Fine structure and long duration of a flare coronal X-ray source with RHESSI and SDO/AIA data

1.1 MB·

by  Sylwester Kolomanski

Astronomy&Astrophysicsmanuscriptno.kolomanski-mrozek-chmielewska (cid:13)cESO2017
February1,2017
Fine structure and long duration of a flare coronal X-ray source
with RHESSI and SDO/AIA data
S.Kołoman´ski1,T.Mrozek1,2,andE.Chmielewska1
1 AstronomicalInstituteofUniversityofWrocław,Poland
e-mail:[email protected]
e-mail:[email protected]
2 SolarPhysicsDivision,SpaceResearchCentre,PolishAcademyofSciences,Poland
7 e-mail:[email protected]
1
0 Received.../accepted...
2

n ABSTRACT
a
J Context.Coronal X-raysources(CXSs)arephenomenon veryoftenoccurringinsolarflaresregardlessofaflaresize,durationor
1 power. Thenature of the sources was difficult touncover for many years. It seems that at last, combining datafrom RHESSI and
3 SDO/AIA,thereisaunprecedentedpossibilityto’lookinside’CXSsandtoanswerthequestionsabouttheirformation,evolutionand
structure.
] Aims.Wepresent astudy of a CXS of the SOL2011-10-22T11:10 long-duration flareobserved simultaneously with RHESSI and
R SDO/AIA.Wefocusourattentiononthefollowingquestions:WhatwasresponsiblefortheCXSpresenceandlongduration?Was
S thereanyfinestructureintheCXS?
Methods.TheAIAinstrumentdelivershighqualityimagesinvariousEUVfilters.RHESSIdatacanbeusedtoreconstructimagesin
.
h X-raysandtoperformimagingspectroscopy.SuchacomplementarydataenablestostudyarelationbetweentheCXSandstructures
p observedinEUVduringthedecayphaseoftheflare.
- Results.X-ray emission recorded by RHESSI during the decay phase of the flarecame from about 10 MK hot CXS. The source
o
wasobservablefor5hours.Thislongpresenceofthesourcecouldbesupportedbymagneticreconnectionongoingduringthedecay
r
t phase. Supra-arcade downflows, whichareconsidered tobeamanifestation of magnetic reconnection, wereobserved at thesame
s timeastheCXS.Thesourcewasco-spatialwiththepartofthehotsupra-arcaderegionthathadthehighestemissionmeasureand
a
simultaneouslythetemperaturewithintherangeofRHESSIthermal-response.However,whilethesupra-arcaderegionwasadynamic
[
regionconsistingofsmall-scalestructures,theCXSseemedtobesmooth,structureless.Werunsimulationsusingrealandsynthetic

1 RHESSIdata,butwedidnotfindanystrongevidencethattheCXShadanysmall-scalestructure.
v
Keywords. Sun:corona–Sun:flares–Sun:X-rays,gammarays–Sun:UVradiation
7
2
1
91. Introduction (≈ 1MK)post-flareloopsandtheacronymCXRs willbeused
0 throughoutthe paper in this particular sense. It is worth to no-
.HighenergyradiationfromtheSunishighlynon-uniform,con- ticethataterm“post-flareloops”isconsideredbysomeauthors
1
fined to areas where hot plasma or non-thermal particles are
0 asmisnomer(Priest&Forbes2000;Švestka2007).Infactthose
present.Duringsolarflaresthepatchesofemission becomefar
7 loops are visible when a flare is sill ongoing. Thus, instead of
more localized and fewer. Energy contained in magnetic fields
1 “post-flare loops” we will use a term “post-reconnection flare
isinthecourseofaflareconverted(viamagneticreconnection)
: loops”,PFLs.
vine.g.thermalandnon-thermalenergyofplasmawhichcanbe
CXSs are phenomenon very often occurring in solar flares
i
Xradiated,interalia,inextremeultravioletandX-rays.Thesepro- regardlessofaflare size,durationorpower.Theywerediscov-
cessesareconfinedtorelativelysmallvolumesbuttheyarevery
r ered in 1970sin observationstaken from the Skylab space sta-
aeffectiveandpowerful.
tion(Kahler1977).SincethattimetheknowledgeofCXSshas
As the result, X-ray radiation of the Sun is dominated by
increasedsignificantlyduetothenextgenerationofsolarspace
smallbrightcentersoftwotypes:foot-pointsourcesandcoronal instruments.InvestigationofCXRscarriedoutbymanyauthors
sources.Thefirsttypeisplacedinthelowestpartsofmagnetic
led to many important conclusions (e.g. Vorpahletal. 1977;
archesi.e.atthechromosphereandthetransitionlevel.Thesec-
Actonetal. 1992; Doscheketal. 1995; Feldmanetal. 1995;
ondtypeislocatedhighinthesolar corona.Based onobserva- Doschek&Feldman 1996; Jakimiecetal. 1998; Whiteetal.
tionscoronalX-raysources(CXSs) canbedividedintoseveral 2002;Jiangetal.2006;Kołoman´skietal.2011):
classes due to their emission properties (thermal, non-thermal,
mixed),locationinaflarestructureoraccelerationmechanisms. – CXSs are filled with hot (≥ 10 MK) and relatively dense
FordetaileddiscussionofdifferentclassesofCXSsobservedin plasma(1010−1011cm−3)
hardX-raysseeKruckeretal.(2008).However,asitispointed – their emission is dominated by thermal emission, however
out by the authors, such a classification is not straightforward sometimesanon-thermalcomponentisalsopresent
due to present day observational limitation. In our work we – physical parameters of plasma in CXSs (e.g. temperature,
willfocusonCXSswhichareobservedjustaboveabovewarm density)changesmoothlywithtime
Articlenumber,page1of16
A&Aproofs:manuscriptno.kolomanski-mrozek-chmielewska
– continuousenergyinputmustbepresentas thesourceslast as in typical TRACE images. In the Ca XVII image, there was
longerthancharacteristictimeofcoolingofhotplasma an additional set of sharp loops with a distinct, diffuse CXS
(Pres´&Kołoman´ski2009,Fig.5)-animagesimilartothatwhat
Thereislong-standingdiscussionconcerningthefinestruc- was seen in the SXT data. Nevertheless, this observations and
ture of CXSs. Yohkoh (Ogawaraetal. 1991) observations similar ones of EIS, cannot give us key information about the
showed CXSs as a coherent region of size of the order of natureofCXSs.EISimagesareobtainedbyrasteringwithaslit.
