Investigating the Dynamics and Density Evolution of Returning Plasma Blobs from the 2011 June 7 Eruption

7.6 MB·

by  Jack Carlyle

DRAFTVERSIONJANUARY21,2014
PreprinttypesetusingLATEXstyleemulateapjv.5/2/11
INVESTIGATINGTHEDYNAMICSANDDENSITYEVOLUTIONOFRETURNINGPLASMABLOBSFROMTHE
2011JUNE7ERUPTION
JACKCARLYLE1,2,DAVIDR.WILLIAMS1,LIDIAVANDRIEL-GESZTELYI1,3,4,DAVINAINNES2,ANDREWHILLIER5,
SARAHMATTHEWS1
DraftversionJanuary21,2014
ABSTRACT
ThisworkexaminesinfallingmatterfollowinganenormousCoronalMassEjection(CME)on2011June7.
The material formed discrete concentrations, or blobs, in the corona and fell back to the surface, appearing
4 as dark clouds against the bright corona. In this work we examined the density and dynamic evolution of
1 these blobs in order to formally assess the intriguing morphology displayed throughout their descent. The
0 blobswerestudiedinfivewavelengths(94,131,171,193and211A˚)usingtheSolarDynamicsObservatory
2 Atmospheric Imaging Assembly (SDO/AIA), comparing background emission to attenuated emission as a
n function of wavelength to calculate column densities across the descent of four separate blobs. We found
a the material to have a column density of hydrogen of approximately 2 × 1019 cm−2, which is comparable
J with typical pre-eruption filament column densities. Repeated splitting of the returning material is seen in a

0 mannerconsistentwiththeRayleigh-Taylorinstability. Furthermore,theobserveddistributionofdensityand
2 its evolution are also a signature of this instability. By approximating the three-dimensional geometry (with
data from STEREO-A), volumetric densities were found to be approximately 2 × 10−14 g cm−3, and this,

] along with observed dominant length-scales of the instability, was used to infer a magnetic field of the order
R
1Gassociatedwiththedescendingblobs.
S
Subjectheadings:Sun:activity—Sun:atmosphere—Sun:coronalmassejections(CMEs)—Sun:filaments,
.
h prominences—Sun: magneticfields—Sun: UVradiation
p
-
o 1. INTRODUCTION flux ropes thought to support prominences may become un-
r
stable, causingthematerialto riseand, ultimately, erupt, of-
t Eruptions occurring on the surface of the Sun cast huge
s tenresultinginwhatisknownasaCME.Notallofthepromi-
a amounts of matter and magnetic field out into the solar at-
nencematerial,however,willnecessarilyescapetheSun,with
[ mosphereandinterplanetaryspace. Thesebundlesofplasma
the magnetic field reconfiguring beneath the eruption and a
with frozen-in magnetic fields can interact with the Earth’s
1 newprominenceoftenforming(perhapsaportionofflux-rope
magnetosphereandhavethepotentialtointerferewithmany
v remains).
differentaspectsofmoderntechnologyonlargescales;forex-
4 On 2011 June 7, an immense CME took place; the active
ample,thegeomagneticstormwhichoccurredinMarch1989
2 region precursor filament near the west solar limb rose and
caused the collapse of Hydro-Que´bec’s electricity transmis-
8 erupted,hurlinganenormousamountofmaterialintotheso-
sionsystem,leaving6millionpeoplewithoutelectricityfor9
4 laratmosphere. Whilethefilamentitselfdidnotappearpar-
hoursandcostingtheCanadianeconomy$2billion. Forsuch
1. reasons, ‘space weather’ has become a focus of government ticularlyunusualpriortoeruption,thesheervolumeofejected
0 interestinrecentyears,andabetterfundamentalunderstand- materialseenreturningtothesolarsurfaceisunparalleledin
4 ingoftheprocessesinvolvedcouldleadtotheabilitytobet- modern observation. The huge lateral expansion of the fil-
1 ament during the eruption is extremely eye-catching – the
ter predict events and protect our technologically-dependent
: vastareaoftheejectaappearsatleastanorderofmagnitude
v worldfromthisharsh,unforgivingenvironment.
