Coronal Dimmings and the Early Phase of a CME Observed with STEREO and Hinode/EIS
C. Miklenic, A.M. Veronig, M. Temmer, C. Möstl, H. K. Biernat
SSolar PhysicsDOI: 10.1007/ ••••• - ••• - ••• - •••• - • Coronal Dimmings and the Early Phase of a CMEObserved with STEREO and
Hinode /EIS
C. Miklenic · A.M. Veronig · M. Temmer , · C. M¨ostl , · H. K. Biernat c (cid:13) Springer ••••
Abstract
We investigate the early phase of the 13 February 2009 coronal massejection (CME). Observations with the twin STEREO spacecraft in quadratureallow us to compare for the first time in one and the same event the temporalevolution of coronal EUV dimmings, observed simultaneously on-disk and above-the-limb. We find that these dimmings are synchronized and appear duringthe impulsive acceleration phase of the CME, with the highest EUV intensitydrop occurring a few minutes after the maximum CME acceleration. During thepropagation phase two confined, bipolar dimming regions, appearing near thefootpoints of a pre-flare sigmoid structure, show an apparent migration awayfrom the site of the CME-associated flare. Additionally, they rotate around the‘center’ of the flare site, i.e. , the configuration of the dimmings exhibits thesame ’sheared-to-potential’ evolution as the postflare loops. We conclude thatthe motion pattern of the twin dimmings reflects not only the eruption of the fluxrope, but also the ensuing stretching of the overlying arcade. Finally, we find that:(1) the global-scale dimmings, expanding from the source region of the eruption,propagate with a speed similar to that of the leaving CME front; (2) the massloss occurs mainly during the period of strongest CME acceleration. Two hoursafter the eruption
Hinode /EIS observations show no substantial plasma outflow,originating from the ‘open’ field twin dimming regions.
Keywords:
Coronal Mass Ejections, Flares
1. Introduction
Coronal dimmings are regions in the solar corona that undergo an abrupt inten-sity drop roughly co-temporal with the launch of a coronal mass ejection (CME).They are observable in soft X-rays (SXR; Sterling and Hudson, 1997) and mostconspicuously in the extreme UV (EUV; e.g. , Zarro et al. , 1999; Thompson etal. , 2000) both on the solar disk and above the limb. Space Research Institute, Austrian Academy of Sciences,Schmiedlstraße 6, A-8042 Graz, Austria Kanzelh¨ohe Observatory-IGAM, Institute of Physics,University of Graz, Universit¨atsplatz 5, A-8010 Graz, Austriaemail: [email protected]
SOLA: mikl_v2_jul11.tex; 20 October 2018; 0:36; p. 1 a r X i v : . [ a s t r o - ph . S R ] O c t lobal-scale dimmings originate from the source region of the eruption andmay expand across almost the entire solar disk ( e.g. , Zhukov and Auch`ere, 2004;Mandrini et al. , 2007). They follow the nearly spherically propagating frontsof global coronal waves (Moreton waves and EIT waves, Moreton and Ramsey,1960; Moses et al. , 1997; Thompson et al. , 1998). On the other hand, small-scale twin dimmings, which typically exhibit a more dramatic intensity decrease,occur in confined areas near the ends of a pre-flare, S-shaped sigmoid structure(Sterling and Hudson, 1997).The generally accepted physical interpretation of coronal dimmings is thatthey reflect the strong density depletion of the inner corona due to the expansionor ‘opening’ of magnetic field lines in the wake of a CME. Two major lines ofevidence support this view: (1) Measurements of Doppler shifts in emission linesformed in the lower transition region and corona revealed material outflows co-spatial with small-scale dimming regions, with outflow velocities of several tensof km s − (Harra and Sterling, 2001; Harra et al. , 2007; Jin et al. , 2009). Insideglobal-scale dimmings mass outflows with velocities of ≈
100 km s − were found(Harra and Sterling, 2001). (2) Based on the idea that due to mass conservationthe mass accumulated in the leading edge of the CME has to be balanced bythe density depletion in the dimming region, coronal dimmings have been usedto estimate CME masses ( e.g. , Harrison and Lyons, 2000; Harrison et al. , 2003;Aschwanden et al. , 2009a,b; Jin et al. , 2009). In some cases, the consistency withCME masses based on white-light measurements was good ( e.g. , Harrison et al. ,2003; Aschwanden et al. , 2009a), in other cases, the CME mass based on EUVobservations was 30–70% smaller than the white-light estimate ( e.g. , Harrisonand Lyons, 2000; Jin et al. , 2009). Jin et al. (2009) attributed this discrepancypartly to transition region outflows, which refill the dimming regions. Never-theless, these studies strengthen the link between coronal dimmings and CMEs.Therefore, the investigation of the dimming process is considered as a powerfuldiagnostic of the early phase of a CME, especially with regard to the physics ofCME onsets (Harrison et al. , 2003). This phase cannot be examined using white-light observations, since the occulting disk of the coronagraph obscures both theSun and the lower layers of the corona, where the CME is usually launched andaccelerated (Temmer et al. , 2010, 2008).Previous studies on coronal EUV dimmings addressed their temporal andspatial relationship to the associated CME, the amount of the EUV intensitydrop as well as the timescale and duration of the dimmings ( e.g. , Sterling andHudson, 1997; Aschwanden et al. , 1999; McIntosh et al. , 2007; Bewsher et al. ,2008). McIntosh et al. (2007) examined both the temporal evolution of the sizeof particular dimming areas and the apparent motion of these regions. Due toobservational limitations, however, in all of these studies a particular dimmingevent was observed either on the solar disk or above the limb, depending on theparticular line of sight from the observing instrument to the source region of theeruption. Therefore, it was impossible to examine in one and the same event thetemporal evolution of coronal dimmings from different vantage points.With the launch of the twin STEREO spacecraft ( Solar-Terrestrial RelationsObservatory , Kaiser et al. , 2008) these observational limitations are temporarilyover, and more importantly, the Sun was cooperative enough to produce a
SOLA: mikl_v2_jul11.tex; 20 October 2018; 0:36; p. 2
ME/flare event when STEREO Ahead (STA) and STEREO Behind (STB)were in quadrature on 13 February 2009. The resulting unprecedented dataset allows us to investigate the coronal EUV dimming process in this eventsimultaneously both on the solar disk and above the limb. Due to the line-of-sight integration of the optically thin emission, the combination of on-diskand above-the-limb observations is particularly instructive. We compare bothdimming observations and relate them to the kinematics of the associated CME.Finally, we combine EUV observations obtained from both the STEREO andthe
Hinode spacecraft (Kosugi et al. , 2007) to examine the relationship betweendimming locations and mass flows.In Section 2 we describe the observations as well as the data sets used andthe data reduction procedures applied. The analysis and results are presented inSections 3 and 4. In Section 5 we discuss the results.