100 − 1000 arcsec2 (see e.g. Pres´&Kołoman´ski (2009)). De- Thisprocesstakessignificantamountoftime.Inthecaseofthe
spitethisrelativelylargesize,CXSswereseenasdiffusewithout SOL2006-12-17T17:12flarethescanningthroughtheCXStook
anysmall-scalestructure.However,datafromtheSXTtelescope about15minutes. Itmay be longenoughto blur anydetailsof
(Tsunetaetal.1991)haveanangularresolutiononly3.7arcsec thesource.Thus,theEIStemporalresolutionmaybenotappro-
andalowthermalresolution(broadbandfilters).Thesecharac- priateinthecaseofCXSs.
teristicsmaynotbesufficienttoanswerthequestionaboutfine RHESSI (Linetal. 2002) observations offer another pos-
structure of CXSs. As a consequence it is hindered to answer sibility to study nature of CXSs (e.g. Jiangetal. 2006;
other questions concerning formation of CXSs and their slow Väänänen&Pohjolainen2007;Caspi&Lin2010).Thesatellite
andgradualevolution. enablesus to obtain images and spectra of solar X-ray sources
TRACE (Handyetal. 1999), launched in 1998, cast doubts with a goodangular(2.3arcsec) and spectral(1 keV for imag-
on the diffuse nature of CXSs. This EUV telescope had higher ingspectroscopy)resolution.Kołoman´skietal.(2011)analyzed
spatial resolutionthan Yohkoh/SXT.In TRACE images of solar three long-duration flares. The authors used RHESSI imaging
flaressetsofsharploopsareusuallyseeninsteadofdiffusecoro- spectroscopytoobtainphysicalparametersofCXSsandtocal-
nalsources(seeFig.2inWarrenetal.(1999)).Atfirstglanceit culate the energy balance of the observed sources. One of the
wouldseemthatthediffuseappearanceofCXSscomesfromthe conclusions is that each CXS observed was smooth. No CXS
insufficientspatialresolutionofSXTandthatthesourcesarein was visible in the images reconstructed for the four narrowest
fact composed of multitude of small elements, e.g. bright tops grids(angularresolution2.3−12arcsec).Thissuggeststhatei-
of filamentary loops. However,TRACE is the most sensitive to ther CXSs are diffusive with no internalstructure or they are a
plasmaatquitedifferenttemperatures(1−2MK)thanSXT(≈10 superpositionofsmallunresolvedsubsourceswithseparationof
MK).Thus,thehotplasmaofCXSsisbarelyvisibleinTRACE the subsourcessmaller than the angular resolution of the finest
images. RHESSI grid. Thus, the question about the structure of CXSs
Nevertheless, some TRACE observations of solar flares, wasleftwithoutanswer.
especially in the 195 Å band, reveal diffuse structures lo- RHESSI is an excellent instrument to observe coronal X-
cated above narrow PFLs. Location of these diffuse structures raysources.Itcoversverywiderangeofradiationenergyfrom
corresponds well with the Yohkoh CXSs (see e.g. Fig. 2 in 3 keV to 17 MeV with good resolution and high sensitivity.
Warrenetal. (1999)). Thus, the diffuse nature of CXSs cannot However it has some limitations. First, it is a Fourier imager
beexplainedonlybythelowspatialresolution.Thereisanother thus, images are reconstructedbased on recordedflux modula-
instrumentalfactorthatmaybluranimageofCXSsandveilthe tionadditionallydisturbedbynoise.Second,limiteddynamical
truenatureofthesources.Plasmainsolarflaresismultithermal. range(usually10:1)andlackofsensitivitytoplasmacolderthan
A thermal response of the TRACE 195 Å channelhas two dis- ≈7MKmakeitdifficulttorelatedirectlyCXSstocolderand/or
tinctmaxima(seeFig.3inPhillipsetal.(2005)).Thehigher(in fainter structures in flares. Therefore,a good idea is to supple-
response)maximumcomesfromFeXIIline,thatformsattem- mentRHESSI withadditionaldatafromanEUVdirectlyimag-
peratures from 1 to 2 MK (“warm plasma”), whereas the sec- ingtelescope.
ondoneisproducedbyFeXXIVlineintemperaturesfrom10to Gallagher et al. (2002) used combined observations of
30MK(“hotplasma”).Duetodifferentthermalwidthsofthese RHESSI and TRACE to analyze the SOL2002-04-21T01:50
twomaxima,thecoolermaximumselectsmuchlessemissionel- flare. Relation between EUV structures (TRACE) and X-ray
ements(e.g.loops)fromthemultitudeofelementswithdifferent sources(RHESSI)wasinvestigated.Theauthorsfoundthatdur-
temperaturesthanthehottermaximum.Thus,diffuseappearance ing the rise phase of the flare CXS was at the same or greater
of CXSs mightbe a resultof blendingof fine elementsof very altitude in the corona as diffuse and hot emission recorded by
differenttemperature.Itseemsthat,despitetheverygoodspatial TRACEabovewarmPFLs.TheCXSwasobservablealsoduring
resolution,theTRACEthermalresolutionforhot,flareplasmais the decay phase of the flare for 11 hours and it was always at
insufficient to study CXSs. Moreover,TRACE response to hot higheraltitudethanthetopsofPFLs.Theauthorsnotethatthere
plasma is almost two orders of magnitude lower than to warm wasnoevidenceofanyfinestructureoftheCXS.
plasma.Duetothisfact,hotcoronalsourcesarehardtoobserve Some instrumentalcharacteristics of TRACE, especially, as
inapresenceofbrightpost-reconnectionflareloops. mentionedabove,relativelylow sensitivity to hotplasma, limit
TheEUVImagingSpectrometer(EIS)(Culhaneetal.2007) analysisthatcanbeperformedoncombinedRHESSI –TRACE
is one of the three scientific instruments aboard Hinode dataset.Fortunately,atpresentthereisaninstrumentwhichcan
(Kosugietal.2007).EISprovidessimultaneousobservationsin significantly improve the situation. The Atmospheric Imaging
several spectral lines. The instrument can record emission of Assembly(AIA)onboardtheSDOsatellitedelivershighquality
hot plasma in e.g. a band that includes the line of CaXVII at imagesinvariousEUVfilters.Theinstrumenthasgoodangular,
192.82Å.Thelineformsattemperature≈ 4.5−7.5MK,i.e.it thermalandtimeresolutionandishighlysensitivetohotplasma.
can partially revealhot plasma that is present in a flare. More- Thermalresponsesof AIA and RHESSI overlapin the range ≈
over,thethermalresolutioninthislineishigherthantheTRACE 7−16MK allowingto studya relationbetweencoronalX-ray
thermal resolution for hot plasma. Pres´&Kołoman´ski (2009) sourcesandstructuresobservedinEUV.
analyzedEISobservationsofthelong-durationflareSOL2006- We present the study of a coronal source of the SOL2011-
12-17T17:12.The authors reported that the flare emission was 10-22T11:10long-duration flare observed simultaneously with
seen in the form of sharp filamentary loop structures in lines RHESSIandSDO/AIA.Usingsuchcomplementarydataweare
withformationtemperaturesbelow3MK(e.g.FeXII195.12Å), abletostudyarelationbetweentheCXSandEUVemissionof
Articlenumber,page2of16
S.Kołoman´ski etal.:FinestructureandlongdurationofaflarecoronalX-raysourcewithRHESSIandSDO/AIAdata
the flare in greater detail than it was possible earlier. We focus
GOES 1-8 A
on the following questions:What was responsiblefor the CXS
presenceandlongduration?Wasthereanyfinestructurein the 10-5
CXS?
