsÕresponsetoincreasingmoisture.Partofthepresentconfusionintheliteraturestemsfromthefactthatdifferentauthorshavefocusedondifferentmetrics(e.g.,pressurevs.winds),whichmakesitdifÞculttocomparethestudies.HerewegowellbeyondthetraditionalEKE,anddocumentbarocliniclifecyclechangesinseveralimportantaspects:somearemoretraditional,likethecentralpressureminimum(widelyusedinthesynopticweathercommunity),whileothersemphasizethestormimpacts(suchassurfacewindsandextremeprecipitation).Thispaperisdividedintothefollowingsections.Sec-describesthenumericalmodel,theexperimentalsetup,andthestormmetricsusedinthisstudy.TheresultsarereportedinSect.,whichisseparatedinto3experi-ments:theÞrstfocusesonstormsensitivitytoinitialRHasinB2011,thesecondonstormsensitivitytochangingthemoisturecontentfromdrytocurrentlevels,andthethirdreportsonthesensitivitytochangingmoisturefromcurrenttotwotimescurrentmoistureconditions.SectionisadiscussionoftwospeciÞcissues,andSect.offersasummaryandsomeconclusions.2MethodsInthispaper,unlikemanyoftheclassicstudiesofidealizedbarocliniclifecycles,weuseanumericalmodelwithÔÔfullphysicsÕÕ,i.e.,arealisticboundarylayerscheme,parame-terizedcumulusconvection,cloudmicrophysics,andsur-facemoistureandheatßuxes.Nonetheless,wewishtoretaintheßavoroftheearlierstudies,notablytheirsim-plicityandreproducibility.Inthatspirit,wehave(implementedanalyticallyspeciÞedinitialconditionsforboththedynamicalvariablesandwatervaporand(2)wehaveadoptedafreelyavailableandverywidelyused(non-proprietary)model.Inthissectionweinclude:Þrstthedetailsofourmodel,thentheinitialconditionswehaveadopted,andÞnallythedifferentmetricsthatwillbeusedinlatersectionstodescribethestorms.2.1ThemodelWehereusetheWeatherResearchandForecast(WRF)(Skamarocketal.)modelversion3.0.1.WRFsolvesthenon-hydrostaticprimitiveequationsandisconÞguredasachannel:theeastandwestboundariesareperiodic,andthenorthandsouthboundariesaresymmetric.Nearthemodeltop,verticallypropagatinggravitywavesareabsorbedwitha5km-thicklayerwithanimplicitdampingscheme,whichpreventsunphysicalwavereßection.Thehorizontalandverticaladvectionschemesare5thorderand3rdorderaccurate,respectively.MoistureandotherscalarsareadvectedusingapositivedeÞnitescheme(Skamarocketal.ThechannelistakentobeontheCartesianplanewithconstantCoriolisparameter,correspondingto45latitude.UnlessotherwisespeciÞed,themodeldomainhas8150gridpointsinthex,yandzdirections,withhorizontalgridspacing(DX)equalto50km.Thiscorrespondstoacubicdomainof23km,roughly81degreesoflatitudeby51degreesoflongitude.Totestsensitivitytomodelresolution,wealsocarryoutintegrationsusingDX100and200km,whilemaintainingthesamephysicaldomainsize.Fortheparameterizationschemes,weoptforthosewhicharewidelyusedintheweatherforecastingcommunity.CumulusconvectionisparameterizedusingtheKain-Fritsch(KF)scheme(KainandFritschBoundarylayerturbulenceandverticalsub-gridscaleeddydiffusionaretreatedwiththeYSUscheme(Hongetal.);thisisaÞrstorderclosureschemethatincludesanon-localcountergradienttransport.Horizontalsub-grideddymixingisparameterizedusingthe2DSmagorinskyÞrstorderclosurescheme,performedinphysicalspace.Surfacemoistureandheatßuxesareparameterizedfol-lowingMoninÐObukhovsimilaritytheory.Thebulk Midlatitudestormsinamoisterworld microphysicsschemeisthePurdue-Linscheme(Linetal.;ChenandSun).Thisschemehassixspecies:watervapor,cloudwater,cloudice,rain,snowandgraupel.Noradiationisincludedinourmodelintegrations.Finallywenotethat,unlikemosttraditionallifecycleexperimentsbutfollowingB2011,thelowerboundaryconditionistreatedhereasaseasurface.Therefore,itactsasasourceandsinkofsensibleheatandmoisture,aswellasmomentum.Theseasurfacetemperatureischosentobetime-independent,andissettobe0.5Csmallerthantheinitialatmospherictemperatureatthelowestmodellevel.2.2TheexperimentsAlloftheintegrationspresentedinthispaperuseidenticalinitialconditionsfortemperature()andzonalwind(similartothoseinPolvaniandEsler(),exceptthatourinitialjetisplacedonanplanewithCartesiangeometry,asinWangandPolvani().Theseinitialconditionsyieldwindsthataresimilartotheobservedmidlatitudejets,whileremainingbalancedandsimpleenoughtobedescribedanalytically.ToinitiatethelifecycleweapplyaÞniteamplitudeperturbation(1K)tothetemperatureÞeldatallmodellevels,asinPolvaniandEsler().Becausewearenotinitializingthelifecyclewithnormalmodes,theinstabilitytakesseveraldaystoreachthematurestage.Asalreadymentioned,thispaperreportsonthreesetsofnumericalexperiments,eachconsistingof6integrations.TheÞrstsetofintegrationsattemptstoreproducetheresultsofB2011.Todothis,weperformintegrationswithdifferentinitialrelativehumidity(RH),butidenticalwindsandtemperature.Intheseintegrations,theinitialRHisgivenby:15forforZTð1ÞZT,themoisturescaleheight,equals12km.ThisformulaisverysimilartotheoneusedionB2011,withRHthekeyparametertobevaried.Notethat,inourcase,theinitialRHvariesonlywithheight,whereasB2011andPavanetal.