2. Observations and Data Reduction
The 13 February 2009 CME/flare event occurred in NOAA active region AR11012. At this time, the separation angle of the two STEREO spacecraft was91 ◦ . STA observed the event above the solar limb, while STB observed it nearthe center of the solar disk. The CME had a typical three-part structure, i.e. , abright front, a dark cavity, and a bright core. The flare (GOES class B2.4) startedaround 05:30 UT (GOES maximum at 05:47 UT). The event was associated witha global coronal wave observed in EUV (Cohen et al. , 2009; Kienreich et al. , 2009;Patsourakos and Vourlidas, 2009).Although in this paper we focus on the investigation of the early phase of theCME, it is interesting to note that the Heliospheric Imagers (HI, Eyles et al. ,2009) onboard STA tracked the CME’s leading edge up to the distance of about1 AU. In addition, the STB/IMPACT (Luhmann et al. , 2008) and PLASTIC(Galvin et al. , 2008) instruments registered a weak shock on 18 February 2009,10:00 UT, followed by the complex magnetic field and bulk plasma signaturestypically associated with an interplanetary coronal mass ejection (ICME). Apossible encounter of the ICME with Venus was observed with the Venus Expressmagnetometer (Zhang et al. , 2006). A detailed study on the ICME using theSTA/HI observations along with the in situ measurements provided by STB andthe Venus Express is underway and will be presented elsewhere (M¨ostl et al. ,2011).2.1. STEREO DataFrom the STA and STB spacecraft we use EUV 171 and 195 ˚A filtergrams,provided by the Extreme Ultraviolet Imager (EUVI; Wuelser et al. , 2004; Howard et al. , 2008), to investigate the dimming process both on the solar disk and abovethe limb. In addition, we use the STA/EUVI 171 ˚A filter in combination withSTA/COR1 white-light images to track the CME front, and thus, determine itsvelocity and acceleration profile. The observing cadence in the EUVI 171 ˚A filteris 2.5 min for STA and 5 min for STB, while in the 195 ˚A filter it is 10 min SOLA: mikl_v2_jul11.tex; 20 October 2018; 0:36; p. 3 or both STEREO spacecraft. The white-light STA image cadence is 10 min forCOR1 and 15 min for COR2.The SECCHI PREP routine in the SolarSoft software package (SSW; Freelandand Handy, 1998) was used to reduce the images. All images were rescaledto Earth distance to compare observations from STA and STB. STB/EUVIfiltergrams were differentially rotated to the same reference time. In addition,the gamma-log function, a compound of the logarithmic and gamma functions,was applied to the STB/EUVI filtergrams to enhance faint structures in theimages while retaining detail in the bright features.2.2.