We selected a long-duration flare (LDE – long-duration -2m
s
event) for our analysis and gave our attention mainly to its de- att
W
cayphase.Thereasonsareasfollows.LDEflaresarecharacter-
izedbyaveryslowevolution,especiallyduringthedecayphase.
Moreover, LDEs occur in large magnetic structures – loops in
such flares may reach the height of 105 km. Any instrumental 10-6
drawbackslikeinsufficientangularortemporalresolutionsmay 10:00 12:00 14:00 16:00 18:00
Start Time (22-Oct-11 09:00:00)
belesslimitinginthecaseofslowlyevolvinglarge-scalestruc-
tures of LDEs. Basic characteristics of coronal X-ray sources, 104
like slow evolution, smooth appearance and resistance, are ex- RHESSI 6-12 keV
ceptionally conspicuousand surprising during the decay phase 103
-1V
ofLDEs,wherethesourcesmaylastforhoursonend. e
k
-2m 102
c
-1s
2. Observationsanddataanalysis ns 101
o
ot
h
The SOL2011-10-22T11:10 LDE flare occurred in the ac- p 100
tive region NOAA 11314, close to the west solar limb.
The GOES/SEM (Space Environment Monitor, Donnellyetal. 10-1
10:00 12:00 14:00 16:00 18:00
(1977)) and RHESSI light curves of the flare are presented in Start Time (22-Oct-11 09:00:00)
Fig. 1. According to GOES data the flare started at 10:00 UT.
Fig.1.GOESandRHESSI lightcurvesoftheSOL2011-10-22T11:10
The rise phase lasted very long and the maximum of bright-
flare.GapsintheRHESSI curvearecausedbythesatellitenightsand
nessin1−8Åband(M1.3)wasreachedat11:10UT.Theso- the South Atlantic Anomaly. Combined RHESSI - AIA analysis pre-
lar X-rayfluxreturnedto thepre-flarelevelatabout19:30UT. sentedinthispaperwasperformedfor10timeindicatedmarkedbythe
Theflarewasatypicalexampleofso-calledlong-durationevent dottedlines.Thelengthoftheintervalsis16to180seconds.
with the slow rise phase (sLDE) (Hudson&McKenzie 2000;
Ba¸k-Ste¸s´lickaetal.2011).Thedataanalyzedherewereobtained
by the Atmospheric Imaging Assembly (AIA) (Lemenetal. the instrument’s capabilities thanks to annealing1. The last an-
2012)installedonboardtheSolarDynamicsObservatory(SDO) nealing,beforetheOctober2011,wasperformedinMarch2010,
(Pesnelletal. 2012) and by Reuven Ramaty High Energy So- thereforewe may expectthat, for the analyzedevent, detectors
lar SpectroscopicImager(RHESSI).TheSolarSoftWare (SSW, areinagoodconditionandstillprovideuswithvaluablescien-
Freeland&Handy(1998))systemwasusedtoanalyzethedata. tificdata.
The SDO/AIA consists of a set of four 20 cm, normal- The RHESSI light curve of the analyzed flare is shown in
incidence telescopes. The field of view coversentire solar disk Fig. 1. The light curve is not continuous due to the satellite
with 4096×4096CCDs. AIA providesobservationswith high nights and the South Atlantic Anomaly. Thus, the flare is not
angular (≈ 1.5 arcsec) and temporal (≈ 10 s) resolutions in 10 fully covered by RHESSI observations. Dotted lines in the fig-
bandsincluding7EUVbands.TheAIAobservationscoverthe ureindicate10shorttimeintervals,forwhichwereconstructed
wholedurationoftheanalyzedflare.Forourstudyweselected RHESSI images. There are currently several algorithms avail-
able for RHESSI image reconstruction. Images for our analy-
sixAIAbands,i.e.94Å,131Å,171Å,193Å,211Åand335Å.
All six bandswere used to calculate differentialemission mea- sis were reconstructedusing PIXON algorithm(Piña&Puetter
1993).Theenergyresolutionchosenforreconstructionwasusu-
suredistributions.Twoofthesebands,94Åand131Å,arethe
ally1−2keV.Inafewcaseswiderenergyranges(4−6keV)
best choice to study morphology and dynamics of hot plasma
werenecessarybecauseofverylowsignal(latedecayphase,en-
inregionsofcoronalsources.Eachofthese twobandshastwo
narrow maxima in thermal response: warm (≈ 1 MK) and hot ergyabove10keV).WechosePIXONalgorithm,whichisgen-
(≈ 10 MK), and both maxima are almost equal in sensitivity erallyconsideredasveryreliabletechniqueinimagereconstruc-
tion2.Moreover,basedonourexperience,wethinkthatPIXON
(O’Dwyeretal.2010;Boerneretal.2012).Thus,ahotcompo-
algorithm combined with the grid selection method (described
nentofflareemissionisnotoverwhelmedbyawarmcomponent
below) is the best option to study coronalsources. However,it
as it was in the case of the TRACE 195Å band.Moreover,the
should be kept in mind that reconstructionof images based on
94 Å and 131 Å bands have high thermal resolutions for hot
flux modulation recorded by the RHESSI detectors is compli-
maxima.Theresolutionisasgoodasforthewarmmaxima,and
cated.Therecordedfluxisdisturbedbynoiseofsolarandnon-
muchhigherthan the resolutionof TRACE 195Å bandfor hot
solar origin. In result we deal with an ill-posed inverse prob-
plasma.
