ebook img

Oscillatory dynamos and their induction mechanisms PDF

0.55 MB·English
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Oscillatory dynamos and their induction mechanisms

Astronomy&Astrophysicsmanuscriptno.oscillatory˙1712 (cid:13)c ESO2011 January11,2011 Oscillatory dynamos and their induction mechanisms M.Schrinner1,L.Petitdemange2,andE.Dormy1 1 MAG(ENS/IPGP),LRA,EcoleNormaleSupe´rieure,24rueLhomond,75252ParisCedex05,France e-mail:[email protected] 2 Max-Planck-Institutfu¨rAstronomie,Ko¨nigstuhl17,69117Heidelberg Received;accepted ABSTRACT 1 1 Context.Large-scalemagneticfieldsresultingfromhydromagneticdynamoactionmaydiffersubstantiallyintheirtimedependence. 0 Cyclicfieldvariations,characteristicforthesolarmagneticfield,areoftenexplainedbyanimportantΩ-effect,i.e.bythestretching 2 offieldlinesduetostrongdifferentialrotation. Aims.Thedynamomechanismofaconvective,oscillatorydynamomodelisinvestigated. n Methods.Wesolvethe MHD-equations for a conducting Boussinesq fluidinarotating spherical shell. Foraresulting oscillatory a J model,dynamocoefficientshavebeencomputedwiththehelpoftheso-calledtest-fieldmethod.Subsequently,thesecoefficientshave beenusedinamean-fieldcalculationinordertoexploretheunderlyingdynamomechanism. 0 Results.Theoscillatorydynamomodelunderconsiderationisofα2Ω-type.Althoughtheratherstrongdifferentialrotationpresentin 1 thismodelinfluencesthemagneticfield,theΩ-effectaloneisnotresponsibleforitscyclictimevariation.IftheΩ-effectissuppressed theresultingα2-dynamoremainsoscillatory.Surprisingly,thecorrespondingαΩ-dynamoleadstoanon-oscillatorymagneticfield. ] R Conclusions.TheassumptionofanαΩ-mechanismdoesnotexplaintheoccurrenceofmagneticcyclessatisfactorily. S . h p 1. Introduction implemented in a large number of solar dynamo models (e.g. - Steenbeck&Krause1969;Roberts1972;Roberts&Stix1972; o The study of the solar cycle has motivated dynamo theory for Stix 1976; Ossendrijver 2003; Brandenburg&Subramanian r many decades. Hence, the solar dynamo has become the pro- t 2005; Chanetal. 2008) corresponds to the isotropic compo- s totype of oscillatory dynamos.However,its explanationis still nent of a in relation (2), while the Ω-effectresults from the φ- a controversial (Jonesetal. 2010). Most solar dynamo models [ componentofthe∇×(V×B)terminequation(1). have been built on the assumption of an αΩ-dynamo mecha- However, a strong differential rotation is not a neces- 1 nism(Ossendrijver2003);thatis,thepoloidalfieldresultsfrom sary condition for oscillatory solutions of the dynamo equa- v the interaction of helical turbulence with the toroidal field (α- 7 effect)whereasthetoroidalfieldisthoughttooriginatefromthe tion (1), as has been demonstrated in several papers (see e.g. 3 shearingofpoloidalfieldlinesbystrongdifferentialrotation(Ω- Ra¨dler&Bra¨uer 1987; Schubert&Zhang 2000; Ru¨digeretal. 8 effect).Thisattemptisattractiveformainlytworeasons: 2003;Stefani&Gerbeth2003).Theseauthorsconstructmodels 1 inwhichthetoroidalfieldislikewisegeneratedfromthepoloidal First, the existence of a strong shear layer at the bottom of . field by anα-effect(α2-models)andinvestigatenecessarycon- 1 thesolarconvectionzoneisobservationallywellestablished,and 0 theimportanceofaresultingΩ-effectisnon-controversial. straintsona,theboundaryconditionsforthemagneticfieldand 1 Second,Parker’splanelayermodel(Parker1955)andinpar- the geometry of the dynamo region in order to obtain oscilla- 1 ticular mean-fieldelectrodynamics(Steenbecketal. 