NonamemanuscriptNo. (willbeinsertedbytheeditor) The Structure and Dynamics of the Corona – Heliosphere Connection SpiroK.Antiochos · JonA.Linker · Roberto Lionello · ZoranMikic´ · ViacheslavTitov · ThomasH.Zurbuchen Received:date/Accepted:date Abstract DeterminingthesourceattheSunoftheslowsolarwindisoneofthemajorun- solvedproblemsinsolarandheliosphericphysics.First,wereviewtheexistingtheoriesfor theslowwindandarguethattheyhavedifficultyaccountingforboththeobservedcomposi- tionofthewindanditslargeangularextent.Anewtheoryinwhichtheslowwindoriginates fromthecontinuousopeningandclosingofnarrowopenfieldcorridors,theS-Webmodel, is described. Support for the S-Web model is derived from MHD solutions for the quasi- steady corona and wind during the time of the August 1, 2008 eclipse. Additionally, we performfullydynamicnumericalsimulationsofthecoronaandheliosphereinordertotest the S-Webmodel as well asthe interchange model proposed by Fiskand co-workers. We discusstheimplicationsofoursimulationsforthecompetingtheoriesandforunderstanding thecorona–heliosphereconnection,ingeneral. Keywords Sun:corona·Sun:solarwind 1 Introduction Since the pioneering theoretical work of Parker (1958, 1963) and the discovery observa- tionsofNeugebauer&Snyder(1962)ithasbeenknownthattheSun’satmospherestreams continuouslyoutwardintheformofasupersonicwind.Thiswindcarriesbothplasmaand magneticfieldtotheboundaryofthesolarsystem,theheliopause.Atabasiclevel,theori- gins of the wind are straightforward. As argued by Parker, the difference in gas pressure between the Sun’s hot, 1 MK, corona and the tenuous interstellar gas causes the coronal S.K.Antiochos NASAGoddardSpaceFlightCenter,Code674,Greenbelt,MD20771,USA Tel.:+1-301-286-8849 Fax:+1-301-286-1648 E-mail:[email protected] J.A.Linker,R.Lionello,Z.Mikic´,V.Titov PredictiveScience,Inc.,9990MesaRimRd.,Ste.170,SanDiego,CA92121,USA T.H.Zurbuchen DepartmentofAtmospheric,OceanicandSpaceSciences,CollegeofEngineering,UniversityofMichigan, 2455HaywardSt,AnnArbor,Michigan48109 2 gastoexpandoutward.Iftheheatingtothecoronaisconstant,thenthewindcanadoptthe steadystatedescribedbyParker’soriginaltheory. Ofcourse,ontherealSuntherearemanyaddedcomplicationstoParker’ssimplesteady- statemodel.Thefirstandforemostisthepresenceofthesolarmagneticfield.Thisnaturally divides the corona into two physically distinct regions. In thoseregions wherethe fieldis strong,suchasdeepinsideanactiveregion,thefieldpreventstheplasmafromexpanding; therebyresultinginanapproximatelystaticcoronalplasma.Theseregionsarereferredtoas “closed”, because all fieldlines areconnected tothe photosphere at twoends. Theclosed flux is truly coronal in that it does not connect to the heliosphere, but appears instead as thewell-knownX-raycoronalloops(e.g.,Orrall1981).Ontheotherhand,inthoseregions wherethefieldisweakandthegasdominates,thegaspressuredragsthefieldlinesoutward indefinitely.Theseregionsarereferredtoas“open”inthatthefieldlineshaveonlyoneend connected to the photosphere. The outward mass and energy flow in open regions results inadecreased densitythere, sothat theyappearas “coronal holes”inX-rayimages (e.g., Zirker1977).Notethatforatruesteadystate,thesolarwindandtheheliosphericmagnetic fluxoriginatesolelyfromphotospheric/coronalopen-fieldregions. Inadditiontointroducing thecomplication oftopology tothecoronaandheliosphere, themagneticfieldalsoforcesthemtobefullytimedependent.Sincethedistributionofflux at the photosphere is constantly changing due to flux emergence/cancellation and abroad rangeofphotosphericflows,thedistributionofopenandclosedfluxmustchange,aswell, implyingthatthesolarwindisinherentlydynamic.ConsequentlyParker’stheorycanbe,at