lem and uncertainty of synthesized images. In order to verify
RHESSI is a rotating Fourier imager with nine germanium
reliability of the obtained PIXON images we reconstructed for
detectors. Detectors are large (7.1 cm in diameter x 8.5 cm in
some of the 10 selected time intervals also control images us-
height)andcooledtoabout75K,thusthesensitivityofthein-
ingotheralgorithmsandthe gridselectionmethod.These con-
strumentisgreat,andallowsforinvestigationofverylowsolar
flare fluxes that are observed during the decay phase of long- 1 http://sprg.ssl.berkeley.edu/˜tohban/nuggets/?page= arti-
durationflares. The instrumentwas launchedin February2002 cle&article_id=69
whichresultsinlowersensitivityafterseveralyearsofobserva- 2 http://hesperia.gsfc.nasa.gov/rhessi3/software/imaging-
tions. However,there isa possibility fora partialrestorationof software/image-algorithm-summary
Articlenumber,page3of16
A&Aproofs:manuscriptno.kolomanski-mrozek-chmielewska
trolimagesaresimilartoPIXONimagesexcepthigherenergies andSAHR2werevisibleasareasofconvergenceofSADs,giv-
where recordedflux is low. In such a conditionsother used re- ing them the appearanceof two hands. The SAHR was visible
construction algorithms usually failed while PIXON algorithm in the 131 Å band untilalmost 17 UT. The 94 Å band is more
gavegoodqualityimages.Weusedimagestoperformalsoimag- sensitivetohotthantowarmplasma.Thus,imagesinthisband
ingspectroscopy.Forthispurposeweneedaswideenergycov- were dominated by hotter plasma of the SAHR and the PFLs
erage as possible. PIXON images meet this requirement most werefainterthaninthe131Åband.
satisfactorily. The images reconstructed using RHESSI data (contours in
X-ray emission observed several hours after a flare maxi- Fig.2)showonlycoronalX-raysource(CXS).Despitethefact
mum is extremely weak, and usually X-ray images cannot be that the whole structure of the flare was visible (no occulting
reconstructed with standard parameters. We used a method of bythesolardisk),therewerenodetectableX-raysourcesatthe
gridselectiondescribedinKołoman´skietal.(2011).CoronalX- loops’footpoints.ThepositionoftheCXSfollowedthebrightest
ray sources observed by the authors were large and diffuse. In
partofhotcomponentintheAIA131Åimagesformostofthe
such a case, the reconstructionmethod does not convergewith
durationoftheflare(seeFig.2).Beforetheflaremaximumthe
“standard” grid selection, i.e. a set consisting of grids No. 3-
CXS was located at or very near the brightapexesof the HLs.
9, producing more or less noisy distribution of small sources.
Such a situation lasted for about an hour until the HLs cooled
The situation may be improved. We should keep in mind that
down.HowevertheCXScontinueditsexistencebecauseanew
a signal from a large source is modulated only by grids which
hotregionemerged,i.e.theSAHR.AfterthemaximumtheCXS
haveresolutionworsethantheactualsourcesize.Thedecision,
was located just above the PFLs, in the SAHR. Before about
whichgridsshouldbeused,ismadeonabasisofBackProjec-
13:00UTtherewasonlyoneX-raysourceandafter13:00UT–
tion (BP) imagesreconstructedfor individualgrids. Only grids
thereweretwo(CXS1andCXS2).Thedivisiontookplaceatthe
showing clear modulation of source signal are selected for the
sametimeasthedivisionoftheSAHRintheAIA131Åband.
finalimagereconstruction.Inthe case ofthe analyzedflare the
ThelastimageoftheCXS, whichwewere abletoreconstruct,
modulationwasnotpresentingridsNo.1-5inanyselectedtime
isfor16:00UT.NomorethanonehourlatertheSAHRbecame
interval. Thus, we used only grids with lower resolution, start-
practicallyinvisible.
ing from grid No. 6, to reconstructimages for our study. Later
Goodenergyresolutionofthereconstructedimagesenabled
inthispaperwetesthowagridselectionmethodbehavesinthe
us to perform an imaging spectroscopy of the observed CXS.
casewhereanCXSconsistsofseveralsmall-scalesources.
Theimagingspectroscopyhasa few advantagesin comparison
DuetolowX-rayemissionduringthedecayphaseofflares
with a spectroscopic analysis of the entire solar signal, with-
longertimeintervalsarerequiredtoreconstructreliableRHESSI
out a spatial resolution. First, during an image reconstruction
images,i.e.imageswithhighenoughphotonstatistics.Weused
the backgroundis naturally subtracted. The problem of appro-
timeintervalsfrom16to180secondslong.Slowtemporalevo-
priately subtracted background is severe in the case of long-
lutionoftheanalyzedflare,includingslowchangesoftheCXS
durationflares,whicharecharacterizedbyverylowsignalsdur-
position,enablestousesuchalongtimeintervals.Lengthofthe
ingthedecayphase.Second,duringthelongdecayphase,other
intervalswereselectedaccordinglytokeepnumberofcountsin
flaresmayoccurandmasktheoriginalemissionfromanLDE.
the range 103−104 per one image. Furthermore,we used data
Theflareobservedaround15:20UT(seeFig.1),whichwasin
phasestacking,i.e.combiningdataonthebasisofRHESSI roll
anotheractiveregion,farfromtheanalyzedflare,isagoodex-
angle.Thestackingimprovesstatisticsandenablesimagingover
ample.Insuchacasetheimagingspectroscopyallowstodistin-
longtimeintervals.
guishtheactualemissionfromtheflaredespitethefactthatthe
Weselected10timeintervalsforacombinedRHESSI-AIA
additionalemissionfromanotherflarewaspresent.
analysis.TheintervalsareindicatedinFig.1bydottedlines.A
The physical parameters of the observed coronal X-ray
set of AIA images with RHESSI contours is shown in Fig. 2.
source obtained from imaging spectroscopy are shown in Ta-
With help of the figure we can take a look at the flare evolu-
ble 1. This analysis could be done for the eight first time in-
tion.TheSOL2011-10-22T11:10eventbeganwithaneruptionat
tervals. For the last two intervals there was not enoughimages
about10:00UT.Asetofbrightandhotloops(HLs,T ≈10MK)
in different energy ranges (too low signal above 10 keV) to fit
appearedjustafterthestartoftheeruption.Theloopswerevis-
thespectrum.TheCXSspectrumcouldbefittedwithtwother-
ibleonlyintheAIAbandswithsignificantsensitivitytoplasma
malcomponents.Nonon-thermalcomponentwasdetected.The
at temperatures around 10 MK or more (i.e. 94 Å and 131 Å temperaturesofthecomponentswere8−9and13−23MKre-
bands). The apexes of the HLs were their brightest parts. Just spectively,whichcorrespondswelltothehotstructuresobserved
after the maximum of the flare (11:10 UT according to GOES byAIA,i.e.theHLsandSAHR.
