No Trace Left Behind: Stereo Observation of a Coronal Mass Ejection without Low Coronal Signatures
aa r X i v : . [ a s t r o - ph . S R ] M a y Draft version November 1, 2018
Preprint typeset using L A TEX style emulateapj v. 08/22/09
NO TRACE LEFT BEHIND:STEREO OBSERVATION OF A CORONAL MASS EJECTION WITHOUT LOW CORONAL SIGNATURES
Eva Robbrecht , Spiros Patsourakos and Angelos Vourlidas Draft version November 1, 2018
ABSTRACTThe availability of high quality synoptic observations of the EUV and visible corona during theSOHO mission has advanced our understanding of the low corona manifestations of CMEs. TheEUV imager/white light coronagraph connection has been proven so powerful, it is routinely assumedthat if no EUV signatures are present when a CME is observed by a coronagraph, then the eventmust originate behind the visible limb. This assumption carries strong implications for space weatherforecasting but has not been put to the test. This paper presents the first detailed analysis of afrontside, large-scale CME that has no obvious counterparts in the low corona as observed in EUVand H α wavelengths. The event was observed by the SECCHI instruments onboard the STEREOmission. The COR2A coronagraph observed a slow flux-rope type CME, while an extremely faintpartial halo was observed in COR2B. The event evolved very slowly and is typical of the streamer-blowout CME class. EUVI A 171 ˚A images show a concave feature above the east limb, relativelystable for about two days before the eruption, when it rises into the coronagraphic fields and developsinto the core of the CME. None of the typical low corona signatures of a CME (flaring, EUV dimming,filament eruption, waves) were observed in the EUVI-B images, which we attribute to the unusuallylarge height from which the flux-rope lifted off. This interpretation is supported by the CME massmeasurements and estimates of the expected EUV dimming intensity. Only thanks to the availabilityof the two viewpoints we were able to identify the likely source region. The event originated along aneutral line over the quiet sun. No active regions were present anywhere on the visible (from STEREOB) face of the disk. Leaving no trace behind on the solar disk, this observation shows unambiguouslythat a CME eruption does not need to have clear on-disk signatures. Also it sheds light on the questionof ‘mystery’ geomagnetic storms, storms without clear solar origin (formerly called problem storms).We discuss the implications for space weather monitoring. Preliminary inspection of STEREO dataindicates that events like this are not uncommon, particularly during the ongoing period of deep solarminimum. Subject headings:
Sun: coronal mass ejections (CMEs), Sun: activity, Sun: streamer, Sun: cavity,Space weather INTRODUCTION
The relationship of coronal mass ejections (CMEs) toother forms of solar activity has been the subject of nu-merous studies (see Pick et al. 2006). Nevertheless, nosimple causal relation has been found and it seems thatno such simple relation exists. CMEs were discoveredin the early 1970’s with space-borne white light coro-nagraphs (Tousey 1973; Gosling et al. 1974), long afterflares and prominences had been observed. It was as-sumed, then, that the CMEs were simply a product of aflare and/or filament eruption. As CME observations be-came more common, this causal relationship came underdispute (Gosling 1993; Harrison 1996) with importantimplications for Sun-Earth studies. Up to that point,flares and H α filament disappearances had been inter-preted as the direct cause of large nonrecurrent geomag-netic storms. We now know that nonrecurrent geomag-netic storms are due to the interaction of the magneticfield of CMEs with the terrestrial magnetic field. Whilethe causal relationship among these forms of solar ac- Electronic address: [email protected] George Mason University, 4400 University Dr., Fairfax, VA22030, USA Naval Research Laboratory, 4555 Overlook Ave SW, Washing-ton, DC 20375, USA tivity is still under debate, there is a strong belief inthe community (including the operational space weathercommunity) that CMEs are strongly intertwined withflares and prominence eruptions to such an extent thatone expects to observe either or one of them whenevera CME is observed. For example long duration flares,EUV dimming regions and filament eruptions are rou-tinely used by the NOAA forcasters as proxies for de-termining the source region and probable propagationdirection of a CME. The recent emphasis on the EUVlow corona counterparts of CMEs has actually led to theadoption of an EUV full disk imager (but without a coro-nagraph) for the next generation of operational satellites.But if this is to be a reliable operational strategy, CMEsmust always have discernible low corona counterparts. Isthis true?There is enough evidence to believe that CMEs arebest correlated with erupting prominences and fila-ments (Munro et al. 1979; Webb & Hundhausen 1987;Alexander 2006, and references therein). These comprisequiet sun as well as active region prominences and fila-ments. However, the reverse is generally not true: fil-aments can disappear thermally and prominences oftenerupt in a confined fashion, thus not leading to a CME.On the other hand, flares are more numerous than CMEs. E. Robbrecht et al.
