• Nenhum resultado encontrado

Braz. J. Phys. vol.32 número4

N/A
N/A
Protected

Academic year: 2018

Share "Braz. J. Phys. vol.32 número4"

Copied!
13
0
0

Texto

(1)

Compressible Kelvin-Helmholtz Instability

at the Terrestrial Magnetopause

Alejandro G.Gonzalez

InstitutodeFsiaArroyoSeo, UniversidadNaionaldelCentrode la

ProviniadeBuenosAires,Pinto399, 7000Tandil,BuenosAires,Argentina

JulioGratton

,Fausto T. Gratton

,

INFIPInstitutode FsiadelPlasma,Faultadde CieniasExatasyNaturales,

UniversidaddeBuenosAires, PabellonI,CiudadUniversitaria,1428, BuenosAires,Argentina

and CharlesJ. Farrugia

Spae SieneCenter,Universityof NewHampshireatDurham,NH,USA.

Reeivedon30June,2002

TheompressiblemagnetohydrodynamiKelvin-Helmholtzinstability oursintwovarieties,one

that anbe alled inompressible as it exists inthe limit of vanishing ompressibility (primary

instability),while the other exists onlywhenompressibility is inludedinthemodel(seondary

instability). In previous work we developed tehniques to investigate the stability of a surfae

of disontinuity between two dierent uniform ows. Our treatment inludes arbitrary jumpsof

theveloity andmagnetieldsaswellas ofdensityandtemperature,withnorestrition onthe

wavevetorofthemodes. Thenitallowsstabilityanalysesofomplexongurationsnotpreviously

studiedindetail. Hereweapplyourmethodstoinvestigatethestabilityofvarioustypialsituations

ourring at dierent regions of the front side, and the near anks of the magnetopause. The

physialonditionsofthevetorandsalareldsthatharaterizetheequilibriuminterfaeatthe

positionsonsideredareobtainedbothfromexperimentaldataandfromresultsofsimulationodes

ofthemagnetosheathavailableinthe literature. Wegive partiularattentiontotheompressible

modesinongurationsinwhihtheinompressiblemodesarestabilizedbythemagnetishear. For

ongurationsofthefrontofthemagnetopause,whihhavesmallrelativeveloities,wendthatthe

inompressibleMHDmodelgivesreliableestimatesoftheirstability,andompressibilityeetsdo

notintroduesignianthanges. However,attheanksofthemagnetopausetheourreneofthe

seondaryinstabilityandtheshiftoftheboundaryoftheprimaryinstabilityplayanimportantrole.

Consequently,ongurations thatarestableifompressibilityisnegletedturnouttobeunstable

whenitisonsideredandthestabilitypropertiesarequitesensitiveonthevaluesoftheparameters.

Thenompressibilityshouldbetakenintoaountwhenassessingthestabilityproperties ofthese

ongurations, sine the estimates basedon inompressible MHDmay be misleading. A areful

analysisisrequiredineahase,sinenosimpleruleofthumbanbegiven.

I Introdution

The Kelvin-Helmholtz Instability (KHI) an our in

laboratory and astrophysial plasmas when two (or

more) regionsarein relativemotion. Otherproperties

oftheplasma(likemagnetield,densityand

tempera-ture)mayalsohangearossthetransitionlayer.These

hangesmodifythestabilityandotherpropertiesofthe

modes,thusompliatinganalysis. TheKHImayour

attheterrestrialmagnetopause,theboundarythat

sep-arates theinterplanetarymagnetield (IMF) present

inthemagnetosheath,andthegeomagnetieldinside

the magnetosphere. The relative motion is generated

by thesolarwindowingthroughthe magnetosheath.

Theow veloity and the magneti eld are assumed

parallelto the transitionlayer, andthe magnetield

mayhangeitsdiretion(magnetishear)andits

mag-nitudearossthetransition. Thelowfrequeny

magne-tohydrodynami(MHD)modesshallbeourmain

on-ernheresinetheirinstabilitymayproduelarge-sale

turbulene. The KHI is onsidered a major soure of

(2)

anomaloustransportofmomentumfromthesolarwind

intothemagnetosphere,andaauseofvisousdragat

themagnetopause. Thisissopartiularlyduring

peri-odsofnorthwardIMF,whenthereonnetion of

mag-netilinesislesslikelyto our. Weshallnotonsider

other unstablemodes that analsoarise loseto, and

inside the magnetopause, in frequeny ranges higher

thanthose treatedbyMHD. These modes leadto

mi-rosopiturbulenethat aetsthetransport

proper-tiesoftheplasmaatsmallerwavelengths. Weshallalso

negletdissipativeeetssothatourtreatmentshallbe

basedonidealMHD.TheKHIatthemagnetopausehas

beenexaminedsine the1950's, andalargeliterature

hasgrownon thissubjet. Wequote heresomereent

reviews[?℄,[?℄,[? ℄among severalothers.

The theory of the KHI for arbitrary geometry is

exeedingly ompliated, but fortunately in many

ir-umstanes,thewavelengthoftheperturbationisvery

small as ompared to the urvature radii of the

tran-sition layer. Then the problem an be treated in a

planeslabgeometry,inwhih theunperturbed

quanti-ties(density, pressurep,magneti eld B andmass

owveloityu)dependonlyontheyoordinate

(per-pendiulartothetransitionlayer). Adierential

equa-tionforthelinearMHDperturbationsofstratied

plas-mas [?℄,and its generalization to inlude the eet of

gravity[?℄,anbeusedasastartingpointtoinvestigate

theseproblems.

If the wavelength of the perturbation is large as

ompared to the thikness of the transition layer, we

anignorethestrutureofthetransitionandassumeit

isamathematialsurfae(aty=0)separatingtwo

in-nite, uniform plasmaregions. Theproblem anthen

be treated by means of standard normal mode

anal-ysis. The equations of ideal MHD and the

parame-ters of the onguration do not involve either a

fre-queny or a length sale. Therefore the theory only

preditsthe (omplex) phaseveloityv of the

pertur-bation(v =!=k

t

), as afuntion of the parametersof

theplasma,ofthemassveloitiesoftheuniformregions

and of thewavenumberk

t (k

x ;0;k

z

) ofthe

pertur-bation of the interfae. Then an algebrai dispersion

relationisobtained,whoserootsyield v.

Within the inompressible MHD approximation

(IMHD) it is possible to derive a simple formula for

the phase veloity. Beause of this, the IMHD result

isoftenemployedinspae physistointerpretthe

ob-served data and to examine the stability of the

on-guration. Sometimes the tait assumption is made

that the ompressibility eets thus negleted should

(ifany)improvestability. Thisassumptionisbasedon

wellknowntheorems(seeforexample[?℄)derivedfrom

a variational priniple, aording to whih

ompress-ibilityleadstoapositiveontributiontotheenergyof

theplasmaandsotendstostabilizeit. However,these

resultsannotbeappliedhere,sineduetothepresene

ofmassowtheplasmaansustainperturbationswith

a negative energydensity that anlead to instability.

This isaonsequeneofthetransformationproperties

oftheenergydensityW ofaperturbationof

wavenum-berk

t

andfrequeny!underGalileantransformations.

Let theprimes denote thequantities alulated in the

refereneframeoftheplasmathatmoveswitha

velo-ityu(u

x ;0;u

z

)withrespettotheobserver,andthe

quantities withoutprimes refer to theobserverframe.