data)theHLsstartedtofadeduetoplasmacooling.Theyreap-
peared later, roughly at 12:00 UT, as an arcade of warm post-
reconnectionflareloops(PFLs,T ≈1−2MK).Simultaneously 3. Results
with the PFLs, a brightsupra-arcadehotregion(SAHR) began
3.1.CXSstructure
tobevisible.AsinthecaseoftheHLs,theSAHRwasalsodis-
tinctlyvisibleonlyintheAIA94Åand131Åbands,i.e.ithad The coronal X-ray source observed by RHESSI during the
temperature≈10MK. SOL2011-10-22T11:10 flare showed typical characteristics as
The SAHR showed clearly a well known phenomenon – described in the previous papers on coronal sources of long-
supra-arcade downflows (SADs) (McKenzie&Hudson 1999; durationflares(e.g.Kołoman´skietal.2011).RHESSIimagesre-
McKenzie 2000), i.e. dark features moving downwards in the constructedwiththeuseofthementionedgridselectionmethod
SAHR towards the PFLs. The SADs were also visible in the showlargeandsmoothCXS,withoutanyinternalstructure.The
94Åband.DuringtheflaredecayphasePFLsgothigherandthe positionoftheCXSduringthedecayphasewascoincidentwith
SAHR movedslowly upwardsand becamefainter. Since about thesupra-arcadehotregionseeninAIAimages.However,unlike
13:00 UT the SAHR is seen divided into two parts (SAHR1, theCXS,theSAHRwasnotstructureless.TheSAHRconsisted
SAHR2) separated by the fainter region between. The SAHR1 of many brighter and fainter small-scale areas (see e.g. Fig. 2,
Articlenumber,page4of16
S.Kołoman´ski etal.:FinestructureandlongdurationofaflarecoronalX-raysourcewithRHESSIandSDO/AIAdata
Fig.2.SDO/AIAimagesillustratingtheevolutionoftheSOL2011-10-22T11:10flare.ContoursshowthecoronalX-raysource(CXS)observed
withRHESSIintheenergiesgivenineachimage.Intheimagestheflarehotloops(HLs),thesupra-arcadehotregion(SAHR)withsupra-arcade
downflows(SADs),andthepost-reconnectionflareloops(PFLs)arevisible.Thedashedlinesmarkthecuts(1and2)forwhichdifferencedynamic
mapswereconstructed(seeFig.7).
Table1.ThegeometricalandphysicalparametersoftheobservedcoronalX-raysourceobtainedfromRHESSIdata(imagesandimagingspec-
troscopy).
time fitted temperature emission cross-section altitude
component measure area ofcentroid
[UT] [MK] [1049cm−3] [arcsec2] [103km]
low 8.6 1.66
10:24 7980 61.0
high 23.1 0.0020
low 7.5 6.42
10:50 6740 67.3
high 16.4 0.010
low 6.7 7.67
11:10 8770 59.7
high 13.7 0.031
low 8.7 1.64
11:22 4840 62.8
high 17.8 0.0022
low 8.6 1.22
12:00 4820 60.4
high 15.3 0.0062
low 8.6 0.88
12:35 6510 68.9
high 13.9 0.0022
low 8.5 0.14
12:55 9050 76.4
high n/a n/a
low 7.8 0.071
13:37 2410 85.6
high 12.6 0.0042
Notes.Parametersforthelowandhightemperaturecomponent,thatwerefittedtotheobservedspectra,aregiven.CombinedparametersT and
EM aregivenforbothpartsoftheCXSvisibleat13:37UT(CXS1andCXS2).TheprojectedareaandthecentroidaltitudeoftheCXSwere
determinedfromtheRHESSIimagesintheenergyrange5.5−7.5keV.TheCXSwasdefinedbyintensityisoline0.5withrespecttothebrightest
pixeloftheCXS.
Articlenumber,page5of16
A&Aproofs:manuscriptno.kolomanski-mrozek-chmielewska
12:35UTimage).Thus,weshouldanswerthequestionwhether thathotplasmavisibleintheAIAimage(T≈10MK)canbealso
thesmoothappearanceoftheCXSisrealoriscausedbycharac- observedbyRHESSI.
teristicsofreconstructionmethodsusedandtheinstrumentitself WedefinedasetofelevensmallX-raysourceswithsizes,lo-
(Fourierimager). cations,andrelativeintensitiessimilartoEUVsourcesvisiblein
Wedecidedtochecksensitivityofthegridselectionmethod theAIA131ÅimagewithintheCXS.Ratioofbrightnessofthe
to a scenario in which an CXS consists of several small-scale this small sources is ≤1.5:1. The ratio is smaller than RHESSI
sources.SuchanapproachismotivatedbytheAIAobservations dynamic range (typically 10:1, Linetal. (2002)). Thus, all the
oftheanalyzedflare.Aspreviouslymentioned,theobservations sourcesshouldbevisibleinRHESSIimagessimultaneously.The
showmanyEUVstructuresspatiallycorrelatedwithX-rayemis- simulated set of sources is presented in Fig. 4, the right panel.
sion. Having such a synthetic CXS consisting of sub-sourceswe re-
constructeditsRHESSIimagesusingthePIXONalgorithmwith
If X-ray emission comes from a number of small sources,
the grids No. 3-6, 8, and 9. Althoughthe reconstructedimages
occupyingarelativelysmallarea,thensignalmodulationisnot
show a set of small sub-sources but their sizes, locations, and
presentinfinergrids,and,asa consequence,we misinterpretit
relative intensities are different from the ones in the assumed
as one large source. The example of such a case is presented
model.Moreoverimagesreconstructedforslightlydifferenten-
in Fig. 3. The first column shows several of simulated scenar-
ergyrangesshowasignificantlydifferentsetofsub-sources(see
ios:asinglesmall(3arcsec)source,asinglelarge(20arcsec)
images in the right column of Fig. 5). Thus, we can conclude
source, two small sources spaced at 10 arc sec, and ten small
that the sub-sources and overall fine structure in reconstructed
sources (3-7 arc sec) spread on elliptical area. Back Projection
imagesarenotreal.ThefinestructureofthesyntheticCXSwas
(BP)singlegridimagesreconstructedforthescenariosarepre-
notpossibletorecover.