1966) pro- torysolutionsof(1).Recently,oscillatorydynamomodelshave : alsobeeninvestigatedbymeansofdirectnumericalsimulations. v vide a very elegant theoretical framework for this approach. Mitraetal. (2010) performed dynamo simulations in a wedge- i Within mean-field theory, attention is focused on large scale, X shapedsphericalshellwith anappliedforcinganddemonstrate i.e. averaged fields, only, and the induction equation may be r replaced by a mean-field dynamo equation (Krause&Ra¨dler againtheexistenceofoscillatoryα2-dynamomodels. a 1980) However, the success of mean-field models in reproducing solar-like variations of the magnetic field relies partly on the ∂B =∇×(E+V×B−η∇×B), (1) largenumberoffreeparameters,i.e.onthearbitrarydetermina- ∂t tionofthedynamocoefficientsaandb.Analternativeapproach inwhich BandV denotetheaveragemagneticandtheaverage is presented by Pe´tre´lisetal. (2009). They constructamplitude velocityfield,ηstandsforthemagneticdiffusivityandEisthe equationsguidedfromsymmetryconsiderationsandanalysepo- meanelectromotiveforce.Moreover,itisassumedthatEisho- larity reversals and oscillatons of the magnetic field resulting mogeneousinthemeanmagneticfieldandmaybereplacedby fromtheinteractionbetweentwodynamomodes. aparameterisationintermsofBanditsfirstderivatives Self-consistent, global, convective dynamo models with cyclic magnetic field variations have been bublished by E=aB+b∇B. (2) Busse&Simitev (2006), and Goudard&Dormy (2008). In(2), the so-calleddynamocoefficientsa andb are tensorsof Convective dynamo simulations with stress-free mechanical secondandthirdrank,respectively,anddependonlyontheve- boundary conditions (Busse&Simitev 2006) exhibit a strong locityfieldandthemagneticdiffusivity.Thetraditionalα-effect andaweakfieldbranch,dependingontheinitialconditionsfor 2 M.Schrinneretal.:Oscillatorydynamosandtheirinductionmechanisms themagneticfield.Ifthemagneticfieldisinitiallyweak,stress inwhichthelinearoperatorDisdefinedas free boundary conditions enable the development of a strong zonal flow carrying most of the kinetic energy and rendering 1 DB=V×B+aB+b∇B− ∇×B. (7) convectionineffective. The magnetic field resulting from these Pm dynamosisrathersmallscaled,oftenofquadrupolarsymmetry Thetimeevolutionofeachmodeisdeterminedbyitseigenvalue and weak. Oscillatory solutions of the induction equation are σandproportionaltoexp(σt).Formoredetailsconcerningthe typicalforthisdynamobranch. eigenvaluecalculation,werefertoSchrinneretal.(2010b). A transition from steady to oscillatory dynamos may Wealsoconsidertheevolutionofakinematicallyadvanced also be governed by the width of the convection zone; magneticfield,B ,governedbyasecondinductionequation Goudard&Dormy(2008)foundoscillatorymodelsbydecreas- Tr ingthe shellwidth.Inthisstudy,we followtheirapproachand ∂B 1 analysethedynamomechanismfortheseoscillatorymodels.In ∂tTr =∇×(v×BTr)+ Pm∇2BTr. (8) particular, we address the question whether an Ω-effect is re- sponsibleforthecyclicvariationofthemagneticfield.Different ThetracerfieldB experiencestheself-consistentvelocityfield Tr frompreviouswork,wedeterminethedynamocoefficientsaand at each time step but does not contribute to the Lorentz force bfromdirectnumericalsimulationswiththehelpofthetest-field andispassiveinthissense(seealsoSchrinneretal.2010a).Its method(Schrinneretal.2005,2007).Theapplicationofaandb evolution will be compared with mean-field results originating inamean-fieldcalculationrevealstheirimportanceforthegen- likewisefromakinematicapproach.Moreover,akinematically erationofthemagneticfield. advanced tracer field allows us to test for the influence of the Ω−effectindirectnumericalsimulations.Inanumericalexperi- ment,wesubtractthecontributionoftheΩ−effectandthemean 2. Dynamocalculations meridionalflowintheequationforthetracerfield, WeconsideraconductingBoussinesqfluidinarotatingspheri- ∂B 1 calshellandsolvetheequationsofmagnetohydrodynamicsfor Tr =∇×(v×B )+ ∇2B −∇×(V×B ), (9) ∂t Tr Pm Tr Tr the velocityv, magneticfield B andtemperatureT as givenby Goudard&Dormy (2008) with the help of the code PaRoDy andstudyinthiswaytheoutcomeofakinematicα2-dynamo. (Dormyetal.