larger than the initial foot-point separation, and suggests the
i These phenomena are closely coupled with the eruption
X filament carried a very large amount of mass. The material,
of prominences – condensations of chromospheric material
whichemergedasasinglelargecloud,appearedtorepeatedly
r suspended by magnetic forces in the corona. Since chromo-
a undergotheRayleigh-Taylor(RT)instabilityabovethesolar
sphericmaterialisrelativelycoolcomparedtothelow-density
limb (Innes et al. 2012), highlighting a difference in density
corona, thesestructuresappeardarkwhenviewedontheso-
betweentheejectaandthesurroundingcorona. Asthiscloud
lar disc (colloquially referred to as filaments), but are seen
expanded, some of the material from the flanks of the CME
inemissioninvisiblelightwhenprotrudingoff-limb(promi-
appearedtostopmovingoutwardandstartedfallingbackto-
nences)(seeChen2011andreferencesthereinforadetailed
ward the solar surface, fragmenting into discrete condensa-
description of these structures). The magnetic arcades and
tionsofmatter: “blobs”. Thisin-fallingmaterialpassesback
overthesolardisc,appearinginabsorptionintheExtremeUl-
[email protected]
1MullardSpaceScienceLaboratory,UniversityCollegeLondon,Holm- traviolet(EUV)wavelengths,whichindicatesafinitedensity
buryStMary,Surrey,RH56NT,UnitedKingdom of lower-temperature material (neutral or singly-charged hy-
2Max-PlanckInstitutfu¨rSonnensystemforschung,37191Katlenburg- drogenorhelium),comparedwiththesurroundingandback-
Lindau,Germany ground atmosphere. A snapshot of this fallback is shown in
3LESIA-ObservatoiredeParis, CNRS,UPMCUniv. Paris06, Univ. Figure1,withoneblobhighlighted.
Paris-Diderot,92195Meudon,France
4KonkolyObservatory,Budapest,Hungary Whilst blobs such as these have not been observed in the
5KwasanandHidaObservatories,KyotoUniversity,Kyoto,Japan solaratmosphereuntilnow(perhapsduetothelackofhigh-
2 Carlyleetal.
FIG.3.—SimulationsoftheRayleigh-Taylorinstabilitybetweenmaterials
ofdifferentdensityintheabsence(left)andpresence(right)ofamagnetic
field(paralleltotheinitiallyhorizontalinterfaceandintheplaneoftheim-
age). A material of higher density lies above a material of lower density
andisacceleratedbygravity,demonstratinghowamagneticfieldmodifies
themorphologyoftheRayleigh-Taylorinstability. GeneratedbytheAthena
MHDCode.
FIG.1.—Imageofin-fallingmaterialasitcrossesontothesolardiscat
07:05UT-themorphologyinthehighlightedsquaremaybecomparedwith theAthenamagnetohydrodynamic(MHD)code(Stoneetal.
thatofthematerialindicatedinFigure2. Highlightedblobisshownfrom 2008). Thisdemonstrateshowtheundularmodeofthemag-
STEREOinFigure4andisanalysedinFigure9.
neticRayleigh-Taylorinstabilitysuppressestheformationof
small-scalestructurealongthemagneticfield.
A quantitative assessment of this instability may be made
byconsideringitsgrowthintheincompressiblelimit.Assum-
ingthatthemagneticfieldispurelyinthex-direction(wherex
isparalleltotheinterfacebetweenthetwolayers),thegrowth
rateforsuchaperturbationisgivenas(Chandrasekhar1961):
k2cos2θB2
γ2=gkA− x (1)
2π(ρ +ρ)
u l
wheregistheaccelerationduetogravityandisperpendicular
tox,kisthewavenumber,AistheAtwoodnumber(givenas
(ρ −ρ)/(ρ +ρ)),θ istheanglebetweenk andB ,andρ
u l u l x
isdensity(usignifiestheupperlayerandlsignifiesthelower
layer). Takingthederivativewithrespecttokgives:
∂γ 2kcos2θB2
2γ =gA− x (2)
∂k 2π(ρ +ρ)
u l
Themostunstablegrowthratewillbeatthepeakofthedistri-
FIG.2.—TheRayleigh-TaylorinstabilityasobservedintheCrabNebula. butionwhere∂γ/∂k=0,thereforetheinstabilityisdescribed
Credit:NASA,ESAandAllisonLoll/JeffHester(ArizonaStateUniversity).