sented in the remainingcolumnsof Fig. 3. It is clear that for a
Thefactthatwedonotseethelargestofthesimulatedsub-
singlesourceordoublesourcesthegridselectionmethodworks
sources constitutes an unexpected result of this test. We per-
properly.Namely,thesmallsource(3arcsec)givesmodulation
formedseveralothertestsinwhichwetriedotherscenarioswith
visible on single grid images starting from grid the No. 3 (res-
onelargersub-sourceandvariouscombinationsofsmallerones.
olution 6.79 arcsec). The size of the larger source is compara-
Evenforatwo-sourcescenario(onelarge,andonesmall)weob-
blewiththeresolutionofthegridNo.5.Therefore,weobserve
tainedanimagewiththesmallersub-sourceonly.Thelargersub-
clearmodulationstartingfromthegridNo.6whichhasresolu-
sourceiscompletelyinvisible,itwaslostduringthereconstruc-
tion35.27arcsec.Thedoublesourceconsistingofsmallsources
tion process. Thisunexpectedresult was not mentionedin pre-
separated by 10 arc sec is barely visible in the grid No. 3 im-
vious papers testing performance of various image reconstruc-
age, but both sources are noticeable. The multiple-source sce-
tion algorithms (Aschwandenetal. 2004; Schmahletal. 2007;
nario in single grid imagesis similar to the single large source
Dennis&Pernak 2009). From the interpretation point of view
scenario. There is a weak periodicity visible in the grid No. 5
sucha resulthasseriousconsequences.Itispossiblethatinre-
imagewhichisconnectedtoellipticalshapeoftheareacovered
constructedimagesofX-rayemissionbasedonrealobservations
by small sources, but, generally,a fine structure cannotbe rec-
wemaynotsee large,diffusesourceswhensmallersourcesare
ognizedinsingle-gridimages.Thus,itispossiblethatthelarge
presentnearby.
andsmoothCXSoftheanalyzedflarehadsmall-scalestructure,
Fromthispaper’spointofview,suchartificialdisappearance
similartothatseenintheAIAimages.
ofalargersourcecanhelpustoanswerthequestionaboutfine
SimulationsofRHESSIsourceswereperformedwiththeuse
structure of the CXS of the analyzed flare. The presence of a
of standard software available in the SSW package. It allows
large,diffuseCXSinreconstructedimagesoftheflaremaysug-
to include typical background count rates for a given segment
gest that there are no small sources, namely a fine structure is
of detector. We tried several levels of background and photon
less likely. We checked this supposition by comparing images
countratestoverifyhowmuchitmayaffectourresults.Signal-
reconstructedforrealdatatoimagesreconstructedforoursimu-
to-noiseratiosfrom3to1000(countrates101−106)weresim-
lateddistributionofsub-sources.Therealimageswereobtained
ulated.We concludedthatnoisedonotaffectreconstructionre-
forthesameenergyintervals(6-7,6.5-7.5,and7-8keV)andthe
sults significantly even for S/N ratios as low as 3. It is obvi-
samesetofgrids(No.3-6,8,and9)assimulatedones.Keeping
ousifwerememberthatbackgroundisalmostnotmodulatedby
inmindtheresultsofoursingle-gridtest,wereconstructedthese
RHESSI rotation while solar flare flux is (Hurfordetal. 2002).
imageswiththeuseoffinegrids,downtothegridNo.3,despite
Inrealobservationsweareabletoregisterweak4smodulation
lack of modulation in those grids. Using overlapping intervals
of X-ray flux reflected by Earth but the level of such signal is
wewantedtoseeifanyofthereconstructedsourcesisvisiblein
far below solar flare signals. The rangeof countrates tested in
thesamelocationinenergy-neighboringimages.
ournumericalsimulationscoverstherangeofcountstatisticsin
The resulting images are presented in Fig. 5 (the left col-
imagesreconstructedforrealRHESSIdata(103−104countsper
umn), and compared to the images reconstructed for the syn-
oneimage).
thetic CXS (the right column of the same figure). The sources
Weranthefollowingsimulationtocheckthepossibilitythat reconstructedfortherealdatashowalmostnorepeatability,i.e.
analyzed flare had small-scale structure. An AIA 131 Å image theimagesreconstructedforadjacentenergyrangesarenotsimi-
wasusedasaproxyofapossibledistributionofafinestructure lar,suggestingthatafinestructureisabsent.Inordertoincrease
of the CXS. In Fig. 4 (left panel) we present the AIA 131 Å the certainty of this result we reconstructed additional images
image taken at 12:35:33UT. The white contoursare drawnfor with the same parameters and real data using five other algo-
EUVemissionwithlevelsof0.6,0.7,0.8,and0.9ofthebrightest rithmsavailableinRHESSIsoftware.Outcomewassimilarasin
pixel.Thegraycontourrepresents0.5levelofthemaximumof thePIXONimages,finestructurewasnotvisible.Thereisonly
theRHESSIsource(CXS)reconstructedfor2-minutelongtime onesourcethatispresentinallthreePIXONimagesaroundpo-
interval(12:35-12:37UT).WeassumedthattheCXSmighthave sition x = 925,y = 560. However, we doubt that the source
the fine structure similar to the structure observed in the AIA is real due to the following reasons. The source is not in the
131 Å image. Such an assumption can be justified by the fact same position in energy-neighboringimages. Its centroid in 6-
Articlenumber,page6of16
S.Kołoman´ski etal.:FinestructureandlongdurationofaflarecoronalX-raysourcewithRHESSIandSDO/AIAdata
Fig.3. Leftcolumn: simulateddistributionsofX-raysources.Remainingcolumns: single-grid, backprojectedimagesofsimulatedsources. A
singlelargesource,anddistributionofsmallsourcesinellipticalareaaresimilarintermsofmodulationseeninsingle-gridimages.Namely,the
modulationofsignalisvisiblestartingfromthegridNo.6whichsuggestthatsourcehasdiameterbetween20-30arcsec.Asinglesmall,andtwo
smallsourcesarebothvisiblestartingfromthegridNo.3whichisthefinestgridwithresolutionworsethantheactualsizeofthesesources.
7keVisshiftedwithrespectto7-8keVbyavaluesimilartothe source was about 8 seconds (about two orders of magnitude
source size. Moreover, there is no bright structure in AIA im- shorterthanthedurationofthecoronalsourceoftheflare).Thus,
ages norin differentialemission measure maps(see subsection ifRHESSIimagesarereconstructedfortimeintervalsofe.g.tens
3.3) around x = 925,y = 560. Furthermore, the source is not ofsecondsorevenseveralminutes,CXSsinsuchimageswillbe
presentin any image reconstructedwith BP, Clean, EM, MEM asuperpositionofmanytransientsub-sources.
NJIT,UVSmoothalgorithms.