(1998)andfurtherdevelopments), ∂v E +v·∇v−∇2v +2z×v+∇P= 3. Results ∂t ! The modelunderconsiderationhas been previouslystudied by r 1 Ra T + (∇×B)×B, (3) Goudard&Dormy (2008). It is defined by E = 10−3, Ra = ro Pm 100(= 2.8Rac), Pm = 5, Pr = 1 and an aspect ratio of 0.65. ∂B 1 Except for the stress-free mechanical boundary condition ap- =∇×(v×B)+ ∇2B, (4) ∂t Pm pliedatr = ro andanincreasedaspectratio,thegoverningpa- ∂T 1 rameters are those of a rather simple, quasi-steady benchmark ∂t +(v·∇)(T +Ts)= Pr∇2T. (5) dynamo(Christensenetal.2001).However,Goudard&Dormy (2008) report a transition from steady, dipolar to oscillatory Governingparameters are the Ekman number E = ν/ΩL2, the models for these parameter values. Note that the model re- (modified) Rayleigh number Ra = α g ∆TL/νΩ, the Prandtl quires a rather high angular resolution up to harmonic degree T 0 numberPr = ν/κ andthemagneticPrandtlnumber Pm = ν/η. l =112. max In these expressions, ν denotes the kinematic viscosity, Ω the Figure1 displaystheradialcomponentofthe velocityfield rotationrate, L theshellwidth,α the thermalexpansioncoef- at a given radial level. A typical columnar convection pattern T ficient,g isthegravitationalaccelerationattheouterboundary, is visible, even though the convection columns are noticeably 0 ∆T standsforthe temperaturedifferencebetweenthe spherical disturbedbytheinfluenceofthecurvedboundariesandastrong boundaries,κisthethermalandη=1/µσthemagneticdiffusiv- zonalflowcarryingabout50percentofthekineticenergy.The ity withthe magneticpermeabilityµandthe electricalconduc- magnetic Reynolds number based on the rms-velocity and the tivity σ. Furthermore,the aspectratio is definedas the ratio of shellwidth,Rm = v L/η,isabout90.Theflowissymmetric rms the innerto the outershell radius, r/r ; it determinesthe shell with respect to the equatorialplane and convectiontakes place i o width. onlyoutsidetheinnercoretangentcylinder. In our models,convectionis drivenby an imposed temper- Theevolutionofthemagneticfieldiscyclic.Infigure2(top), ature gradient between the inner and the outer shell boundary. contours of the azimuthally averaged radial magnetic field at Themechanicalboundaryconditionsarenoslipattheinnerand the outer shell boundaryvarying with time are plotted in a so- stressfreeattheouterboundary.Moreover,themagneticfieldis called butterfly diagram. A dynamo wave migrates away from assumedtocontinueasapotentialfieldoutsidethefluidshell. the equator until it reaches mid-latitudes where the inner core Time-averaged dynamo coefficients for an axisymmetric tangent cylinder intersects the outer shell boundary. The mag- mean magnetic field have been determined from direct numer- neticfieldlooksrathersmallscaledandmultipolar.Thisiscon- icalsimulationsasdescribedindetailbySchrinneretal.