ResearchinAstron.Astrophys. 2012Vol.XNo.XX,000–000 R esearchin http://www.raa-journal.org http://www.iop.org/journals/raa A stronomyand A strophysics Period ratios for standing kink and sausage modes in magnetized 6 1 structures with siphon flow on the Sun 0 2 n HuiYu1,Shao-XiaChen1,BoLi1,&Li-DongXia1 a J ShandongProvincialKeyLaboratoryofOpticalAstronomyandSolar-TerrestrialEnvironment, 8 InstituteofSpaceSciences,ShandongUniversity,Weihai,264209,China;[email protected] ] R Abstract Standing oscillations with multiple periods were found in a number of atmo- S sphericstructuresontheSun.Theratiooftheperiodofthefundamentaltotwicetheone . h of its first overtone,P /2P , is importantin applications of solar magneto-seismology. 1 2 p We examine how field-aligned flows impact P /2P of standing modes in solar mag- - 1 2 o netic cylinders.For coronalloops, the flow effectsare significantfor bothfast kinkand r sausagemodes.Forkinkones,theyreduceP /2P byupto17%relativetothestaticcase 1 2 t s evenwhenthedensitycontrastbetweentheloopanditssurroundingsapproachesinfinity. a Forsausage modes,the reductionin P /2P duetoflow istypically. 5.5%compared [ 1 2 with the static case. However,the threshold aspect ratio, only abovewhich can trapped 1 sausage modes be supported, may increase dramatically with the flow magnitude. For v photospherictubes, the flow effect on P /2P is not as strong. However,when applied 1 2 3 tosausagemodes,introducingfield-alignedflowsoffersmorepossibilitiesininterpreting 0 the multiple periodsrecently measured. We concludethat field-alignedflows should be 8 taken into accountto help better understandwhat causesthe departureof P /2P from 1 1 2 0 unity. . 1 Key words: magnetohydrodynamics (MHD) – Sun: corona – Sun: magnetic fields – 0 waves 6 1 : v 1 INTRODUCTION i X The frequently measured waves and oscillations can be exploited to deduce the physical param- r eters of the structured solar atmosphere that are otherwise difficult to yield, thanks to the di- a agnostic power of the solar magneto-seismology (SMS) (see e.g., the reviews by Roberts 2000; Aschwanden 2004; Nakariakov&Verwichte 2005; Nakariakov&Erde´lyi 2009; Erde´lyi&Goossens 2011; DeMoortel&Nakariakov 2012). In the context of SMS, multiple periodicities inter- preted as a fundamental standing mode and its overtones detected in a substantial number of oscillating structures are playing an increasingly important role (see Andriesetal. 2009; Ruderman&Erde´lyi2009,forrecentreviews).Inthecaseofstandingkinkoscillations,bothtwo(e.g., Verwichteetal.2004;VanDoorsselaereetal.2007)andthreeperiodicities(DeMoortel&Brady2007; vanDoorsselaereetal. 2009; Inglis&Nakariakov 2009; Kupriyanovaetal. 2013) have been found. Moreover, the ratio between the period of the fundamentaland twice the period of its first overtone, P /2P ,deviatesingeneralfromunity.ThiswasfirstfoundbyVerwichteetal.(2004)intwoloopsina 1 2 post-flarearcadeobservedbytheTransitionRegionandCoronaExplorer(TRACE)inits171A˚ pass- bandon2001April15,wherevaluesof0.91and0.82weremeasuredforP /2P .Asimilarvalue(0.9) 1 2 2 HuiYuetal. wasfoundforTRACE 171A˚ loopson1998November23(VanDoorsselaereetal.2007),andalsoin flaringloopsasmeasuredwiththeNobeyamaRadioheliograph(NoRH)on2002July3whereP /2P 1 2 isdeducedtobe0.83(Kupriyanovaetal.2013).InthislatterstudythedeviationP /2P fromunityis 1 2 likelytobeassociatedwithwavedispersionatafiniteaspectratiooftheflaringloop.However,loops seemthininEUVimages,therebypromptingAndriesetal.(2005)toattributethefinite1 P /2P to 1 2 − thelongitudinalstructuringinloopdensities,giventhatwavedispersionisexpectedtobeminimalfor kinkmodessupportedbyastaticlongitudinallyuniformloopwithtinyaspectratios.Whenitcomesto standing sausage modes, fundamentalor globalmodestogetherwith their first overtoneswere identi- fied.Inflare-associatedquasi-periodicpulsationsmeasuredwithNoRH on2000January12,theglobal (fundamental)sausage mode was foundto have a P of 14 17 s, and its first overtonecorresponds 1 − to a P of 8 11s (Nakariakovetal. 2003;Melnikovetal. 