Hinode
DataTo derive plasma flow velocities in the active region we use the earliest availableEUV raster scan for this event, provided by the EUV Imaging Spectrometer(EIS; Culhane et al. , 2007) onboard the
Hinode spacecraft. The scan covers theperiod immediately after the decay phase of the CME-associated flare (07:25 –08:12 UT). The EIS instrument provides high-resolution EUV spectra within twowavebands: 166 – 211 ˚A and 246 – 292 ˚A, covering emission lines in the transitionregion and corona. Four slit/slot positions (1 (cid:113) , 2 (cid:113) , 40 (cid:113) , and 266 (cid:113) ) allow a widechoice of operation modes. EIS raster scans are carried out in the following way.In each emission line the intensity is measured by sampling the line across the1 (cid:113) or 2 (cid:113) width of the slit. This sampling is carried out at up to 512 positionsacross the length of the slit. After this, the slit is moved to the next positionof the region of interest until the entire region is scanned. Observations can becarried out in up to 25 lines simultaneously. Images in a particular emissionline are constructed as follows. To obtain a pixel column of the image, at eachposition across the length of the slit, the intensity is integrated over the spectralline width. Repeating this procedure for all slit positions within the region ofinterest yields the entire image.After the decay phase of the B2.4 flare, the EIS instrument scanned the activeregion from west to east in 90 steps, each step 2 (cid:113) wide, with a slit length of320 (cid:113) , yielding a field of view (FOV) of 180 (cid:113) × (cid:113) . The observations compriseeight spectral windows, each 32 pixels wide ( i.e. , the intensity in each line wassampled at 32 positions across the line). We use the Fe xii xii Hinode /EIS ImagesAs aforementioned, EIS images, derived from raster scans, are not snapshotstaken at a particular moment in time. When we look at such an image we look at
SOLA: mikl_v2_jul11.tex; 20 October 2018; 0:36; p. 4 tringed-together pixel columns taken at different instants in time. Therefore, ifthe scanned region is a flaring region, the features in the FOV may have changeddramatically during the scan. Accordingly, the co-alignment of EIS images andEUVI filtergrams is a tricky issue. Even though there will be similar featuresin EIS and EUVI images taken in the same passband, the images will not lookexactly the same. To solve the problem of co-alignment, we used a method similarto that described by Jin et al. (2009). We compared EIS and EUVI images at195 ˚A. We set the mid-time of the raster scan as the time of the EIS image, chosean EUVI image observed closest to this time, visually compared it with the EISimage by identifying similar features, and then shifted the values of XCEN andYCEN in the EUVI image to roughly co-align it with the EIS raster image.After this, we cross-correlated both images to refine the visual co-alignment. Weestimate the uncertainty of the co-alignment achieved with this method to ± (cid:113) .
3. Analysis
STEREO/EUVI 171 and 195 ˚A filtergrams reflect the emitting state of thecoronal plasma at approx. 1 and 1.5 MK, respectively (Wuelser et al. , 2004;Howard et al. , 2008). To investigate the coronal dimming process, we calculateintensity light curves from these filtergrams by (1) summing up at each time allpixel intensities within a region of interest (ROI) and (2) dividing this sum bythe number of pixels in this region. For a global light curve the ROI is a largerectangular box surrounding the dimming regions, for a local light curve theROI is determined as follows. To enhance the contrast of the dimming regionswe use base difference images, i.e. , we subtract a pre-event image taken around13-Feb-2009 04:45 UT from all successive frames. We then count among thedim pixels all pixels with intensities in the range [ I min , f · I min ], where I min isalways negative, and is the intensity of the darkest pixel in a difference imageand f (cid:15) [0 , I min = −
500 (in units of normalized detectedphotons) and a factor of f = 0 .
8, pixels with intensities in the range [ − − i.e. , an 80%-threshold is used). Hence,the number of pixels forming the ROI of a local light curve varies in time. Thisoffers the opportunity to track the size evolution of local dimmings as well. Wenormalize all light curves to the pre-event intensity level.The drop in the EUV intensity I during coronal dimmings is usually inter-preted as mass loss in the corona in the wake of a CME. For instance, Jin etal. (2009) showed that the EUV intensity variation in coronal dimming regionsreflects the variation of the mass density. Therefore, we consider the intensitychange rate dI/dt as a proxy for the mass loss rate, i.e. , the time evolution ofthe mass loss. We calculate dI/dt as the time derivative of a spline fit to thenormalized dimming light curve.We derive the temporal evolution of the CME velocity and acceleration ( i.e. ,the profiles v CME ( t ) and a CME ( t ), respectively) as follows. We (1) constructrunning difference images from STA/EUVI 171 ˚A and COR1 white light framesto enhance the contrast of the CME front, (2) track the fastest-moving part ofthe front, (3) fit the obtained height-time measurements by a spline fit, and (4)calculate the first and second time derivative of the fit curve. SOLA: mikl_v2_jul11.tex; 20 October 2018; 0:36; p. 5 . Results
Figure 1 shows the source region of the CME observed with STB/EUVI 171 ˚A.The CME-associated flare occurs at the site of an S-shaped sigmoid structure(Figure 1(a)). During the impulsive phase of the flare, two elongated, roughlynorth-south aligned flare ribbons separate (Figure 1(b)). The first postflare loops,resulting from the reconnection of magnetic field lines, appear at the beginningof the decay phase (Figure 1(c)). Furthermore, two small-scale dimming regions,designated as DR1 and DR2, appear near the ends of the pre-flare sigmoid.These regions grow and exhibit an apparent motion during the decay phase(Figure 1(d)). The postflare loops in between the dimmings form an east-westaligned arcade.In the top row of Figure 2 we plot the STB/EUVI 195 ˚A intensity light curvecalculated from the FOV of the images presented in the middle row (global lightcurve). At the beginning of the presented time interval we use the full imagecadence available (10 min) to capture the abrupt intensity drop. After 13-Feb-2009 08:25 UT the cadence is reduced to 40 min, as a lower cadence is sufficientto capture the subsequent gradual changes in the EUV intensity. The intensitydrops down to roughly 83% of the pre-flare EUV level in approx. 30 min. Therecovery of the EUV intensity back to the pre-event level in the event understudy takes approx. 16 h. We note that in extreme cases, recovery times ofcoronal dimmings up to 48 h have been observed (Attrill et al. , 2008). Themiddle and bottom rows show four representative EUVI 195 ˚A direct and basedifference images, respectively, illustrating the onset of the dimmings and theensuing gradual intensity increase, as the rarefied regions are slowly refilled.