()usedinitialRHthatvarieswithbothheightandlatitude.Wechosenottoincludealatitudinaldependenceforsimplicity.ThisanalyticalRHdistributionappearstohaveoriginatedinWeismanandKlempTheRHvaluesinthesixintegrationsofExperiment1are:0,0.2,0.4,0.6,0.8and0.95.WehavenotincludedthevalueRH1.0,becauseitgeneratesaratherstrangestorm,havingtodowithsomepeculiarbehaviorofthecumulusscheme.TherearetwocharacteristicsoftheintegrationsinExperiment1wewishtoemphasize.First,inalloftheexperimentsinwhichwevaryRH,moistureenterstheatmosphereatthelowerboundaryviaevaporationfromtheseasurface.Thus,theRHintegrationisonlydryattheonset,butmoistureenterstheatmosphereasthelifecycleevolves.Second,theintegra-tioninExperiment1withRH0.8willbeconsideredtheÔÔreferenceÕÕintegration,asitcorrespondstoconditionsmostsimilartothoseinobservations,andwillbereferredtowiththattermhereafter.Infact,forthesecondandthirdsetsofintegrationswesetRH0.8,andalterthemoisturecontentbychangingthesaturationvaporpressuredeÞnition.Thisapproachteststhesensitivitytomoistureinamannerthatisrele-vanttoglobalwarming,asdocumentedintheIPCCAR4projections(e.g.,Sherwoodetal.2010),sinceitallowsustokeepRHÞxedwhileincreasingwatervaporintheatmosphere.FollowingFriersonetal.(2006),wemodifytheClausius-Clapeyronequationbymultiplyingthesat-urationpressureofwateratitstriple-pointbyacoefÞ-cient,CSVP,sothat: LvRv 1273 isthelatentheatofvaporization,isthegasconstantofwatervapor,andistemperature.Whencorrespondstotherelationshipobservedinnature,andthemodelÕsmoisturecontentisequaltothereferencecaseinExperiment1.Thesecondsetofintegrations,Experiment2,isananalogtotheRHexperiment,andconsistsofsixinte-grationswithCsettovaluesfrom0to1,inincrementsof0.2.WhenC1.0,thesaturationvaporpressureforanygiventemperatureislessthanthecurrentconditions.SinceweinitializeallintegrationsinthisexperimentwithanidenticalRHproÞle,usingC1.0causestheactualvaporpressuretobelessthancurrentconditions.For0.0,thewatervaporissetzeroatalltimes(sothemodelisentirelydry).AthirdandÞnalsetofintegrations,Experiment3,consistsofsixintegrationswithCrangingfrom1to2inincrementsof0.2.Thisexperimentseekstoshedlightonhowincreasesinmoisturecontentbeyondthecurrentamountwillimpactthedevelopmentandintensityofmidlatitudestorms.Tablesummarizesthe3setsofintegrations. Table1ExperimentdetailsInitialrelativehumidityRH(seeEq.MoisturecontentparameterC(seeEq.Experiment1[0.0,0.2,0.4,0.6,0.8,0.95]1.0Experiment20.8[0,0.2,0.4,0.6,0.8,1]Experiment30.8[1,1.2,1.4,1.6,1.8,2] J.F.Boothetal. 2.3ThemetricsAsmentionedintheintroduction,partofthecurrentcon-fusionastotheimpactofglobalwarmingonmidlatitudestormscanbetracedtothefactthatdifferentstudieshavebeenusingdifferentmetricstodeÞnethestrengthofthestorms[e.g.,therelativevorticityinBengtssonetal.),orextremewindinGastineauandSoden(2009Inlightofthis,wedecidedtoofferacomprehensiveviewofthestormsinourpaper,soastocapturemostoftheimportantaspects.Tothisend,ouranalysiswillfocusonÞvemetrics:(1)theeddykineticenergy(EKE);(2)thestormcentralpressureminimum;thestrongest99thpercentileof(3)thesurfacewindspeedand(4)theprecipitationrates;and(5)theaccumulatedprecipitation.TheÞrstmetric,EKEisthetraditionalmetricreportedinlifecyclestudies(e.g.,SimmonsandHoskins).Thesecondmetric,thecentralpressureminimumatsealevel,istypicallyusedtotrackthepathofstorms,andcapturesaverticallyinte-gratedresponseoftheatmosphere.Thestrongestsurfacewindsandprecipitationareincludedbecausetheyareofgreatinterestfortheextremeeventsthatmidlatitudestormscanproduce.Wecalculatethemetricsasfollows.TheEKEiscal-culatedfromavolumeintegralofthemassweightededdies,wheretheeddiesarewithrespecttothezonalmean.Thestormcentralpressuremetric(CTR_PRES)isdeÞnedastheminimumsealevelpressurewithintheentiredomain;simplyÞndingtheminimumsufÞces,becauseSLPisasmoothspatialÞeld.Ontheotherhand,thesurfacewinds(WIND99)canincludelargeoutliersfortheirmaxima.Therefore,weusethevalueatthelargest99thpercentiletocapturetheextremewinds,whichwecalcu-lateusingadistributionoftheinstantaneouswindsoverthewholedomainateachoutputinterval.Forthesethreemetrics(EKE,CTR_PRESandWIND99),weanalyzeoutputat12-hintervals.Fortheprecipitation,weconsidertwometrics:(1)thestrongest99thpercentileofhourlyaccumulatedprecipita-tionrates,and(2)thetotalaccumulatedprecipitationoverthefullbarocliniclifecycle.Forthelatter,weoutputtheaccumulatedprecipitationhourly,andcalculatetheratesfromthedifferencesbetweensuccessivevalues.Then,foreachhourlyrate,thetop99thpercentileiscalculatedusingallpointsinthedomain.Whenplottingtherates,weonlyshowtheresultsevery12h,soastominimizethenoisinessoftheplots.Furthermore,weconsiderseparatelytheratesforthelarge-scalescaleandcumulusschemeprecipitation,denotedPRCP99andPRCP99,respectively.ThisispartlymotivatedbytheresultofB2011,whoshowthatWCBprecipitationisdominatedbythelarge-scales,whilecumulusprecipitationmostlyoccurssouthoftheWCB.Forthecomparisonoftheprecipitationfromrunswithdifferentgridsize(DX),wereducealloftheoutputtothe200-kmgridtoensurealike-for-likecomparisonoftheprecipita-tionratesacrossthedifferentDXvalues.Forthesecondprecipitationmetric,wedeÞnethetotalaccumulatedprecipitationasthelarge-scaleplusconvectiveprecipitationfromday0today12,averagedoverthemodeldomain.Wepresenttheresultsforthismetricinthedis-cussionsection,tobettercomparethedifferentexperiments.Finally,forsomeoftheseÞeldsthemaximumormin-imumvalueovertheentirelifecycleisofinterest.ThisisdenotedwiththesubscriptsÔÔMAXÕÕorÔÔMINÕÕ.So,forinstance,theminimumcentralpressureovertheentirelifecycleislabeledCTR_PRES3ResultsBeforereportingontheeffectofmoistureonbarocliniclifecycles,wedescribetheevolutionofourreferenceinte-gration,forwhichthetwokeyparameterstakethevalues0.8andCSVP1.InFig.,weshowtheprecipi-tation(color)andSLP(blackcontours)asthelifecycleevolvesfrom1/4EKEonday6.5(panela),tonear1/2onday8(panelb),andontoEKEonday9.5(panelc).Todisplaythelarge-scaleandcumulusprecipita-tiononthesameÞgure,wemultiplythecumulusprecipita-tionrateby1.Thereissomesmalloverlapinthelocationofthecumulusandlarge-scaleprecipitation,sothecumulusoverlaymaybecoveringoverlarge-scaleprecipitation.AsonecanseeinFig.,thestorminourreferenceintegrationdevelopswithastructurethatistypicalofmidlatitudestorms,asevidencedbythespatialdistribu-tionsoftheSLPandtheprecipitationinFig..Theregionwiththestrongestprecipitationrates(darkred)isfoundbetween35and50,approximately5degreeseastofthedashedgreenlineinFig.b;thisisaregionbetweenthecycloneÕscoldandwarmfronts,inwhichalargeamountofcondensationoccurs.Asdiscussedintheintroduction,themoistprocessesinthisregionaffectthestormstrengththroughdiabaticheating(e.g.,Davisetal.ThegreencurveinFig.ashowstheEKEoftheref-erencebarocliniclifecycleinFig..Thestormbeginsitsexponentialgrowthonday4.5oftheintegration,andattainsitsmaximumstrengthonday9.5,afterwhichitbeginstodecay.Thelengthoftimeittakesourstormtodevelopisatypicaldurationforabarocliniclifecycle(e.g.,B2011).3.1Experiment1:varyingRHHavingestablishedthereferenceintegration,wenowconsiderhowvaryingtheinitialrelativehumidity,RH Midlatitudestormsinamoisterworld showninEq.),affectstheevolutionofthebarocliniclifecycle.Forsimplicity,wewillrefertothissetofintegra-tionsasExperiment1(cf.Table).OnegoalofthisexperimentistovalidateresultsreportedinB2011:wehereuseaverydifferentmodelandsetofparameterizations,soitisnotaprioriobvioushowrobusttheresultstoB2011mightbe,especiallyinviewofthewellknownsensitivityofconvectiveparameterizations.Wealsowishtoextendtheirresults,byanalyzingalargersetofmetricsofthestormresponsetoincreasedrelativehumidity.Tostart,letusconsidertheEKEtime-seriesforeachvalue,from0to0.95.First,itisclearthatEKEincreaseswithRHa).Second,wenotethattheEKEgrowsfasterasRHisincreased.ThekeypointofthisÞgureisthattheEKEdependenceoninitialRHisthesameasthatreportedbyBoutleetal.()andB2011,includingthesmallchangesinEKEforRHRH0.6,andnoapparentchangeforRHRH0.8.Hence,theresultsofthosestudiesarehereindependentlyreproduced.Thephysicalmechanismthroughwhichmoistureaffectsstormstrengthhasbeendiscussedinpreviouscasestudies(e.g.,Reedetal.1993;Stoelinga1996),andweonlysum-marizeithereforcompleteness:higherinitialRHallowstherisingairwithinthestormtosaturateearlierandingreaterabundance.Thisincreasesthediabaticheatinginvicinityofthecyclonecenter,whichstrengthenstheverticalgradientofthediabaticheating,generatingapositiveanomalyinlow-levelpotentialvorticity,whichleadstoafasterandstrongerstormdevelopment.Anupper-level,negativePVanomalyassociatedwiththelatentheatreleaseisalsopresent(e.g.,Martin2006,pp.294Ð300):however,itsimpactissecond-ary,andthereforewedonotdiscussithere. Fig.1Sealevelpressurecontours()andprecipitationrate)fortheintegrationwithRH0.8.ThestormisshownattimesrelativetoitsEKEmax,day6.5(max,day8(andmax,day9.5().SLPcontoursincreasemonotonicallyoutwardsfromthestormcenter,contourinterval:10hPa.Thethickestcontouris1,000hPa.Thedashedgreenlinein()helpsidentifytheWCB(seetext).Precipitationgeneratedbythecumulusschemeisplottedasnegative.Precipitationunits:mm/hour.TheÞguresarezoomedinonthestormcirculationinsteadofshowingthefullmodeldomain.Theordinateandabscissahavebeenlabeledwiththelatitudeandlongitudeequivalenttothegridsize Fig.2TimeevolutionofEKEfortheRHexperiment()andtheintegrationswithdifferentgridspacing().Inbothpanelssolidgreenlineshowsthereferenceintegration.In(isthelowestandisthehighestRH.In(),theshowtheRHcasesandthedashedcurvesshowthecaseswithRHUnitsforEKE:10 J.F.Boothetal. Havingvalidatedpreviousresults,wenowaddressthequestionofsensitivitytohorizontalresolution.Severalversionsofthesameexperiment,i.e.,withRHfrom0to0.95,werecarriedoutatcoarser(DX100and200km)aswellasÞner(DX25km)resolution.TheresultsarepresentedinFig.b.TokeeptheÞgurereadableweonlyshowtwocasesforeachresolution:RHthedriestmemberoftheset,andRH0.8,thereferencecasealreadydiscussed.Differentcolorsdenotedifferentresolutions;thesolidcurvesshowthereferencecase,whiledashedcurvesshowthedriestcase.Twopointsareworthnoting:Þrst,oneseesaconvergenceoftheEKEmaximumnearDX50km.Second,comparingthedashedandsolidcurvesofthesamecolor,oneseesthatinallcasesthereferencestormisstrongerthantheoneinitializedwith0.0,regardlessofthehorizontalgrid-size.Fur-thermore,foreachDX,thechangeinEKEwithRHqualitativelyidenticaltothosewithDX50km(notshown).ThisclearlydemonstratesthatthestrengtheningofthestormÕsEKEwithincreasedwatervaporcontentisarobustresult.Wealsoexaminedthesensitivityoftheresultstochangesintheverticalresolution(notshown),andagainfoundsimilarbehavior.BeyondEKE,wenowwishtoreportontheeffectofincreasingtheinitialRHonothercharacteristicsofadevelopingstorm.InFig.,wediscussthe:stormcentralpressure(CTR_PRES)andthestrongest99thpercentileforsurfacewindspeed(WIND99).ConsiderÞrstFig.a:itisclearthattheminimumpressureovertheentirelifecycledeepensgraduallyandmonotonicallyfrom961to951hPaasRHincreasesfromzeroto0.95.Thus,thecentralpressurerespondstomoistureincreasesinamannersimilartoEKE,exceptthatatlargeRHtheEKEappearstosaturate,whereasthestormCTR_PREScon-tinuestodeepen.NotethatthedifferenceforCTR_PRESbetween0and0.8isabout10hPa,whichissmallerthanchangesreportedbypreviouscasestudiesinwhichthe Fig.3TimeevolutionofCTR_PRESandWIND99,fortheRHexperiment()andtheintegrationswithdifferentgridspacing().InallÞgures,thegreensolidlineisthesamereferenceintegration.In(isthelowestandisthehighestRH.In(),thesolidcurvesshowtheRH0.8casesandthedashedcurvesthecaseswithRH0.UnitsforCTR_PRES:hPa,forWIND99: Midlatitudestormsinamoisterworld latentheatingwasturnedoff(e.g.,20hPainReedetal.andStoelinga).However,onlytheinitialcon-ditionsaredryinourstudy,astheseasurfaceboundaryconditionactsasacontinuousmoisturesource.ThismeansthatthereisatleastsomelatentheatinginallofthestormsinFig.a,evenforRHTheeffectofincreasingtheinitialRHontheextremesofsurfacewindspeedisshowninFig.b,whereweplotWIND99forallthesixintegrations.Aswithcentralpressure,WIND99growsfasterandstrengthensastheinitialmoistureisincreased,evenintheintegrationwith0.95.Hence,inalltheserelatedyetdifferentquantities,moisterstormsdevelopfasterandaregrowstrongerthandryerones.Weagainaddressthequestionofnumericalresolutioninc,d,whereCTR_PRESandWIND99areplottedfor25,20,100and100km,asinFig.b.Acompari-sonofthesolid(referenceRH)anddashed(initiallydry)curvesforanycolorshowsthatthemoisterstormshavedeepercentralpressureminimaandstrongerextremesur-facewinds,forallvaluesofDX(Fig.c,d).Thus,themoisterinitialconditionsgeneratesnotonlylargerEKE,butalsodeepercentralpressureandstrongerwinds,andtheseresultsarerobusttochangesinthegridspacing.Finally,weturnourattentiontotheextremeprecipita-tionmetrics:asmentionedabove,wedocumentseparatelythelarge-scaleprecipitationextremes(PRCP99)andthecumulusprecipitationextremes(PRCP99).Figurea,bshowthetimeevolutionforPRCP99andPRCP99the6integrationsinExperiment1,atthestandardresolu-tionofDX50km.Notsurprisingly,thestrongestlarge-scaleandcumulusprecipitationratesoccurintherunswithlargerRH.Asbefore,Fig.c,dillustratethesensitivityofthesequantitiestoDX.Inaccordwiththestormstrength,bothextremeprecipitationratesincreasewithRHforallhorizontalresolutions(Fig.c,d).WeconcludethediscussionofExperiment1byshow-ing,inFig.,thehourlyaccumulatedprecipitationÞelds Fig.4TimeevolutionofextremeprecipitationPRCP99,fortheRHexperiment()andtheintegrationswithdifferentgridspacing().InallÞgures,thegreensolidlineisthesamereferenceintegration.In(isthelowestandisthehighestRH.In(),thesolidcurvesshowtheRH0.8casesanddashedcurvesshowthecaseswithRH0.Unitsforprecipitationrates:mm/hour J.F.Boothetal. withdifferentDX,atthepointwhenthestormEKEishalfofEKE.AlongtheregionofstrongprecipitationinthecycloneÕswarmsector,thestrengthoftheprecipitationincreasesasDXdecreasesfrom200to100to50km,butisquitesimilarinthemodelswithDX50and25km(panelsaandb).Inlightofthis,Experiments2and3werecarriedoutusing50km.WenotethatsuchhorizontalresolutionissimilartotheonesintheGCMexperimentsanalyzedbyBengtssonetal.(2009)andCattoetal.(3.2Experiment2:varyingCfrom0to1ThesecondandthirdsetsofintegrationsexaminethestormsensitivitytomoisturebyvaryingthesaturationvaporpressurecoefÞcient(C),asdescribedinSect.).TheadvantageofdoingthisresidesinthefactthatmoisturecanbeincreasedwithoutchangesinRH,inaccordancewiththeconditionsprojectedtooccurwithglobalwarming(e.g.,Sherwoodetal.).Also,byusingthismethodtoincreasethemoisturecontent,oneavoidsthecomplicationsinvolvedinchangingthetemperatureinthetroposphere,asdiscussedintheintroduction.Fortheseintegrations,weÞxthevalueofRHat0.8.Thus,theintegrationwithC1isidenticaltothereferencecasedescribedabove.WhenC0,themoisturecontentiszero,andtheatmosphereremainscompletelydryfortheentirelifecycle.Westartbyconsideringtherange01,termedExperiment2(seeTable).Thissetofintegrationsstudiestheeffectthatincreasingwatervaporfromcompletelydrytopresentconditionshasonamidlatitudestorm,usinganalternatemethodtothatofExperiment1.Itservestoval-idatetheconclusionsofExperiment1withadifferentmethod,andtosetthestageforunderstandinghowbarocliniclifecyclesmaychangeunderfuture,moisterconditions.InthethreepanelsofFig.,weplotthetime-evolutionoftheEKE,CTR_PRESandWIND99fortheintegrationsfromExperiment2.WeplottheresultsforallsixintegrationsonthesameÞgure,usingadifferentcolorforeachvalueofCSVPfrom0to1,instepsof0.2.Foreachmetric,thedirectionoftheresponsetomoistureisthesameasExperiment1:thestormstrengthincreasesasweincreasethemoisture,withanappearanceofsaturationinEKEnearCThetakehomemessagehereisthat,evenwhentherelativehumidityiskeptconstant,thestormbecomesstrongerwhenmoisturecontentisincreased.InExperiment2,thestormsalsogrowfasterasmoistureincreases,however,thechangeingrowthrateisnotasdrasticasitisExperiment1(e.g.,Fig.a).ThisisbecauseinExperiment2thetimeittakesfortherisingairtoreachsaturationisnearlythesameineachintegration,sinceallhavethesameinitialRH.NotealsothatthedifferencesinstormstrengthbetweenthedryandthemoistintegrationsarelargerinExperiment2thaninExperiment1(e.g.,compareFigs.a).ThisisbecausethedrieststorminExperiment2(with0)isremainscompletelydrythroughoutitslifecycle,whilesettingRH0inExperiment1onlysetstheinitialmoisturetozero(butlatentheatingstilloccursinthelifecycleasmoistureisaddedviasurfaceevaporation).TheextremeprecipitationratesforExperiment2areshowninFig.,inthesamemannerasthestormstrength.Thelarge-scaleprecipitationPRCP99increaseinstepwithC(panela).ThecumulusprecipitationPRCP99alsoincreaseswithmoisturecontent,eventhoughiszeroforC0.6.Thisisbecausecon-vectiveinstabilityistooweaktoinitiatesub-gridscaledeepconvectioninthecaseswithsmallC Fig.5Sealevelpressure()andprecipitationrate()fordifferentgridspacing(DX),shownatthetimethateachstormÕsEKEofitsmaximum.SLPcontoursincreasemonotonicallyoutwardsfromthestormcenter,startingwith970hPa,contourinterval10hPa.Thethickercontouris1,000hPa.Precipitationgeneratedbythecumulusschemeisplottedasnegative.Precipitationunits:mm/hour.TheaxesusethesameconventionasFig. Midlatitudestormsinamoisterworld