by:
Acknowledgement:DavideDeMartin(ESA/Hubble)
2kcos2θB2
gA= x (3)
cadenceobservationsratherthanalackofsimilareruptions), 2π(ρ +ρ)
u l
theseobserveddynamicsarenotentirelyunique. Itisaston-
ishinghowmorphologicallysimilartheCrabNebulaappears whichrearrangestogive:
inimagescapturedbytheHubbleSpaceTelescope,ascanbe
2π 2cos2θB2
seeninworkbyHesteretal.(1996), whichincludesarigor- =λ = x (4)
u
ous study of these dynamics. This work concludes that the ku g(ρu−ρl)
observedstructuresinthisexpandingsupernovaremnantare
whereλ isthedominantgrowthscaleoftheinstability. Itis
likely formed as a result of the RT instability, an image of u
thentrivialtorearrangethisequationtocalculateB froman
whichispresentedinFigure2.Assuch,itwouldbeintriguing x
observedgrowthscale:
toinvestigatewhethertheobservedblobsmovingthroughthe
lowersolaratmosphereundergothisinstabilityasthehigher (cid:114)gλ(ρ −ρ)
u l
materialdoes(Innesetal.2012). Bx= 2cos2θ (5)
In a nonmagnetic RT instability, the growth rate increases
with decreasing scales, resulting in tightly bunched, long, These equations have been successfully applied to the Crab
thin fingers of high-density material progressing into low- NebulabyHesteretal.(1996)andtoprominenceplumeson
density material, and vice versa. However, in the presence theSun(Ryutovaetal.2010;Bergeretal.2010).
ofamagneticfieldthereisacriticalwavelengthbelowwhich However, prominence mass (and therefore density, both
magnetictensionsuppressestheinstability,resultinginmuch before and after eruption) is not a trivial quantity to de-
wider fingers (with respect to their length) which also ap- termine, as, historically, techniques have required spectro-
pear more cohesive. Figure 3 shows simulations of the RT scopicobservationsinopticallythicklinesandradiativetrans-
instabilityunderdifferentmagneticconstraints,generatedby fer modelling of these lines, leading to order-of magnitude
DensityEvolutionofReturningFilamentBlobs 3
estimates (see Labrosse et al. 2010 and references therein
for a more detailed history). Using EUV imaging observa-
tions, Gilbert et al. (2005) applied temporal– and spatial–
interpolative approaches to determine the column density of
erupting and quiescent prominences, respectively, using ab-
sorptionintheFeXII(195)spectralband,andcalculatedthe
massofaneruptingprominencefrom1999July12tobeap-
proximately6×1014 g.? Gibertetal.(2010)thenexpanded
thistechniquetoconducttheanalysisinthreedifferentwave-
length regimes, covering three different species’ photoioni-
sation continua. They concluded that the total prominence
mass estimate is lower for the longer wavelengths analysed,
attributedtothehigheropacityinhigherwavelengthscausing
asaturationofcontinuumabsorptionintheselinesandthusa
potentially large underestimation of the mass. This suggests
FIG.4.—Theviewofthein-fallingmaterialfromSTEREO-Aat07:05UT.
that such column density diagnostics are best conducted at Thewhitelinesmarktheline-of-sightofthetopandbottomoftheboxshown
shorterwavelengths,whereH0,He0andHe+areallionised. inFigure1. Sincethereareonlytwoviewsofthematerial,precisevolume
calculationscannotbeperformed;howeverthegeometryofthematerialfrom
Analysingthefallingprominenceblobsinthe2011June7
twoperspectivescanaidestimatesoftheshapesandvolumesoftheblobs.
filamenteruption,Gilbertetal.(2013)findthatthebrighten-
steps were chosen to be at roughly equal intervals, although
ings observed when the blobs impacted the solar surface are
the requirement for ‘clean’ background images also guided
likely due to compression of the plasma rather than recon-
thischoice.Fourblobswerestudiedintotal,withtheblobex-
figurationofthelocalmagnetictopology(i.e.,reconnection).
hibitingthemostobviousapparentRTinstabilitybeingstud-
Landietal.(2013)havealsoinvestigatedtheemissionandab-
iedatthehighestcadence.
sorptioninthiseruptedmaterialandfindthatthetemperature
is likely to be 33,100 ± 2,200 K with an electron density of
3. METHOD
3.6+1.1 × 1019cm−2.
−0.7 In this study, we use the polychromatic opacity imaging
Inthiswork,weanalysethedensityoftheseblobsandex-
method developed by Williams et al. (2013) to obtain esti-
amine how this evolves as they fall through the solar atmo-
matesofthecolumndensityoftheeruptedneutralhydrogen.?
sphere. We also examine the dynamics and distribution of
This technique works by measuring the absorption depth of
massintheblobsinordertodeterminewhetherthematerial
the cooler material in five co-temporal AIA images, each at
undergoes the magnetic Rayleigh-Taylor instability. Finally,
a different wavelength, using intensity measurements of the
weuseEquation5toinferamagneticfieldstrengthassociated
target image compared to a background image (a co-spatial
withtheblobsundertheinstability.
image some minutes beforehand). The measured absorption
depths,afunctionofwavelength,canthenbefittedtoafunc-
2. OBSERVATIONS
tionofonlytwovariables,oneofwhichiscolumndensity.