We use the STB/EUVI 195 ˚A filtergrams to track the apparent motion of thelocalized dimming regions DR1 and DR2 ( cf.
Figure 1, panels (c) and (d)) andto investigate the evolution of their size. We determine the pixel size of DR1 andDR2 in each base difference image, using a factor of f = 0 .
8. This 80%-thresholdproved to be most suitable for tracing the dim areas in an image. After this, wecalculate the ‘center of gravity’ (CoG) of DR1 and DR2 at each time ( i.e. , foreach frame we weight the location of dim pixels by their intensity, and then wedivide the sum of these products by the number of dim pixels). Figure 3 showsthe locations of the CoGs for both dimming regions at each time superimposedon a base difference image, taken during the decay phase of the flare. The CoGof DR1 moves to the north-east, while the CoG of DR2 is directed to the south-west. The straight lines connecting the CoGs of both dimming regions at thebeginning and end of the analyzed time interval demonstrate that the CoGsrotate clockwise around the ‘center’ of the flare site. Also, taking the east-westalignment of the postflare loops around 07:45 UT as a reference, the lines indicate
SOLA: mikl_v2_jul11.tex; 20 October 2018; 0:36; p. 6
71 Å STEREO-B 05:21 UT 05:31 UT171 Å STEREO-B05:51 UT171 Å STEREO-B 07:06 UT171 Å STEREO-BDR1 DR2 DR1 DR2(b)(a)(c) (d) flareribbonflareribbon
Figure 1.
Evolution of the CME-associated flare and the small-scale coronal dimmings inSTB/EUVI 171 ˚A. (a) Pre-flare S-shaped sigmoid structure. (b) Bright, separating flare rib-bons shortly after flare onset. (c) The first postflare loops, resulting from the reconnection ofmagnetic field lines, appear at the beginning of the decay phase of the flare. The white arrowspoint to two small-scale dimming regions, designated as DR1 and DR2, which appear nearthe ends of the pre-flare sigmoid structure. (d) During the decay phase, the dimming regionsexhibit an apparent motion. The postflare loops in between the dimmings form an east-westaligned arcade. The dashed box shows the field of view (FOV) of
Hinode /EIS. – North is up,west is to the right. that the observed rotation reflects the evolution of the dimmings from a ‘sheared’to a more ‘potential’ configuration, as the event progresses. The successivelyappearing postflare loops exhibit the same sheared-to-potential evolution ( cf.
Figure 1, panels (c) and (d)).In Figure 4 we consider the apparent motion of the CoGs of the two dimmingregions separately for the east-west ( x ) and north-south ( y ) direction. In addi-tion, we compare the motion of the CoGs with both the kinematics of the CMEand the temporal evolution of the soft X-ray emission of the associated flare. TheCoGs of the dimming regions separate in the x -direction (Figure 4(a)), whilethey approach each other in the y -direction (Figure 4(b)). Given the east-westalignment of the latest appearing postflare loops, the approaching in y reflectsthe aforementioned evolution of the dimming regions from a ‘sheared’ to a more‘potential’ configuration. The separation in x indicates that the extent of theregion, affected by the expelled CME, grows, as the event progresses. The mostconspicuous movement of the CoGs is detected between 05:50 – 06:30 UT ( i.e. , SOLA: mikl_v2_jul11.tex; 20 October 2018; 0:36; p. 7 igure 2.
Evolution of the on-disk EUV dimmings. Top: STB/EUVI 195 ˚A light curve.The intensity values are normalized to the pre-flare EUV flux. The shaded area highlights theperiod of the dimmings and the EUV flux recovery back to the pre-event level. The times of theimages are marked with triangle symbols. The times of the four selected images shown beloware marked with dashed lines. Middle: STB/EUVI 195 ˚A images covering a 600 (cid:113) × (cid:113) FOVcentered around the dimmings. Bottom: Corresponding base difference images. At 06:05:49 UTtwo distinct small-scale dimming regions are visible. They slowly disappear during the next16 h. after the impulsive acceleration phase of the CME). In this period, the CME isin its propagation phase, moving with a constant speed of around 350 km s − (Figure 4(c)). After approx. 07:00 UT, the CoGs of both dimming regions seemto stall in place.In Figure 4(d) we present the CME acceleration profile a CME . The impulsiveacceleration phase of the CME lasts for approx. 40 min and reaches its maximumof 250 m s − at 05:36 UT. Around this time, the dimming regions appear andthe CoGs start to move. In Figure 4(d) we also investigate the relationshipbetween the acceleration phase of the CME and the energy release phase of theflare. The maximum CME acceleration occurs approx. 11 min before the flare-related maximum of the soft X-ray (SXR) emission. In other words, the highestCME acceleration occurs in the impulsive phase of the flare, during which theflare energy is released and particles are accelerated to nonthermal energies.Since observations of nonthermal emission (hard X-rays, microwaves producedby these particles) are not available for the event under study, we assume thatthe Neupert effect is valid in this event, i.e. , we assume that peaks in the timederivative of the SXR flux coincide with peaks in the nonthermal hard X-ray SOLA: mikl_v2_jul11.tex; 20 October 2018; 0:36; p. 8
Figure 3.