Insummarythen,Experiment2showsthatincreasingmoisturestrengthensthestormandtheaccompanyingprecipitationevenwhentheinitialRHisÞxed,andthisresultisrobustacrossallofourstormstrengthmetrics.ThequalitativeagreementbetweentheresultsforExperiments1and2showsthatalteringthemoisturecontentbychangingCdoesnotcreateanyunexpectedoddbehavior,butinfactyieldsmeaningfulresults.Withthisinmind,wenowconsiderhowthestormstrengthisaffectedwhenCC1,tohelpunderstandhowfutureglobalwarmingmightaffectbarocliniclifecycles.3.3Experiment3:varyingCfrom1to2Itiswellestablishedthatatmosphericmoisturecontentwillincreasewithglobalwarming(e.g.,HeldandSodenandthischangeshouldprovideanincreasedmoisturesourceformidlatitudestorms.Toquantifytheprojectedchangeinmoisture,westartbyÞrstcalculatingthechangeinspeciÞchumidityby2100foreachGCMintheCMIP3ScenarioA2(seeSect.AppendixfordetailsonthemodelsusedandCMIP3ScenarioA2).InFig.aweshowboththemultimodelmean1980Ð2000(blackcontours)andthe Fig.6TimeevolutionofEKE(),CTR_PRES()andWIND99)forExperiment2.InallÞgures,thegreenlineisthereferencerepresentsmallerCvaluesandlargerC.Units:EKE:10,CTR_PRES:hPa,WIND99:m/s Fig.7Timeevolutionoftheextremeprecipitationratesatthelarge-scalescale,PCPC99),andfromthecumulusscheme,PCPC99).InallÞgures,thegreenlineisthereferenceintegration,representsmallerCvaluesandrepresentlargerC.Units: J.F.Boothetal. projectedchanges(i.e.,the2080Ð2100meanminusthe1980Ð2000mean,incolor);welimitthelatitudestothedomainusedinourlifecycleexperiments.TheÞgureshowsthatthechangeinmoistureismostpronouncednearthesurface,atlowlatitudes.Inagreementwiththesepro-jections,varyingCcreatesthelargestchangesinmoisturenearthesurfaceinthelowlatitudes.Tostudytheimpactsofthemoistureincreasewithglobalwarming,wecreateathirdsetofintegrations,withvaluesrangingfrom1to2inincrementsof0.2:thissetwillbereferredtoasExperiment3(seeTable).Asbillustrates,theCvalueof1.2createsthemoisturechangesthatarestrongertothosefoundintheCMIP3(Fig.a).However,theCMIP3resultshowsthezonalmeanforalllongitudesaveragedoverwinter,whileourmodelinitialconditionsrepresentasnapshotpriortostormformation.WealsorunintegrationswithlargervaluesofCtodocumentthestormresponsetolargermoistureincreases.NotethatFriersonetal.()usedvaluesofCaslargeas10;however,wefoundthatourmodelbecomesnumericallyunstablewhenlargervaluesofareused.WeÞrstconsiderhowthedynamicalmetricsEKECTR_PRESandWIND99changeasCisincreasedbeyond1.FigureashowsthattheEKEdecreasesasapproaches1.4,levelsoffatC1.6andthenincreasesbeyondthatpoint,suchthattheEKE1and2arethesame.Thisnon-monotonicresponseappearstobeuniquetotheEKE,andwediscussthisinSect.4below.Inaddition,notethatthechangesinforthecaseswithCC1arerelativelysmall10Ð15%ofEKEinthereferenceintegration),comparedtotheresponsefoundinExperiment2(morethan70%ofthereferencecaseinFig.UnlikeEKE,thecentralpressuredeepensmonotonicallywithC,withC2havingthedeepestlow(approximately945hPa,asseeninFig.b).Similarly,thesurfacewindmaximumincreasesmonotonicallyfrom25m/stonearly30m/swithCincreasingfrom1to2c).Thus,inExperiment3,theCTR_PRESandWIND99respondtothemoistureincreaseinthesamemannerastheydidinExperiments1and2.FortheextremeprecipitationratesinExperiment3,theresponsesofthelarge-scaleandcumulusprecipitationdiffer,asillustratedinFig..Thelarge-scaleprecipita-tionrates,PRCP99,increasemonotonicallywithCduringtheinitialgrowthofthestorm(atday6inFig.buttheoverallmaximainPRCP99,whichoccuraroundday8.5,areverysimilarforalloftheintegrations.Incontrast,theextremeprecipitationfromthecumulusscheme,PRCP99,increasesmonotonicallywithCb),whichisinagreementwiththeresponseoftheCTR_PRESandWIND99.4Discussion4.1TheaccumulatedprecipitationHereweconsidertheaccumulatedprecipitation,whichincludesboththelarge-scaleandthecumulusprecipitation,averagedovertheentiredomainandovertheentirelife-cycle.Thismetricoffersatimeandspaceintegratedresponseofthestormtochangesinmoisture,andallowsustoeasilycomparetheresponseinthedifferentexperimentsusingasinglequantity.Tableshowstheaccumulatedprecipitationforeachoftheintegrations.ForExperiment1,theÞrstresulttonoteisthatthe0casehasanon-zerovalue.Thisservesasareminderthatthisintegrationisonlydryintheinitialconditions.Second,asRHisincreased,theaccumulatedprecipitationincreasesgradually,suchthatthereferencecase,RH0.8,hasdoubletheaccumulatedprecipitationofthecasewithRH0.FinallywithRH00.95theaccumulatedprecipitationincreasesdrastically.Interest-ingly,thisismostlyduetoalargeincreaseinthespatialextentofcumulusprecipitationinthelowerlatitudesofthedomain(notshown). Fig.8ZonalmeanspeciÞchumidity()anddifferences(),fortheCMIP3ensembleaverageforwinter(DJF)()andforthemodel().In()thecontoursshowthe1980Ð2000mean;theshadingshows2080Ð2100minus1980Ð2000.In(),thecontoursshowtheinitialconditionsforC1.0,theshadingshowinitialconditionsforC1.25minus Midlatitudestormsinamoisterworld