2005). Interestingly,sausagemodeswere 2 − alsoseenincoolpost-flareloopsinhighspatialresolutionH imagesandcorrespondtoaP 587s α 1 ≈ andaP 349s(Srivastavaetal.2008).Actually,sausagemodeshavebeendirectlyimagedinmag- 2 ≈ neticpores(Mortonetal. 2011)andchromosphere(Mortonetal. 2012).UsingtheRapid Oscillations in the Solar Atmosphere (ROSA) instrument situated at the Dunn Solar Telescope, the former study employedanEmpiricalModeDecomposition(EMD)analysisto revealanumberofperiods,someof whichseemtocorrespondtoafundamentalmodeanditshigherovertoneswiththestandingmodeset upbyreflectionsbetweenthephotosphereandtransitionregion. Inthesolar atmosphere,flowsseemubiquitous(e.g.,Aschwanden2004), andhavebeenfoundin oscillating structures in particular (e.g., Ofman&Wang 2008; Srivastavaetal. 2008). In the coronal case,wheretheflowspeedstendtobe.100 kms−1 andthereforewellbelowtheAlfve´nspeed,they arenotnecessarilyalwayssmall.Asamatteroffact,speedsreachingtheAlfve´nicrange( 103kms−1) ∼ havebeenseenassociatedwithexplosiveevents(e.g.,Innesetal.2003;Harraetal.2005).Inthecontext of standing modes supported by loops, a siphon flow causes their phases to depend on the locations, which is true even for the fundamental mode where only two permanent nodes are present and are locatedatloopfootpoints.Actuallythislocation-dependentphasedistributionwasseenforthestanding kinkmodemeasuredwith TRACE andSOHO on2001September15 (Verwichteetal. 2010), and yields a flow speed indeed in the Alfve´nic regime (Terradasetal. 2011). The authorswent on to find thatneglectingtheflowsleadstoanunderestimationoftheloopmagneticfieldstrengthbyafactorof three. Giventhatmultipleperiodicitieshavereceivedconsiderableinterest,andthatasignificantflowmay play an important role as far as the applications of the solar magneto-seismologyare concerned,one naturallyasks:howdotheflowsaffectmultipleperiodicitiesfroma theoreticalperspective?andwhat wouldbetheobservationalimplications?Inaslabgeometry,thesequestionswereaddressedbyLietal. (2013) (hereafter paper I) where a rather comprehensive analytical and numerical examination was conducted. In cylindrical geometry, the flow effect on the period ratio for standing kink modes was assessed by Ruderman (2010) for thin coronal loops. The present work extends both paper I and the one by Ruderman (2010) by examining how the flows affect the dispersion properties and hence the periodratiosofbothstandingkinkandsausagemodessupportedbyamagnetizedcylinderofarbitrary aspectratio.Besides,inadditiontoacoronalenvironment,aphotosphericonewillalsobeexaminedin detailtodemonstratehowintroducingaflowhelpsoffermorepossibilitiesininterpretingtherecently measuredmultipleperiodsinoscillatingphotosphericstructures. Thispaperisorganizedasfollows.Section2presentsabriefdescriptionofthecylinderdispersion relation.Section3isconcernedwithcoronalcylinders,wherewefirstgiveanoverviewofthedispersion diagrams,briefly describe a graphicalmeansto computethe period ratios, and examinehow the flow affects the period ratios for standing kink and sausage modes. Likewise, section 4 examinesin detail isolatedphotosphericcylinders.Finally,section5summarizestheresults,endingwithsomeconcluding remarks. Periodratiosforstandingkinkandsausagemodes 3 Fig.1 Schematic diagram illustrating the magnetic cylinder (denoted by subscript 0) and its environment(subscripte). The variablesρ ,c ,v and U (i = 0,e) representthe mass i i Ai i density,adiabaticsoundspeed,Alfve´nspeed,andtheflowspeed,respectively. 2 CYLINDERDISPERSIONRELATION Consider a cylinderof radiusa with time-independentfield-alignedflows. As illustrated in Fig.1, the cylinderisinfiniteinthez-direction,andisborderedbytheinterfacer = ainacylindricalcoordinate system(r,θ,z).Thephysicalparameterstaketheformofastepfunction,characterizedbytheirvalues externalto (denotedby a subscripte) and inside (subscript0) the cylinder.The backgroundmagnetic fields(B andB ), togetherwith theflow velocities(U andU ),are inthe z-direction.Letρ andp 0 e 0 e denote the mass density and thermal pressure. It follows from the force balance condition across the interfacethat ρ 2c2+γv2 e = 0 A0, (1) ρ 2c2+γv2 0 e Ae whereγ = 5/3istheadiabaticindex,c = γp/ρistheadiabaticsoundandv = B2/4πρisthe A Alfve´nspeed.Itisalsonecessarytointroducethetubespeeds,c (i=0,e), Ti p p c2v2 c2 = i Ai , (2) Ti c2+v2 i Ai 4 HuiYuetal. andthekinkspeedc , k c2 =ρˆ v2 +ρˆ v2 , (3) k 0 A0 e Ae whereρˆ =ρ /(ρ +ρ )isthefractionaldensitywithi=0,e. i i 0 e The dispersion relation (DR) for linear waves trappedin a cylinder with flow has been examined by a number of authors(e.g., Narayanan1991; Somasundarametal. 