STB EUVI 195 ˚A base difference image of the flaring and localized dimming regions.Bright east-west aligned postflare loops are visible in the center of the image. Gray contourstrace the total dimming area, i.e. , the area that exhibited a drop in intensity within the timeinterval indicated by the color bar. The contour level represents the 80%-threshold used toidentify the dimming regions. The color-coded diamonds mark the positions of the ‘centerof gravity’ (CoG) of dimming regions DR1 and DR2 at each time. The dotted/dashed lineconnects the CoGs of both dimming regions at the beginning/end of the time interval. (HXR) emission ( e.g. , Neupert, 1968; Dennis and Zarro, 1993; Dennis et al. ,2003; Veronig et al. , 2005). Consequently, we fit a spline curve to the GOESSXR light curve and take the time derivative of the fit to obtain a proxy forthe HXR emission, and thus, a proxy for the energy release rate in the flare( cf. dashed profile in Figure 4(d)). The maxima in the SXR time derivative andCME acceleration profile are roughly co-temporal. The time difference betweenthe peaks is less than 4 min. This indicates that the CME acceleration phase isclosely synchronized with the energy release phase of the associated flare ( e.g. ,Zhang et al. , 2001; Mariˇci´c et al. , 2007; Vrˇsnak et al. , 2007; Temmer et al. , 2010).In the upper panel of Figure 5 we present the temporal evolution of the sizeof DR1 and DR2. Both dimming regions grow steadily until approx. 07:20 UT,but DR2 grows faster than DR1. After this, the size of DR1 remains almostconstant within the time interval under study, while DR2 starts to contract asit is slowly refilled. The bottom panel of Figure 5 shows the time history ofthe intensity change of DR1 and DR2, averaged over all pixels in the respectivedimming region (local light curves). Both regions undergo an abrupt intensitydrop on a timescale of approx. 25–30 min. The intensity drops down to roughly
SOLA: mikl_v2_jul11.tex; 20 October 2018; 0:36; p. 9 [ m s ] C M E - a r cs e c CoG -distance of DR1 and DR2 x r CME : EUVI 171 ÅCOR1spline fit a r cs e c v CME SX R f l u x [ W m ] - - . r C M E / R o v [ k m s ] C M E - r CME a CME
EIS F SXR dFdt
SXR
CoG -distance of DR1 and DR2 y (a)(b)(c)(d) Figure 4.
Apparent motion of dimming regions DR1 and DR2 in east-west ( x ) and north–south ( y ) direction related to the CME kinematics and the soft X-ray (SXR) emission of theflare. (a) and (b): The x - and y -distance of the ‘centers of gravity’ (CoG) of DR1 and DR2observed in STB/EUVI 195 ˚A. (c): Height-time measurements of the CME front observed inSTA/EUVI 171 ˚A and COR1 white light plus spline fit ( r CME ) and derived velocity v CME . (d):Derived CME acceleration a CME (solid thick), GOES12 1 – 8 ˚A SXR flux (dotted) plus splinefit ( F SXR , solid thin), and time derivative of the fit (dashed). The time derivative is scaled insuch a way that its amplitude is equal to the peak of a CME . The vertical bar marks the timedifference of 3.5 min between the maxima of a CME and GOES derivative. The horizontal barshows the period of the EIS raster scan.
SOLA: mikl_v2_jul11.tex; 20 October 2018; 0:36; p. 10
R1DR2 [ M illi on t h s o f s o l a r d i sk ] Dimming area [ k m ] DR1Normalized IntensityDR2
EIS
Figure 5.
Temporal evolution of the dimming areas. Top: Size of dimming areas DR1 andDR2 observed in STB/EUVI 195 ˚A. Bottom: Normalized intensity, averaged over all pixels inthe dimming region. The horizontal bar shows the period of the EIS raster scan.
20% of the pre-flare EUV flux and remains virtually unchanged during the next2.5 h.
In the top panel of Figure 6 we plot: (1) the SXR emission profile of the flarealong with its time derivative; (2) the normalized STB/EUVI 171 and 195 ˚Aintensity profiles, calculated over a FOV of 640 (cid:113) × (cid:113) surrounding both the flare SOLA: mikl_v2_jul11.tex; 20 October 2018; 0:36; p. 11 ite and the small-scale twin dimmings (global light curve); and (3) the splinefits I to the data points. During the impulsive phase of the flare, the intensitydrops down to a level of approx. 90% of the pre-flare EUV flux in 30 min. Inthe bottom panel of Figure 6 we investigate the temporal relation between theacceleration phase of the CME (represented by the acceleration profile a CME )and the EUV intensity change rate dI/dt , which is considered as a proxy forthe evolution of the mass loss rate ( cf.
Section 3). In both wavelengths, theintensity change is most conspicuous during the acceleration phase of the CME.The peak of a CME appears at 05:36 UT. The peaks in the mass loss rate occur afew minutes later, namely, at 05:41 UT in the 195 ˚A filter and 05:43 UT in the171 ˚A filter, respectively. This indicates that large amounts of mass are removedfrom the analyzed region while the CME is expelled from the Sun. Moreover,the timing of the highest mass loss coincides roughly with the maximum of theSXR time derivative (05:40 UT, cf.