InExperiment2,theaccumulatedprecipitationappearstoincreaseevenmorerapidlywithincreasingC(exceptwithRH0.95,whichisexceptional).Note,however,thatthevaluesofaccumulatedprecipitationareheresmallerthaninExperiment1:sincethemoisturecontentintheCintegrationsislimitedthroughoutthelifecyclewhereas,intheRHintegrations,surfaceevaporationprovidesacontinuousmoisturesourceduringstormdevelopment.Thus,basedontheaccumulatedprecipita-tion,theRH0lifecyclewouldbeapproximatelyequivalenttotheintegrationinExperiment2with0.5.InExperiment3,theaccumulatedprecipitationcontin-uestorisewithC.Thisisaconsequenceofanincreaseinprecipitationatboththecumulus-andlarge-scale(notshown).Notethat,forthelarge-scaleprecipitation,thisresultisdifferentfromtheresponseoftheextremepre-cipitationrates(Fig.),whichappearedtobelargelyinsensitivetoC.Theconclusion,therefore,isthatthechangeinaccumulatedlarge-scaleprecipitationiscausedbyanincreaseinthespatialextentofmoderateprecipita-tionrates,ratherthantheextremes. Fig.9TimeevolutionofEKE(),CTR_PRES()andWIND99)forExperiment3.InallÞgures,thegreenlineisthereferencerepresentsmallerCvaluesandlargerC.TheC1.6integrationisplottedin.Units:EKE:10,CTR_PRES:hPa,WIND99:m/s Fig.10Timeevolutionoftheextremeprecipitationratesatthelarge-scalescale,PCPC99),andfromthecumulusscheme,),forExperiment3.InallÞgures,thegreenlineisthereferenceintegration,representsmallerCvaluesandrepresentlargerC.TheC1.6integrationisplottedinUnits:mm/hour J.F.Boothetal. 4.2HorizontalandverticalscalesInExperiment3,itwasfoundthattheEKEnon-monotonicallytoincreasingCa);thisbehaviorcontraststhatoftheothermetricswehavecon-sidered,allofwhichgrowmonotonicallywithincreasingmoisture.ToshedlightonthisuniqueEKEresponse,weÞrstintroduceanewmetricthatreßectsthevolumeinte-gratedwindÞeld:thetotalkineticenergy(TKE).TheTKEiscalculatedinthesamemannerastheEKE,butthefullwindsareusedinthemassweightedintegral.a,bshowTKEforExperiments2and3.Clearly,TKEincreasesmonotonicallywithmoistureacrossbothexperiments,i.e.,forCincreasingfrom0allthewayupto2.Thus,themoistureresponseofTKEisqualitativelysimilartotheothermetrics.ThemonotonicincreaseofTKEindicatesthatmorekineticenergyisconvertedfrompotentialenergywithincreasedmoisture:however,thenon-monotonicincreaseofEKEsuggeststhatonlyafractionisconvertedtoeddykineticenergy.Clearly,ananalysisoftheenergycycleisneededforfurtherclar-iÞcation,ataskbeyondthescopeofthisstudy.However,theresponseoftheEKEalsomanifestsitselfinchangesinthestructureofthestorm,whichwenextdiscuss.ConsiderÞrstthesea-levelpressureÞeld,shownbytheblackcontoursinFig.,fortheintegrationswith1(panela)andC2(panelb).Comparingthe1,000hPaisobar(thickblackcontour),itisclearthatthemeridionalextentofthestormdecreases,asCchangedfrom1to2.Thiscontractionisintimatelyrelatedtothenon-monotonicbehaviororEKE(Fig.a),sincetheEKEmeasurestheasymmetryofthecirculationwithrespecttothezonalmean.Incontrasttothehorizontalscale,theverticalscaleofthestormisfoundtoincreasewithmoisturemonotonically.ToshowthisweanalyzetheverticalproÞleofthewindÞeld,andplotWIND99versusheight,inFig.,fortheintegrationsinExperiments2and3.TheseextremewindproÞlesarecalculatedatthetimeswheneachstormreachesanEKEvaluethatishalfofitownmaximumEKE.Acrossthetwoexperiments,theheightofthewindmaximumincreaseswithC.Weinterpretthisasanincreaseintheverticalheightofthecyclone.Wereachthesameconclu-sionafteranalyzingtheheightofthetropopause(notshownhere),intheregionofthestormlocatedbetweenthewarmfrontandthecoldfrontandabovethelocationofmaximumprecipitation.UnderstandingtheresponseofthestormÕshorizontalandverticalscalestoincreasedmoistureisstillelusive.Previoustheoreticalstudieshavefoundthatincreasingmoistureleadstoanexcitationofasmallerlengthscales(e.g.,Emanueletal.1987),andthismightexplainthechangeinhorizontalscaledocumentedaboveinourintegrations.However,thesametheoriespredictthattheverticalscaleofthestormshouldalsodecrease.Henceitisnotclearhowtoreconcilethoseearliertheorieswithourresults.EvenmoreperplexingaretheÞndingsofFriersonetal.),whorecentlyaddressedthesesamequestion(inthecontextofanidealizedGCM)byvaryingCaswehavedone.Theyfoundonlyasmallchangeinthehorizontal Table2Domain-averagedaccumulatedprecipitation(mm)Experiment1Experiment2Experiment306.200112.50.27.60.22.21.213.30.49.50.44.91.414.90.611.70.68.51.617.00.812.50.811.31.819.10.9520.4112.5220.9 Fig.11TimeevolutionofTKEforExperiment2()andExperiment3().Thegreencurveisthereferenceintegration,representsmallervaluesandrepresentlargerCvalues.ThecasewithC1.6isshownin.Units:10 Midlatitudestormsinamoisterworld eddylength-scalewhenvaryingCoveramuchbroaderrangethanwehavedonehere.TheysuggestthatthemostlikelyexplanationforthesmalllengthscalechangesintheirmodelmightbeduetoashiftinthelatitudeoftheirmodelÕssubtropicaljet,anexplanationthatdoesnotimmediatelyseemapplicabletoourresults,sinceweusethesameinitialwindsinallintegrations.Insummarythen,theresponseofmidlatitudestormsÕlengthscalesasmoistureisincreasedremainsanopenquestion.Nonetheless,aswehaveamplydocumentedabove,theresponsetoincreasingmoistureinallotherimportantmetrics(notablywindspeedandprecipitation)isclearandrobust:asmoistureisincreasedthestormbecomesstronger.5SummaryandconclusionThepurposeofthisstudyhasbeentoexaminetheresponseofmidlatitudestormstoincreasesinmoisturecontent,ascenarioexpectedundertheglobalwarming.Threenumericalexperimentswithidealizedbarocliniclifecycleswerecarriedout.TheÞrstoneexaminedthestormresponsetodifferentinitialrelativehumidity.InthesecondandthirdonestheinitialRHwasheldÞxed,andmoisturecontentwasvariedusingthecoefÞcientCintheClausius-Clapeyronequation,eitherfromdrytopresentconditions,orforpresenttofutureconditions.Themain,robustresultthatemergesfromtheseexper-imentsisthatasmoistureisincreasednearlyalltheimportantmetricsofstormstrengthincrease.SpeciÞcallythestormÕsintensiÞcationrate,centralpressureminimum,extremesurfacewindsandprecipitation,increasesmono-tonicallywithmoisture.WehavealsoveriÞedthattheseconclusionsarerobusttochangesinthemodelÕshorizontalgridsize,includingmodelswitharelativelycoarsereso-lution,typicalofcurrentgenerationmodelsusedforcli-matechangeprojections.Thissuggeststhatthestormenhancementassociatedwithincreasedmoisturecanbe