Fig. 1.—
CME shown in background removed composite imagefrom the STEREO A spacecraft. The ‘ × ’ indicates the featurewe tracked from EUVI through COR2. The two inserts show theeast limb of the Sun in COR1A before and after the eruption (notto scale with the larger image). The first insert (left) shows abackground subtracted image, the contours indicate the positionof the streamer. The second insert (right) is a difference imagewith the pre-event image subtracted, showing the disappearanceof the southern part of the helmet streamer. Almost all long duration events (LDE) have an associ-ated CME, because the LDE is caused by the heatingof loops due to reconnection below the CME. Again,the reverse is not true. There are large CMEs with-out flaring and large flares (even X-class) without CMEs(e.g. Feynman & Hundhausen 1994). SOHO observa-tions have shown EUV dimmings associated with CMEs(e.g. Thompson et al. 1998). It is generally believed thatthe dimming is due to the evacuation of coronal massand as such it is a reliable proxy for a CME. Dimmingshad been observed also in soft X-ray images from Skylab(Rust & Hildner 1976) and Yohkoh (Sterling & Hudson1997). Bewsher et al. (2008) confirmed the close rela-tionship between CMEs and CDS dimmings, but againno one-to-one correspondence was found; only up to 84%of the CMEs could be traced back to a CDS dimming.For the majority of CMEs, especially fast ones, it isgenerally easy to identify a number of associated lowcoronal and chromospheric signatures. But there existseveral examples of white light halo CMEs with no suchassociation, even though the in situ data suggests arrivalat Earth. Using a comprehensive set of data, Zhang et al.(2003, 2007) and Schwenn et al. (2005) searched forsource regions of geomagnetic storms (respectively withDst ≤ −
100 nT and Dst ≤ −
50 nT). The identifica-tion process itself did not seem straightforward: onlyhalf of the geomagnetic storms had a clear associationwith a unique disk-signature and CME. The remain-der of the storms had multiple candidate sources or nocandidate. 11% of the storms studied by Zhang et al.(2007) could not be linked to a signature on the disk,but they were all caused by slow partial halo CMEs.Schwenn et al. (2005) reported that about 20% of thegeoeffective ICMEs were not preceded by an identifiablefrontside halo CME (they used SOHO/EIT data to sep-arate backsided CMEs). The standard explanation has always been that halo CMEs lacking on-disk signaturesmust be backsided and were catalogued as such. Whilethis is correct for some events, there have been casesof geomagnetic storms associated with apparently back-sided CMEs (e.g. in Schwenn et al. 2005). These as-sociations have always been controversial because it ishard to imagine how a CME directed away from the Sun-Earth line could have any geoeffective potential. Appro-priately, these geomagnetic storms are called ‘problemstorms’. An early example is the January 6-10 1997 event(Webb et al. 1998) for which the corresponding whitelight halo CME was only identified post-facto. Two otherearly examples of problem storms were the April 22-231997 and the June 9 1997 storms that reached a Dst of-107 nT and -84 nT, respectively. These storms weredriven by magnetic clouds with flux-rope characteristics,but could not be associated with any frontside (halo)CME (Webb et al. 2000). The problem would disappearif the origins of these storms have been misidentified sim-ply due to the lack of observable EUV or other low coro-nal signatures. The implications for space weather stud-ies are obvious.In this paper, we analyze STEREO/SECCHI observa-tions of a streamer blowout CME without a clear sourceregion other than the quiet sun. Thanks to the wide an-gle separation (53 ◦ ) of the STEREO spacecraft we wereable to study the CME and its source region edge-on inSTEREO A and face-on in STEREO B ( § ≈ −
70 nT (Yan Li, private communication).This event is thus a good example of a ‘problem storm’.In STEREO B only an extremely faint halo CME wasobserved lacking any obvious disk counterparts. We sug-gest that the CME originated from high in the coronaand therefore caused no observable dimming. We con-clude in § STEREO OBSERVATION AND MEASUREMENTS
Our analysis is based on data from the Extreme Ultra-Violet Imaging Telescope (EUVI) and white-light imagesfrom the COR1 and COR2 coronagraphs onboard SEC-CHI (Howard et al. 2008). Details of the instruments aregiven in Table 1. The COR1 and COR2 images are to-tal brightness images. Figure 1 is a composite image ofthe SECCHI A observations (a composite movie is avail-able online). Following a streamer swelling that lastedfor about two days, the CME entered the COR2 field ofview on June 2, 2008. Figure 2 shows a running differ-ence snapshot of the event taken in the two COR2 tele-scopes (as in all the Figures in this paper, the STEREOA perspective is on the right). The CME has a classicalflux-rope morphology (Vourlidas et al. 2000) in COR2Aand is an extremely faint halo in COR2B. Halo CMEs arefaint because the Thomson scattering, which forms thewhite-light image, is most effective in the plane of the sky.Therefore Figure 2 immediately suggests that the CMETEREO observation of a CME without disk signature 3
Fig. 2.—
SECCHI/COR2 observations of the June 2, 2008 CME. The images are running differences of total brightness images. Only avery faint halo was observed in COR2B (better seen in the movie). To enhance the contrast in the COR2B image we averaged two imagesbefore subtracting a previous image. The times at which the original images were taken are printed at the bottom in each figure.The blackcircles mark the position and size of the Sun. The white sector in the COR2A image indicates the region used to calculate the CME mass(see § Fig. 3.—