ThenwehaveW 0

=! 0

=W=!. Notiethatthe

Doppler-shiftedfrequeny in the plasmaframe! 0

=! uk

t

anbenegativeifu

k =uk

t =k

t

issuÆientlylarge(the

phaseveloityintheplasmaframeisthenv 0

=v u

k ).

Sine W 0

is always positive, the energydensity of the

perturbation as measured in the laboratory frame is

thennegative. Anegativeenergydensityperturbation

mayleadtoinstabilityifitouplestoapositiveenergy

density perturbation, sineboth angrowwithout an

externalsoureofenergy.

InIMHD, theinstabilityarises from perturbations

oftheinterfaethat areexponentiallydampedinboth

plasma regions (like surfae gravity waves in water).

The penetration depth of these perturbations is equal

totheirwavelength2=k

t

. Theinstabilityourswhen

u

k

exeeds a ritial value u

i

(see below), suh that

their energy density (in the laboratory frame) is

pos-itive in one region and negative in the other. These

perturbationsgrowbeauseenergyistransferredaross

theinterfaefromthenegativeenergydensityregionto

theother,andremainsloalizedasitannotbe

trans-ported away aross the magneti eld. This

instabil-itywillbealledprimary KHI.Notiethat theAlfven

wave onlytransports energyalong theeld lines, and

doesnotpartiipateintheKHIifthetransitionregion

hasavanishingthikness.

The essential dierene between IMHD and

om-pressibleMHD (CMHD)isthat inthelatterthereare

perturbations (the fast andslowmagnetosoni waves)

that propagate in the bulk of the plasma and

trans-portenergy arossthemagnetield (seefor example

[?℄) Consequently, theeet of ompressibility on the

Kelvin-Helmholtzinstabilityisomplex,leadingtothe

stabilization of ertain perturbations, the

destabiliza-tionof others, and theourreneof modes that have

noanalogueinIMHDandthatmayleadtonew

insta-bilities. Letusbrieydisussthebasiphysisinvolved.

Considerrst the primary KHI. Due to

ompress-ibility, ifu

k

is suÆiently largethe perturbation may

be able to propagate in both regions as fast

magne-tosoniwaves. Whenthishappens,theperturbationis

stable, and itis seenby the observerasapair of fast

magnetosoni wavesradiated awayfrom the interfae.

Oneofthesewaveshasanegative,andtheothera

pos-itiveenergydensity[?℄,[?℄. Inthisway,perturbations

that areunstable aordingto IMHDare stabilized by

ompressibility. However,in addition to this large-u

k

stabilization, there is another eet of ompressibility

(3)

for the onset of the instability is lowered (u

u

i ).

This is a destabilizing eet, sine perturbations that

are stable aording to IMHDbeome unstable when

ompressibilityistakenintoaount. Asweshallshow

later,thiseetisimportantinthemagnetopause. The

penetrationdepthoftheunstablemodesisalsoaeted

byompressibility,andisnolongergivenby2=k

t .

Yet,therearefurthereetsofompressibility. Due

to the existene of the slow magnetosoni waves new

kindsofperturbationsoftheinterfaearepossible,that

havenoounterpartinIMHD.Someofthese

perturba-tions are stable evanesent osillations, and other are

slowmagnetosoniwavesin bothregions(radiationof

apairof slowmagnetosoniwaves). However,someof

these new perturbationsmay be unstable. These new

instabilities (alledseondary KHI)arefoundin

inter-vals of u

k

that orrespond to stable perturbations

a-ording to IMHD. They lie totally, or partially, below

the ritialvalueu

fortheonset oftheprimaryKHI.

The ourreneofthe seondaryKHI is anadditional

destabilizing eet ofompressibility. Thegrowthrate

of theseondary modesis usually small. However, we

shall see that these modesannot be ignored,sinein

someongurationsof themagnetopausetheyare the

onlyunstablemodespresent.

Asubstantialontribution to theunderstandingof

theompressibilityeetsontheKelvin-Helmholtz

in-stability was given by Miura and Prithett [?℄.

How-ever,theseeets havenotyet beenfully explored for

general ongurations with density and temperature

disontinuitiesaswellasmagneti eld shear,suh as

those ourring in the magnetopause. For an

exten-sive revision of the literature and of the main results

on this subjet, see refs. [?℄, [?℄. Reently, we have

developedtehniques[?℄,[?℄,[?℄ to analyzethe

stabil-ity of an interfae separatingtwo uniform ows, with

arbitraryjumps intheveloity,magnetield, density

andtemperature,andwithnorestritionsonthewave

vetoroftheperturbation.

Theontentof this paper isas follows. InSetion

II we outline the basi theory. In Setion III we

ex-plain thetehniquesfor analyzing thedispersion

rela-tionandtheharateristisoftheunstablemodes.

Se-tionIVdisussestheseondaryKHI.Ausefulgraphial

tehniquethatsimpliestheomplexityofthestability

analysisisgiveninSetionV.Thephysialparameters

used in theappliationsto themagnetopausestability

areintroduedinSetionVI.Theresultsfortwo

exam-ples, (i) amodel of the front side magnetopause,and

(ii) an event observed at the near equatorial ank of

the magnetopausearegiven in Setion VII.We

deter-mine the stability of these ongurations, and study

the unstable modes that may our in them. The

nal remarks are in Setion VIII. We onlude that

theeetsof ompressibility mayintrodueimportant

hangeswithrespettothepreditionsofIMHD.

II Theory

Weonsidertwosemiinnite,uniformplasmaregionsin

relativemotion. Thegeometryoftheproblemisshown

in Fig. 1. Region 1 (y > 0), moves with aonstant

veloity u 0

1

, and region 2 (y < 0) moves with a

on-stant veloity u 0

2

. The veloities u 0

1 and u

0

2

and the

magnetieldsB

1 andB

2

areparalleltotheinterfae

(suÆxesi=1;2denote quantities pertaining toregion

1 and 2, respetively). At the interfae, p, , and B

anhavearbitrarydisontinuities,subjetto the

equi-libriumondition

p

1 +

B 2

1

8 =p

2 +

B 2

2

8

(1)

u

1

=

u

B

1

k

a

j

1

x

z

u

2

=–

u

B

2

k

a

y

2

x

z

region 1 (

y

> 0)

region 2 (

y

< 0)

j

2

p

1

p

2

r

1

r

2

y

1

Figure1. Geometryoftheproblem.

WeshallassumeidealCMHDandonsiderthe

lin-earadiabatiperturbationsofthisonguration. Sine

the growth rate of the instability does not depend

on u 0

1 and u

0

2

separately, but only on their dierene

2u =u 0

1 u

0

2

, weuse a referene frame moving with

the average veloity u 0

= (u 0

1 +u

0

2

)=2. In this frame

u

1

= u=ue

x , u

2

= u, and is the angle between

uand the wavenumberk

t (k

x ;0;k

z

) ofa

perturba-tion whose frequeny is !; the phase veloity is then

v=!=k

t

. Onethephaseveloityoftheperturbation

isfound,weanobtainitsvaluein theobserverframe

bymeansoftheGalileantransformationv 0

=v u 0

k .

The (Doppler-shifted) frequenies in the plasma

framesare!

1

=! u

k k

t

inregion1and!