Section 4.1.2), suggesting that the energyrelease in the flare is synchronized with the mass loss due to the CME.4.2. Plasma Flows in the CME Source RegionLarge amounts of mass expelled from the Sun in the course of the CME should bereflected in upflowing plasma at the source region of the eruption, in particular,at the locations of the on-disk dimming regions DR1 and DR2, observed withSTB/EUVI 195 ˚A ( cf.
Figure 3). To measure the plasma flow velocities in thedimming regions we use the first available
Hinode /EIS raster scan and derive aDopplergram from observations in the Fe xii cf. dashed box in Figure 1(d)). The contours mark the total area of DR1 and DR2(white) and the EIS flow velocities between −
25 and +15 km s − (blue/red).The same contours are superimposed on the Dopplergram in Figure 7(b). InDR1 we detect only downflows, and these downflows are relatively weak (around5 km s − ). The strongest downflows with velocities around 15 km s − are ob-served in the arcade of postflare loops, predominantly at the loop footpoints( cf. red contours in Figure 7(a)). In DR2 we find both downflows and upflows,but the upflow velocities are low (less than approx. 5 km s − ). The strongestupflows with velocities around 20–25 km s − are co-spatial with a large-scale,coronal loop structure south-east of DR2 ( cf. dark-blue contours in Figure 7(a)).This implies that during the EIS raster scan ( cf. Figure 4(d)) the strongestupflows of the 1.5 MK plasma seem to be rather flows along large, closed coronalloops than outflows originating from ‘open’ field regions. This is consistent withthe finding presented in Section 4.1.3: The main mass loss, inferred from theSTEREO/EUVI intensity drop occurs between roughly 05:25 and 05:55 UT, i.e. , approx. 90 min before the beginning of the EIS raster scan ( cf.
Figure 6and 5). During the time interval of the EIS observations, the mass loss rateapproaches zero, indicating that the coronal mass loss in the wake of the CMEhas stopped.
SOLA: mikl_v2_jul11.tex; 20 October 2018; 0:36; p. 12 O ES f l u x [ W m ] - - a [ m s ] C M E - d I/ d t [ ] - N o r m a li z ed I n t en s i t y a CME dIdt dIdt
GOES fluxspline fit toI spline fit to I GOES derivative
Figure 6.
Top: GOES12 1 – 8 ˚A soft X-ray flux (SXR, solid thick), GOES time derivativescaled to the SXR peak flux (solid thin), normalized STB/EUVI 171 and 195 ˚A light curvesintegrated over the FOV of Figure 9b, and spline fits I and I to the data points. Thethick × -symbol around 05:32 UT highlights the temporary increase in the 171 ˚A intensity,caused by the onset of the flare. The vertical line at 05:39:30 UT indicates the maximum ofthe GOES derivative. Bottom: CME acceleration profile a CME (solid) and intensity changerates (time derivatives of the spline curves I and I ). The vertical lines at 05:40:49 and05:43:19 UT indicate the times of the highest change in intensity for both filters. The verticalbar marks the time difference of 1.3 min between the maximum of the GOES derivative andthe minimum of dI /dt . i.e. , in the early decay phase of the flare).A 70%-threshold was used to determine the pixel size of DR-L at each time( f = 0 . SOLA: mikl_v2_jul11.tex; 20 October 2018; 0:36; p. 13
Time (UT)STEREO-B/EUVI 195 Å(a)
DR1 DR2
Intensity (b)
DR2DR1
Time (UT)VelocityHinode EIS Fe XII 195.12 Å -1 Figure 7. (a): STB/EUVI 195 ˚A intensity image taken at 07:45:49 UT ( i.e. , around themid-time of the
Hinode /EIS raster scan), showing the region of the EIS FOV ( cf. dashedbox in Figure 1(d)). The dashed line marks the position of the EIS slit at 07:49:11 UT. (b)Dopplergram derived from observations in the
Hinode /EIS Fe xii cf.
Figure 3). The contour level represents the 80%-threshold used to identify thedimming regions. The colored contours mark
Hinode /EIS 195.12 ˚A plasma flow velocities withcontour levels of [ − − − −
10 ] km s − (blue) and [+10, +15] km s − (red). for about an hour, and then its size remains virtually unchanged within the timeinterval under study ( i.e. , until 08:25 UT).Figure 8(c) shows the time derivative of the local intensity profile I DR − L . Thecurve progression indicates that the intensity drop, and therefore mass loss, ismost distinct during the period highlighted by the shaded box in Figure 8(c).The width of this box is the FWHM of the CME acceleration profile shownin Figure 4(d) and in the bottom panel of Figure 6. We use the FWHM torepresent the period of strongest CME acceleration. The end of this periodroughly coincides with the end of the impulsive phase of the flare ( cf. verticalsolid line at around 05:48 UT in Figure 8(c), marking the GOES maximum).The peak time of the intensity change rate at 05:37 UT is close to the middle of SOLA: mikl_v2_jul11.tex; 20 October 2018; 0:36; p. 14 igure 8.