Potential Field Source Surface magnetic field extrapolations based on MWO data. The extrapolations are centered as viewedfrom STEREO B (left) and STEREO A (right) on May 31, 2008 00:00 UT. The black box marks the probable source region, it is the samebox as the one in Figure 4. Green (blue) refers to open field lines of negative (positive) polarity, and orange (red) refers to long (short)closed field lines: the red field lines reach heights up to 1.5 R ⊙ , the yellow field lines between 1.5 R ⊙ and 2.5 R ⊙ . At the photosphere,light grey areas show positive magnetic flux (0 < B r <
10 G), and dark grey areas show negative flux ( − < B r < erupted close to the plane-of-sky (POS) of STEREO A.The CME erupted from below a helmet streamer atthe east limb, as seen by the STEREO A coronagraphs.The two inserts in Figure 1 show the initial streamerand its partial disappearance after the CME erupted.The streamer swelling, its disappearance, slow evolutionand flux-rope structure clearly identify this event as astreamer-blowout CME (Vourlidas et al. 2002). The onlypartial disappearance of the streamer can give us a clue as to the origin of the CME with the help of the poten-tial field source surface (PFSS) magnetic field extrapo-lation shown in Figure 3. The extrapolation method isdescribed in Wang & Sheeley (1992, 1995). The preex-isting helmet streamer can be readily identified with theoverlying closed (red and orange) field lines. The partthat erupted overlies the red field lines that close belowa height of 1.5 R ⊙ . These field lines only occupy thelower latitudes of the streamer. E. Robbrecht et al. TABLE 1STEREO/SECCHI instrument details (References are in text)
Instrument
EUVI COR1 COR2
Type
EUV Telescope Coronagraph Coronagraph
Bandpass
171 ˚A 195 ˚A 284 ˚A 304 ˚A White-Light (650-660 nm) White-Light (650-750 nm)
Cadence
FOV from sun center − . R ⊙ . − R ⊙ . − R ⊙ Pixel Size
Figure 4 shows the full disk corona in 171 ˚A prior tothe eruption. This is a typical solar minimum coronalacking big active regions, and is dominated by the smallscales characteristic for the quiet sun. The arrow in-dicates the bright structure that traveled outward andeventually developed into the CME core. The morphol-ogy of the feature (concave shape, bright core) suggeststhat it is the bottom of a flux-rope, so large that the topis outside the field of view of the EUVI image. We drawa parallel with the cavity/bright rim that is observed inthe north west (top right) corner of the EUVI B image.To compare the sizes, we measured the widths of bothcavities, similarly to Gibson et al. (2006). The cavitieswere measured in the EUVI 171 ˚A images at 1.15 R ⊙ from sun center and resulted in widths of around 25 ◦ and 11.5 ◦ , for the A and B cavity, respectively. We thusestimate that the cavity of interest here is a factor ∼ . § Estimation of the true CME direction ofpropagation
At the time of the eruption, the separation angle be-tween the STEREO A and B spacecraft was 53 degrees.This separation allowed us to observe the CME simul-taneously edge on (in A) and face on (in B). Figure 5shows a schematic representation of the position of theCME relative to the STEREO spacecraft. We used sev-eral methods to derive the true propagation direction ofthe CME: • The observation of a bright CME in the COR2Aand an extremely faint halo in the COR2B images(see Figure 2) means that the CME propagatedtowards STEREO B and lies close to the POS ofSTEREO A. • Polarization analysis of COR2 polarized brightness(pB) images suggests that the CME lay in the POSof STEREO A ± ◦ . • Applying a forward modeling technique to theCME, Thernisien et al. (2009) estimate an angle of26 ± ◦ out of the POS of STEREO A (frontsided). • The in situ observation of the arrival of a MC onJune 6 in STEREO B (Yan Li, private communica-tion) confirms that the CME propagation path hada large component in the STEREO B direction.From the above, we estimate that the CME propagatedat approximately 40 ◦ east from the Sun-Earth line. At the time of maximum acceleration (around 20:00 UT on1 June 2008) this direction corresponds to a Carringtonlongitude of 65 ◦ . This also gives us an initial estimate ofthe position of the source region. The derived Carringtonlongitude corresponds to the position of the east limb asseen from STEREO A at 00:00 UT on 31 May 2008 (seeFigure 4) and suggests indeed that this CME eruptedfrom the pre-existing flux-rope indicated with the whitearrow. CME kinematics derived from STEREO A
During its evolution, the CME front is barely visiblein COR1, but it can be clearly seen when it enters theCOR2 field-of-view (FOV) around 19:00 UT on 1 June2008. The evolution of the event in the EUVI-COR1-COR2 combined FOV spans more than three days start-ing on 31 May 2008. The slow evolution of the eventhas allowed us to measure the kinematics of the eventin great detail (Figure 6, a-b). Because of the lack ofa front in the COR1 images, we traced the back of theCME core, which is the best obsered feature in all threeinstruments. It is indicated with an ‘ × ’ in Figure 1. Be-cause of the CME expansion the core travels slower thanthe leading edge. The same feature can be seen off-limbin EUVI A (Figure 7) as it leaves the Sun. We only showthe 171 ˚A component where it is best observed. It is alsovisible in 195 ˚A, but not in the other 2 channels. We canidentify three stages in the CME evolution: (1) a slowrise phase that starts around 20:36 UT on 31 May 2008,(2) a quasi-constant acceleration phase starting around21:00 UT on 1 June 2008 and (3) a constant velocityphase. The asymptotic velocity of around 300 km/s isonly reached at around 20 R ⊙ in the STEREO/HI (He-liospheric Imager) FOV. Here we focus on the first twophases.Several hours before the actual eruption, EUVI A im-ages show off-limb a helical structure rolling around itsaxis (see top panel in Figure 7 and online movie). Thisactivity may be interpreted as flux-rope activation orformation. Then a concave structure below the helicalstructure starts to rise very slowly in the EUVI FOVat around 10 km/s (shown in the next two panels inFigure 7). This quasi-steady phase can be best approx-imated by a linear fit, until the feature reaches ∼ ⊙ .During this phase, the white light streamer brightensand swells. Once in the coronagraph FOV, the featurestarts to accelerate gradually until it reaches a velocityof around 200 km/s at 13 R ⊙ . At this stage it has devel-oped into the bright core of the CME. We find that thisacceleration phase is well described by a second ordercurve (red curve in Figure 6, a-b).Under certain conditions different initiation modelspredict different height-time profiles (e.g. Schrijver et al.TEREO observation of a CME without disk signature 5 Fig. 4.—