2

=!+u

k k

t

inregion2,andtheorrespondingphaseveloitiesare

v

1

=v u

k ; v

2

=v+u

k

(2)

We shall denote by

i

the angles betweenB

i and

k

t

,andby'

i

theanglesbetweenB

i

andu. Theangle

betweenk

t

anduis,sothatu

k

=uos. Weemploy

(4)

A i = B i p 4 i

=Alfvenveloity ; S

i = r p i i

=Soundveloity(=5=3) (3)

d i = p A 2 i +S 2 i ; a i =A i os i ; b i =(A i S i =d i )os i (4) q i = q 1 2 (d 2 i + p d 4 i 4d 2 i b 2 i ; m i = q 1 2 (d 2 i p d 4 i 4d 2 i b 2 i (5) i = r (v 2 i q 2 i )(v 2 i m 2 i ) d 2 i (v 2 i b 2 i )

; withRe(

i

)>0;orIm(

i

)>0 ifRe(

i )=0

(6)

d

Ineahoneoftheuniformregions,thegeneralform

oftheperturbation is

i =e i(k t x !t) (C i e k t i y +D i e k t i y ) (7)

Inthisformula,

i

denotesthey-omponentofthe

dis-plaement,andC

i andD

i

areonstants. The

penetra-tiondepthoftheperturbationinregioniisproportional

to1=Re(

i ).

When Re(

i

) = 0, it an be reognized that eq.

(6) is the dispersion relation for the fast and slow

magnetosoni wavesin a semiinnite uniform moving

plasma; their wavevetoris (k

x ;k t Im( i );k z ). When Re( i

)6=0,eq. (6)isthedispersionrelationfor

evanes-ent perturbations, andof oursein this asethe

am-plitudesoftheexponentiallygrowingperturbations(D

1

andC

2

)ineq. (7)mustvanishsinetheplasmaextends

to(+ or )innity.

Theboundary onditionsat y=0(ontinuityof

andofthenormalstress)oupletheperturbationsof

re-gions1and2. Theouplingleadstoamatrixequation

that relatestheamplitudesC

i ,D i : C 1 D 1 = 1 2 1 (v 2 1 a 2 1 ) T S S T C 2 D 2 (8) Here T= 1 (v 2 1 a 2 1 ) 1 2 (v 2 2 a 2 2 ) 2

; S= 1 (v 2 1 a 2 1 ) 1 + 2 (v 2 2 a 2 2 ) 2 (9) d

The matrix equation (8) may be used to

investi-gatemanyproblemsinvolvingloalizedmodesaswellas

thereetionandtransmissionofmagnetosoniwaves.

Sine unstable modes and stable surfae osillations

have Re(

1;2

) 6= 0, we must have C

1 = D

2

= 0 and

thentheorrespondingdispersionrelationis

T= 1 (v 2 1 a 2 1 ) 1 2 (v 2 2 a 2 2 ) 2

=0 (10)

Thepropagating modeshaveRe(

1;2

)=0,so that

there isno restritionontheC

i , D

i

. Inthis ase,the

onditionsT=0andS=0determinethepolesandthe

zerosofthereetionoeÆient. Notiethatthesignof

they-omponentofthegroupveloityofmagnetosoni

wavesmaybedierentfromthesignofRe(

1;2

). Then

one must examine thegroupveloities in eah region,

to asertain whih of these onditions yields the

dis-persionrelationforradiationofapairof waves,orfor

thetotaltransmissionofwaves(andtheirtime-reversed

proesses).

III Classiation of the modes

Toinvestigatetheroots ofthedispersionrelation(10)

itisusefultointrodueapolynomialP assoiatedtoit.

It is obtainedbytaking thenumeratorof theprodut

TS,thuseliminatingthesquareroots:

(5)

Clearly, the roots of P = 0 enompass the

solu-tionsof T=0('true'roots)aswell asthoseofS =0

('spurious'roots),andwemustdisardthelatterwhen

welookforloalizedmodes. Aordingly,weall'true

branh'(T)thesetofsolutionsofT=0,and'spurious

branh'(S) thesolutionsof S=0. TheT branhand

theSbranhonsistofseveralpiees(thatweall

'seg-ments'), eah ofwhihis aontinuous manifoldin the

parameterspaeoftheproblem. Theyarearrangedin

suhawaythatanygivenTsegmentisboundedbyone

ormoreS segments(andvie-versa). Letusonsidera

rootv (realoromplex)ofP thatbelongstoaertain

segmentT

j

. If theparametershange, v movesalong

T

j

untileventuallyitarrivesto theboundarybetween

T

j

andasegmentS

l

ofS. Asitrossesthisboundary,

veasestobelongtothetruebranhandbeomes

spu-rious. ItisimportanttonotiethattheT-Sboundaries

ouronlyforv real,at thepointswhere

1;2

=0;1.

This is fortunate, sine it simplies onsiderably the

analysis ofthetopologyoftheTandS branhes.

A onvenientwayto visualize the real roots of the

dispersionrelation(10)andsounravelthetopologyof

theTandSbranhesistouseagraphialmethod

intro-duedbyChandrasekhar[?℄. Thismethodtakes

advan-tageofthefatthattheproblemofndingtherootsv

ofP =0isequivalenttosolveforv

1 andv

2

theoupled

equations

P(v

1 ;v

2

)=0 (12a)

v

2 =v

1 +2u

k

(12b)

Fromanygivensolution(v

1 ;v

2

)ofthesystem(12),

one an obtainaroot v =(v

1 +v

2

)=2of P =0. The

realsolutionsofthissystemanbeeasilyfound

graph-ially in a(v

1 ;v

2

) diagram,astheintersetions of the

lineL givenby(12b)withtheurvesthatareobtained

solvingthebiubi(inv

1 andv

2

)equation(12a)forv

2 .

In Fig. 2 we show a typial Chandrasekhar

dia-gram, orresponding to the plasma onguration (d)

investigatedin Setion VII(notie thatthesaleof v

2

isdierentfromthatofv

1

foronvenieneindrawing).

ItsuÆestoonsiderasinglequadrant,sinetheurves

v

2 =v

2 (v

1

)are symmetrialunder reetionsonboth

axes. In Fig. 2, thelines v

2 =v

1 and v

2

= v

1

rep-resentthev- and u

k

- axes (exeptfrom asale fator

p

2).

Theperturbationintheregioni(i=1;2)isaslow

magnetosoniwaveifb

i j v

i jm

i

,andafast

magne-tosoniwaveifq

i jv

i

j. Itisevanesentif0jv

i jb

i

orm

i jv

i jq

i

(weshallallthelatterthe'slow'and

'fast'evanesentperturbations,nottobeonfusedwith

theslowandfastmagnetosoniwaves).Arealroot

be-longstotheTbranhifSign(v 2

1 a

2

1

)6=Sign(v 2

2 a

2

2 )(i.

e.,ifjv

1 j<a

1 andjv

2 j>a

2 ,orjv

1 j>a

1 andj v

2 j<a

2 );

R

2

R1

R1

R3

–q

1

–a

1

–m

1

–b

1

q

2

a

2

m

2

b

2

R3

T

2

T1

T1

S2

S1

S3

S3

L

A

A

B

B

200

(6)

u

k = u

d2

when L is tangent to T

2

. For u

k >u d2 the rootsv B , v B 0

areomplexonjugate,buttheystill

be-longtotheTbranh(sine rootsanpasstotheother

branh only for v real). Then u

k = u

d2

orresponds

to marginal stability, and one of the roots v

B , v

B 0 is

unstablefor u

k >u

d2

(the other isdamped). This

in-stability persists untilu

k =u

d6

, whenL is tangentto

thesegmentR

2

. Thenu

d6

orrespondsalsotomarginal

stability,andforu

k >u d6 ,v B ,v B 0

arestableradiation

modes. On the other hand, if oneonsiders valuesof

u

k

larger than that represented in Fig. 2, the points

A, A 0

movealongT 0

1

and eventuallypassintothe

seg-mentsR

1 ,R

0

1

,andthenv

A ,v

A 0

orrespondtoradiation

modes.