(2007) firmed by the magnetic energy spectrum which is essentially andasrecentlydiscussedfortime-dependentdynamomodelsby white,exceptforanegligibledipolecontribution.Furthermore, Schrinner(2011).Inasecondstep,thesecoefficientshavebeen themagneticfieldisweak,asexpressedbyanElsassernumber appliedin a mean-fieldmodelbased on equation(1) written as ofΛ= B2 /(µρηΩ)=0.13. rms aneigenvalueproblem, The kinematically advanced tracer field grows slowly in time,i.e.themodelunderconsiderationiskinematicallyunsta- σB=∇×DB, (6) ble according to the classification by Schrinneretal. (2010a). M.Schrinneretal.:Oscillatorydynamosandtheirinductionmechanisms 3 Fig.1.Snapshotoftheradialvelocityoftheconsidereddynamo model at r = 0.79r . The velocity component has been nor- o malised by its maximum absolute value, v = 24.46ν/L. r,max Hence, the colour coding ranges from −1, white, to +1, black. Contourlinescorrespondto±0.2and±0.6. But,deviationsofthetracerfieldfromtheactualfieldarehardly noticeable in the field morphology. Moreover, the very same dynamo wave persists in the kinematic calculation (see also Goudard&Dormy(2008)),asvisibleinfigure2(middle).Note thatthe tracerfield in figure 2 hasevolvedfromrandominitial conditions. A mean-field calculation based on the dynamo coefficients a, b and the mean flow V determined from the self-consistent model is presented in the bottom line of figure 2. The fastest growing eigenmodes form a conjugate complex pair and give rise to a dynamo wave which compares nicely with the direct numericalsimulations.Since thismodeldependsonthe fulla- tensorandthemeanflow,werefertoitasanα2Ω-dynamo. Theinfluenceofthedifferentialrotationmaybesuppressed inthekinematiccalculationofthetracerfieldwithoutchanging anyother componentof the flow. A butterflydiagramresulting fromakinematicallyadvancedfieldaccordingtoequation(9)is presentedinfigure3(top).Theevolutionofthemagneticfieldis again cyclic. Apart from small-scale variationson shorter time scales,adynamowavemigratesfrommid-latitudestowardsthe equator. This is in agreement with a correspondingmean-field calculationinwhichthemeanflowV in(7)hasbeencanceled: Thebottomchartoffigure3providesthebutterflydiagramstem- ming fromthe fastest growingeigenmodesof the resulting α2- dynamo.An explanationwhy directnumericalsimulationsand meanfieldcalculationscomparesomewhatbetterinfigure2than infigure3isprovidedinappendixA. Fig.2. Azimuthally averaged radial magnetic field at the outer ThetimeevolutionoftherelatedαΩ-dynamoisoffurtherin- shell boundary varying with time (butterfly diagram) resulting terest.Astheα-effectisnotdirectlyaccessibleindirectnumer- from a self-consistent calculation (top), kinematic calculation icalsimulations,thecorrespondingαΩ-dynamocanberealised according to (8) (middle) and mean-field calculation (bottom). ina mean-fieldcalculation,only.Ina firstattempt,we haveset Thecontourplotshavebeennormalisedbytheirmaximumabso- arr = aθθ = 0 in order to suppress the generation of toroidal lutevalueateachtimestepconsidered.Thecolourcodingranges fieldfrompoloidalfieldbyanα-effect.Bothcomponentsmake from−1,white,to+1,black. major contributionsto this process. The leading eigenmodere- sulting fromthis calculationis shown in figure 4; it is real, i.e. weintroduceaCartesiancoordinatesystem(x,y,z)correspond- non-oscillatory, and close to marginal stability. The results re- ingtothe(φ,θ,r)directionsanddefinemeanquantitiestobex- mainsimilarifweneglectfurther,non-diagonalcomponentsof a. independent.Moreover,wewriteB= Bex+Bp= Bex+∇×Aex andreformulate(1)inthefollowingsimplifiedmanner ∂A 1 4. Discussion = α B+ ∇2A, (10) ∂t xx Pm The Frequency and the propagation direction of the dynamo ∂B ∂ ∂A dV ∂A 1 wave visible in figure 2 depend strongly on the differential ro- = − (α )+ x + ∇2B. (11) ∂t ∂y zz∂y dz ∂y Pm tation, in agreement with Busse&Simitev (2006). We