best,aquasi-steadyapproximationtotheactualstateofthecoronaandwind. Thedynamicsintroducedtothesolarwindbythephotosphericfieldevolutionnaturally breaks upinto different regimesdetermined bythetimerequired for establishingasteady wind.ThisisoforderafewtraveltimestotheAlfve´nradius,∼20R ,whichimpliesatime ⊙ scaleoftenhoursorso.Sincethephotosphericevolutionhasamore-or-lessconstantspeed of1km/s orless,thetimescaletranslatesdirectlytoasizescaleforthephotospheric dy- namics.Large-scalephenomena,suchasthedifferentialrotationortheemergence/dispersal of large active regions occurs over days, and so could be incorporated within the quasi- steadyapproximation.Small-scalephenomena,however,suchasthemagneticcarpet(Har- vey1985;Schrijveretal.1997)orgranularflowshavetimescalestypicallylessthanafew hoursand,hence,canbeconsideredasaconstantsourceoffluctuationsornoisetothequasi- steadystate.Intermediate betweenthesespatialscalesisthetypical sizeofsupergranules, whose lifetimes are of order that required to establish a steady state. It is likely that their effectonthewindcanbedeterminedproperlyonlybyexplicittime-dependentcalculations. Intheheliosphere,thephotosphericdynamicsappeartostructurethesolarwindviathe mediationofthemagneticfieldintotwodistinct forms,theso-calledfastandslowwinds. Thefastwindhasspeedsgenerallyinexcessof500km/s,butthatisnotitsdistinguishing feature. This wind has 3 defining features: (a) itstemporal variations, (b) spatial location, and(c)plasmacomposition. (a)AsshownclearlybytheUlyssesmeasurements,thehigh-latitudefastwindexhibits nearconstantspeed(McComasetal.2008)andcomposition(Geissetal.1995;vonSteiger etal.1995;Zurbuchen2007).ItsobservedvariabilityconsistsprimarilyofAlfve´nicfluctu- ationsthatmaybedueentirelytotheexpecteddynamicsinducedbythesmall-scalephoto- sphericevolutiondescribedabove. (b)Thefastwindoriginatesinsidenon-transient(lifetime>1day)coronalholes,where thefieldhasbeenopenforasufficientdurationtoestablishasteadystate. (c)Thefastwindhaselementalabundancesclosetothatofthephotosphere(vonSteiger etal.1997,2001;Zurbuchenetal.1999,2002).ItdoesnotexhibittheFIPbiasoftheclosed- 3 fieldcorona(Meyer1985;Feldman&Widing2003).Furthermore,itsioniccompositionis steadyandimpliesafreeze-intemperatureneartheSunaroundorbelow1MK(Zurbuchen 2007). Fromtheseproperties,weconcludethatthefastwindisthetruequasi-steadywindofthe originalParkertheory.Theslowwind,ontheotherhand,iscompletelydifferent.Itsspeeds are generally <500 km/s, but again, this is not the distinguishing feature. The 3 defining featuresofthiswindaremarkedlydifferentthanthoseabove: (a) The slow wind is intrinsically variable, both inspeed and, especially, composition (Zurbuchen &vonSteiger2006; Zurbuchen2007). Thevelocitystructureconsistsofperi- odsoffastflowsintermingledwithslow.ThisvariationisnotsimplyAlfvenicfluctuations superimposedonaquasi-steadystate.Certainlythelargecompositionalvariability,bothin elementalandion-temperatures,cannotbeduetoturbulenceintheflow,butreflectsinstead anintrinsicdifferenceintheoriginsofthefastandslowwind. (b) The slow wind is associated with the heliospheric current sheet (HCS), which is alwaysembeddedinslowwind,notfast(Burlagaetal.2002).Ontheotherhand,theslow windisobservedtoextendintheheliospheretoanglesfarfromtheHCS,upto30◦orso. (c)Theslowwindhaselementalcomposition(FIPbias)closetothatoftheclosedfield corona. Its ionic composition implies a freeze-in temperature near the Sun (∼1.5 MK), considerablyhigherthanthatofthefast(Zurbuchenetal.2002).Furthermore,theelemental andioniccompositionsarehighlyvariable,unlikethesteadycompositionofthefast.Akey