Let usnowonsider smallervaluesof u

k . As u

k is

redued,L movesdownwards. ThepointsB, B 0

move

on T

2

until they pass into S

2

. It an be shown that

for u

k

> 0, no new instability arises from the roots

v

B , v

B

0. On the other hand, the points A, A 0

move

along T 0

1

until theyoalese foru

k =u

d3

. Whenthis

happens,thedoublerootv

A =v

A

0 ismarginallystable.

Foru

d1 <u

k <u

d3

wehaveinstability,andv

A ,v

A 0

,are

omplexonjugate. Marginalstabilityisagainahieved

atu

k =u

d1 . Foru

k <u

d1

thepointsA,A 0

movealong

T

1

untiltheypassintoS

1

. Nootherinstabilityrelated

tov

A ,v

A 0

oursforu

k

>0. UsingdiagramslikeFig. 2

itispossibletoinvestigatethestabilityofany

ongu-rationofinterest. However,thismethodisunpratial,

sineweneedanewdiagramforeahorientationofk

t ,

and in addition, weannot determinethegrowthrate

oftheunstablemodes.

In theaseshownin Fig. 2thereare twounstable

intervals(foru

k

>0). Therst(in orderofinreasing

u

k

) involves v

A , v

A

0 and ours for u

d1 < u k < u d3 .

The seond ours for u

d2 < u

k < u

d6

, and involves

v

B , v

B 0

. This resultisdue tothespeialhoieof

pa-rametersof Fig. 2. In general,there anbeup tosix

positivevaluesof u

k

orrespondingto signiant(i.e.,

belongingto the Tbranh)double roots of P(v)= 0,

that we label u

d1 ;:::;u

d6

. Correspondingly, there an

beuptothreeunstableintervals,namely

u d1 <u k <u d3

seondaryintervalA (13a)

u d2 <u k <u d4

seondaryintervalB (13b)

u d5 <u k <u d6

primaryinterval (13)

As theparametersare varied,itmayhappen that the

pair(u

d3 ,u

d5

),orthepair(u

d4 ,u

d5

),oalesesand

be-omes omplex. In therst ase,the seondary

inter-valA mergeswiththeprimaryinterval. Intheseond

ase,theseondaryintervalBmergeswiththeprimary

interval(as in Fig. 2). In addition the seondary

in-tervals (13a, b) may also overlap partially or totally.

The seondary intervals an be extremely narrow for

somevaluesoftheparameters,inpartiulartheyvanish

when theperturbations areutesin either region1or

2. Fornegativeu

k

, aompletelysymmetrialresultis

obtained,inwhihthesignsofu

d1 ;:::;u

d6

arehanged,

theinequalities(13a-)arereversed,andtherolesofthe

root pairs (A, A 0

) and (B, B 0

) are exhanged. There

arenosimpleformulaeforu

d1 ;:::;u

d6 .

Tondthedoublerootsv

d1 ;:::;v

d6

ofP(v)andthe

orresponding u

k

valuesu

d1 ;:::;u

d6

wemust solve

nu-meriallyfor(u;v)thesystem

P(u;v)=0 ; P(u;v)=v=0 (14)

and disardtheroots belongingto the Sbranh. The

unstableintervalsanthenbemappedintheparameter

spaeoftheonguration(seeSetion V).

IV The seondary

Kelvin-Helmholtz instability

With thehelp ofdiagrams likeFig. 2it iseasyto

un-derstand the role of the ompressibility on the KHI.

As S i inreases, b i and m i approah a i

, and q

i

in-reases without bound, so that in the inompressible

limit (S

1 ;S

2

! 1) the seondary unstable intervals

disappear, and we are left only with the primary

un-stable interval,that now extends to innity. The

se-ondary KHI is then a onsequene of ompressibility,

aswellasthelargeu

k

stabilizationoftheprimaryKHI

at u

d6

. It isenlighteningto seein moredetailthe

ori-ginoftheseondaryKHI.NotiethatP(v

1 ;v

2

)anbe

written intheform

P(v

1 ;v

2 )=S

2 1 S 2 2 P in (v 1 ;v 2 )+S

2 1 P 1 (v 1 ;v 2 )+S

2 2 P 2 (v 1 ;v 2 )+P

3 (v 1 ;v 2 ) (15) In(15), P in (v 1 ;v 2 )=(v

(7)

and P 1 , P 2 , P 3

, arepolynomialswhose oeÆientsare

funtions of a

1 , a 2 , A 1 , A 2 , 1 and 2

. Then, in the

limitS

1 ;S

2

!1,ifweassumethat v

1;2

remainnite,

eq. (12a)reduesto P

in

=0,whosesolutionsare:

v 1 =a 1 (16a) v 2 =a 2 (16b) v 2 = r a 2 2 1 2 (v 2 12 a 2 1 ) (16) v 2 = r a 2 2 + 1 2 (v 2 12 a 2 1 ) (16d)

Thesesolutionsyield,respetively,theroots

v =v

a1 =u k a 1 (17a)

v=v

a2 = u k a 2 (17b)

v=v

K =u k 1 2 1 + 2 p (a 2 1 1 +a 2 2 2 )( 1 + 2 ) 4u 2 k 1 2 1 + 2 (17)

v=v

S =u k 1 + 2 1 2 p (a 2 1 1 a 2 2 2 )( 1 2 )+4u

2 k 1 2 1 2 (17d) d

The roots (17a) and (17b) are stable, and orrespond

toAlfvenwavespropagatinginregions1and2,

respe-tively. Theroots (17)havebeendisussedbyAxford

[?℄,[?℄, Chandrasekhar [?℄ and Southwood [?℄. They

yield the inompressible (primary) Kelvin-Helmholtz

instabilityfor

ju

k j>u

i = 1 2 r 1 2 1 2 (a 2 1 1 +a 2 2 2 ) (18)

Finally,theroots(17d)arespurious.

Someoftheroots(17a d)anoinideforspeial

valuesofu

k

. Forexample,when

u k =u D = 1 2 (a 1 +a 2 ) (19)

therootsv

a1 ,v

a2+ ,v

K

( or+aordingtothesign

ofa 2 1 1 a 2 2 2 )andv

S

oinideandtheirvalueis

v=v

D = 1 2 ( a 1 +a 2 ) (20)

This degeneray orresponds to the point D

1 (v 1 = a 1 , v 2 = a 2

), where the lines (16a d) ross. It

an be veried that u

i

u

D

, and that the

equal-ity holds only for a 2 1 1 a 2 2 2

= 0. Let us now

as-sume that S

1 ;S

2

are large, but nite. Then the term

S 2 1 P 1 (v 1 ;v 2 )+S

2 2 P 2 (v 1 ;v 2

) in (15) an be treated as

asmallperturbationthatouplesthedegenerateroots

v a1 , v a2+ ,v K+ (orv K ),v S

. Itanbeshownthat

partofthe degenerayisremoved,andsomeroots

a-quireanimaginarypart. Theunstablemodessoarising

(seondaryKHI) have no analogue among the purely

inompressiblemodes. It isinterestingto observethat

this instability appearsaround u

k u

D

, that is, in a

rangeofu

k

that isstable aordingtoIMHD, sineit

liesbelowu

i .