follow theirapproachandgiveanestimateforthecyclefrequencyap- In the above equations, we have considered only the dominant plyingParker’splanelayerformalism(Parker1955).Tothisend, diagonalcomponentsofa,a anda correspondingtoα and rr φφ zz 4 M.Schrinneretal.:Oscillatorydynamosandtheirinductionmechanisms Fig.4. The leading dipolar eigenmode resulting from a mean- field calculation with a = a = 0. Contour plots of all three rr θθ componentsare presented, each normalised separately by their maximumabsolutevalues.Maximaandminimaarewrittennext to each plot. The colour coding ranges from −1, white, to +1, black, and contour lines correspond to ±0.1, ±0.3, ±0.5, ±0.7 and±0.9. and ω = ℑ(σ)= Fig.3. Azimuthally averaged radial magnetic field at the outer 1/2 shell boundary varying with time (butterfly diagram) resulting 2 α dV ∂α apfrloocmtosrraaerskepipnorenemdsieanntgitcemdcaaelsacnuin-lfiafiteiglodunrcewa(li2cthu).lsautibotnrac(bteodttoΩm-e)ff.eTchte(tocopn)taonudr ±r 2xx vt(αzzky2)2+dz − ∂yzz ky2−αzzky2 . (17) Thesignin(17)isdeterminedbythesignofk (dV/dz−∂α /∂y). y zz Ifwefurtherassumethatthefrequencyisdominatedbydifferen- tialrotationandneglecttheα-termsin(17),weestimatesimilar α ;allcomponentsofbandthemeanmeridionalflowhavebeen xx toBusse&Simitev(2006) neglected.Furthermore,V hasbeenassumedtodependonlyon z.Then,theansatz π 1/2 ω=ℑ(σ)≈± α 2E . (18) (A,B)=(Aˆ,Bˆ)exp(ik·x+σt), (12) L2 xxq T! In(18),E denotesthethekineticenergydensityduetotheax- T leadsto isymmetrictoroidalvelocityfieldandk ≈ 2π/Lhasbeenused. y Approximating α by the rms-value of a , α = 11.50ν/L, pAˆ = αxxBˆ, (13) and with E = 7x1xν2/L2, we find ω ≈ ±φ1φ03.7xxη/L2 which is T dV ∂α surprisingly close to ω = ±100.95η/L2 in the full calculation pBˆ = α k2+ik − zz Aˆ, (14)  zz y ydz ∂y  presentedin figure 2. Note that αxx and dV/dzare of the same    orderofmagnitudeandcontributeequallytoω.Thesignin(18) with p = σ+|k|2/Pm.From(13),(14),we derivea dispersion is determined by the sign of the product aφφ∂Vφ/∂r which is relation positiveinthenorthernandnegativeinthesouthernhemisphere. Therefore,ourestimatein(18)predictsadynamowavemigrat- dV ∂α p2 =α α k2+ik − zz , (15) ingawayfromtheequator.Thisisinagreementwiththesimu- xx zz y ydz ∂y  lationsshowninfigure2.    However, the attempt to describe the model under consid- from which the real and the imaginarypart of σ can be calcu- erationasanαΩ-dynamofails.Anoscillatorymodewithafre- lated.Ifα ispositive(e.g.inthenorthernhemisphere),itfol- xx quencyclosetotheaboveestimateturnsouttobeclearlysubcrit- lows icalinamean-fieldcalcuation,ifa anda areomitted.Instead, rr θθ thismodelisgovernedbyareal,dipolarmodeclosetomarginal λ = ℜ(σ)=−|k|2/Pm stability(seefigure4).Hence,theΩ-effectisonlypartlyrespon- 1/2 2 sibleforthegenerationofthemeanazimuthalfield,asconfirmed α dV ∂α +r 2xx vt(αzzky2)2+dz − ∂yzz ky2+αzzky2 (16) rbByrfi∂g(ru−r1eV5φ.)/T∂hre+crh−1arstininθBthθe∂(msiindθd−le1Vcφo)m/∂pθarinesgrtheyescΩa-leeffweictth, M.Schrinneretal.:Oscillatorydynamosandtheirinductionmechanisms 5 the mean azimuthal field displayed by superimposed contour lines.Inparticular,theelongatedfluxpatchesclosetotheinner coretangentcylinderare,ifatall,negativelycorrelatedwiththe Ω-effect.Consistentwiththisfinding,thepoloidalaxisymmetric magneticenergydensityexceedsthetoroidaloneby20%. Differential rotation alone is not responsible for the cyclic timeevolutionofthemagneticfield,despiteitsinfluenceonthe frequency and the propagation direction of the dynamo wave. Thisismostclearlyvisibleinfigure3.Simulationswithoutdif- ferentialrotationstillleadtoadynamowaveeventhoughitsfre- quency