pointisthat,asdefinedbythecomposition,theboundarybetweenthefastandslowwinds isnarrow,oforderafewdegreesorso(Zurbuchenetal.1999),whichissmallcomparedto theangularextentofeitherthesloworfastwinds. The results that the slow wind is associated with the HCS, which maps down to the Y-line at the top of the helmet streamer belt, and that the slow wind has the composition oftheclosedcoronasuggestthatitsomehoworiginatesfromnearorinsidetheclosedfield region. The most obvious scenario is that it is due to the interaction between closed and openfields,whichreleasesclosedfieldplasmaontoopen fieldlines.Thiswouldnaturally accountforbothitsobservedvariabilityandcomposition.Theproblem,however,isthatthe slowwinddoessometimesextendfarfromtheHCSinlatitude,whichimpliesthatitssource at the Sun is inside the open field region, far from the open-closed field boundary. But in that case, it is difficult to understand why its composition should resemble that of closed fieldplasma.Theseapparentlyconflictingobservationshavelongmadetheidentificationof thesolarsourcesoftheslowwindoneofthemajorunsolvedproblemsinHeliophysics.We describebelowthebasictheoriesthathavebeenproposedfortheslowwindsources,present somenumericaltestsofoneofthesetheories(Linkeretal.2011),anddescribeanewtheory fortheslowwind,theS-Webmodel(Antiochosetal.2011). 2 TheoriesfortheSourcesoftheSlowWind There areessentiallythreedifferent possibilities for theslow windsource: It originates in theopen field region, just like thefast; itsomehow originates from insidetheclosed field region;oritoriginatesfromthestreamertops,attheboundarybetweenopenandclosed.We discuss eachof them, in turn, below, but focus on the streamertop model and propose an extensionofthistheory,theS-Webmodel,thatcanreconcilethetheorywithobservations. 4 2.1 TheExpansionFactorModel Perhaps,thesimplesttheoryfortheslowwindisthatitoriginatesfromopenfieldnearthe boundary between open and closed (Suess 1979; Kovalenko 1981; Withbroe 1988; Wang &Sheeley1990;Cranmer&vanBallegooijen2005;Cranmeretal.2007).Thisscenariois highlyappealinginthatitimpliesaunifiedtheoryfortheoriginsofthefastandslowwinds. Bothareduetoasinglemechanism:photospherically-drivenMHDwavesdepositheatand momentum to coronal plasma, resulting in aParker-like solar wind outflow. The key idea underlyingthemodelisthatthespeedofthewindissensitivetotheexactlocationsofthe heatandmomentumdepositioninanopenfluxtube;inparticular,heatinglowdownbelow the critical point leads to a hotter, slower wind (Holzer & Leer 1980). The observed dif- ferencesbetweenthefastandslowwinds,therefore,mayarisesolelyfromthegeometrical differencebetweenfluxtubesnearthecoronalholeboundaryversusthosedeepintheinte- rior.Fluxtubesneartheboundaryexpandsuper-radiallyfromthephotospheretoaheightof orderafewsolarradii;whereasthoseneartheinteriorexpandradiallyorevensub-radially. Evenifthephotosphericfluxofwavesintoallfluxtubesinacoronalholeisthesame,the evolutionofthewavesinthecorona,(theresultingturbulenceanddissipation),willdepend on thegeometry ofthe fluxtube. Cranmer andco-workers have argued that allthedistin- guishingfeaturesoftheslowwind,includingthevariabilityandelementalcompositioncan beexplainedbytheeffectofthefluxtubegeometry,theso-calledexpansionfactor,onthe waveevolution. The challenge for the expansion factor theory is that the speed of the solar wind is sometimesobservedtobeslow,butthewindstillhasthevariability,composition,andother featuresindicativeofthefastwind.Zhaoetal.