Fig.12Sealevelpressure)andprecipitationrate),fortheintegrationwith)andCThestormsareshownattimewhenEKEishalfofeachstormÕsEKE:thiscorrespondstoday8for()andday6.5for().SLPcontoursincreasemonotonicallyoutwardsfromthestormcenter,contourinterval10hPa.Thethickercontouris1,000hPa.Precipitationgeneratedbythecumulusschemeisplottedasnegative.Precipitationunits:mm/hour.TheaxesusethesameconventionasFig. Fig.13TimeevolutionofWIND99versusheightforExperiment2()andExperiment3().TheshowsthereferencesmallerCvaluesandrepresentlargerCThecasewithC1.6isshownin.Units:m/s J.F.Boothetal. capturedevenwhentherisingmotionwithinthestormÕswarmsectorandtheindividualfrontsarenotwellresolved.Inaddition,wefoundthatincreasingmoisturebeyondcurrentlevels(asinExperiment3)resultsinstormsthataresmallerinsize,eventhoughthestormÕsstrength,asmea-suredbythecentralpressureorextremesofthesurfacewindsandprecipitationincrease.Whileincreasingmois-turereducesthehorizontalscaleofthestormitalsohelpsgeneratemorevigorousrisingmotioninthewarmsector,whichcausestheheightofthestormtoincrease.Whethersimilarbehaviorcanbefoundinmodeloutputfromclimatechangeprojectionsusingstate-of-the-artgeneralcircula-tionmodelsremainsanopenquestion(Kidstonetal.BarnesandHartmann2012Intermsoftheglobalwarmingprojections,ourresultssuggestthatthemoistureforcingonstormswillnothaveahugeimpactforthestrongeststorms.First,theincreaseinmoistureinGCMsprojectionsisweakerthanthechangecausedbyusingtheCvalueof1.2inourExperiment3).Second,forthatintegrationweÞndonlymodestdifferences(lessthan10%)inEKE,CTR_PRESorWIND99,ascomparedtoC1.Hencethisstudysuggeststhatincreasesinthemoisturecontentunderglobalwarmingcouldonlyleadtoarelativelysmallstrengtheningofthemidlatitudestorms.ThisisinagreementwiththeGCMresultsreportedbyBengtssonetal.()andCattoetal.(),whofoundnoincreaseinthefrequency,orstrength,ofextremestormsinglobalwarmingprojections.Weintentionallylimitedthescopeofthisstudytomoistureintheatmosphere,toallowforafocusedexam-inationoftheimpactofthatonevariableonmidlatitudestorms.Thus,bydesignourstudycannotfullyanswerthequestion:willthestrengthofmidlatitudestormsincreasewithglobalwarming?However,ourstudydoesprovideoneimportantpieceneededtoanswerthatquestion.Needlesstosay,increasingsurfacetemperatures,changingstaticstabilitiesatdifferentlatitudes,stratosphericprocessandothervariablesarelikelytoaffectmidlatitudestormsinthefuture.Wehopetoreportonsomeoftheseaspectsinfuturepapers.WeacknowledgetheWorldClimateResearchProgrammeÕsWorkingGrouponCoupledModeling,whichisresponsibleforCMIP,andwethanktheclimatemodelinggroups(listedintheSect.)forproducingandmakingavailabletheirmodeloutput.ForCMIPtheU.S.DepartmentofEnergyÕsPCMDUprovidescoordinatingsupportandleddevelopmentofsoftwareinfrastructureinpartnershipwiththeGlobalOrganizationforEarthSystemSciencePortals.TheworkofLMPisfunded,inpart,byagrantfromtheUSNationalScienceFoundation.TheworkofJFBisfundedbytheNationalAeronauticsandSpaceAdministration(NASA)postdoctoralprogram.SWthanksJianLufordiscussionsofexperimentdesignatearlierphase.WethankHeiniWernliandananonymousreviewerforusefulsuggestionsthathelpedtoclarifythepresentationofthemainpointsofthiswork.Appendix:DetailsforFig.ForFig.,wedownloaddatafromtheProgramforCli-mateModelDiagnosisandIntercomparison(PCMDI),CMIP3archivefortheIPCCAR4.WeusethehistoricalmodelintegrationstorepresentthetwentiethcenturyandtherunsfromtheGlobalWarmingA2scenarioforthetwentyÞrstcentury.ScenarioA2fromCMIP3correspondstoanincreaseinglobaltemperaturesintherangefromC.TheincreaseinCOassociatedwiththisscenarioseemedpessimisticwhenitwascreatedin2000,butnotanymore.WechoseScenarioA2becauseitcorrespondstoaprojectioninwhicheconomiesmaintainthestatusquo,whichdoesnÕtseemunreasonable.ModelsusedinFig. BCCR:BergenClimateModel(BCM)projectattheBjerknesCentreforClimateResearch. CCCMA:CanadianCentreforClimateModelingandAnalysis,Victoria,BC,Canada. CCSM:CommunityClimateSystemModelproject,supportedbythe,DirectorateforGeosciencesoftheNationalScienceFoundation,andtheOfÞceofBiologicalandEnvironmentalResearchoftheU.S.DepartmentofEnergy. CNRM:CentreNationaldeRecherchesMeteoro-logiques,Meteo-France,Toulouse,France. CSIRO:Atmo-sphericResearch,Melbourne,Australia. ECHAM:MaxPlanckInstituteforMeteorology,Hamburg,Germany. GFDL:USDeptofCommerce/NOAA/GeophysicalFluidDynamicsLaboratory,Princeton,NJ,USA. GISSModelE:NASAGoddardInstituteforSpaceStudiesNewYork,NY,USA. INMCM:InstituteforNumericalMathematics,Mos-cow,Russia. IPSL:InstitutPierreSimonLaplace,Paris,France. CCSR/NIES/FRCGC:CenterforClimateSystemResearch,Tokyo,Japan/NationalInstituteforEnvironmen-talStudies,Ibaraki,Japan/FrontierResearchCenterforGlobalChange,Kanagawa,Japan. MRI:MeteorologicalResearchInstitute,Tsukuba,Ibaraki,Japan. PCM:NationalCenterforAtmosphericResearch,Boulder,CO,USA.ReferencesBarnesEA,HartmannDL(2012)Theglobaldistributionofatmo-sphericeddy-lengthscales.JClim25:3409Ð3416BengtssonL,HodgesKI,KeenlysideN(2009)Willextratropicalstormsintensifyinwarmerclimate?JClim22:2276Ð2301BoothJF,ThompsonL,PatouxJ,KellyKA(2012)SensitivityofmidlatitudestormintensiÞcationtoperturbationsintheseasurfacetemperatureneartheGulfstream.MonWeatherRevBoutleIA,BeareRJ,BelcherSE,BrownAR,PlantRS(2010)Themoistboundarylayerunderamid-latitudeweathersystem.BoundaryLayerMeteorol134:367Ð386 