EUV dimmings above the limb. (a): STA/EUVI 195 ˚A difference image with apre-CME frame around 04:45 UT subtracted from a post-eruption frame at 06:05:30 UT. Thesolid contour enclosing DR-L represents the 70%-threshold used to identify the dimming region.The dashed arc marks the solar limb. The solar disk has been artificially occulted. (b) Solid:temporal evolution of the size of DR-L observed in STA/EUVI 195 ˚A. Dotted: spline fit I FOV to the ∆-symbols, which represent the normalized intensity calculated from the FOV of panel(a) at each time. Dashed: spline fit I DR − L to the ♦ -symbols, which represent the normalizedintensity calculated from all pixels forming DR-L at each time. (c): Time derivative of thespline fit I DR − L . The gray vertical bar highlights the period of strongest CME acceleration,represented by the FWHM of the CME acceleration profile shown in Figure 4(d). The verticallines mark the time of the minimum of the intensity change rate at 05:37:30 UT (dashed) andthe time of the GOES maximum of the flare at 05:47:39 UT (solid). The horizontal black barhighlights the period of the EIS raster scan. the FWHM box (05:36 UT). After the period of strongest CME acceleration theintensity change rate approaches zero, in agreement with the intensity changerate derived from the on-disk observations ( cf. Section 4.1.3). This indicates thatthe mass loss is synchronized with the CME acceleration phase and substantialonly during the period of strongest CME acceleration.4.4. CME Velocity in the Initial Stage of the EruptionIn Figure 4(c) we presented the evolution of the CME velocity, derived fromcombined STA/EUVI and COR1 above-the-limb observations, extending overthe period between roughly 05:20 and 06:45 UT. In the following, we exploit theavailability of both above-the-limb and on-disk EUVI observations in one andthe same event to compare two methods of determining the velocity of the CMEin the initial stage of the eruption ( i.e. , until approx. 05:56 UT).In the first method we use above-the-limb observations. We apply the pro-cedure described in Section 3, i.e. , we track the fastest part of the CME frontin STA/EUVI 171 ˚A running difference images and calculate the velocity froma linear fit to the height-time measurements. Figure 9(a) shows an example ofsuch a measurement. To estimate the error we repeat the measuring procedureseveral times and determine the mean and standard deviation of the calculatedvelocities. Thus, we obtain a linear CME velocity of 212 ±
10 km s − between05:26 and 05:46 UT. SOLA: mikl_v2_jul11.tex; 20 October 2018; 0:36; p. 15 d DR1 DR2 E U V f l u x ( no r m a li z ed ) D i s t an c e ( R ) o . EUVI-A CME front distanceEUVI-B 171 Å light curve (c) Dt
EUVI-B 195 Å light curve
Figure 9. (a): STA/EUVI 171 ˚A running difference image. Dashed line: direction, in whichthe CME front was tracked; × -symbol: one of the height-time measurements; solid arc: solarlimb. (b): STB/EUVI 171 ˚A filtergram taken at the same time as (a). The dashed line fromthe flare site to the most remote vertex of the FOV marks the distance d , used to estimate theaverage CME velocity. Arrows point to dimming regions DR1 and DR2. FOV: 640 (cid:113) × (cid:113) .(c): Solid/dotted: STB/EUVI 171/195 ˚A flux integrated over the FOV of panel (b) andnormalized to the pre-flare intensity. The ∆-symbols represent the averaged STA height-timemeasurements of the CME front. The dashed line is the linear fit to the data points inside thegray vertical bar that highlights the period ∆ t of the EUV intensity drop. In the second method we use STB/EUVI 171 and 195 ˚A on-disk observationsof the region surrounding both the flare site and the small-scale twin dimmings.To select an appropriate FOV, i.e. , a FOV covering the range over which theoutermost border of the expanding global-scale dimmings is discernible, we vi-sually track these dimmings in difference images. Figure 9(b) shows the FOVselected according to this criterion. We associate the expanding global-scaledimmings with the 1 MK and 1.5 MK plasma, rarefied or ejected in the courseof the CME ( cf. , Aschwanden et al. , 1999). We calculate the intensity profiles forboth wavelengths, integrating over the entire FOV of Figure 9(b) (global lightcurve). We then normalize both profiles to the pre-flare intensity. Figure 9(c)shows that the intensity drops down to a level of 87–90% of the pre-flare EUVflux on a timescale of ∆ t ≈
30 min (05:26 – 05:56 UT). This is the periodover which the CME clears the FOV, because afterwards the intensity level inboth wavelengths stabilizes. The average velocity of the CME in this period iscalculated as v CME = d/ ∆ t , where d ( ≈ ,
000 km) is the distance shown inFigure 9(b). To estimate ∆ t we use both the EUVI 171 and 195 ˚A light curvesand determine the times when the intensity drop in both profiles starts and ends SOLA: mikl_v2_jul11.tex; 20 October 2018; 0:36; p. 16 cf. shaded box in Figure 9(c), highlighting the time interval ∆ t ). This yieldsan average speed of v CME ≈
192 km s − . Due to the image cadence of 5 min forSTB 171 ˚A and 10 min for STB 195 ˚A, however, ∆ t is likely to be overestimated,since the actual intensity drop may start later, namely, between the time of thelast pre-event intensity image and the time of the first ‘intensity-drop’ image.Similarly, the actual intensity drop may end earlier, namely, between the timeof the last ’intensity-drop’ image and the time of the first image with stabilizedintensity. Assuming that the overestimation of ∆ t is up to 10 min, we obtain anaverage speed of v CME ≈
265 km s − . Taking the mean of both velocity valueswe obtain an average speed of v CME ≈ ±
36 km s − .Both the ‘above-the-limb’ method (CME tracking) and the ‘on-disk’ method(expanding global-scale dimming) yield comparable results.