(2009)haveshownthatthewindfromsmall low-latitude coronal holes with large expansion factors is, indeed, slow, ∼500 km/s; but thiswindstillhasallthetemporalandcompositionalcharacteristicsofthefastwind.These observations are in direct conflict with any model proposing that the differences between thefastandslowwindresultsolelyfromdifferencesinfluxtubegeometry.TheZhaoetal. (2009)resultsdemonstratethatalargeexpansionfactorinanopenfluxtubedoesslowdown thewind,aspredicted,butitdoesnotleadtothevariabilityandcompositionobservedinthe slowwind.Weconclude,therefore,thattheexpansionfactormodel,aspresentlydescribed, isnotconsistentwiththeobservations. 2.2 TheInterchangeModel AnothertheoryfortheslowwindistheinterchangemodelproposedbyFiskandco-workers (Fisk et al. 1998; Fisk 2003; Fisk & Zhao 2009), which in many ways is the diametric oppositeoftheexpansionfactormodel.Intheinterchangemodeltheslowwindispostulated tooriginatefrom theclosedfieldregionviacontinuous interchange reconnection between openfieldlinesandtheclosedflux.Consequently,thismodelisintrinsicallydynamic;there isnosteadystatesolarwind,atleast,fortheslowcomponent.Notealsothatthereisnotruly closedfieldregion,becausetheopenfluxispostulatedtodiffusethroughouttheapparently closed field regions outside coronal holes. The key idea underlying the model is that the evolutionofthecoronalopenfluxisdominatedbythecontinuoussmall-scaledynamicsof the photosphere, such as emergence/cancellation of magnetic carpet bipoles, which drive reconnection between open flux and closed. In addition to being fundamentally dynamic rather than steady, the interchange model also proposes a completely different magnetic topology than the expansion factor model. In the latter, the topology is smooth with well 5 separatedopenandclosedfieldregions,butintheformerthetopologyisessentiallychaotic withopenandclosedfluxmixingindiscriminately. In terms of accounting for slow wind observations, the advantages of the interchange modelareobvious.Itnaturallyproducesacontinuouslyvariablewindwithclosedfieldcom- positionandlocatedaroundtheHCS,but withlargeextent.Theprimarychallengeforthe model is to verify that interchange reconnection induced by photospheric dynamics does, indeed, produce the required diffusion of open flux into closed field regions. Arguments have been presented by Antiochos et al. (2007) that basic Lorentz force balance consid- erations prohibit the mixing of open flux with closed. In fact, Antiochos et al argued that the open-closed topology remains smooth even during interchange reconnection and have proposed theorems that severely constrain the possible topologies of the Sun’s open field regions. Theseauthors, however, didnot perform anactual dynamic calculation oftheef- fect of interchange reconnection withmagnetic carpet bipoles on open flux evolution. We presentjustsuchacalculationinthefollowingsection.Aswillbeevidentbelow,theresults ofthiscalculationdonotsupporttheunderlyingassumptionsoftheinterchangemodel.Our results, however, do support key aspects of the interchange model in that dynamics and a statisticalapproacharelikelytobeessentialformodelingtheslowwind. 2.3 TheStreamer-TopModel Thethirdtheoryforthesourceoftheslowwind,thestreamer-topmodel,isinsomeways, acompromisebetweentheexpansionfactorandinterchangemodels.Thebasicideaisthat the boundary between the open and closed flux, the edge of a streamer, is either unstable (Suess etal. 1996; Endeve et al.2004; Rappazzo et al.2005) orsensitivetoperturbations and,consequently,undergoescontinuousdynamics(Mikic´ etal.1999).Inresponsetopho- tospheric changes or other disturbances, closed flux near the streamer boundary expands outwardandbecomesopenandopenfluxreconnectsattheHCStobecomeclosed.Further- more,interchangereconnectioncanoccuratthestreamertopsinresponsetophotospheric