and propagation direction have changed. In the frame- work of Parker’s plane layer formalism, the frequency of this oscillatoryα2-dynamodependscruciallyon−∂α /∂yinsteadof zz dV /dz. Note the additional minus sign, which might explain x thereversedpropagationdirectioniftheassumethat∂α /∂yis zz predominantlypositive. But differentfromthe radialderivative of the mean azimuthal flow, (1/r)∂arr/∂θ is highly structured, Fig.5. Left:∂Vφ/∂rinunitsofν/L2.Middle:Ω-effectasgiven changessigninradialdirectionandexhibitslocalisedpatchesof by rB ∂(r−1V )/∂r + r−1sinθB ∂(sinθ−1V )/∂θ (greyscale) lownegativevalues(seefigure5).Therefore,wedonotattempt r φ θ φ and B (superimposed contour lines, solid [dashed] lines indi- togiveanestimateforthefrequencysimilarto(18). φ cate positive[negative]values).Right:(1/r)∂a /∂θ in unitsof rr ν/L2.Thecontourplotsarepresentedinthesamestyleasinfig- In order to better understand the influence of the mean ure4. flow on the frequency of the dynamo wave, we have gradu- ally changed the amplitude of V in a series of kinematic cal- culations. Results are presented in figure 6. Stars denote fre- quenciesobtainedfromeigenvaluecalculationsaccordingto(6), wherastrianglesstandforfrequenciesestimatedfromkinematic results due to equation (8). In both cases, the amplitude of the meanflowV hasbeenvariedbymultiplicationwithascalefac- tor f. For f = 1, the originalcalculation is retained, while for f = 0, we reproduce the α2-dynamo already discussed above. Frequencies of dynamo waves resulting from direct numerical simulationsaccordingto(8)havebeenmeassuredfor f =1, 0.7, and0.5.Owingtotheturbulencepresentinthesimulations,these are rather rough estimates and error bars have been included. Nevertheless,the resultsobtainedare in satisfactory agreement with the eigenvalue calculations. The frequencies in figure 6 decrease continously with decreasing scale factors. If the am- plitude of V is reduced to 25 per cent of its original value, ω changessignandthepropagationdirectionofthedynamowave isreversed.Thedashed-dottedlineinfigure6givesωaccording to (18) as predicted for an αω-dynamo. It matches the numer- Fig.6. Frequencies resulting from kinematic calculations in ical results if dV/dz dominates in (17) but deviates clearly for whichtheamplitudeofthemeanflowhasbeenchangedbymul- smaller amplitudes. On the other hand, it is illustrative to use tiplicationwithascalefactor, f.Starsdenotefrequenciesstem- relation (17) to model the dependence of ω on the mean flow. ming from an eigenvaluecalculation accordingto (6), whereas triangles are estimates obtained from kinematic results due to If we set ∂α /dy = 0.25dV/dz anddeterminea representative zz (8). The dashed-dotted line gives frequencies as predicted for valueforα inverting(17)fordV/dz=0andω=−29.15η/L2, zz anαω-dynamoby(18),whilethedashedlinerepresentsωasa thedashedlineinfigure6resultsfrom(17).Itfitsthenumerical functionofVforanα2ω-dynamoaccordingto(17). dataratherwellandconvergestowardsthefrequenciespredicted foranαω-dynamo,iftheamplitudeofVissufficientlyhigh. Let us stress againthat some cautionis neededin applying 5. Conclusions the present mean-field analysis to non-linear direct numerical simulations, as the the dynamo modelconsidered here is kine- A particulardynamomechanismdoesnotseemto beresponsi- maticallyunstable.Strictlyspeaking,ourmean-fieldresultsare blefortheoccurrenceofperiodicallytime-dependentmagnetic only relevant for the kinematically advanced tracer field. But, fields. It turns out, that the influence of the large-scale