This work uses images collected by the Solar Dynamics Optical depth is related to density and intensity, respec-
Observatory’s Atmospheric Imaging Assembly (SDO/AIA: tively,usingthefollowingequations:
Lemenetal.2012)between06:40and08:40UTon2011June
7 in the 94, 131, 171, 193 and 211 A˚ passbands. The erup- τ(λ)=Nσ(λ) (6)
tion occurred from NOAA active region 11226, which was where τ is optical depth (a function of wavelength, λ), N
locatedinthevicinityofthesouth-westlimb,andmostofthe is column density (particles per unit area) and σ is cross-
in-fallingmaterialpassedoverthisquadrantuponreturningto sectional area (also a function of wavelength); then, the ob-
theSun(seetheonlinecontentforamovieoftheeruptionand servedintensity,I,isgivenby
fall-back).
Each image in each wavelength was deconvolved using a I=I0e−τ (7)
Richardson-Lucy (RL) algorithm and wavelength-dependent
whereI istheintensitythatwouldbereceivedintheabsence
point-spread functions as described by Grigis et al. (2012). 0
ofanyobscuringmaterial.Equation7maybere-writtenusing
Thisistominimisetheeffectofdiffractionandstraylightin
observedintensityinthepresenceoftheblob, I (givenby
theimagesarisingfromtheentrancefilterandthefocalplane obs
the blob image), background intensity, I (the intensity from
filteronSDO/AIA. b
material obscured by the blob), foreground intensity, I (the
Also used were images collected by STEREO-A, which f
emitted intensity from all material between the blob and the
givesadifferentperspectiveontheeruptionandamorecom-
observer)andtheunattenuatedintensity,I (givenbytheun-
plete idea of the blobs’ geometry. A snapshot of the post- 0
occultedimage,whichisapproximatelythesumoftheback-
eruptionfallbackisshowninFigure1asseenfromSDO,and
groundandforegroundintensity)and f,thepixel-fillingfac-
fromSTEREO-AinFigure4. Notethatthedescentofthese
tor:
blobshasalsobeenanalysedbyGilbertetal.(2013). I =I [fe−τ+(1−f)]+I (8)
Thetargetsofourstudywerechosenbasedonalong, un- obs b f
obscureddescentforeachblob,inordertomaximisetheevo- Rearranging,weareleftwiththefollowingequation:
lution timescale observed. The images used were taken at
I I
points in the descent where the blob appeared to lie above a 1− obs = f b(1−e−τ)) (9)
relatively ‘quiet’ region of the solar surface, in order to en- I0 I0
surethattheimageusedasthebackground(co-spatialtothe or
blobimage)wasasuniformaspossibleovertime. Thetime d(λ)=G(1−e−τ(N,λ)) (10)
4 Carlyleetal.
FIG.6.—171A˚ imagesofablob(at07:06,left)anditsassociatedback-
groundimage(at07:03,right).Thisblobisseentomovefromthesouth-west
oftheimagetothenorth-east.
to-follow descent through the solar atmosphere. The large
separation suggests that these successive branches separate
parallel to the plane of the sky (or close to); statistically, no
orientationofmagneticfield(paralleltotheinterfaceandper-
pendicular to the acceleration) should be favoured over an-
other,andweexpectscalesinallblobstobeofthesameorder
FIG.5.—ThisfiguregivesagraphicalrepresentationofEquation10with ofmagnitude(sincethephysicalparametersshouldallreflect
thefreeparameters(NH andG)constrainedsuchthattheplotbestliesover those of the progenitor cloud). It is therefore a trigonomet-
measuredabsorptiondepthasafunctionofwavelength(representedbythe ricargumentwhichstatesthatthelargestobservedprojected
circles, calculated using the intensity ratio between the target and back-
separations are likely to be (almost) in the plane of the sky
ground).