changes(e.g.,Wangetal.2000).Theseprocesseswillnaturallyreleaseclosedfieldplasma intothesolarwindandproduce avariablewindwiththeobservedcomposition. Notethat inthismodelthesourceoftheslowwindistheboundary regionbetweenopenandclosed flux, similartotheexpansion factormodel, but theboundary inthiscaseisfullydynamic andinvolvesthecontinualinterchangeofopenandclosedflux.Inthisrespectthemodelis similartotheinterchangemodel,buttheopen-closedinterchangeinthestreamer-topmodel occursonlynearthestreamerboundary. Thereisnodiffusionofopenfluxdeepinsidethe closed-fieldstreamer. There area number of observations that provide compelling support for the streamer- top model. Movies of coronal evolution often show the upward expansion and eventual opening of closed loops (Hundhausen et al. 1984; Howard et al. 1985; Sheeley & Wang 2002).Conversely,coronagraphobservationsfrequentlyshowblobsstreamingoutwardnear the current sheet and reconnection events at the HCS (Sheeley 1997), as required for the streamer-top model. In the heliosphere the HCS is observed to be clearly dynamic with no evidence for a simple field reversal as would be the case for a quasi-steady wind. We conclude, therefore, that the boundary between the open and closed flux in the corona is, indeed,highlydynamicand,therefore,willbethesourceofavariableslowwind. Theproblem,however,isthatthesedynamicsareexpectedtobeconfinedtoarelatively narrowregionaboutthestreamerboundary,orequivalently,abouttheHCS.Assumingthat theopen-closedboundaryis“blurred”bythephotosphericdynamicsoverascaleofordera 6 supergranuleradius,∼30,000km,wederiveanangularextentforthisdynamicwindofno morethan5◦.Infact,thiswidthiswhatisobservedforthestreamerstalksemanatingfrom thetopsofstreamersandfortheso-calledplasmasheetintheheliosphere(Winterhalteret al.1994;Wangetal.2007).Butthisangularwidthisfartoosmalltoexplaintheslowwind whichhasbeenobservedtoextendoutto30◦ fromtheHCS.Explainingthelargeangular extent of the slow wind is the fundamental challenge for the streamer-top model. Since the boundary between open and closed flux maps directly to the HCS, it seems unlikely that blurring this boundary by a small distance on the Sun would ever produce effects in the heliosphere far from the HCS. We describe below our theory, the S-Web model, that accomplishesexactlythisunlikelyresult. 3 TheS-WebModel In previous work (Antiochos et al. 2007) we proposed the uniqueness conjecture, which states that any unipolar region on the photosphere can contain at most one coronal hole. However,low-latitudecoronalholesthatappeartobedisconnectedfromtheircorresponding polarholearefrequentlyobservedontheSun(e.g.,Kahler&Hudson2002).Wearguedthat inthesecases,thelow-latitudesatelliteholeisactuallyconnectedtothemainpolarholeby anarrowopenfieldcorridorwhosewidthisbelowthespatialresolutionoftheobservations. Notealsothatifthecorridorissufficientlynarrow,itwillbeobscuredbyneighboringclosed fieldregionswithmuchhigherbrightness. Let us consider how such a corridor would map into the heliosphere. For illustrative purposes, Figure 1 shows such a mapping in the extreme case where the polar hole and satelliteholeshavenearequalflux.Theinnerhemispherecorresponds tothephotosphere, with the dark yellow region representing the closed field and the light blue representing the two coronal holes connected by an open-field corridor. The light orange hemisphere correspondstosomesurfaceintheheliospherewherethefieldisallopen,sayat10R .On ⊙ thissurfacetheradialfluxisapproximatelyuniformlydistributed,asintherealheliosphere, andthereisasingleHCS,indicatedbytheblacklinerunningaroundtheequatorofthe10 R surface. Note that this HCS maps down to the boundary of the single, (topologically ⊙ connected), openfieldregiononthephotosphere. Fourfieldlineswithfootpoints nearthe “endpoints”oftheopenfieldcorridoraredrawn,illustratingthismappingfromtheHCSto theopenfieldboundary. Iftheholeshaveroughly equalflux,theirfluxmustdividethe10R surfaceintotwo ⊙ nearequalhalves.InFig. 