1999; Terra-Homemetal. 2003; VasheghaniFarahanietal.2009;Zhelyazkov2009,2012).Itsderivationstartswiththeansatzthatany perturbationδf(r,θ,z;t)totheequilibriumf(r)takestheform δf(r,θ,z;t)=Re f˜(r)exp[i(kz+nθ ωt)] , (4) − n o where Re( ) means taking the real part of the function. Besides, k and n are the longitudinal and ··· azimuthalwavenumbers,respectively.Thephasespeedv isdefinedasv = ω/k.Oneproceedsby ph ph defining c2 (v U )2 v2 (v U )2 m2 =k2 i − ph− i Ai− ph− i , (5) i h (c2+v2 ) c2ih (v U )2 i i Ai Ti− ph− i h i wherei=0,e.Toensurethewavesaretrapped,m2hastobepositive,meaningthat e c2 <(v U )2 <c2 or(v U )2 <c2 , (6) m,e ph− e M,e ph− e Te where c = min(c ,v ) and c = max(c ,v ). On the other hand, the spatial profiles of the m,e e Ae M,e e Ae perturbationsinther directionaredeterminedbythesignofm2.Whenm2 <0(m2 >0),thewaves − 0 0 0 arebody(surface)ones,correspondingtoanoscillatory(aspatiallydecaying)r-dependenceinsidethe cylinder. With Eq.(4) inserted into the linearized, ideal MHD equations, the DR follows from the require- mentsthattheradialcomponentoftheLagrangiandisplacementξ andthetotalpressureδp becon- r T tinuousatr =a.TheDRreads ρe m0 vA2e−(vph−Ue)2 In′(m0a) = Kn′(|me|a) (7) ρ0 me hv2 (v U )2iIn(m0a) Kn(me a) | | A0− ph− 0 | | h i forsurfacewaves,and ρe n0 vA2e−(vph−Ue)2 Jn′(n0a) = Kn′(|me|a) (8) ρ0 me hv2 (v U )2iJn(n0a) Kn(me a) | | A0− ph− 0 | | h i for body waves (n2 = m2 > 0). Furthermore,kink and sausage waves correspondto the solutions 0 − 0 totheDRwithnbeing1and0,respectively.TheprimedenotesthederivativeofBesselfunctionwith respecttoitsargument,e.g.,J′(n a) dJ (x)/dxwithx=n a.Onemaynotethatm appearsonly n 0 ≡ n 0 e asabsolutevaluestoensurethatthewavesexternaltothecylinderareevanescent. Itprovesnecessarytoexaminetheimportanceofdensityfluctuationrelativetothetransversedis- placement.ThisisreadilydonebyevaluatingX (ρ˜/ρ )/(ξ˜/a) , 0 r ≡ r=a (cid:12) (cid:12) X = (m20a)(ω−kU0)2 p˜T (cid:12) . (9) [k2c2 (ω kU )2]dp˜ /dr 0− − 0 T (cid:12)r=a (cid:12) (cid:12) (cid:12) Periodratiosforstandingkinkandsausagemodes 5 Forbodywaves,p˜ insidethecylinderisproportionaltoJ (n r)forakinkwave,andtoJ (n r)fora T 1 0 0 0 sausageone,resultingin (n a)J (n a)/J′(n a) 0 1 0 1 0 X = (vph−U0)2 kink, (10) (vph−U0)2−c20 (n0saa)uJs0a(gne0.a)/J0′(n0a) Likewise,forsurfacewaves,insidethecylinderp˜ I (m r)forakinkwave,and I (m r)fora T 1 0 0 0 ∝ ∝ sausageone,leadingto (m a)I (m a)/I′(m a) 0 1 0 1 0 X = (vph−U0)2 kink, (11) c20−(vph−U0)2 (m0saau)Is0a(gme.0a)/I0′(m0a) The dispersion relations (7) and (8) possess three symmetric properties that allow us to simplify ourexaminationofthestandingmodes.Thefirsttwodictatethatif[v ,k;U ,U ]representsasolution ph 0 e to the DR, then so does [v , k;U ,U ]; if [v ,k;U ,0] is a solution, then so is [ v ,k; U ,0] ph 0 e ph 0 ph 0 − − − (seeEq.(5)withU = 0).TheyweredetailedintheappendixofpaperIwhichadoptsaslabgeometry, e and can be readily shown to hold in the cylindrical case if one recognizes that xZ′(x)/Z (x) is an n n evenfunctionforBesselfunctionsZ ofintegerordern,whereZ isJ orI .Theyaresummarized n n n n here for one to realize that as long as the external medium is at rest (U = 0), as will be assumed e throughoutthisstudy,thenforthe purposeof examininghowthe periodratio dependson the internal flowU ,oneneedsonlytoconsiderpositiveU .ThethirdsymmetrypropertyreflectssimplyaGalilean 0 0 transformation,which relates the phase speed v (k;U ,U ) in one frame,where the speeds read U ph 0 e 0 andU ,tov (k;U†,U†)inadifferentonewherethespeedsreadU† andU†.Certainlyonerequires e ph 0 e 0 e thatU† U† = U U . One thensees that v (k;U†,U†) = v (k;U ,U )+(U† U ), and in e − 0 e − 0 ph 0 e ph 0 e 0 − 0 particular,v (k;U U ,0) = v (k;U ,U ) U . Whatthismeansisthat, eventhoughthe wave ph 0 e ph 0 e e − − dispersionpropertiesexpressedasaseriesofanalyticalexpressionsinanumberofphysicallyinteresting limitsinbothcoronalandphotosphericenvironmentsaretobederivedinaframewhereU = 0,they e canbeeasilyextendedtoanarbitraryframeofreference. 