InthecaseoftheSOL2011-10-22T11:10flareweusedtimes
However, the same situation is observed in the case of the ofintegrationaslongas2-4minutestoreconstructRHESSIim-
images reconstructed for the synthetic CXS which consists of ages.MuchshorterintervalsshouldbeusedtocheckiftheCXS
sub-sources.Nevertheless,theimagesreconstructedforthereal isasuperpositionoftransient,randomlyoccurringsub-sources.
and simulated data differ. In the case of the synthetic CXS the However, this can not be done due to the low level of signal.
PIXON algorithm managed, to some extent, to reconstruct its InsteadofRHESSIimageswecanuseAIAimages.Asanexam-
finestructure,i.e.thesizesoftheobtainedsub-sourcesarecom- plewecompareagaintheRHESSIimagereconstructedfortime
parable to their simulated input values, despite different posi- interval 12:35-12:37 UT with the AIA images. As mentioned,
tions.Fortherealdatawedidnotobservesuchabehavior.The therewereEUVsourcesvisibleintheAIA131Åimageswithin
images reveal big sources that are significantly larger than the theareaoftheX-rayCXS(seeFig.4).IfthelargediffuseCXS
resolutionofthefinestusedgrid(No.3).Wealsoreconstructed wasthesuperpositionoftransientsub-sourcesthisshouldbevis-
PIXON imagesforrealdata addingthegridNo.1 tothe setof ibleintheAIAimages.WecomparedalltenAIA131Åimages
grids. The resulting images are almost the same as without the takenduringthe12:35-12:37UTintervalandfoundnoevidence
gridNo.1.We interpretthisresultasalackoffinestructurein of such transient sub-sources. All the visible EUV sources are
the observed source. However, it should be noticed that a sim- quitestableintheirlocationandbrightness.NonewEUVsource
ilar result may be produced in the case of many small sources appearedandnoexistingsourceextinguished.Thus,iftheCXS
with fast changes of intensity (fast in comparison with time of wassuperpositionoftransientsub-sourcesitcannotbedetected
integrationusedforimaging). byneitherRHESSInorAIA.
Such a model of an CXS was presented by e.g. However,a small-scalestructureoftheCXScannotbede-
Longcopeetal. (2010). In the modelthe CXS consists of a set cisively excluded.Firstly, spectra of CXS emission suggestthe
oftransient,randomlyoccurringsub-sources.Inthecaseofthe presenceoftwothermalcomponents.Thecomponentsmayori-
flare analyzed by the authors, a typical lifetime of each sub- ginfromtwothermodynamicallydifferenttypesofareaswithin
Articlenumber,page7of16
A&Aproofs:manuscriptno.kolomanski-mrozek-chmielewska
Fig.4.Left:SDO/AIA131imagewiththeintensitycontours(0.6,0.7,0.8,and0.9ofbrightestpixel).MoreoverthecoronalX-raysourceinthe
energy6-7keVisshownby0.5contour (ingray).TheEUVemissionwithintheCXSwasnotuniform,thereweremanysmallbrightregions.
Right:syntheticCXSusedinoursimulation.ThesyntheticCXSconsistsofasetofelevensmallX-raysourceswithsizes,locations,andrelative
intensitiessimilartoEUVbrightestregionsvisibleintheAIA131ÅimagewithintheCXS.
theCXS.Secondly,AIAimagescombinedwiththeRHESSIdata inthe SDHR. Dueto thiswedecidedtocombinethe 94Å and
showthattheCXSwasco-alignedwiththesupra-arcadehotre- the131Åmapsintoone94+131DDmap(seeFig.6).Itisworth
gionwhichwasfullofsmall-scalestructures.Oneofthesesmall- tonoticethatthethirdsensitivetohotplasmaband,i.e.the193Å
scale structures were supra-arcadedownflows. In the next sub- band,doesnotprovideadditionalinformationaboutthe SADs.
section we analyze SADs dynamicsand possible physicalcon- Althoughthe 193 Å band’smaximal sensitivity for hot plasma
nectionbetweenthedownflowsandthecoronalX-raysource. isaround17−19MK,theSADtracksarevisibleatroughlythe
samealtitudeasonthe131Åmapatagiventime.
Fig.7showsdifferencedynamicmapsfortwocuts.Thecuts
3.2.CXSandSADs
startatthefootpointsofthePFLsandrunuptofollowthechang-
We carefully prepared and studied difference dynamic maps ing position of the CXS including its division into LTS1 and
(space-timemaps)toverifythispossibleconnection.SADsare LTS2(seeFig.2).Eachofthetwocutsfollowsuponeofthese
rather weak when comparedto other parts of a flare. Thus, we sub-CXS. The positionsof the CXS centroidsestimated on the
RHESSIimagesaremarkedintheDDmaps.Colorscodeenergy
took the AIA images in 193 Å, 94 Å, and 131 Å bands, and
constructedrunningdifferenceimagesforeachbandseparately. for which an image of the source was reconstructed.The CXS
Thenweextractedanarrowcutfromeachrunningdifferenceim- movedhorizontallybeforetheflaremaximum(seetwofirstim-
ages in Fig. 2). Due to this the source is outside the selected
ageandstackthemintimesequence(Sheeleyetal.2004).After
cuts for two first time intervalsfor which we reconstructedthe
this operationwe gotan image with time runningalong x-axis
RHESSI images.Thepositionofthe CXS arenotshownin the
and distance measured along the extracted path on y-axis. On
such difference dynamic (DD) maps all the moving structures figureforthesetimeintervals.
(e.g.SADs)arevisibleasbrightanddarktracks. The bright and dark tracks of the SADs reveal that the
downflows were decelerated from almost 100 km s−1 to about
The SADs were very faint in the 193 Å DD maps. Fortu-
2 km s−1. These are typical values observed for SADs in
nately, they are distinctly visible in the 94 Å and 131 Å ones
many flares (see e.g. Savage&Mckenzie (2011); Warrenetal.