1,thesatelliteholemapstothenearhalfandthepolartothefar. Separatingthesehalvesisathinarc,bluecurveonthe10R surface,thatmapstotheopen ⊙ fieldcorridor.Notethatthisarcdividestheheliosphericsurfaceand,hence,mustextendto near90◦fromtheHCS.Iftheopen-fieldcorridorisverynarrow,thenthefield-linemapping from thebluearcdown tothesolarsurface isquasi-singular, sotopologically, thearcisa so-calledquasi-separatrixlayer(Priest&De´moulin1995;De´moulinetal.1996;Titovetal. 2002). The HCS,ontheother hand, isatrueseparatrix, becausethefield-line mapping is singularthere. Figure 1 illustrates a steady model, but to obtain the slow wind we need to add the temporalvariability.Assumethatduetotherandomphotosphericevolution,theopen-closed boundaryatthephotospherebecomesdynamicallyblurredbycontinuousfieldlineopening and closing over some finite width, such as a supergranule scale. This narrow boundary regionisnowasourceofslowwind.Duetothefieldlinemapping,theHCSalsoacquires afiniteangularwidthoforderseveraldegrees,andbecomesalocationforslowwindinthe 7 Fig.1 Illustrationofanopenfieldcorridorconnectingtwocoronalholes(blueshading)atthephotosphere (yellowinnersurface),anditsmagneticfieldline(reddashed)mappingtoaheliosphericsurface(paleor- ange). heliosphere.Butiftheopenfieldcorridorisnarrowerthanasupergranulescale,thenallthe fluxinthecorridor willbedynamically blurred byfieldlineopening andclosingandwill beasourceofslowwind.Sincethecorridor fluxmapstothebluearc,thismustalsobea location ofslow windintheheliosphere. Thekeypoint isthat thearcextends well above theHCS,sothat inthecaseofFig. 1,slowwindoccurs upto90◦ abovetheHCS,atthe heliosphericpole! Notethataslongasitisnarrowerthanasupergranulescale,thewidthoftheopen-field corridorisirrelevanttoourarguments.Dependingonthedistributionofphotosphericflux, Titovetal.(2011)haveshownthatthecorridorcanshrinkdowntothepointitbecomesthe zero-widthfootprint ofaseparatrixdome.Regardlessofwhetherafinitecorridororasin- gularityispresent,thisisaregionwhereelectriccurrentconcentrationswouldbeexpected toformwhenthefootpoints ofthefieldsarestressedbyphotospheric motions,andwhere themagneticfieldismostsusceptibletoreconnection.Reconnectioninturnallowsplasma formedonclosedfieldlinestoaccessopenfieldsandescapeaspartoftheslowsolarwind. Notethattheinstantaneousopen-fieldboundary andresultingHCSarealsoofzerowidth, andyet,theyproduceaslowwindregionoffiniteextent. We conclude from the discussion above that open-field corridors and the topology of Fig.1maybeabletoreconciletheobservedextentoftheslowwindwiththestreamertop model.Ifthisisthecase,themainchallengetothemodelwouldbeovercome.Ofcourse,for asinglecorridorasinFig.1,theslowwindoccursonlyinanarrowarcofseveraldegrees width,whichisnotsufficienttoaccountfortheobservedslowwind,irrespectiveofwhere thearclies.Theimportantquestion,thereforeisthenumberofsuchcorridorsandarcsfora truesolarphotosphericfluxdistribution. Toanswerthisquestionwecalculatedwithveryhigh numerical resolution, thesteady state MHD solution for the corona and wind during a Carrington rotation centered about the August 1, 2008 eclipse(Rusˇinet al. 2010). Figure 2 shows the open and closed field distributionatthephotospherecalculatedfromthemodel,alongwiththepolarityinversion 8 Fig. 2 Distribution of open magnetic flux (black) and closed (grey) at the