3 PERIODRATIOSFORSTANDINGMODESSUPPORTEDBYCORONALCYLINDERS 3.1 OverviewofCoronalCylinderDispersionDiagrams Consider first the coronal case, where the ordering v > v > c > c holds. To be specific, we Ae A0 0 e choosev = 4c andc = 0.72c ,theobservationaljustificationofwhichwasgiveninpaperI.For A0 0 e 0 theexternalAlfve´nspeed,unlessotherwisespecified,we willdiscussindetailareferencecasewhere v =2v .Evidently,thelargertheratiov /v ,thestrongerthedensitycontrast. Ae A0 Ae A0 Figure 2 presents the dependence on longitudinal wavenumber k of the phase speeds v for a ph seriesofU =M c ,wheretheinternalMachnumberM reads0,0.8,1.2and3.2,respectively.Kink 0 0 0 0 andsausagewavesareplottedwiththedashedandsolidcurves,respectively.AsshowninFig.2b,they are labeled by combinationsof letters b/f+F/S+K/S, representingbackward or forward, Fast or Slow, KinkorSausage.“Fast”or“Slow”isrelatedtothemagnitudeofthephasespeed,while“backward”or “forward”derivesfromthesignofthephasespeedswhentheflowisabsent,andwastermed“originally backward-(forward-)propagating”byAndriesetal.(2000)inthesamesense.Thenumberappendedto thelettersdenotestheorderofoccurrence,meaningthatfFK1representsthefirstbranchofforwardFast Kink wave. The characteristicspeedsexternal(interior)to the cylinderare givenon the left(right)of eachpaneltoaidwavecategorization.InagreementwithTerra-Homemetal.(2003)(hereafterTEB03), Fig.2indicatesthatallwavesinsuchacoronalenvironmentarebodywaves. 6 HuiYuetal. AclearflowdependencecanbeseeninFig.2.Considerfirsttheslowwaves.Thepropagationwin- dowsalwaysencompass( c +U , c +U )and(c +U ,c +U ),whichisreadilyunderstandable 0 0 T0 0 T0 0 0 0 − − whenoneexaminestheslenderandthickcylinderlimits.Intheformerlimit(ka 1), ≪ c4 k2a2 v U c 1+ T0 , (12) ph ≈ 0± T0s c20vA20 h2l,± whereh hasaninfinitenumberofvalues.Forkinkwaves,h isanarbitraryrootofthetranscen- l,± l,± dentalequation xJ′(x) ρ v2 c2 1 = 0 A0− T0 , (13) J (x) −ρ v2 c2 1 e Ae− T0,± wherec = c +U ,andxdenotestheunknown.Forsausagewaves,h canbeapproximated T0,± T0 0 l,± ± by h j . (14) l,± 1,l ≈ Whentheoppositelimitholds(ka 1),onefinds ≫ c2 g2 v U c 1 0 l , (15) ph ≈ 0± 0s − vA20−c20k2a2 where j kink g = 1,l (16) l j sausage 0,l (cid:26) in which l = 0,1,2, , and j denotes the l-th zero of J . The plus and minus signs in Eqs.(12) n,l n ··· and(15)correspondtotheupperandlowerbands,respectively.However,theslowwavesinthecoronal casearenotofinterestasfarastheperiodratioP /2P isconcerned,fortheyarenearlydispersionless 1 2 duetothenearlyindistinguishablevaluesofc andc ,andthedeviationofP /2P fromunityinthe 0 T0 1 2 presentstudyderivesentirelyfromwavedispersion. Inviewoftheirstrongerdispersion,letuspayacloserlookatfastwaveswhosepropagationwin- dowsencompass( v ,U v )and(U +v ,v ).Onemayreadilyunderstandthisbyexamining Ae 0 A0 0 A0 Ae − − thethickcylinderlimit(ka 1),whereonefinds ≫ v2 h2 v U v 1+ A0 l,±. (17) ph ≈ 0± A0s v2 c2k2a2 A0− 0 WiththeexceptionofbFK1andfFK1,thereexistwavenumbercutoffsforbothkinkandsausagewaves, andthesearegivenby (ka) =g Λ , (18) c l ± where (c2+v2 )[(v U )2 c2 ] Λ = 0 A0 Ae∓ 0 − T0 . ± s[(vAe∓U0)2−c20][(vAe∓U0)2−vA20] Ontheotherhand,forbFK1andfFK1intheslendercylinderlimitka 1,v maybeapproximated ph ≪ by ρˆ v2 (d U )2 v± d 1 0 A0− ±− 0 (λ ka)2K (λ k a) , (19) ph ≈ ± ± h 2d d i ± 0 ±| | ± k Periodratiosforstandingkinkandsausagemodes 7 where d =ρˆ U d , (20a) ± 0 0 k ± d = c2 ρˆ ρˆ U2, (20b) k k− 0 e 0 q (d2 c2)(v2 d2) λ = ±− e Ae− ± . (20c) ± s(c2e +vA2e)(d2±−c2Te) Moreover,v+ andv− representtheupperandlowerbranches,respectively. ph ph Compared with available ones, our study offers some new analytical expressions for the phase speed v in a numberof physicallyinteresting limits. Equations(12) and(15) offerthe approximate ph expressionsofv forslowwavesintheslenderandthickcylinderlimits,respectively.Forthefastones, ph Eq.