(see Fig. 6). The maps show clearly the difference in the ther-
(2011); Liuetal. (2013)).For detailed analysisof SADs of the
mal response of the AIA bands and the multithermalcharacter
SOL2011-10-22T11:10flareseeSavageetal.(2012).Velocities
of the flare emission. In each of the three DD maps the post-
of the downflowsestimated by us are similar to those obtained
reconnection flare loops are visible in similar position because
bySavageetal.(2012).
allthethreebandshavethe(local)maximumofthethermalre-
The coronal source was located at the altitude where the
sponsearound1MK.ThePFLsgothigherwiththeaverageve-
SADsslowdownandaccumulateabovethePFLs.Constantand
locity of 3−4 km s−1. On the other hand, at a given time the
distinctstreamoftheSADswaspresentsincethemaximumof
SAD tracks were visible at higher position in the 131 Å band
the flare for about four hours. After 15:00 UT the SADs be-
than in the 94 Å band. Both bands have the significant sensi- came virtually invisible. Since the SADs were visible as voids
tivity to hot flare plasma. However,the 131 Å band’smaximal inthesupra-arcadehotregion,theirvisibilityduringthelatede-
sensitivity for hot plasma is around 11−12 MK while for the cayphaseoftheflaremaybehinderedbyaverylowbrightness
94 Å band it is around 7−8 MK. The DD maps in these two oftheSAHR.Both,theSAHRandtheCXS,vanishedsoonaf-
bandsgivecomplementaryinformationabouthotplasmaofthe ter the SADs, about16–17 UT. Thus, a presence of the SAHR
SAHRandthereforeabouttheSADswhichwerevisibleasvoids andtheCXScoincidesintimewiththeSADs.Ifthedownflows
Articlenumber,page8of16
S.Kołoman´ski etal.:FinestructureandlongdurationofaflarecoronalX-raysourcewithRHESSIandSDO/AIAdata
Fig.5.Leftcolumn:PIXONimagesoftherealX-rayemissionoftheSOL2011-10-22T11:10flarereconstructedforthe6-7,6.5-7.5,and7-8keV
energybands,andforthetimeinterval12:35-12:37UT.Rightcolumn:PIXONimagesreconstructedforthesameenergybandsandgridsbutfor
simulateddistributionofX-raysources(syntheticCXS)showninFig.4(therightimage).
Articlenumber,page9of16
A&Aproofs:manuscriptno.kolomanski-mrozek-chmielewska
Fig.6.Comparisonofthedifferencedynamic(DD)mapsobtainedfromtheSDO/AIAimagesmadeintheAIA193Å,131Åand94Åbands.
ThelastDDmapisthecombination ofthe131Åand94Åmaps.EachDDmapismadeforthesametimeperiodandthecut2(seeFig.2).
Supra-arcadedownflows(SADs)arevisibleasbrightanddarktracksmovingdownwards.Darkandbrightsmudges,visiblebelowtheSADs,are
thetopsofrisingpost-reconnectionflareloops(PFLs).
are manifestation of magnetic reconnectionas they are consid- DeterminationofDEMisnotstraightforward.Observedin-
ered(Asaietal.2004;Khanetal.2007)thenthiscoincidencein tensitiesine.g.broadbandfilters, constitutethe convolutionof
timecanexplainthelong-lastingcoronalX-raysource.Thishot the response function of an instrument and DEM, additionally
source needs constant energy supply and the reconnection is a disturbed by errors. In result we deal with an ill-posed inverse
sourceoftheenergy.SeeSect.4.fordetaileddiscussionofthis problem (Tikhonov 1963; Berteroetal. 1985; Craig&Brown
topic. 1986; Schmittetal. 1996; Pratoetal. 2006). Thus, we can not
be sure that an obtained solution is actual. In order to increase
reliabilityoftheobtainedDEMitisrecommendedtousemore
3.3.CXSandSAHRthermalstructure than one method. For this reason we calculated DEM with the
useoftwomethods.
The first method (hereafter method-1) applies the
A comparison of RHESSI and AIA images indicates that what
xrt_dem_iterative2.pro routine in SSW package
the first instrument sees as CXS during the decay phase of the
(Weberetal. 2004; Golubatal. 2004) modified for use
flareisapartofthesupra-arcadehotregionobservedbythesec-
withthe AIA filters(see detailsinthe appendixof Chengetal.
ondone.Hence,a questionarises. WhytheCXSwas observed
(2012)). This is an iterative forward fitting method in which
inthelowerpartoftheSAHR?Toanswerthequestionweneed
the algorithm minimizes differences between the fitted and
to determine the thermal structure of the flare above the PFLs.
the observed intensities measured in the six EUV AIA bands.
Firstly,wecanlookatthepositionoftheCXScentroidsmarked
Errors in the method-1 are estimated using the Monte Carlo
in Fig. 7. The relation the higher the energy the higher the al-
(MC) approach. Each of 100 MC simulations is disturbed by
titude is not clearly shown by the centroids. Such a behavior
a randomly drawn noise within an uncertainty in the observed
may be caused by an absence of distinct vertical stratification
flux in each AIA band. The uncertainty is computed using
ofthetemperatureintheSAHR(namely,thehigherthealtitude
aia_bp_estimate_error.proprocedureinSSWpackage.
thehigherthetemperature).Temperaturedistributionintheana-
lyzedflareshouldbeknowntoverifythissupposition. The second used method, regularized inversion technique
(hereafter method-2), was introduced by Hannah&Kontar
Using a proper set of AIA images such distribution in (2012).Themethodiscomputationallyveryfastandcalculates
a form of differential emission measure (DEM) distribution also both the vertical (emission measure) and horizontal (tem-
can be prepared. In order to calculate the DEM in tem- perature)errorbars.Inbothmethodserrorsarehigherfor(a)the
perature range logT = 5.5 − 7.5, the six EUV bands of verylowsignalinsixusedAIAbandsandfor(b)temperatures
AIA are used, i.e. 131 Å (peak of temperature response log closetoedgesofrangeofthetemperatureresponsefunctionsof
T=7.05, O’Dwyeretal. (2010)), 94 Å (log T=6.85), 335 Å theseAIAbands(1-20MK).
(log T=6.45), 211 Å (log T=6.30), 193 Å (log T=6.20), Again, for a detailed analysis we took AIA observations
and 171 Å (log T= 5.85). The EUV band 304 Å is op- taken at 12:35 UT, i.e. the same time for which we analyzed
tically thick and has a small response to flare-like temper- a structureof the CXS. Using the method-1we preparedDEM
atures, thus is disregarded. The input data are processed maps for the selected time in the following way. Firstly, AIA
to the level 1.6 by deconvolving the point spread function imageswereresizeddownbyfactoroftwo,i.e.eachpixelinre-
withtheaia_deconvolved_richardsonlucy.proprocedure sizedimageissumofabox2by2originalpixels.Thisreduces
in SSW package before data calibration by aia_prep.pro. spatial resolution but simultaneously reduces also an influence
Checking of co-alignment of images is made by using ofnoiseonDEM.Secondly,DEMwascalculatedforeachnew
aia_coalign_test.proroutine. pixelbased on intensities measured in six EUV bands. The re-
Articlenumber,page10of16