Simultaneous images taken by EUVI A and B showing the corona in the 171 ˚A line before the start of the eruption. Thefield-of-view in both images is cropped at 1.35 R ⊙ (from suncenter). The white box marks the probable source region of the CME, itextends from 30 ◦ to 100 ◦ in Carrington longitude and from − ◦ to 0 ◦ in latitude. The arrow indicates a bright structure, which is thebottom part of the erupting flux-rope. The images were contrast enhanced using a wavelet algorithm (adapted from Stenborg et al. 2008). Fig. 5.—
Schematic view of the CME direction projected onthe ecliptic plane (top view). We assume a circular cone modelwith an opening angle equal to the CME width as measured fromSTEREO A (54 degrees). The separation angle between the A andB spacecraft was 53 degrees.
Localization of the source region in EUVI B
From our estimation of the true propagation direction( § ◦ to 100 ◦ in Carrington longitudeand from − ◦ to 0 ◦ in latitude. The latitudinal ex-tent of the box is derived from the edge-on view seen inEUVI A. In longitudinal direction, the box is centered at65 ◦ , which is the direction of the CME derived in § α just after the CMEerupted. Its shape is outlined in turquoise in the bottom E. Robbrecht et al.frames of Figure 7. The filament forms fast, is not veryconspicuous and disappears quickly between 21:00 and22:00 UT (S. Martin, private communication). Whilethis filament is not related to the observed CME, it indi-cates the presence of a filament channel at the time of theeruption (e.g. Martin 1998). Further, a small dimmingarea is indicated (magenta box in f5). While these smallfeatures may have a relation to the observed CME, theirscale is much smaller than the CME itself. Moreover, themovie illustrates that similar activity is observed outsidethe box, thus it is difficult to decide which features areconnected to the observed CME.The most surprising result is the lack of observation ofan eruptive filament in the EUVI B images. We expectedto see something given the clear helical structure seen inboth EUVI A and COR2A. To make sure, we examinedH α images from varied sources, but no H α filament wasobserved prior to the CME. H α images are only availablefrom Earth, which solar view is between that of STEREOA and B. On May 31, 2008 00:00 UT, the east limb ofthe sun, as seen from the Earth, has Carrington longi-tude 39 . ◦ , which is 25 . ◦ east of 65 ◦ , our initial estimatefor the source region location from § α images. CME mass measurement
Because of the detailed coronagraph observations inSTEREO A, we can assess the height from which theCME material originates by measuring the mass flux inthe COR1 and COR2 fields of view (shown in Figure 6(d)). The CME mass is estimated by integrating over aregion of interest (ROI), the excess brightness of a givenimage relative to a pre-event image. We defined the ROIas a sector bounded by the CME edges in position angleand by 1.8 - 3.7 R ⊙ in COR1 and 3 - 14 R ⊙ in COR2 inradial direction (the COR2 sector is drawn in Figure2),thus the curves in Figure 6 (d) represent the mass evolu-tion in these ROIs. The maximum mass measurement inthe COR2 FOV is a representative number for the totalCME mass, being ∼ . × g which is normal for suchevents (e.g. Vourlidas et al. 2002). The COR1 FOV istoo small (in radial direction) to capture the whole CMEat one instant of time, therefore the COR1 curve doesnot reach the full CME mass. To obtain a rough cross-calibration between the two instruments, we comparedthe mass measurement in a common sector (from 2.4 to3.7 R ⊙ ) while the CME traveled through this ROI. TheCOR1 mass curve was on average 20% lower than theCOR2 mass curve.The streamer swelling is represented by the initial slowrise in the mass curves. When the CME enters the COR2FOV a sharper increase can be seen, until the leadingedge of the CME reaches the outer edge of the COR2ROI after which the mass decreases back to its pre-eventlevel. As we can see from the dashed curve, the coronain the COR1 FOV does not recover its pre-event levelshowing an apparent ‘negative mass’. This indicates thatthe COR1 corona was depleted. As a check, we did thesame exercise for the 26 April 2008 CME, which wasnot a streamer blowout and had a clear source regionand dimming. The COR1 mass curve for that event didreturn back to its pre-event level, implying that most ofthe CME mass originated from below the COR1 FOV. Fig. 6.— (a) Height (from sun center) and (b) speed profile ofthe CME. The feature indicated with ‘x’ in Figure 1 is traced fromEUVI-171 to COR2. Three functions (second order, exponentialand power law) are fitted to the acceleration phase. The initialslow rise is fitted with a constant speed profile. (c) The GOESX-ray flux shows no flaring activity during the period of interest.(d) CME mass as a function of time through the STEREO A coro-nagraphs FOV. The mass is calculated by integrating the electrondensity in a fixed sector in base-difference images (illustrated inFigure 2). When the CME leaves the COR1 FOV (dashed line)a mass depletion (‘negative mass’) can be observed indicating apartial streamer blowout.