InadditiontoD

1

,therearethreeotherpointswhere

degeneray of the roots (17a d) ours, namely D

2

(u = u

D

, v = v

D ), D

3

(u = v

D

, v =u

D

) and D

4

(u= v

D

, v = u

D

). Theremovalof thedegeneray

atD

2

duetoompressibilityleadstooverstablemodes

likeat D

1

. Onthe otherhand,the removalof the

de-generaiesatD

3 andD

4

doesnotleadtoinstabilities.

V Stability diagrams

The problem of nding the loalized eigenmodes is

straightforward,iftedious,sinewemustalulate

nu-meriallytherootsofthepolynomialP(ofdegreeten),

thatdependsonsevenindependentparameters. Six of

them haraterize the plasma onguration; they an

be taken as the ratios r

b = B 2 =B 1 , r s = S 2 =S 1 and r d = 2 = 1

, the magnitude u of the relative veloity,

andtheangles'

1

from B

1

tou,and fromB

1 to B

2

(magnetishear angle). Theremainingistheangle

1

from B

1 to k

t

, and identiesthe perturbation we are

onsidering.

Notie,however,thatP (anditsroots)dependson

uonlythroughtheombination

(8)

Aordingly,weshallrepresenttheresultsofthe

anal-ysisbymeansofdiagramsinwhihweplotthegrowth

rate Im(v) of the unstable modes as funtions of

1

andu

k

,keepingxedtheremainingparameters(r

b ,r

s ,

r

d

, ) that determine P. An example is the stability

diagram shown in Fig. 3, in whih we have drawn a

ontourplot of Im(v) forthe parametersof aase

in-vestigatedin SetionVI(ase(d),r

b

=1:5, r

s

=2:67,

r

d

=0:117, = 80 Æ

). InFig. 3wehavealso drawn

theurveu=u

i (

1

) thatgivesthemarginalstability

ondition(18) aordingto the inompressible model.

Allthepointsofthestabilitydiagramaboveu

i (

1 )

or-respondto instabilityaordingto IMHD.Notiethat

theontourplotdoesnotrefertoauniqueplasma

on-guration, sine we have not yet speied u and '

1 .

Then it an be used for all the ongurations having

thesame valuesof r

b , r

s , r

d

and. This is adistint

advantagesinewithasinglegraphweanstudythe

ef-fetsofhangingthemagnitudeoftherelativeveloity

anditsorientationwithrespettothemagnetield.

In Fig. 3, we an see the main eets of

om-pressibility: thelarge-u

k

stabilizationandthesmall-u

k

destabilizationoftheprimaryinstability,andthe

pres-ene of the seondary instability in u

k

intervals that

are stable aordingto IMHD.It should also be

men-tionedthatforu

k

appreiablylargerthanu

i (

1 )(i. e.,

not very lose to it) the growth ratesalulated with

CMHD are always smaller thanthose alulatedwith

IMHD(17).

Toinvestigatea partiularonguration bymeans

ofthestabilitydiagramwemustdrawtheurveCgiven

by eq. (21), that depends on the two remaining

pa-rameters(u, '

1

)that haraterizeit. FollowingC we

nd the value of Im(v) for this onguration, for any

perturbationharaterizedbythe angle

1

. Sinethe

problemisinvariantunderthetransformation(v! v,

u

k

! u

k

),itsuÆestoalulatetheontourplotfor

u

k

0;ofourse,wemustthendrawtheurveC 0

given

byu

k

= uosin additiontoC.

In the example shown in Fig. 3 we see that the

ongurationdesribedbyCandC 0

(u=1:7A

1 , '

1 =

89:9 Æ

) is stable aording to IMHD, but is unstable

if ompressibility is taken into aount. Theunstable

modes our in the (approximate) range 40 Æ

1

50 Æ

andtheybelongtotheseondaryKHI.Their

max-imum growth rateis small (Im(v)0:04A

1

), but

sig-niant. It anbe observedthat other ongurations

withthesamevaluesofr

b ,r

s ,r

d

andanbemore

un-stable. TheextremeofIm(v)inFig. 3is2:115A

1 ;it

oursfor

1 25

Æ

andu

k 5A

1

,avaluethatrequires

averylargeu.

The meaning of the large-u

k

stabilization an be

laried by means of an example. Let us onsider in

Fig. 3ahypothetialongurationwithu=10A

1 and

'

1

= 89:9 Æ

. Thenew urves C

n and C

0

n

are similar

to CandC 0

, onlysaled bythefator10=1:7. Partof

C

n (for45

Æ

113

Æ

)liesin theupperpartofthe

stability diagram (u

k > u

d6

) orresponding to stable

perturbations: theseare thelarge-u

k

modes stabilized

by ompressibility. However, this onguration is

un-stable,sineC

n andC

0

n

passthroughtheunstablepart

ofthediagram(atually,itsmostunstablemodeshave

Im(v)verylosetotheextremevalueforthediagram).

Clearly, ongurations having verylarge u arealways

unstable,althoughlarge-u

k

stabilizationmayourfor

somemodes.

0.2

0.4

0.6

0.8

1

y

1

/p

2.5

5

7.5

u

k

/

A

1

C

C’

C

n

C

n

'

u

i

Figure3. Stabilitydiagramforase(d)ofSetionVII.The

linesCandC 0

orrespondtothevaluesofu

k

ofthe

ongu-ration,givenbyeq. (21). TheontoursofIm(v)arespaed

by0:2A

1

. Thelarge whiteareasatthe topandbottomof

thediagramorrespondtostableperturbations.

A few omments must be made onerning

stabil-ity diagrams like Fig. 3. To draw them we alulate

therootsofthedispersionrelationatdisretepointsof

a grid in the (

1 ;u

k

) plane. Sometimes, at the same

point (

1 ;u

k

) there are twounstable roots (this

hap-pens when the seondary intervals A and B overlap,

or when any one of them overlaps parts of the

pri-mary interval). In these instanes we show only the

largerIm(v). Theseondaryintervalsare verynarrow

in some (

1 ;u

k

) regions, and may not show up, due

to theoarsenessofourgrid(forthesameausesome

ontours appear wavy in Fig. 3). Nothwithstanding

thesedrawbaks,stabilitydiagramsareextremely

use-fultools,sinetheyprovideaneonomialandfastway

for ndingthe mostinteresting domains in parameter

spae. It isthen possibleto performmorepreiseand

detailed alulationsin these regions only, with

(9)