(17) presentsan explicitexpressionforv in the limitof ka 1, therebyextendingthe original ph ≫ discussionof static cylindersin thissituation byEdwin&Roberts(1983) (hereafterER83)where the authorsemphasizedtheanalogywiththeLovewavesofseismologyandPekeriswavesofoceanography (seeEq.(13)inER83).Moreover,Eq.(19)examinesfastkinkwavesintheslendercylinderlimitka 1, ≪ andextendsavailableresultsinthreeways.First,neglectingthefirstordercorrection,Eq.(19)reduces to d , which agrees with Eq.(70) in Goossensetal. (1992). Second, taking U = 0, we recover the ± 0 expression for a static cylinder, namely Eq.(15) in ER83. Our expression also shows that Eq.(15) as giveninER83isinfactnotrestrictedtothecoldplasmalimit(c =c =0),butvalidforarathergeneral e 0 coronalenvironmentaslongasλisgeneralizedtoincorporatec andc ,asgivenbyourEq.(20).Third, e Te theplusversionv+ reducestoEq.(5)inVasheghaniFarahanietal.(2009)wherethetransversewaves ph propagatinginsoftX-raycoronaljetsareexamined,whenonenotesthat(d2 c2)/(d2 c2 ) 1, + − e + − Te ≈ andv2 c2.However,itturnsoutthatexceptforextremelysmallka,retainingtheoriginalformin Ae ≫ e termsof the modifiedBessel functionK is more accuratethanthe logarithmicformgivenbyEq.(5) 0 in VasheghaniFarahanietal. (2009). Furthermore, the expression v− gives the phase speed for the ph wavesthatarebackwardpropagatingintheabsenceofflow. 3.2 ProceduresforComputingStandingModes By“standing”,werequirethattheradialLagrangiandisplacementξ (r,θ,z;t)iszeroattheinterface r r = aatbothendsofthecylinderz = 0,L,irrespectiveofθandt.Onerequirementforthistobetrue forarbitraryθ isthatonlypropagatingwaveswithidenticalazimuthalwavenumbersncancombineto formstandingmodes.Apairofpropagatingwavescharacterizedbyacommonangularfrequencyωbut differentlongitudinalwavenumbersk andk thenleadtothat r l ξ (r,θ,z;t)= Re ξ˜ (r)exp[i(k z+nθ ωt)] r r,l l − n o + Re ξ˜ (r)exp[i(k z+nθ ωt)] . (21) r,r r − n o Specializing to (r,z) = (a,0), one finds ξ˜ (a) = ξ˜ (a), meaning that one is allowed to choose r,l r,r − ξ˜ (a)=A tobereal.Itthenfollowsthat r,l ξ ξ (a,θ,z;t) r = A [cos(ωt k z nθ) cos(ωt k z nθ)] ξ l r − − − − − k k k +k r l l r = 2A sin − z sin ωt z nθ . (22) ξ − 2 − 2 − (cid:18) (cid:19) (cid:18) (cid:19) Forξ (a,θ,L;t)tobezeroatarbitraryt,thisrequires r 2πm k k = ,m=1,2, (23) r l − L ··· 8 HuiYuetal. Byconvention,m=1correspondstothefundamentalmode,andm=2toitsfirstovertone. Atthispoint,itsufficestosaythattheprocedureforcomputingtheperiodratiosofstandingmodes isidenticaltotheslabcase,whichwasdetailedinpaperI.Basicallyitinvolvesconstructinganω k − diagram where each propagating wave in a pair to form standing modes corresponds to a particular curve, meaning that a horizontal cut with a constant ω would intersect with the two resulting curves attwopoints.Iftheseparationbetweenthetwopointsis2π/L,thenonefindsthefundamentalmode. If it is twice that, one finds the first overtone. Let the angular frequency of the fundamental mode (firstovertone)bedenotedbyω (ω ),theperiodratioissimplyP /2P = ω /2ω .Theexistenceof 1 2 1 2 2 1 cutoffwavenumbersforsausagewavestobetrappedtranslatesintotheexistenceofcutoffaspectratios (a/L) forstandingsausagemodestobenon-leaky.AsemphasizedbypaperI(seeFig.3therein), cutoff this (a/L) is not determinedby the differencebetween the two cutoffsof the two ω k curves cutoff − dividedby2π,butlargerthanthat. When computingthe coronalstanding modes, we consideronly bFK1 and fFK1 for kink modes, andbFS1andfFS1forsausagemodes.BrancheswithlargermodenumberslikebFK2orbFS2would formstandingmodesonlyforrelativelythickcylinderswherea/Lis oftheorderunity.Forthe same reason, we discard the combinationsbetween slow and fast sausage propagatingwaves. On the other hand,combinationsof slow with fast kinkwave, such as bFK1 plus fSK, turn outextremelyunlikely