Thus, we can say with confidence that a large fraction ofthe CME mass in our event originated from the COR1FOV.Standard assumptions in mass measurements are (1)that all of the CME material is located in the planeof the sky (therefore, it gives a lower limit for the trueCME mass) and (2) that the corona is completely ion-ized (Vourlidas et al. 2000). Since our CME lies closeto the plane of the sky of STEREO A (where the massmeasurements are made) these measurements are closeto the real values.TEREO observation of a CME without disk signature 7
Fig. 7.—
Five snapshots from EUVI 171 ˚A illustrate the low-coronal evolution during the event (A is on the right, B is on theleft, time runs from top to bottom). The images were contrast en-hanced using a wavelet technique (based on Stenborg et al. 2008).Several stages are seen in EUVI A: helical motion in f1 (white ar-row), rising flux-rope in f2 and f3 and detaching flux-rope f4 andf5. Movies are made available on the website. The white box is thesame as in Figure 4. Several small features are indicated in color:orange in f2: a faint filament that was observed in EUVI A in 304˚A (the footpoint is indicated in B), yellow in f3: a small signatureof heating, possibly as part of the eruption, magenta in f5: smalldimming or appearance of coronal hole and finally turquoise in f5:a small filament appeared and disappeared quickly in H α . Source region size estimation
Arguably, the most intriguing aspect of our observa-tions is the lack of a well-defined source region of theevent in EUVI B data (see § L SR and across θ SR the neutral line. Often the length of the eruptingfilament is used as a proxy for the length of the neu-tral line. Based on a large statistical survey of observa-tions of erupting filaments and their associated CMEs,Cremades & Bothmer (2004) found that the observedlengths of the filaments ranged from a few degrees upto more than 40 ◦ and show a Poisson-type distributionpeaking in the range 6 ◦ - 12 ◦ . We do not directly observea filament and therefore cannnot measure its length, butfrom the edge-on orientation of the CME and the PFSSmagnetic field extrapolation we estimate that the ori-entation of the neutral line in question is more or lessin the longitudinal direction. Assuming that the twosmall features indicated in Figure 7 were both related tothe CME, their longitudinal separation gives us a lowerlimit of L SR ≈ ◦ . A second estimation can be madeusing the 3D information derived from a forward mod-eling technique (Thernisien et al. 2009). This techniquemodels the CME as a hollow flux-tube (see schematicview in their Figure 1). The separation angle 2 α be-tween the legs of the flux-tube and the aspect ratio κ ofthe tube give an upper limit for the length of the sourceregion. From the values given by Thernisien et al. (2009)for our source region (see their Table 1b), we find then L SR ≤ α + arctan( κ ) ≈ ± ◦ .Another interesting quantity is the size θ SR of thesource region across the neutral line, which should becommensurate to the footpoint-separation of the associ-ated erupting arcade. Moore et al. (2007) determined ascaling-law relating the CME width θ CME (which is whatwe observe edge-on in EUVI A) to the size of the associ-ated erupting arcade. θ CME corresponds to the CME an-gular span when the CME is fully developed in the COR2FOV and has reached lateral pressure balance with thesurrounding magnetic fields. This is thought to occurin the outer corona (i.e > ⊙ ). The scaling-law (e.g.Equation 20 in Moore et al. 2007) can be written as θ SR ≈ r . < B r > θ CME · (1) < B r > is the absolute radial magnetic field averagedover the extended quiet sun area enclosed within thesmall box of Figure 4. A full disk photospheric mag-netogram from SOLIS (Keller et al. 2003) was used forthis calculation (registered from 20h55 to 21h42 on June1st 2008). We found < B r > ≈ θ CME ≈ ◦ as taken from theonline CACTus catalog (Robbrecht & Berghmans 2004).Using these values, we find a source region size θ SR ≈ ◦ .This value is comparable to the latitudinal width of thebox we derived at first sight from the edge-on view inEUVI A in Figure 4. This width is commensurate withthe footpoint separation of the red magnetic field lines(see Figure 3) that erupted.Summarizing the above, we deduced a source regionsize of about 36 ◦ × ◦ . These values of the source regionshould be viewed only as estimates. They neverthelesspoint to a rather extensive source area. The main conclu-sion is that we expect a large source region, somethingwhich is in concert with the inferences of a large flux- E. Robbrecht et al.rope mentioned in §