VI Parameters of magnetopause

ongurations

Before disussing the appliation of thetheory to the

magnetopause, it is onvenientto review how the

pa-rameters used in our formulae an be obtained from

thetheoretialmodelsorfrom spaeraftdata. We

as-sumethattheoutsideregion(asseenfromtheEarth)is

region1. Intheappliations,theparametersofthe

on-guration(theowveloitiesu 0

1 andu

0

2

andtheangles

B1;u 0

1 and

B1;u 0

2

fromB

1 tou

0

1 andu

0

2

,respetively)

are given in a referene frame at rest with respet of

theEarth. Inourtheoryweusearefereneframe

mov-ing with the average veloity u 0

= (u 0

1 +u

0

2

)=2, then

u

1

= u, u

2

= u, with u = (u 0

1 u

0

2

)=2. In what

follows,allveloitiesaregiveninkm/s. Weemploythe

followingformulae:

u= 1

2 q

u 02

1 +u

02

2 2u

0

1 u

0

2 os

u ; u

0

= 1

2 q

u 02

1 +u

02

2 +2u

0

1 u

0

2 os

u ;

u =

B

1 ;u

0

2 B

1 ;u

0

1

(22)

'

1 =

B

1 ; u

0

1

+Sign(sin

u )aros

u 0

1 u

0

2 os

u

2u

(23)

d

TheangleÆfromutou 0

isgivenby

Æ=Sign(sin

u )aros

u 02

1 u

02

2

4uu 0

(24)

Ineqs. (23)and(24)thefatorSign(sin

u

)isneededto

obtaintheorret determination of'

1

andÆ. Finally,

the phase veloity of the perturbation in the Earth's

frameisgivenby

v 0

=v u 0

os(

1 '

1

Æ) (25)

Toonsidertheremainingparameterswemust

dis-ussseparatelythetheoretialandspaeraftdata.

Data derived from theoretialmodels

The onditions for the exitation of the

Kelvin-Helmholtz instability on the dayside magnetopause

werereentlyinvestigatedintwopapers. One,byMiura

[?℄, reports numerial MHD simulations on the

devel-opmentoftheinstability,startingfromhyperboli

tan-gentproles of the unperturbed eld quantities. One

of themain results is that the magnetopauseappears

tobemoreKHunstablefornorthwardIMFonditions

thanforsouthwardpointingmagnetields. However,

the hoie of parameters for the simulations is more

adequate for the analysis of near ank ongurations

rather than dayside proper. The seond paper [?℄ is

based on a linear inompressible theory, and

spei-allyaddressesthedistributionoftheKHativityover

themagnetopausesurfae. ChartsofKHIgrowthrates,

relyingoninputparametersprovided byaMHD ode

that simulates thephysialonditions of thefrontside

magnetosheath, are given. One of the main results is

thatundernorthwardIMFonditions,theKHativity

is restrited to somewell dened regions of the

mag-netopause. Even when the IMF is due north, the

a-tivitydereasesasthelatitudeinreasesfromthe

mag-netopauseequator. In anthird paper [?℄ the latitude

dependene of the KHI is examined with a linear

in-ompressibletheory based on hyperboli tangent

pro-les, using our inompressible theory for a tangential

veloitydisontinuity.

The data used here are given in Tables 1 and 2.

Those of Table 1 (also used in [?℄) proeed from ref.

[?℄andwerealulatedwiththemagnetosheathmodel

of ref. [?℄, whih is also desribed in [?℄. The IMF

is assumed to point exatly north. The positions are

atequator,middlelatitude,and highlatitude,alonga

meridianabout3hourseastof noon. Themain

dier-eneamong these positions is theinreaseoftheloal

magnetishear angle.

Table1. Parametersofongurationsofthefrontofthemagnetopauseaordingtothemodel.

Case Latitude f

u 0

f

b

f

d

r

b B

1 ;u

0

1

(a) low 0:2352 1:1994 0:7502 1:2670 90 Æ

0 Æ

(b) intermediate 0:2568 1:0875 0:7387 0:8935 81 Æ

6 Æ

() high 0:2945 1:0055 0:7259 0:5288 92 Æ

(10)

Table2. Additionalparametersof ongurationsofthefrontofthemagnetopause(r

s

hasbeenalulatedto

satisfytheequilibrium ondition,veloitiesareinkm/s).

Case Latitude u

1 f

s r

d r

s A

1

(a) low 500 1:2 0:1 1:81 692

(b) intermediate 500 0:463 0:1 4:75 633

() high 500 0:463 0:1 7:00 589

Wehavein this aseu 0

2

=0, thenu=u 0

=u 0

1 =2,

'

1 =

B

1 ;u

0

1 and v

0

= v u

0

os(

1 '

1

). The

model provides

B

1 ;u

0

1

, the shear angle , the ratio

r

b =B

2 =B

1

,aswellasthefollowingquantities:

f

u 0

=u 0

=u

1 ; f

b =B

1 =u

1 p

4

1 ; f

d =

1 =

1

(26)

In(?? ),u

1 and

1

aretheveloityandthedensity

ofthesolarwind,farfromtheinterationregion.

Beausethemagnetosheathmodeldoesnotprovide

magnetospherivaluesfortemperatureanddensity,

Ta-ble2omplementsthesetofparametersneededinthe

theory. Weassumeu

1

=500km/sandusetypial

val-uesofthedensityratioandofthetemperaturesinboth

regions, often observed in spaeraft rossings of the

magnetopause (expressed as r

d =

2 =

1 , r

s = S

2 =S

1

and f

s = S

1 =u

1

). Clearly r

d , r

s and f

s

must

sat-isfy the equilibrium ondition (1). In terms of the f

's and r's this implies2f

d f

2

s (1 r

d r

2

s ) =f

2

b (r

2

b 1).

So we rst assign r

d and f

s

, and then use this

rela-tiontoalulater

s

. Siner 2

s

mustbepositive,weneed

f 2

s > f

2

b (r

2

b

1)=2f

d

. When r

b

> 1 (as in ase (a)

below)this ondtionsetsalowerbound onf

s .

UsingthedataofTables1and2wethennd

u=f

u 0

u

1

2 ; A

1 =

u 0

1 f

b

f

u 0

p

f

d ; A

2 =A

1 r

b p

r

d ; S

1 =

u 0

1 f

s

f

u 0

; S

2 =r

s S

1

(27)

d

and from these quantities we alulate the remaining

parametersofourtheory.

Spaeraft data

At the anks of the magnetopause, the

magne-tosheathowissupersoniandtheeetsof

ompress-ibility on the KHI are expeted to be important. We

shall examine a onguration at the near equatorial

magnetopauseatdusk,observedby Interball/tail

dur-ingtheeventofJanuary11,1997. Thiswasatthetailof

aoronalmassejetionpassingEarth,duringwhihthe

dynamipressureofahighdensitysolarwindstrongly

ompressedthemagnetosphere. As aonsequene,

In-terball/tail was for some time in the magnetosheath,

and reentered the magnetosphere, after the dynami

pressure returned to lower values. During the

ross-ing of the magnetopause, from magnetosheath to low

latitude boundary layer, the spaeraft measured the

physial parametersneessary for a stability analysis.

A detailed study of the event is given in ref. [?℄,

to-getherwithadisussionoftheKelvin-Helmholtz

insta-bility,basedonaninompressiblemodel. Weintendto

revisitthatongurationwithourompressibletheory.

Theinterfaethatweshallonsideristhatbetweenthe

magnetosheathandthelowlatitudeboundarylayer.

Table3. ParametersoftheJanuary11,1997event(spaeraftdata,veloities areinkm/s).