as well. Thisis because, while slow kinkwaves are dominatedby the intensityoscillationsinstead of transversedisplacements(X 1),theoppositeholdsforfastones(X 1).Theendresultisthatif | |≫ | |≪ afastkinkwavedoescombinewithaslowonetoformastandingmode,atransverseloopdisplacement on the orderof the cylinderradiuswill lead to a relative intensity variationthat exceedsunity.To see this,considerslendercylinderssuchthatka 0,andconsiderthecasewherethecomponentstoform → standingmodesarebFK1andanybranchoffSK.ForbFK1,oneseesthatv d andn a 0,and ph − 0 henceX (d−−U0)2 (n a)2.Becaused U d ρˆ U isoftheorde≈rofv ,andv→2 c2, ≈ (d−−U0)2−c20 0 −− 0 ≈− k− e 0 A0 A0 ≫ 0 Xwouldberoughly(n a)2andhenceapproacheszeroaswell.However,forslowkinkwaves,bynoting 0 that(v U )2 c2 whenka 0,onefindsthatX (n a)2(1 v2 /v2 )(ρ v2 )/(ρ c2 ),which ph− 0 → T0 → ≈ 0 − ph Ae e Ae 0 T0 is approximately(n a)2v2 /c2 since ρ v2 ρ v2 and c c . Note thatin coronalconditions, 0 A0 0 e Ae ≈ 0 A0 T0 ≈ 0 h as given by Eq.(13) can be approximatedby (l+3/4)π. When ka 0, with n a approaching l,± 0 → h ,X willbelarge. l,± 3.3 PeriodRatiosforStandingKinkModes Figure3presentsthedependenceontheaspectratioa/LoftheperiodratioP /2P pertinenttostand- 1 2 ing fast kink modes. Here the results for a number of different U are shown with different colors, 0 withU representedbytheinternalAlfve´nMachnumberM = U /v .Onecanseethatallcurves 0 A 0 A0 decreasefromunityatzeroa/L,attainaminimum,andthenincreasetowardsunity.IncreasingU sub- 0 stantially strengthensthe deviationof P /2P from unity relative to the static case (the black curve). 1 2 Take the minima for instance. While in the static case it reads 0.938,attained at a/L = 0.405,when M = 0.8 it is significantly reducedto 0.778 attained at a/L = 0.267. At smaller aspect ratios, the A dispersionintroducedbytheflow,andhencethedeviationfromunityoftheperiodratioP /2P ,isnot 1 2 asstrong.However,atanaspectratioofa/L=0.19,onefindsthatP /2P decreasessignificantlyfrom 1 2 0.953inthestaticcaseto0.79whenM = 0.8.ActuallythisaspectratiocorrespondstotheNoRH A loop that experienced standing kink oscillations on 2002 July 3 with multiple periodicities that yield P /2P = 0.82(Kupriyanovaetal. 2013).Onefindsthatwhilethewavedispersionduetotransverse 1 2 density structuring alone cannot account for this measured value of P /2P , it may be attained with 1 2 theaidoftheadditionalwavedispersionduetoflowshear.Inthisregard,weagreewithAndriesetal. (2009)inthesensethatthecontributionofthedensitycontrastalonetothedeviationofP /2P from 1 2 unityseemstobemarginalforextremelythincylinders.However,wenotethatwhenasubstantialflow shearexistsbetweenthecylinderanditssurroundings,theshear-associatedwavedispersionmaynotbe neglectedfor loopswith finite aspectratios. As a matter of fact, for loopswith a/L as small as 0.05, thefloweffectisstillsubstantialenoughtobeofobservationalsignificance:whileP /2P reads0.989 1 2 Periodratiosforstandingkinkandsausagemodes 9 inthestaticcase,whenM = 0.8itis0.934,whichisalreadybelowtheminimumP /2P canreach A 1 2 whenthe flow is absent. We notethatthis a/Lis notunrealisticbutlies within the rangeofthe mea- sured values of oscillating EUV loops examined in Ofman&Aschwanden (2002)(see their Table 1). Thepointwe wanttomake hereisthatthewave dispersionassociated withthe transversestructuring needstobeconsideredforatheoreticalunderstandingoftheperiodratiosofstandingkinkmodes,and thisisparticularlynecessaryinthepresenceofastrongflowshearandwhentheloopaspectratiosare notextremelysmall. Figure 4 further examines the flow effect by showing (a) the minimal period ratio, (P /2P ) 1 2 min and(b)itslocation,(a/L) ,asafunctionoftheinternalAlfve´nicMachnumberM .Inadditionto min A thereferencecase wherev /v = 2,Fig.4alsoexaminesotherratiosof3,4,10,and20,shownin Ae A0 differentcolors.RegardingFig.4a,oneseesthatthefloweffectontheperiodratiosissignificantforall theconsideredv /v ,orequivalently,thedensitycontrast.Asamatteroffact,atagivenM ,even Ae A0 A though (P /2P ) tends to decrease with increasing density contrast, this