Midlatitudestormsinamoisterworld BoutleIA,BelcherSE,PlantRS(2011)Moisturetransportinmid-latitudecyclones.QJRMeteorolSoc137:360Ð367CampaJ,WernliH(2012)APVperspectiveontheverticalstructureofmaturemidlatitudecyclonesintheNorthernHemisphere.JAtmosSci69:725Ð740CarlsonTN(1998)Mid-Latitudeweathersystems.AmericanMete-orologicalSociety,BostonCattoJL,ShaffreyLC,HodgesKI(2011)NorthernHemisphereextratropicalcyclonesinawarmingclimateintheHiGEMhigh-resolutionclimatemodel.JClim24:5336Ð5352ChampionAJ,HodgesKI,BengtssonLO,KeenlysideNS,EschM(2011)ImpactofincreasingresolutionandawarmerclimateonextremeweatherfromNorthernHemisphereextratropicalcyclones.Tellus63A:893Ð906ChenS-H,SunW-Y(2002)Aone-dimensionaltimedependentcloudmodel.JMeteorolSocJpn80:99Ð118DavisCA,StoelingaMT,KuoY-H(1993)Theintegratedeffectofcondensationinnumericalsimulationsofextratropicalcyclo-genesis.MonWeatherRev121:2309Ð2330EmanuelKA,FantiniM,ThorpeAJ(1987)Baroclinicinstabilityinanenvironmentofsmallstabilitytoslantwisemoistconvection.PartI:two-dimensionalmodels.JAtmosSci44:1559Ð1573FantiniM(1993)Anumericalstudyoftwo-dimensionalmoistbaroclinicinstability.JAtmosSci50:1199Ð1210FriersonDMW,HeldIM,Zurita-GotorP(2006)Agray-radiationaquaplanetmoistGCM.Part1:staticstabilityandeddyscale.JAtmosSci63:2548Ð2566GastineauG,SodenBJ(2009)Modelprojectedchangesofextremewindeventsinresponsetoglobalwarming.GeophysResLett36:L10810.doi:GutowskiWJ,BranscomeLE,StewartDA(1992)Lifecyclesofmoistbarocliniceddies.JAtmosSci49:306Ð319HeldIM,SodenBJ(2006)Robustresponsesofthehydrologicalcycletoglobalwarming.JClim19:5686Ð5699HoltonJR(2004)Anintroductiontodynamicmeteorology,4thedn.ElsevierAcademic,NewYorkHongS-Y,NohY,DudhiaJ(2006)Anewverticaldiffusionpackagewithanexplicittreatmentofentrainmentprocesses.MonWeatherRev134:2318Ð2341KainJS,FritschJM(1993)Convectionparameterizationformeso-scalemodels:theKain-Fritschscheme.In:EmanuelKA,RaymondDJ(eds)Therepresentationofcumulusconvectioninnumericalmodels.AmerMeteorSoc,BostonKidstonJ,DeanSM,RenwickJ,VallisGK(2010)Arobustincreaseintheeddylengthscaleinthesimulationoffutureclimates.GeophysResLett37:L03806LambertSJ,FyfeJC(2006)Changesinwintercyclonefrequenciesandstrengthssimulatedinenhancedgreenhousewarmingexperiments:resultsfromthemodelsparticipatingintheIPCCdiagnosticexercise.ClimDyn26:713Ð728LiF,CollinsWD,WehnerMF,WilliamsonDL,OlsonJG(2011)Responseofprecipitationextremestoidealizedglobalwarminginanaqua-planetclimatemodel:towardsarobustprojectionacrossdifferenthorizontalresolutions.Tellus63A:876Ð883LinY-L,FarleyRD,OrvilleHD(1983)BulkparameterizationsofthesnowÞeldinacloudmodel.JClimApplMeteorol22:1065Ð1092LorenzDJ,DeWeaverET(2007)TropopauseheightandzonalwindresponsetoglobalwarmingintheIPCCscenariointegrations.JGeophysRes112:D10119.doi:MakM(1994)Cyclogenesisinaconditionallyunstablemoistbaroclinicatmosphere.Tellus46A:14Ð33MartinJ(2006)Mid-latitudeatmosphericdynamics.Wiley,WestNaudCM,DelGenioAD,BauerM,KovariW(2010)CloudverticaldistributionacrosswarmandcoldfrontsinCloudSatÐCALIPSOdataandageneralcirculationmodel.JClim23:3397Ð3415OÕGormanPA,SchneiderT(2009)Thephysicalbasisforincreasesinprecipitationextremesinsimulationsof21-centuryclimatechange.ProcNatlAcadSci106:14773Ð14777OÕGormanPA(2011)Theeffectivestaticstabilityexperiencedbyeddiesinamoistatmosphere.JAtmosSci68:75Ð90PavanV,HallN,ValdesP,BlackburnN(1999)Theimportanceofmoisturedistributionforthegrowthandenergeticsofbarocliniceddies.AnnGeophys17:242Ð256PolvaniLM,EslerJG(2007)Transportandmixingofchemicalairmassesinidealizedbarocliniclifecycles.JGeophysRes112:D23102.doi:ReedRJ,GrellG,KuoY-H(1993)TheERICAIOP5storm.PartII:sensitivitytestsandfurtherdiagnosisbasedonmodeloutput.MonWeatherRev121:1595Ð1612RotunnoR,SkamarockWC,SnyderC(1994)Ananalysisoffrontogenesisinnumericalsimulationsofbaroclinicwaves.JAtmosSci51:3373Ð3398SherwoodSC,RocaR,WeckwerthTM,AndronovaNG(2010)Troposphericwatervapor,convectionandclimate.RevGeophysSimmonsAJ,HoskinsBJ(1978)Thelifecyclesofsomenon-linearbaroclinicwaves.JAtmosSci35:414Ð432SkamarockWC,KlempJB,DudhiaJ,GillDO,BarkerDM,DudaM,HuangX-Y,WangW,PowersJG(2008)AdescriptionoftheadvancedresearchWRFVersion3,NCARTechnicalNoteSolomonS,QinD,ManningM,ChenZ,MarquisM,AverytKB,TignorMMillerHL(eds)(2007)Climatechange2007:thephysicalsciencebasis.In:Contributionofworkinggroupitothefourthassessmentreportoftheintergovernmentalpanelonclimatechange.CambridgeUniversityPress,NewYorkStoelingaMT(1996)Apotentialvorticity-basedstudyontheroleofdiabaticheatingandfrictioninanumericallysimulatedbarocliniccyclone.MonWeatherRev124:849Ð874ThorncroftCD,HoskinsBJ,McIntyreME(1993)Twoparadigmsofbaroclinicwavelife-cyclebehaviour.QJRMeteorolSocUlbrichU,LeckebuschGC,PintoJG(2009)Extra-tropicalcyclonesinthepresentandfutureclimate:areview.TheorApplClimatolWangS,PolvaniLM(2011)Doubletropopauseformationinidealizedbarocliniclifecycles:Thekeyroleofaninitialtropopauseinversionlayer.JGeophysRes116:D05108.doi:WeismanML,KlempJB(1982)Thedependenceofnumericallysimulatedconvectivestormsonverticalwindshearandbuoy-ancy.MonWeatherRev110:504Ð520WernliH,DaviesHC(1997)ALagrangianbasedanalysisofextratropicalcyclones:themethodandsomeapplications.QJRMeteorolSoc123:467Ð490WhitakerJS,DavisCA(1994)Cyclogenesisinasaturatedenviron-ment.JAtmosSci51:889Ð907 J.F.Boothetal.