Case u 0

1 u

0

2 M

A;1 M

S;1 r

b r

d

B1;u 0

1

B1;u 0

2

(d) 300 105 5:0 2:0 1:5 0:117 100 Æ

120 Æ

80 Æ

(e) 300 105 5:7 2:0 1:5 0:117 100 Æ

120 Æ

80 Æ

Table4. CalulatedparametersoftheJanuary11, 1997event(veloitiesareinkm/s).

Case u '

1 u

0

Æ r

s A

1

(11)

In therst line of Table3 we give theparameters

of the event (we use the denitions M

A;1

A

1 =u

0

1 ,

M

S;1 S

1 =u

0

1

). The seond line ontainsthe

param-eters of a hypothetial onguration with a slightly

larger value of M

A;1

, to asertain the eet of this

hange.

Wemustreognizethatowingto measurement

un-ertainties,thespaeraftdatadonotomplywiththe

equilibrium ondition (1) at the interfae. To emend

this aw, we replaed the measured r

s

(that is the

lessreliabledatum)bythe valuederivedusing the

re-mainingdataintheequilibriumondition,expressedas

2M 2

A;1 (1 r

d r

2

s )=M

2

S;1 (r

2

b 1).

Theexpressions ofu,u 0

, '

1

,Æ and v 0

aregivenby

eqs. (22)-(25), and the remaining quantities we need

are:

A

1 =u

0

1 =M

A;1 ; A

2 =A

1 r

b p

r

d ; S

1 =u

0

1 =M

S;1 ; S

2 =r

d S

1

(28)

d

Wenowinvestigatethestabilityofthese

ongura-tionsofthemagnetopause.

VII Stability of some

magne-topause ongurations

Congurations near the front of the

magne-topause

In the low latitude ase (a) there is no magneti

shear, so that the ute modes are unstable, both in

theinompressiblelimitandifompressibilityistaken

intoaount. InFig.4weomparethegrowthrate

ob-tainedwiththeompressiblemodelwiththatpredited

byIMHD(eq. 17). Itanbenotied thatthere isno

signiantdierene. Aninreaseoftherelative

velo-itybyafatorof3-4doesnothangethis onlusion.

0.46

0.48

0.5

0.52

0.54

0.02

0.04

0.06

y

1

/

p

Im (

V

)

/

A

1

Figure 4. Growth rate of the unstable modes of the low

latitudeonguration(ase(a)).TheresultsofCMHDand

IMHDare shownwith fulllinesand gray dashedlines,

re-spetively.Thetwonarrowbandsoftheseondary

instabil-ity near

1

=0:48 and

1

=0:52 arebarely disernible

astheirgrowthratesareverysmall.

The intermediate and high latitude ases are

sta-that oinides with the preditions of IMHD. In Fig.

5weshow the stability diagram for the high latitude

parameters. We an see the lowering of the ritial

valueu

, dueto ompressibility(alreadymentionedin

SetionI).Thiseetisofnoonsequeneforthe

on-guration() owingtothe lowrelativeveloity,but it

mayberelevantforongurationswithalargerelative

veloity(say,u1:5A

1 ).

0.2

0.4

0.6

0.8

1

y

1

/

p

0.5

1

1.5

2

u

k

/

A

1

C

u

i

Figure5. Stabilitydiagram fortheparametersofthe high

latitudease. TheurveCorrespondingtothe

ongura-tionindiatesthatase()isstable.

Aonguration in the near ank of the

magne-topause

The stability diagram orresponding to these

pa-rameters is shown in Fig. 3. It an be appreiated

thatthisongurationisstableaordingtotheIMHD

(12)

to theseondarymodes, whih haveno ounterpartin

IMHD.Thegrowthrateoftheinstabilityissmall, but

signiant,asanbeseeninFig. 6.

0.22

0.24

0.26

0.28

0.02

0.04

0.06

y

1

/

p

Im(

V

)

/

A

1

Figure6. Growthrateoftheseondarymodesforase(d).

When the line C of the onguration passes lose

to the marginal stability line for the primary modes,

as happened in this ase, the stability properties of

the onguration are very sensitive to slight hanges

of theparameters. This an beappreiated in Fig. 7,

in whih we show the growth rate of the instability,

foraslightlydierentorientationoftherelative

velo-ity u with respet to the magneti eld ('

1

= 75:5 Æ

instead of 89:9 Æ

). It an be notied that this hange

produes afourfold inreaseof the growth rateof the

seondary instability (and a shift towards larger

1 ).

The primary instability is also present now, although

its growth rate is muh smaller than that of the

se-ondarymodes. Thisisdueto thesmall-u

k

destabiliza-tion phenomenon, sine the modied onguration is

stillstableaordingtoIMHD.

0.3

0.35

0.4

0.1

0.2

y

1

/

p

Im (

V

)

/

A

1

Figure7. Eetof ahange ofthe orientationof the

rela-tiveveloitywithrespettothemagnetield. Thegure

represents the growth rate for ase (d), with '

1 = 75:5

Æ

insteadof89:9 Æ

. Notetheprimarymodesenteredaround

1

0:15.

If we hange the relative veloity, keeping the

re-mainingparametersxed(asifthesolarwindhadbeen

upwards, and the primary instability beomes

domi-nantfora30%(orlarger)inreaseofu.

In Fig. 8 we show the stability diagram for ase

(e), where M

A;1

diersfrom ase(d)by14%. The

ef-fetofthis hangeistoinreasethegrowthrateofthe

seondaryinstabilitywithrespettotheresultsofFig.

6. This modied onguration is stable aording to

IMHD.

0.2

0.4

0.6

0.8

1

y

1

/

p

1

2

3

4

u

k

/

A

C

u

i

1

Figure8.Stabilitydiagramfortheparametersoftheevent

of January11, 1997. TheurveCorrespondsto the

on-gurationofase(e).

VIII Conlusions

CompressibilityaetstheKelvin-Helmholtzinstability

byhangingtheunstabledomainsofthemodesalready

present in the inompressible model as well as their

growthrate. Modeswithlargeu

k

(theprojetionofthe

relativeveloityonthewavevetoroftheperturbation)

arestabilized. Atthesametimethelowerboundaryof

theunstableu

k

intervalshiftsdownwards,thus

destabi-lizingperturbationsthatarestableaordingtoIMHD,

foru

k <u

i

(theritialvalueformarginal

inompress-ibleinstability). Ontheotherhand,thegrowthrateof

the unstable modes above u

i

is reduedby

ompress-ibility. Inaddition,ompressibilityintroduesnew

un-stable modes (the seondarymodes)that do notexist

in IMHD, and that our for u

k

below u

i

. For these

reasons,ompressibilitymayplayanimportantrolein

determiningthestabilityofongurationswith a

rela-tiveveloityoftheorderofu

i .

Forongurationsofthefrontofthemagnetopause,

whih havesmall relative veloities, theIMHD model

gives reliable estimates of their stability, and

om-pressibilityeetsdonotintroduesignianthanges.

(13)

attheanksofthemagnetopause. AsshowninSetion

VII,in these asestheourreneoftheseondary

in-stabilityand theshift ofthe boundary ofthe primary

instabilityplayanimportantrole. Then,ongurations

thatarestableifompressibilityisnegletedturnoutto

beunstable whenitis onsidered. Inthese situations,

the stability properties are quite sensitive on the

val-ues oftheparameters. Theestimates basedonIMHD

maybemisleading,andaarefulanalysisisrequiredin

eahase, sineno simpleruleofthumb an begiven.