tendencyis rather weak, 1 2 min and seems to saturate when v /v exceeds, say, 10, as evidenced by the fact that the two curves Ae A0 correspondingtov /v being10and20canhardlybedistinguished.Consequently,whenv /v Ae A0 Ae A0 is as large as 20, (P /2P ) decreases from 0.914 in the static case to 0.768 when M = 0.8, 1 2 min A amountingto a relative difference of 16%, which is almost the same as in the case when v /v is Ae A0 2wherethisfractionaldifferencereads17.1%.LookingatFig.4b,onenoticesthatforagivendensity contrast,theaspectratioatwhichtheminimumperiodratioisattainedtendstodecreasewithincreasing flow,andthis tendencyis clearerforweakerdensitycontrasts. Whenv /v is atthe two extremes, Ae A0 (a/L) reads0.405and0.31inthestaticcase,andgoesdownto0.267and0.248foranM being min A 0.8,respectively.Thefractionalchangeduetotheflowintheformerreads34%,while20%inthelatter. Itisinterestingtocontrastthecylindercasewiththeslabone.Inbothcasestheminimalperiodratio (P /2P ) andtheaspectratio(a/L) havebeenexaminedanalytically.Notethatintheslabcase, 1 2 min min a refers to the half-width of the slab. For cold static slabs, Macnamara&Roberts (2011) established that (P /2P ) can never drop below √2/2, which is attained for the infinite density contrast at 1 2 min a zero aspect ratio. While this was established by employing the Epstein profile to connect the slab densityandthedensityofitssurroundings,thenumericalresultsinbothMacnamara&Roberts(2011) and paper I demonstrate that this lower limit for P /2P is also valid when the density profile is in 1 2 the form of a step function. When a flow U is introduced, paper I shows that P /2P is no longer 0 1 2 subject to this lower limit and the change in P /2P relative to the static case is typically 20%. 1 2 ∼ Besides,(a/L) tendstoincreasewithincreasingU .Forcoldstaticcylinders,McEwanetal.(2006) min 0 (hereafterM06)andalsoAndriesetal.(2009)establishedthat(P /2P ) also suffersfromalower 1 2 min limitof 0.92whenthedensitycontrastapproachesinfinity,andtheaspectratiowherethislowerlimit ∼ is attained is 0.3 (see Figure 2 in M06, and note that the symbolL therein is the loop half-length, ∼ and hence their a/L corresponds to twice the value of a/L in the present study). The static case in Fig.3 agreesremarkablywellwith Fig.2 in M06 despite thatthe soundspeeds are allowedto be non- zeronow,whichisnotsurprisinggiventhatthesoundspeedsaresignificantlysmallerthantheAlfve´n speeds. However,Fig.3 offersthe new result that in the cylinder case, the introductionof the internal flow provides significant revision to the period ratio, making it no longer suffer from the lower limit establishedforstaticcylinders.Thisistrueevenwhenthedensitycontrastapproachesinfinity,andthe revisiontotheperiodratioistypically 16 17%,similartotheslab case.Ata givenv /v , the Ae A0 ∼ − tendencyfor(a/L) todecreasewithincreasingU inthecylindercaseisoppositetowhathappens min 0 forslabswithflows. 3.4 PeriodRatiosforStandingSausageModes Figure 5 presents the period ratio P /2P as a function of aspect ratio a/L for a series of v /v , 1 2 Ae A0 pertinenttostandingsausagemodes.Thesolid,dotted,anddashedcurvescorrespondtov /v being Ae A0 2,3,and20,respectively.Asindicatedbythedifferentcolors,asetofU isinvestigatedandmeasuredin 0 unitsoftheinternalAlfve´nspeedU =M v .ItisclearfromFig.5thattheeffectofflowontheperiod 0 A A0 ratioP /2P isnotasstrongasforthekinkmodes.Sincethiseffectincreaseswithincreasingv /v , 1 2 Ae A0 10 HuiYuetal. onemayexaminetheextremewherev /v =20,inwhichcaseonefindsthatata/L=0.4,P /2P Ae A0 1 2 reads 0.611 when M = 0.5, which is 5.4% lower than the value 0.646 obtained in the static case. A ThisfractionalchangeinP /2P istypicalinthiscaseatagivenaspectratio.However,thefloweffect 1 2 is much stronger when it comes to the cutoff aspect ratio (a/L) only above which can standing cutoff sausage modes be supported. This effect is substantial even when it is the weakest among the three v /v considered:whenv /v = 2,(a/L) increasesfrom0.456to0.53to0.651withM Ae A0 Ae A0 cutoff A increasing from 0 to 0.1 to 0.2. Regardingthe other extreme v /v = 20, while (a/L) reads Ae A0 cutoff 0.04forthestaticcase,itreads0.083whenM =0.2and0.197whenM =0.4.Thismeansthatata A A givenv /v ,relativetothestaticcase,cylinderswithflowcansupportstandingsausagemodesonly Ae A0 whentheyaresufficientlythickerifthecylinderlengthisfixed. Atthispoint,acomparisonwithstudiesofsausagemodessupportedbymagnetizedslabsisinfor- mative.AsdemonstratednumericallybyInglisetal.