(Hesteretal.1996). Thedepthofthisblobwasestimatedto
be of the same order of magnitude as the diameter using the
wheretheleft-handsideofEquation9istheabsorptiondepth,
STEREO-A data (shown in Figure 4) and the column den-
d(λ)=1−(I /I ),andthegeometricdepthG= f(I /I ).
obs 0 b 0 sity was divided by this value to obtain a volumetric density
The left-hand side of Equation 10 is observable, obtained
estimateinordertoformallyassesstheinstabilityusingEqua-
bycomparingtheintensityinthebackgroundimagewiththe
tion4.
imageoftheblob;theright-handsideisamodeltowhichthe
calculated opacities from the five wavelengths may be fitted
4. RESULTS
withonlytwovariables,GandN (seeFigure5). Thevalue
H
ofGisexpectedtobehighacrossthewholeblob,decreasing Figure 6 shows two images (in the 171 A˚ passband) used
suddenly as the edge is reached as the pixel filling factor, f, to determine column density – the blob image and the back-
tends to zero in the absence of absorbing material Williams ground image (co-spatial, three minutes prior). Figure 7
et al. (2013), and a particular value of G may be used as a shows the same blob image with contours of G = 0.5 over-
definitionoftheedgeoftheblobs. Thefittingcanbedoneus- laid. G is comprised of pixel filling factor, f, and fractional
ing a Levenberg-Marquardt least-squares minimisation algo- backgroundemission,Ib/I0,bothofwhichareexpectedtobe
rithm,whichreturnsaχ2 valuewhichdescribestheaccuracy closeto1inthepixelscontainingcool, densematterhighin
ν thecorona.Thisvaluecanbedescribedasthefractionoflight
offitofthedatatothemodel.Aperfectfitwouldbeindicated
per pixel that interacts with the blob material, and it can be
byavalueof1,withgoodness-of-fitdecreasingasχ2 moves
ν saidthatwhenthisvalueisgreaterthan0.5,thepixelisdomi-
furtherfromunityineitherdirection.
natedbythismaterial. Thereforeavalueof0.5waschosento
The blobs studied were of various sizes, and maps of the
definetheedgeoftheblobs,andinthefiguresshowingGand
column density were calculated at a number of points in
N maps,greypixelsareallzoneswithaGoflessthan0.5(or
timeduringthedescentofeach(generally5time-stepswere H
whereχ2 isgreaterthan10orlessthan0.1). Figure8shows
used, though 15 were analysed for one particularly interest- ν
thecalculatedcolumndensityandGvaluesforthesameblob.
ing blob). To examine how these structures evolved as they
In order to highlight the time evolution of N and G, the
fellthroughthecorona,andhowwelltheycorrespondtothe H
columndensitymapsoftheblobhighlightedinFigure1be-
behaviour of the magnetic Rayleigh-Taylor (mRT) instabil-
tween07:03and07:50UT(whenthedescentisseenagainst
ity,thecalculatedvaluesofcolumndensityineachblobwere
thesolardisc)arepresentedinFigure9. Furthersnapshotsof
monitored throughout their descent (i.e. if a fluid is acceler-
theblobinFigures6and8areshowninFigure11. Notethat
atedbygravityandencountersaninterface,asisthecasewith
columndensity,N ,ispresentedonalogarithmicscale(i.e.,
themRTinstability,enhanceddensityisexpectedtowardsthis H
aslog (N )).
interface as material “piles-up”). Blobs were also compared 10 H
The figures in this work do not include any pixels where
with one another in order to ascertain whether they possess
similar densities, which may indicate whether the physical χν2 < 0.1 or χν2 > 10, an order of magnitude within a per-
conditions for the blob formation are due to similar density fectfit. Thismaskingremovespixelswithlargemeasurement
and/ormassvalues. errors in absorption depth, many of these being outside the
One blob was chosen based on morphology: it demon- visualedgeoftheblobwhichstillappearedtoshowG>0.5.
strated repeated instances of the suspected instability, with
5. ANALYSIS
largeseparationbetweenconsecutivefingersandalong,easy-
DensityEvolutionofReturningFilamentBlobs 5
FIG.7.—171A˚ imagesofblobshowninFigure6withcontoursofG=0.5
overplotted.
ThecontoursdisplayedinFigure7demonstratetheagree-
ment between the visual edge of the blob and the boundary
where G = 0.5. However, it can be seen both from the con-
tours and from comparison with Figure 8 that there is a dis-
crepancytowardsthetail-endoftheblob;whilstthereappears
to be darkmaterial at approximately (948(cid:48)(cid:48),-203(cid:48)(cid:48)), G here is
below0.5,andassuchisnotdefinedaspartoftheblob. This
is due to the fact that the background image contains dark
material in this location, which means there is little differ-
ence in intensity between this image and the blob image at
this point; in other words, the image selected to act as the
unobscured background may not be a true reflection of the
backgroundradiationfieldatthispoint. Thisisthemostsuit-
able co-spatial background image available, but though the
observedblobshavewell-definedleadingedges(i.e.,towards
thedirectionoftravel),mostofthemhavelong,diffusetails,
often appearing to be still connected to the original erupted
cloud. Therefore,thefrontedgewasattributedgreaterimpor-
tanceinbackground-imageselection.