In these respets the stability diagrams introdued in

SetionVmayproveuseful,sineasinglegraphallows

to visualize the eet of hangingthe magnitude and

orientationoftherelativeveloity.

Aknowledgments

We aknowledge grants from CONICET (PIP

4521/96 and PIP 4535/96), the University of Buenos

Aires (Projets X151 and TX32) and NASA (grant

NAG5-2834).

Referenes

[1℄ M. G. Kivelson and S. H. Chen. The magnetopause:

Surfae waves and instabilities and their possible

dy-nami onsequenes. AGU Monograph 90, Physis of

Magnetopause, editedby P.Song, B.U. O.Sonnerup

and M. F. Thomsen. Geophys. Mon. Ser. 90, AGU,

Washington,DC,257(1995).

[2℄ A. Miura. Kelvin-Helmholtzinstability atthe

magne-topause: omputer simulations. AGU Monograph 90,

PhysisofMagnetopause,editedbyP.Song,B.U.O.

Sonnerup,andM.F.Thomsen.Geophys.Mon.Ser.90,

AGU,Washington,DC,285(1995).

[3℄ C. J. Farrugia, F. T. Gratton, and R. B.

Tor-bert. Visous-type proesses in the solar

wind-magnetosphere interation. Speial volume Challenge

toLong-StandingUnsolvedSpaePhysisProblemsin

the20thCentury.SpaeSieneReview,95(1/2),443

(2001).

[4℄ S. Duhauand J. Gratton. Kelvin-Helmholtz

instabil-ityofanisotropiplasmainamagnetield.J.Plasma

Phys.13,451(1975).

[5℄ F.Gratton, J.Gratton, and A.G.Gonzalez.

Conve-tive instability of internal modes inaelerated

om-pressibleplasmas.PlasmaPhys.ControlledFusion30,

435(1988).

[6℄ W.A.Newomb.Compressibilityeetsoninstability

growthrates.Phys.Fluids 26,3246(1983).

[7℄ M. G. Kivelson and D. J. Southwood. Coupling of

global magnetospheri MHDeigenmodes to eld line

resonanes. J.Geophys.Res.91,4345(1986).

[8℄ M.G.KivelsonandZ.-Y.Pu.TheKelvin-Helmholtz

in-stabilityonthemagnetopause.Planet.Spae. Si.32,

1335 (1984).

[9℄ A.G.GonzalezandJ.Gratton.TheKelvin-Helmholtz

instability in ompressible plasmas, in Pro. VIII

Latin-Amerian Workshop on Plasma Physis. Vol. 1

[10℄ A.MiuraandP.L.PrithettP.Nonloalstability

anal-ysisoftheMHDKelvin-Helmholtzinstabilityina

om-pressibleplasma.J.Geophys.Res.87,7431(1982).

[11℄ C.J.Farrugia,F.T.Gratton,L.Bender,J.M.Quinn,

R. B.Torbert, N. V. Erkaev and H. K. Biernat.

Re-ent work onthe Kelvin-Helmholtz instability at the

daysidemagnetopauseandboundarylayer.PolarCap

BoundaryPhenomena,editedbyJ.Moenetal.,Kluwer

AademiPublishers,Netherlands,1(1998).

[12℄ A.G.GonzalezandJ.Gratton.TheKelvin-Helmholtz

instabilityinaompressibleplasma: theroleofthe

ori-entationofthemagnetieldwithrespettotheow.

J.PlasmaPhys.51,43(1994).

[13℄ A.G.Gonzalez andJ.Gratton. Theroleof adensity

jumpintheKelvin-Helmholtzinstabilityofa

ompress-ibleplasma.J.PlasmaPhys.52,233(1994).

[14℄ A. G. Gonzalez, J. Gratton and M. Gonzalez. The

Kelvin-Helmholtz instability in ompressible plasmas

withmagnetieldshear.Pro.IXLAWPP,AIPConf.

Pro.563,47(2001).

[15℄ S.Chandrasekhar. Hydrodynami andHydromagneti

Stability.OxfordUniv.Press(1961).

[16℄ W. I. Axford. Quart. J. Meh. Appl. Math. 13, 314

(1960).

[17℄ W.I.Axford.Can.J.Phys.40,654(1962).

[18℄ D. J. Southwood. The hydrodynami stability of the

magnetospheriboundary.Planet. SpaeSi. 16, 587

(1968).

[19℄ A. Miura. Dependene of the magnetopause

Kelvin-Helmholtzinstabilityontheorientationofthe

magne-tosheathmagnetield.Geophys.Res.Lett.22,2993

(1995).

[20℄ C. J. Farrugia, F. T. Gratton, L. Bender, H. K.

Biernat, N. V. Erkaev, J. M. Quinn, R. B. Torbert

andV. Dennisenko. Chartsof joint Kelvin-Helmholtz

and Rayleigh-Taylor instabilities at the dayside

mag-netopauseforstronglynorthwardinterplanetary

mag-netield.J.Geophys.Res.103,6703(1998).

[21℄ J. E. Contn, F. T. Gratton and C. J. Farrugia. A

studyof Kelvin-Helmholtz modes for hyperboli

tan-genteldprolesappliedtothedaysidemagnetopause.

Fluidos2001,editedbyJ.C.Ferreri.ISBN

927-20148-0-9(2001).

[22℄ J.E.Contn.Ph.D.Thesis,Inestabilidadesdelaminas

de orriente on ujos de plasma y apliaiones a la

magnetopausadiurna.FaultaddeCieniasExatasy

Naturales,UniversidaddeBuenosAires(1999).

[23℄ N.V.Erkaev.ResultsoftheinvestigationofMHDow

aroundthemagnetosphere.Geomagn.Aeron. 28,455

(1988).

[24℄ C.J.Farrugia,F.T.Gratton,J.Contin,C.C.Cohei,

R. L. Arnoldy, K. W. Ogilvie, R. P. Lepping, G.

N. Zastenker, M. N. Nozdrahev, A. Fedorov, J.-A.

Sauvaud, J. T. Steinberg, and G. Rostoker.

Coordi-nated Wind, Interball/tail, and ground observations

ofKelvin-Helmholtzwavesat thenear-tail,equatorial

magnetopauseatdusk: January11, 1997.J.Geophys.

Imagem

Figure 1. Geometry of the problem.
diagram shown in Fig. 3, in whih we have drawn a
Table 1. Parameters of ongurations of the front of the magnetopause aording to the model.
Table 2. Additional parameters of ongurations of the front of the magnetopause (r
+3

Referências

Documentos relacionados

We investigate topologial defets in models involving one and two or more real salar elds.. In models involving a single eld we examine two dierent sublasses of models,

lating blak hole entropy using string theory as refs.

results hold in the energy sale where hiral symmetry is broken and the Pion wavelength is muh.. bigger than the size of the system ( &gt;&gt;&gt;

data of SD stations are sent to the entral data aqui-. sition system (CDAS) by mean of wireless

program inluded basi mehanisms of superondutivity, vortex dynamis, materials preparation and harateri-.. zation, and appliations

and quantum gauge elds propagating in a lassial.. gravitational bakground, and the

materials the line tension and line energy of ux lines. in general are dierent and depend on the angle

Gravitation presented during the XXII Brazilian National Meeting on Partiles and Fields arei.