(2009)andanalyticallybyMacnamara&Roberts (2011),forstatic coronalslabsP /2P may reachaslow as1/2withthe lowerlimitattainablewhen 1 2 thedensitycontrastisinfinite.Besides,thecutoffaspectratiolowerswithincreasingdensitycontrast. WhileananalyticalexpectationofthelowerlimitofP /2P isnotavailableforcylinders,ourstudyof 1 2 an extremelylargedensitycontrast(v /v = 20)representedby the dashed curvesin Fig.5 shows Ae A0 thatthe sausagemodesin a cylindricalgeometryfollowa similar pattern:P /2P is also subjectto a 1 2 lowerlimitof1/2,thecutoffaspectratiodecreaseswithv /v .Likewise,theinfluenceofflowonthe Ae A0 standingmodesisqualitativelysimilarinbothgeometries:introducingaflowinthestructurehasamore prominenteffectindeterminingthecutoffaspectratiothanonthevalueoftheperiodratio.WithM in A theexaminedrange[0,0.6],inbothgeometriesa flowmayalter(a/L) in anorder-of-magnitude cutoff senseandthefractionalchangeismorepronouncedathigherdensitycontrasts;whereasthefractional changeinP /2P withrespecttothestaticcaseis.5%. 1 2 Figure5alsoallowsustopayacloserinspectionoftheobservedperiodratiosP /2P ofstanding 1 2 sausagemodes.While1 P /2P ofstandingkinkmodeshasbeenexaminedinconsiderabledetail(see 1 2 − e.g.,the introductionin Macnamara&Roberts2011, andreferencestherein)andputin seismological applications(e.g.,Andriesetal.2005,2009),theuseof1 P /2P ofstandingsausagemodesseemsnot 1 2 − aspopular(seee.g.,Inglisetal.2009).Beforemakingitsserioususe,onemayfirstaskthequestionthat whatleadstothedepartureofP /2P from1inthefirstplace.TheavailabledataforNoRH flareloops 1 2 yieldavalueofP /2P 15.5s/(2 9.5s)= 0.82atanaspectratioa/L = 0.12(Nakariakovetal. 1 2 ≈ × 2003;Melnikovetal. 2005), whilethose forcoolH alphapost-flareloopsyielda valueofP /2P 1 2 ≈ 587s/(2 349s) = 0.84atana/L = 0.03(Srivastavaetal.2008).InviewofFig.5whichaddresses × trappedmodes,thetwovaluesofP /2P aredifficulttoexplain:whichevervaluea/Ltakes,P /2P 1 2 1 2 isfarfromthemeasuredvalues,whichareactuallyoutsidetherangeoftheverticalextentofthisfigure. Introducingaflowshearmakesthecomparisonofthetheoreticallyexpectedvalueswiththemeasured onesevenmoreundesirable:atagivena/L,P /2P intheflowingcaseisactuallyevensmallerthanin 1 2 thestatic case. Adoptinga slabdescriptionforcoronalloopsaswasdoneinInglisetal.(2009, Fig.6) andinpaperI(Figure6therein)doesnothelp,varyingtheparametersoftheequilibriumdoesnoteither, fortheperiodsofstandingmodesaremostlydeterminedbythedensitycontrast(Inglisetal.2009).On theotherhand,intheleakyregime,theperiods(andhencetheirratios)subtlydependontheparameter range of the problem (Nakariakovetal. 2012). We conclude that for sausage modes, what causes the deviationofP /2P fromunityneedsadedicateddetailedinvestigation. 1 2 Figure6extendsourexaminationontheeffectsofflowspeedonthecutoffaspectratio(a/L) cutoff pertinenttothestandingsausagemodesbyshowingthedistributionof(a/L) withvaryingAlfve´n cutoff speedratiosv /v andAlfve´nMachnumbersM .Thecontoursof(a/L) areequallyspacedby Ae A0 A cutoff 0.02.Itcanbeseenthat(a/L) decreasesmonotonicallywithincreasingv /v atagivenM , cutoff Ae A0 A but increases rather dramatically with increasing M at a given v /v . What is more importantin A Ae A0 thecontextofSMSisthatFig.6helpsconstrainthecombinationsofdensitycontrastandinternalflow, whentrappedstandingsausagemodesareobservedinacoronalcylinderwithknownaspectratio.This pointcanbe illustratedif oneexaminesthe flaringloopreportedinNakariakovetal.(2003),whichis 25Mmlongand6Mmindiameter,resultinginanaspectratioofa/L=0.12.Nowthatthefundamental sausagemodeoccurredinthisloop,itsaspectratiohastobelargerthanthecutoffvalue,meaningthat