Hydrogencolumndensitieswerecalculatedtobeapproxi-
mately2×1019 cm−2 fortheentiredescentofallblobs. All
FIG.8.—Columndensity(top)andG(bottom)mapsfortheblobshowin
blobs appear to decrease in size as time progresses – i.e., as Figures6and171cont.Directionoftravelis∼40◦tothenegativeX-axis.
theyfall–butascanbeseenfromFigures9and11,column
densityvaluesremainreasonablyconstant. However,thege-
ometricdepthGappearstograduallydecreaseacrossthede- is possible that our calculations may include multiple blobs
scentfrom∼0.95to∼0.8. Althoughthisreductionmayalso lyingalongthelineofsight–however,theconsistencyofNH
beduetothepixelfillingfactorfalling,themorelikelycause valuesderivedfromAIAimagessuggeststhatthisisnotthe
is that a greater proportion of the emission measured origi- case,andthevolumetricdensityofhydrogenforallblobshas
natesintheforegroundastheheightoftheblobsdecreases. beenestimatedtobeoftheorder1010cm−3.
Based on an average value of N = 2 × 1019 cm−2, for a The blob used to investigate whether the mRT instability
H
blobofdimensions10”x20”(theaveragesizeoftheblobsto- is at work is presented in Figure 9, with the images taken at
wardsthestartofmeasurement),theinitialmassofeachblob 07:11and07:25UTusedtomeasuretheseparationscalebe-
isapproximately3×1013 g. BycomparingFigures1and4, tween the fingers of material that coalesce into blobs. We
it would seem that the material is elongated roughly in the estimate this scale to be λ ≈ 1 × 109 cm. Figure 4 in-
radialdirection,withsimilargeometriesappearintheprojec- dicates that the width of the blobs appears similar in dif-
tions of the blobs seen by both SDO and STEREO. Unfor- ferent projections (STEREO-A, SDO), suggesting that the
tunately, with observations from only two directions, we are assumption of cylindrical symmetry is a good approxima-
unable to resolve the volumes and positions precisely, and it tion. We therefore estimate a volumetric mass density in
6 Carlyleetal.
FIG.9.—EvolutionofcolumndensityoftheblobhighlightedinFigure1asitpassesacrossthesolardisc.Thefilamentarystructureindicativeofthemagnetic
Rayleigh-Taylorinstabilitycanbeclearlyseen.Directionoftravelistothenorth-east.
this blob to be ρ = ∼2 × 10−14 g cm−3. Using these val- atcertainpoints, mostnoticeablyjustbefore‘forking’, split-
u
ues of λ and ρ in Equation 5, with a coronal density of tingintotwobrancheswhicheventuallybreakintonewblobs
∼10−16 g cm−3 (although the result is not sensitive to this intheirownright;thewholeprocessthenrepeats. Thesedy-
value as it is orders of magnitude smaller than the blob den- namics are similar to the mRT instability in terms of mass
sity)andg=2.74×104cms−2,andassumingthatcosθ =1, distribution and flow, and indicates that the two regimes of
a magnetic field strength of the order 1 G is expected to be differentdensity(i.e., thecoolblobandsurroundingcorona)
associatedwiththeinstability. are being accelerated against one another – a morphological
Figure 9 also demonstrates the change in density distribu- similaritytothemRTinstability.
tionofthisblobasitfallsthroughthesolaratmosphere. The
massappearstocollect,or‘pileup’towardstheleadingedge
6. DISCUSSION
DensityEvolutionofReturningFilamentBlobs 7
07:03:12 07:04:48 07:07:12 07:09:12

-290

c) -300
e
s
Y (arc - 310

-320
900 910 920 890 900 910 920 880 890 900 910 870 880 890 900
X (arcsec) X (arcsec) X (arcsec) X (arcsec)
FIG.10.—Evolutionofblobmorphology:171A˚ imagesofthefirst4imagesinFigure9.Noticetheback-endoftheblobdoesnotappearinthefirstNHmap;
thisisduetodarkmateriallyinginthebackgroundwhichleadstolittleornodifferenceinintensitybetweentheblob/backgroundimages.