2. More refined calculations couldpin down further the dimensions of the elusive sourceregion. For example, we plan to determine the detailedenergetics of our event as well as the magnetic properties(e.g. fluxes) of the associated magnetic cloud. These val-ues would allow some further estimates of the size of thesource region, to be deduced from the conservation of thetotal CME energy and of the magnetic flux in magneticclouds respectively. DISCUSSION
The analysis in § Flaring hypothesis
Flaring is usually expected for CMEs associated withactive regions. As is obvious from Figure 7, no activeregion was present anywhere on the disk as seen fromEUVI B. The CME originated in the quiet sun and thelack of flaring is not surprising. Besides energetic flares,polar crown filament eruptions are often accompanied bylong post eruptive arcades seen in EUV and soft X-rays.This type of flaring was not observed for our event, eitheron-disk (e.g. as in Hudson et al. 1995) or off-limb (e.g. asin Sheeley et al. 2007) which could imply that the fieldis closing at a large height (small densities) and thatthe energy release was small resulting in weak heatingunobservable by EUVI.
Prominence hypothesis
The kinematics of our CME and its morphology char-acterize it as a gradual event usually linked to a promi-nence eruption. Why then is there no clear prominencevisible in the EUVI B images, nor in H α ? EUVI A im-ages provide an important clue: they show a very highflux-rope (as derived in § α filamentis not a necessary condition for the eruption. Filamentchannels coincide with polarity inversion lines and whenchromospheric mass loads into the channel, a filamentis formed. As most CME initiation theories, models andobservations suggest, the important agent for an eruptionis the existence of a filament channel only, not the chro-mospheric mass. For example, Lin (2004) pointed outthat the magnetic configuration of the filament channelis more important than any mass loading for CME initi-ation. However, there is a widespread belief that H α fil-ament disappearances are reliable proxies for CMEs. Wewould like to caution observers and space weather fore-casters against an over reliance on H α data for disk sig-natures of eruptions. Our observation shows that CMEscan erupt without having a mass-loaded (and thus ob-servable) filament. Fig. 8.—
Calculation of the EUV dimming due to coronal mate-rial removed by a hypothetical CME. The curve shows the ratio ofEUV intensities I CME /I BG as a function of the starting height ofthe CME (measured from sun center). A ratio below one indicatesan observable dimming. As can be seen, no dimming is expectedfor CMEs originating from above 1.4 R ⊙ . Dimming hypothesis
Finally, we address the lack of a large-scale dimmingsignature in the EUV. Estimates of the mass associatedwith EUV dimmings showed it can represent a signifi-cant fraction of the total CME mass (e.g. Harrison et al.2003; Zhukov & Auch`ere 2004). If the CME originatedfrom deep in the corona carrying some part of it, wewould have expected to see a large EUV dimming com-mensurate to the amount of mass carried off by the CME.The electron density, n e ( r ) drops dramatically with ra-dial distance from the solar surface. The EUV intensities,proportional to n e ( r ), exhibit an even stronger fall-off. Asmall difference in initiation height of the ejected mate-rial can give a large difference in the signal of the coronaldimming. Conversely, the absence of a dimming suggestsa large initiation height for our CME. To demonstratethis, we calculated proxies of the EUV intensities before, I BG , and during, I CME , the eruption of a hypotheticalCME with its source region located on disk center. Thebackground intensity per pixel is defined as I BG ≡ Z ∞ R ⊙ n e ( r ) dr ≈ Z R ⊙ R ⊙ n e ( r ) dr, (2)where r is distance from sun center. Similarly, the EUVintensity corresponding to the portion of the corona thatis removed by the CME corresponds to I CME ≈ Z R ⊙ r n e ( r ) dr, (3)assuming a starting height of r for the CME. A ratio I CME /I BG significantly below one implies an intensity de-pletion caused by the removal of part of the corona by theCME. If this depletion occurs simultaneously over a largeTEREO observation of a CME without disk signature 9set of continuous observational pixels