photosphere for a Carrington rotationcenteredaboutAug.1,2008.Alsoshownisthepolarityinversionlineonasurfaceslightlyabovethe photosphere,(adaptedfromAntiochosetal2011). line slightly above the photosphere (for clarity). It is evident that during this time period therewerenumeroussatelliteholesextendingfromthemainpolarholes. Figure 3showsthedistributionofthesquashingfactor,Q,ontheheliosphericsurfaceat 10R .Thesquashingfactorisarobusttopologicalmeasurethatallowsforstraightforward ⊙ identificationofquasi-separatrixlayersandseparatrices(Titovetal.2002, 2008).Wenote thatsurrounding andconnected totheHCS(thickdarkredline)isadensewebofhighQ layers,theS-Web. TheS-WebofFig. 3isexactlywhatisneededtoaccountfortheslowwindobservations. At some locations it extends up to 30◦ from theHCS, which explains theobservations of slowwindathighlatitudes.TheHCSisnotalwayssymmetricallylocatedinsidetheS-Web butcanlienearoneedgeand,infact,theHCSisoftenobservedintheheliospheretooccur closetooneoftheslowwindboundariesratherthansymmetricallybetweenthem(Burlaga et al. 2002). Although the Q layers are densely spaced, they are not space filling, so we expectthatthewindintheS-Webregionwillactuallyconsistofamixtureofslowandfast, asobserved.Notealso,thattheS-Webhasaclearlydefinedboundaryseparatingitfromthe polar flux regions, which can account for the observed sharp transition between slow and fastwind(Zurbuchenetal.1999). Weconcludethat,atleast,forthetimeperiodofAugust2008,theopen-fieldcorridors atthephotosphereandtheS-Webintheheliospherehaveallthenecessarytopologicalstruc- turetoproducetheobservedslowwind.Ofcourse,theunderlyingassumptionisthatwhen thephotosphericdynamicsareaddedtothemodel,magneticfieldwillopenandcloseasre- quired,andasufficientamountofclosedfieldplasmawillbereleasedintotheheliosphere. Wedescribebelowasimulationwhichrepresentsourfirststeptowardsaddingdynamicsto theS-webmodel. 4 CoronalandHeliosphericDynamics Itisnotyetpossibletoperformafullytime-dependentcalculationthatadequatelyresolves dynamics such as reconnection occurring on the smallest scales shown in Fig. 2. Even relaxing this configuration to asteady statestrained our computational resources! Adding the magnetic carpet and its complex evolution to the calculation is well beyond present capabilities.Therefore,weconsiderinsteadasimplifiedscenarioinwhichwebeginwithan 9 Fig.3 DistributionofQ-factoronaheliosphericsurfaceat10R⊙forthecoronalholesystemofFig. 2.The HCSappearsasthedarkredlineandthecontoursofhighQaswhite,(adaptedfromAntiochosetal.2011) observedphotosphericfluxdistribution,butlessstructuredthaninFig. 2,andtwo“magnetic carpet” bipoles. Rather than attempting to emerge and submerge the flux, which is also difficultcomputationally,weonlyconvectthebipoleswithasimplephotosphericmotion. AlthoughthissimulationdoesnotallowforafulltestoftheS-webmodel,itdoestest whetheropen-fieldcorridorsformandcoronalholesstayconnectedinafullydynamicevo- lution. It also determines whether the field opens and closes in response to photospheric dynamics,asrequired.Furthermoreitprovides aseveretestoftheinterchangemodel.Ac- cording to this model, interchange reconnection leads to the diffusion of open flux into theclosed.Weexpectthattherewillbesubstantialinterchangereconnectionasthebipoles move; hence, if the interchange model is correct, the open and closed field regions will becomehighlymixed. 