FIG.11.—EvolutionofthedensityofblobpresentedinFigures6and8.Theblobappearstoreduceinsizebutretainsahighcolumndensitythroughoutthe
descent.Directionoftravelistothenorth-east.
FIG.12.—Evolutionofthedensityofathirdblob(colourbarshowninFigure11applies).Thepilingupofthematerialtowardsthefrontoftheblobcanbe
seeninthefirstimageandthefilamentarystructurecanbeseenformingintheremainingimages.Directionoftravelistothenorth-east.
FIG.13.—Evolutionofthedensityofafourthblob(colourbarshowninFigure11applies).ThemRTinstabilityisnotobservedasclearlyhere,andthisblob
appearstohavealowercolumndensityingeneral.Directionoftravelistothenorth-east-east.
8 Carlyleetal.
Thevaluesobtainedforthehydrogencolumndensityofthe modesacrossthisdirection;wemightthereforeexpecttosee
blobsanalysedinthisworkarecomparabletothosefoundby the front of the blobs decreasing in column density as the
Gilbert et al. (2005), who calculated the column density of nonmagnetic RT instability occurs in a direction which we
aprominencetobe1.6±1.0×1019 cm−2 beforeeruption– are unable to observe. We notice this, for example in the
evenafterthematerialinthisstudywasseentoexpandgreatly 07:32imageofFigure13,butthisisnotgenerallyseenacross
intheinitialeruption. the blobs. Therefore, a suppression of the smallest modes
In work carried out on the dynamics of this erupted mate- in all horizontal directions is necessary. This is where shear
rialbyInnesetal.(2012),theinstabilityoccurringintheblobs between the magnetic fields (as found in the simulations of
studiedinthisworkwasunidentified. HoweverthemRTwas Stoneetal.2007andHillieretal.2012)canbecomeimpor-
seentobeatworkinmaterialfromthesameeruptionhigher tant. The presence of magnetic shear means that there is no
in the corona, which was corroborated by a realistic Alfve´n direction that is perpendicular to both the magnetic field in
velocityinferredfromtheseparationoftheforks. Ourfinding the blob and in the corona (whilst also perpendicular to the
thatthedensityevolutionisself-similarinalltheblobsstud- interface), increasing the size of the smallest scales that can
ied strongly implies the occurrence of an instability, and the be formed. If we assume that the result B = 1 G gives the
morphological similarity between these blobs and the work strengthoftheshearedcomponentofthemagneticfield,then
carriedoutbyInnesetal.(2012)pointstowardsthemRTin- this will be a lower limit for the coronal/blob field strength.
stability. Itshouldalsobenotedthattheestimateforλ usedtocalcu-
It is important to note that our results are in fact a lower late |B| is also a lower limit, which would also result in an
limitoncolumndensity,sincetheydependonbound-freeab- underestimateofthefieldstrength.
sorption by hydrogen or helium (Williams et al. 2013). We
alsonotethatasthematerialfalls,theblobsappeartoshrink, 7. CONCLUSION
suggesting a depletion of bound electrons (ionisation), per- Afteranexceptionallylargeamountofmaterialwasthrown
haps by heating. Nonetheless, the fact that the column den- into the solar atmosphere by a filament eruption on the
sities of the blobs remain reasonably constant in the interior 2011June7, discreteblobsofplasmawereseentofallback
is an intriguing result, and suggests that the cool material is to the solar surface. We have calculated the column density
somewhatisolatedfromthesurroundingenvironment. Given of the in-falling blobs to be approximately 2 × 1019 cm−2,
the relatively high density of cool material, thermal conduc-
which is comparable with pre-eruption column densities of
tionintotheinteriorislikelytobeinefficient.
filaments.
Thismethodonlyconsidersdifferencesinintensityarising
We have studied the evolution of the blobs as they fall
from absorption against a model background radiation field.
through the solar atmosphere and find morphological and
However, the possibility that emission could be occurring in
dynamic evidence that they are formed by the magnetic
the blobs is a source of uncertainty in the results, and this
Rayleigh-Taylorinstability. Theshapesthattheplasmatakes
couldbeafactorinfluencingthegoodnessofthefit.However,
aresimilartothoseseeninsimulationsofthisinstability,and
itisunlikelytobeaprofoundeffect,andasthemajorityofthe
thedynamicsofthedensitydistributionwithintheblobsfur-
pixels have low χν2 values (below 10), the drop in intensity ther support this. We therefore conclude that the returning
causedbythepassingmaterialseemsingoodagreementwith blobs from this enormous CME appear to undergo the mag-
photo-absorptionbylow-temperaturematerial. neticRayleigh-Taylorinstability.