radial because the model is close to dynamo onset and only weakly shear(the Ω-effect),isnotnecessaryforcyclic fieldvariations. non-linear,webelievethatourinterpretationisalsovalidforthe Instead, the action of small-scale convection, represented by a fullyself-consistentfield.Thisisinparticularconfirmedbythe spatiallystructureddynamocoefficienta ,happenstobeessen- rr rathergoodagreementofthethreebutterflydiagramspresented tial.Forthemodelpresentedhere,smallconvectivelengthscales infigure2. areforcedbya thinconvectionzone.Furtherinvestigationsare 6 M.Schrinneretal.:Oscillatorydynamosandtheirinductionmechanisms neededtoassesswhetherourfindingisrepresentativeforawider classofoscillatorymodels. Acknowledgements. MSisgratefulforfinancialsupportfromtheANRMagnet project.ThecomputationshavebeencarriedoutattheFrenchnationalcomput- ingcenterCINES. AppendixA: Theuseoftimeaverageddynamo coefficients In the following, azimuthal averages are, as throughout in the paper, denoted by an overbar, time averages are expressed by brackets, < ··· >. Initially, dynamo coefficients have been de- termined for an azimuthally averaged, mean magnetic field B. Hence,theevolutionofthelatterisgivenby ∂B 1 =∇×(aB+b∇B+V×B− ∇×B). (A.1) ∂t Pm But, the dynamo coefficients a, b and the mean flow V vary stochasticallyin time. In orderto describethe averagedynamo action, we take in addition the time averageof these quantities andwriteapproximatively, ∂B 1 ≈∇×(<a> B+<b>∇B+<V>×B− ∇×B). (A.2) ∂t Pm We emphasisethatthereis noa priorirelationbetweenthe left handsideandtherighthandsideofequation(A.2).Theactual, azimuthallyaveragedmagneticfieldwilldeviatefromourmean- field description the stronger, the more a, b and V fluctuate in time. Amongthese three quantities, the mean flow V is almost time independent, whereas a and b vary considerably. This is the reason, why the butterfly diagramsin figure 2 are in better agreementthaninfigure3,forwhichthestabilizinginfluenceof themeanflowhasbeenomitted. References Brandenburg,A.&Subramanian,K.2005,Phys.Rep.,417,1 Busse,F.H.&Simitev,R.D.2006,Geophys.Astrophys.FluidDyn.,100,341 Chan,K.H.,Liao,X.,&Zhang,K.2008,ApJ,682,1392 Christensen,U.R.,Aubert,J.,Cardin,P.,etal.2001,Phys.EarthPlanet.Inter., 128,25 Dormy,E.,Cardin,P.,&Jault,D.1998,EarthPlanet.Sci.Lett.,160,15 Goudard,L.&Dormy,E.2008,Europhys.Lett.,83,59001 Jones,C.A.,Thompson,M.J.,&Tobias,S.M.2010,SpaceSci.Rev.,152,591 Krause,F.&Ra¨dler,K.1980,Mean-fieldmagnetohydrodynamicsanddynamo theory, ed. Goodman,L.J.&Love,R.N.(Oxford,PergamonPress,Ltd., 1980.271p.) Mitra,D.,Tavakol,R.,Ka¨pyla¨,P.J.,&Brandenburg,A.2010,ApJ,719,L1 Ossendrijver,M.2003,A&ARev.,11,287 Parker,E.N.1955,ApJ,122,293 Pe´tre´lis,F.,Fauve,S.,Dormy,E.,&Valet,J.2009,Phys.Rev.Lett.,102,144503 Ra¨dler,K.&Bra¨uer,H.1987,Astron.Nachr.,308,101 Roberts,P.H.1972,Phil.Trans.RSoc.Lond.A,272,663 Roberts,P.H.&Stix,M.1972,A&A,18,453 Ru¨diger,G.,Elstner,D.,&Ossendrijver,M.2003,A&A,406,15 Schrinner,M.2011,inpreparation Schrinner,M.,Ra¨dler,K.,Schmitt,D.,Rheinhardt,M.,&Christensen,U.2005, Astron.Nachr.,326,245 Schrinner, M.,Ra¨dler, K.,Schmitt, D.,Rheinhardt, M.,&Christensen, U.R. 2007,Geophys.Astrophys.FluidDyn.,101,81 Schrinner,M.,Schmitt,D.,Cameron,R.,&Hoyng,P.2010a,Geophys.J.Int., 182,675 Schrinner,M.,Schmitt,D.,Jiang,J.,&Hoyng,P.2010b,A&A,519,A80+ Schubert,G.&Zhang,K.2000,ApJ,532,L149 Steenbeck,M.&Krause,F.1969,Astron.Nachr.,291,49 Steenbeck,M.,Krause,F.,&Ra¨dler,K.1966,Zeit.Nat.A,21,369 Stefani,F.&Gerbeth,G.2003,Phys.Rev.E,67,027302 Stix,M.1976,A&A,47,243

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.