4.1 TheNumericalModel Weusesphericalcoordinatesandadvanceintimethefollowingsetofviscousandresistive MHDequations(incgsunits): 4p (cid:209) ×B = J, (1) c 1¶ B (cid:209) ×E =− , (2) c ¶ t v×B E+ =h J, (3) c ¶r +(cid:209) ·(r v) =0, (4) ¶ t 1 ¶ T +v·(cid:209) T =−T(cid:209) ·v, (5) g −1(cid:18)¶ t (cid:19) ¶ v 1 r +v·(cid:209) v = J×B−(cid:209) p+r g+(cid:209) ·(nr (cid:209) v), (6) (cid:18)¶ t (cid:19) c where B is the magnetic field, J is the electric current density, and E is the electric field. InpracticethevectorpotentialAisadvanced, withB=(cid:209) ×A.Thevariablesr ,v, p,and T are the plasma mass density, velocity, pressure, and temperature, g=−g rˆ/r2 is the 0 10 gravitational acceleration, h the resistivity, n is the kinematic viscosity, and g =1.05 is thepolytropicindex.Thepolytropicapproximationisadequateforthisstudy,sinceweare interestedprimarilyinthemagneticfieldevolutionratherthanthedetailedplasmaenergetics asinLionelloetal.(2009).TheboundaryconditionsarediscussedbyLinker&Mikic´(1997) and Linker et al. (1999). Atthe inner radial boundary, afixed temperature of 1.8×106 K andanelectrondensityof108cm−3areprescribed.Thecomponentofthevelocityalongthe magneticfieldisnotspecifiedbutcalculatedfromthecharacteristicequations.Attheouter radialboundarytheflowissupersonicandsuper-Alfve´nic,andvariablesarecomputedwith theaidofthecharacteristicequations. Thegridisnonuniforminr×q ×f of151×191×291points,withD r≈2.6×10−3R = ⊙ 1.8MmatthelowerradialboundaryandD r≈0.75R at20R .Thelatitudinalmeshvaries ⊙ ⊙ between Dq ≈3.7◦ at the poles and Dq ≈0.5◦ near the equator. The azimuthal (longitu- dinal)meshvariesbetweenDf ≈0.5◦ intheprimaryregionofstudytoDf ≈3.0◦ further away.Thesimulationdomainextendsoutto20R .TheAlfve´ntraveltimeatthebaseofthe ⊙ corona(t =R /V )for|B|=2.205Gandn =108cm−3,whicharetypicalreferenceval- A ⊙ A 0 ues,is24minutes(Alfve´nspeedV =480km/s).Auniformresistivityischosensuchthat A theLundquistnumbert /t is1×105,wheret istheresistivediffusiontime.Auniform R A R viscosityn isalsoused,correspondingtoaviscousdiffusiontimetn suchthattn /tA=500. Again,thisvalueischosentodissipateunresolvedscaleswithoutsubstantiallyaffectingthe globalsolution.Duringthephasesofthesimulationwherereconnectionoccurs,thelength scalesareconsiderablysmallerthan1R andthenumericaldissipationexceedsthespecified S h , sothe Lundquist number is consequently smaller in the regions where these dynamics occur. On the solar surface, the bottom boundary, we prescribe for the magnetic flux distri- bution a smoothed NSO Kitt Peak map for Carrington Rotation (CR) 1913 (August 22 – September18,1996).Theresultingcoronalholepatternshowsalongsouthwardextension ofthenorthern coronal hole, theso-calledElephant’s Trunk (Gibson etal. 1999) that was visibleduringthistimeperiod,Figure 4.Tostudytheeffectsofphotosphericdynamicson this system, we add two small bipoles in the area around the Elephant’s Trunk, top panel inFig. 4.Thenorthernbipoleproducesasmallclosedfieldregioninsidethecoronalhole, while the southern bipole results in a small satellitecoronal hole connected to the rest of the Elephant Trunk by an open field corridor, Fig. 4. Note that although it has consid- erable fine-scale structure, the open field topology in Fig. 4 is topologically smooth and well-connected. Wethendrivethissystemwithasurfaceflowthatisuniforminlongitudeanddirected fromwesttoeast,vf 0≈−1km/sbetween30.5◦≤Lat.≤36.0◦,andbetween7.8◦≤Lat.≤ 15.3◦.Thisflowcausesthebipolestomovethroughtheopenandclosedfieldregionsand, inparticular,acrosstheopen-closedboundaries.Westoptheflowafter70hoursandallow thesystemtorelaxfurtherfor14hours. 4.2 ResultsandConclusions Fig. 4showstheopenandclosedfluxatthephotosphereatfourtimesduringtheevolution. The dynamics of this evolution are dominated by two basic processes: flux opening and closing at the HCS and interchange reconnection at the moving bipoles. Each of them is discussed,inturn,below.