then we observe adimming.Figure 8 shows this ratio for starting heights r in therange 1 - 2 R ⊙ . For n e ( r ), we used the density profiles ofBaumbach-Allen (Aschwanden 2005; Gibson et al. 1999;Guhathakurta et al. 1996). The first profile (solid line)corresponds to an ‘average’ corona, whereas the othertwo correspond to coronal streamers. As expected, thedimming ratio I CME /I BG is appreciably smaller than oneonly for starting heights below 1.4 R ⊙ . For CMEs start-ing higher up in the corona, this ratio is approximatelyone, meaning that for those CMEs no observable coro-nal dimming can be expected. Conversely, an absenceof dimming implies that most of the plasma that con-tributed to the CME mass originated from above 1.4 R ⊙ .For reference, we note that this height corresponds to theinner edge of the COR1 FOV. This limit is quite robustgiven that the dimming curves have little dependence onthe employed density profile as can be seen in Figure 8.Including temperature effects (through temperatureresponse functions R ( T )) into the determination of theEUV intensities would further compress the radial dis-tance range of sensitivity of I CME /I BG . This is becausefor a multi-thermal line of sight only points with tem-peratures around the peak of the R ( T ) of the employedchannel will contribute to the observed intensities. More-over, EUV and SXR observations show small or lit-tle variations of the electron temperature in streamers(e.g. Foley et al. 2002; Warren & Warshall 2002). Fi-nally note that a large-scale dimming, if present, wouldhave manifested at least in one or more of the four EUVIchannels. The response functions of these channels peakat 0.08, 0.90, 1.50 and 2.00 MK, which spans the bulkof the quiet sun temperature domain (e.g. Brosius et al.1996). We found no evidence of significant flaring, whichmeans that no large amount of plasma was heated offquiet sun conditions. Scenario of a large flux-rope
To summarize, we offer the following scenario basedon the above discussion. The event occurs over a largelyempty filament channel, hence no or weak emission in304 ˚A or H α , on the quiet sun. The flux-rope, thateventually forms the CME core, is visible for at least twodays prior to the eruption in emission in the coronal lines(171 and 195 ˚A). We deduce that the flux-rope mostlyconsists of hot material of about 1 MK and the material isconcentrated at the bottom of the feature. The flux-ropeis also situated at 0.15 R ⊙ (bottom) above the surface,which is unusually high. We illustrated the large heightof the overlying loop system, by comparing the cavitybeneath it with the smaller cavity that is visible in EUVIB, and found that our cavity is 2.2 times larger. Thisis further corroborated by the magnetic field propertiesof the postulated source region. It shows that most ofthe overlying field lines have widely separated footpoints,comparable to the widths derived in § CONCLUSION AND IMPLICATIONS
To conclude, we find that the CME erupted from thequiet sun along a polarity inversion line. The very lowCME speed ( <
300 km/s) is in concert with the weakphotospheric field ( < ⊙ and the overlyingloop system exceeds 1.4 R ⊙ (both measured from suncenter). This interpretation is supported by the CMEmass measurements and estimates of the expected EUVdimming intensity.Overall, this event is a typical streamer blowout CMEin terms of its evolution and physical parameters. How-ever, the multi-viewpoint analysis of this observation re-sults in some very important implications:1. We have unambiguously shown that large CMEsare not necessarily associated with clear lowcoronal or chromospheric features. Their disk-signatures may be weak or undetectable. There-fore, the lack of an obvious on-disk signature doesnot imply that a (partial) halo-CME is backsided,as has been assumed in numerous studies. The useof on-disk EUV or H α imaging as proxies of CMEsin the low corona cannot be considered as fully re-liable for operational purposes.2. A CME can erupt from the quiet Sun where thefield is weak. Our observation shows that no largefilament or active region needs to be present in thepre-CME corona in order to initiate an eruption.The magnetic field configuration itself is more im-portant than the plasma for studying CME initia-tion. Correlations of CMEs with prominences andflares will therefore vary depending on what in-struments are used. Imaging instruments can onlyshow structures that contain enough plasma (atthe ‘right’ temperature), for example active regionsand their loops, but they are unable to track tenu-ous features like filament channels. Vector magne-tograms, preferably in the upper chromosphere orcorona, could be used to detect magnetic configu-rations that can drive CMEs.3. This event is a good example of a ‘problem storm’.If this CME had been directed at Earth, but withsouthward Bz, it could have generated a moderategeomagnetic storm. Previous studies have shownthat a significant fraction ( > § × g). Toreach such a mass, the source region size of theseCMEs should be quite extended as we found in § REFERENCESAlexander, D. 2006, Space Science Reviews, 123, 81Aschwanden, M. J. 2005, Physics of the Solar Corona. AnIntroduction with Problems and Solutions (2nd edition) (Pourla Science)Bewsher, D., Harrison, R. A., & Brown, D. S. 2008, A&A, 478,897Brosius, J. W., Davila, J. M., Thomas, R. J., & Monsignori-Fossi,B. C. 1996, ApJS, 106, 143Cremades, H. & Bothmer, V. 2004, A&A, 422, 307Feynman, J. & Hundhausen, A. J. 1994, J. Geophys. Res., 99,8451Foley, C. R., Patsourakos, S., Culhane, J. L., & MacKay, D.2002, A&A, 381, 1049Forbes, T. G. 2000, J. Geophys. Res., 105, 23153Gibson, S. E., Fludra, A., Bagenal, F., Biesecker, D., Zanna,G. D., & Bromage, B. 1999, J. Geophys. Res., 104, 9691Gibson, S. E., Foster, D., Burkepile, J., de Toma, G., & Stanger,A. 2006, ApJ, 641, 590Gosling, J. T. 1993, J. Geophys. Res., 98, 18937Gosling, J. T., Hildner, E., MacQueen, R. M., Munro, R. H.,Poland, A. I., & Ross, C. L. 1974, J. Geophys. Res., 79, 4581Guhathakurta, M., Holzer, T. E., & MacQueen, R. M. 1996, ApJ,458, 817Harrison, R. A. 1996, Sol. Phys., 166, 441Harrison, R. A., Bryans, P., Simnett, G. M., & Lyons, M. 2003,A&A, 400, 1071Howard, R. A., Moses, J. D., Vourlidas, A., Newmark, J. S.,Socker, D. G., Plunkett, S. P., Korendyke, C. M., Cook, J. W.,Hurley, A., Davila, J. M., Thompson, W. T., St Cyr, O. C.,Mentzell, E., Mehalick, K., Lemen, J. R., Wuelser, J. P.,Duncan, D. W., Tarbell, T. D., Wolfson, C. J., Moore, A.,Harrison, R. A., Waltham, N. R., Lang, J., Davis, C. J., Eyles,C. J., Mapson-Menard, H., Simnett, G. M., Halain, J. P.,Defise, J. M., Mazy, E., Rochus, P., Mercier, R., Ravet, M. F.,Delmotte, F., Auchere, F., Delaboudiniere, J. P., Bothmer, V.,Deutsch, W., Wang, D., Rich, N., Cooper, S., Stephens, V.,Maahs, G., Baugh, R., McMullin, D., & Carter, T. 2008, SpaceScience Reviews, 136, 67Hudson, H., Haisch, B., & Strong, K. T. 1995, J. Geophys. Res.,100, 3473 Keller, C. U., Harvey, J. W., & Giampapa, M. S. 2003, in Societyof Photo-Optical Instrumentation Engineers (SPIE) ConferenceSeries, Vol. 4853, Society of Photo-Optical InstrumentationEngineers (SPIE) Conference Series, ed. S. L. Keil & S. V.Avakyan, 194–204Lin, J. 2004, Sol. Phys., 219, 169Martin, S. F. 1998, Sol. Phys., 182, 107Moore, R. L., Sterling, A. C., & Suess, S. T. 2007, ApJ, 668, 1221Munro, R. H., Gosling, J. T., Hildner, E., MacQueen, R. M.,Poland, A. I., & Ross, C. L. 1979, Sol. Phys., 61, 201Pick, M., Forbes, T. G., Mann, G., Cane, H. V., Chen, J.,Ciaravella, A., Cremades, H., Howard, R. A., Hudson, H. S.,Klassen, A., Klein, K. L., Lee, M. A., Linker, J. A., Maia, D.,Mikic, Z., Raymond, J. C., Reiner, M. J., Simnett, G. M.,Srivastava, N., Tripathi, D., Vainio, R., Vourlidas, A., Zhang,J., Zurbuchen, T. H., Sheeley, N. R., & Marqu´e, C. 2006, SpaceScience Reviews, 123, 341Robbrecht, E. & Berghmans, D. 2004, A&A, 425, 1097Rust, D. M. & Hildner, E. 1976, Sol. Phys., 48, 381Schrijver, C. J., Elmore, C., Kliem, B., T¨or¨ok, T., & Title, A. M.2008, ApJ, 674, 586Schwenn, R., dal Lago, A., Huttunen, E., & Gonzalez, W. D.2005, Annales Geophysicae, 23, 1033Sheeley, Jr., N. R., Warren, H. P., & Wang, Y.-M. 2007, ApJ,671, 926Stenborg, G., Vourlidas, A., & Howard, R. A. 2008, ApJ, 674,1201Sterling, A. C. & Hudson, H. S. 1997, ApJ, 491, L55Thernisien, A., Vourlidas, A., & Howard, R. A. 2009, Sol. Phys.,256, 111Thompson, B. J., Plunkett, S. P., Gurman, J. B., Newmark, J. S.,St. Cyr, O. C., & Michels, D. J. 1998, Geophys. Res. Lett., 25,2465Tousey, R. 1973, in Space Research, ed. M. J. Rycroft & S. K.Runcorn, 713Vourlidas, A., Howard, R. A., Morrill, J. S., & Munz, S. 2002, inSolar-Terrestrial Magnetic Activity and Space Environment,ed. H. Wang & R. Xu, 201Vourlidas, A., Subramanian, P., Dere, K. P., & Howard, R. A.2000, ApJ, 534, 456Wang, Y.-M. & Sheeley, Jr., N. R. 1992, ApJ, 392, 310—. 1995, ApJ, 447, L143+Warren, H. P. & Warshall, A. D. 2002, ApJ, 571, 999