The morphology and dynamics of the blobs are not only By using the point of view given by STEREO-A, we can
self-similar, but are more specifically indicative of the mRT reasonably approximate the geometry of the blobs as cylin-
instability. When we assume this is the case, the calculated drical, which leads to an approximate volumetric density of
densityvaluesandlengthscalesobservedintheinvestigation 2 ×10−14 gcm−3. Bymeasuringa separationof109 cmin
leadustoinferamagneticfieldstrengthoforder1Gassoci-
the forking material, we then use Equation 5 to infer a mag-
ated with the material. This seems to be reasonable: values
neticfieldstrengthassociatedwiththemRTinstabilityofthe
of 0.4 – 1.3 G at a height of 1.6 – 2.1 R were found by
(cid:12) order1G.
Cho et al. (2007), and active region prominences have been
showntohavemagneticfieldsupto100G,whichcouldfea-
siblybecome1Gwithintheseblobsfollowingthehugelateral J.C.thanksUCLandMPIforanImpactPhDStudentship.
expansionthatthismaterialundergoesduringeruption. How- The research leading to these results has received funding
ever,Equation4onlyconsidersstructuringformedalongthe from the European Commission’s Seventh Framework Pro-
magneticfield(constrainedbytensionalongthefieldlines). gramme under the grant agreement No. 284461 (eHEROES
Further to this, should the magnetic field be purely in one project).L.v.D.G.’sworkwassupportedbytheSTFCConsol-
direction,wewouldexpecttheretobenosuppressionofsmall idatedGrantST/H00260X/1andtheHungarianresearchgrant
OTKAK-081421.A.H.issupportedbyKAKENHIGrant-in-
AidforYoungScientists(B)25800108.
REFERENCES
Berger,T.E.,Slater,G.,Hurlburt,N.,Shine,R.,Tarbell,T.,Title,A.,Lites, Gilbert,H.R.,Inglis,A.R.,Mays,M.L.,etal.2013,ApJ,776,L12
B.W.,Okamoto,T.J.,Ichimoto,K.,Katsukawa,Y.,Magara,T., Grigis,P.,Su,Y.,&Weber,M.2012,AIAPSFCharacterizationandImage
Suematsu,Y.,&Shimizu,T.2010,ApJ,716,1288 Deconvolution,TechRep.,SDO/AIATeam,NASA,LMSAL,SAO
Chandrasekhar,S.1961,HydrodynamicandHydromagneticStability(New http://hesperia.gsfc.nasa.gov/ssw/sdo/aia/idl/psf/DOC/psfreport.pdf
York:Dover) Hester,J.J.,Stone,J.M.,Scowen,P.A.,etal.1996,ApJ,456,225
Chen,P.F.2011,LivingReviewsinSolarPhysics,8,1 Hillier,A.,Berger,T.,Isobe,H.,&Shibata,K.2012,ApJ,746,120
Cho,K.S.,Lee,J.,Gary,D.E.,Moon,Y.J.,&Park,Y.D.2007,ApJ,665, Innes,D.E.,Cameron,R.H.,Fletcher,L.,Inhester,B.,&Solanki,S.K.
799 2012,A&A,540,L10
Gilbert,H.,Holzer,T.E.,&MacQueen,R.M.2005,ApJ,618,524 Labrosse,N.,Heinzel,P.,Vial,J.C.,etal.2010,SpaceScienceReviews,
Gilbert,H.,Kilper,G.,Alexander,D.,&Kucera,T.2010,ApJ,727,25 151,243
DensityEvolutionofReturningFilamentBlobs 9
Landi,E.,&Reale,F.2013,ApJ,772,71 Stone,J.M.,&Gardiner,T.2007,ApJ,671,1726
Lemen,J.R.,Title,A.M.,Akin,D.J.,etal.2012,SolPhys,275,17 Stone,J.M.,Gardiner,T.A.,Teuben,P.,Hawley,J.F.,&Simon,J.B.,2008,
Ryutova,M.,Berger,T.,Frank,Z.,Tarbell,T.,&Title,A.2010,SolPhys, ApJSS,178,137
267,75 Williams,D.R.,Baker,D.,&vanDriel-Gesztelyi,L.2013,ApJ,764,165