5. Discussion
Taking advantage of the quadrature of the STEREO Ahead and Behind space-craft on 13 February 2009, we presented the first direct comparison of CME-associated, coronal EUV dimmings observed simultaneously on the solar diskand above the solar limb. Both data sets yield a consistent picture of the eruptionand mass loss during the early phase of the CME, and thus, support the viewthat above-the-limb dimmings and global-scale, on-disk dimmings essentiallyreflect the same process, namely, the density rarefication in the corona due tothe expelled CME.We compared the dimming process with the kinematics of the CME andfound that during the period of strongest CME acceleration (approx. 05:25–05:50 UT) the EUV intensity in both data sets dropped abruptly. Also, themass loss rates, inferred from both on-disk and above-the-limb EUVI intensityobservations, showed a distinct extremum in this period, suggesting that the bulkof the coronal mass was ejected, as the CME left the Sun. Moreover, the extremaof the CME acceleration profile and mass loss rates were only a few minutesapart in time, reflecting the close temporal relationship between the kinematicsof the CME and the dimming process. Two hours after the eruption, both theEUVI 195 ˚A mass loss rates and the Dopplergram derived from Fe xii
Hinode /EIS instrument indicated that the loss of coronalplasma with a temperature of 1.5 MK had virtually stopped.The main characteristics of the dimmings in the event under study werecomparable with other findings. For example, an intensity decrease within 20–30 min and a gradual recovery of the EUV brightness lasting several hours arecharacteristic features of the coronal dimming process ( e.g. , Aschwanden et al. ,1999; McIntosh et al. , 2007; Reinard and Biesecker, 2008; Aschwanden et al. ,2009b). Further, the observed amount of intensity drop, including the differencein brightness decrease between small-scale and global-scale dimmings, is typicalfor the two types of dimmings (Aschwanden et al. , 1999; Zarro et al. , 1999;Thompson et al. , 2000; McIntosh et al. , 2007; Harra et al. , 2007). Also, theinitial growing of dimming areas as well as the apparent motion of these regionsaway from the source region of the eruption was reported earlier (McIntosh et SOLA: mikl_v2_jul11.tex; 20 October 2018; 0:36; p. 17 l. , 2007). In the 13 February 2009 CME event, however, we observed for thefirst time an additional motion pattern of the small-scale twin dimming regions,namely, an apparent clockwise rotation around the source region of the eruptionor the ‘center’ of the flare site, respectively. Taking the orientation of the first andlast appearing postflare loops as a reference, this rotation can also be describedas a ’sheared-to-potential’ migration, superposed on the outward motion.How can we explain such an apparent rotation? Small-scale twin dimmings areoften interpreted as the confined footpoint areas of a CME-associated, eruptingflux rope (Sterling and Hudson, 1997; Webb et al. , 2000). The footpoints of a fluxrope, however, are stationary. Therefore, moving bipolar dimmings represent notonly the footpoints of the flux rope, but also other regions, where field lines arestretched. From MHD simulations the destabilization and subsequent eruptionof flux ropes/CMEs is often achieved by twisting magnetic field structures (see e.g. Chen et al. , 2002; Fan and Gibson, 2004; T¨or¨ok and Kliem, 2003). Associ-ated dimming regions can as well be located at large distances from the sourceregion of the eruption. For example, Delann´ee (2000) reported dimmings at thefootpoints of transequatorial loops that connected a flaring active region to anactive region located in the opposite hemisphere. They proposed that the appear-ance of such dimmings is strongly related to the magnetic field topology: Whenlow-lying, active-region loops rise and erupt, thus forming a characteristic twindimming pattern, the overlying large-scale loops rise and erupt as well. Mandrini et al. (2007) described a scenario in which the arcade above the erupting fluxrope expands significantly before reconnecting. In this case, twin dimmings areexpected to appear not only at the footpoints of the erupting flux rope, but alsoat the footpoints of the sheared magnetic arcade, as the expansion occurs.Bearing in mind this idea we propose a magnetic field topology in the eventunder study that may explain the observed rotation and outward migration ofthe small-scale twin dimming regions DR1 and DR2 as well as the appearance ofless and less sheared postflare loops, as the eruption progresses ( cf.
Figure 10).Before the eruption, a magnetic separatrix layer divides the flux rope fromthe overlying arcade that spans the polarity inversion line (PIL). The innerloops of this arcade, i.e. , loops that are rooted closer to the PIL, are stronglysheared, low-lying loops, whereas the loop tops of the less and less sheared outerloops are at successively higher altitudes. Once the magnetic configuration hasbecome unstable the flux rope erupts. As a consequence, the field lines of theflux rope are stretched and the first dimmings appear near its footpoints, i.e. ,at strongly sheared positions. When the erupting flux rope encounters the fieldlines of the arcade, it stretches them while it continues to rise. Shortly beforethe stretched field lines reconnect, dimmings appear near their footpoints. Sincethe loop tops of the overlying arcade are located at different heights, the low-lying, inner loops, rooted at strongly sheared positions, are stretched before thelarge-scale, outer loops, whose footpoints are less and less sheared. Therefore, therespective dimmings appear at different times and positions, and the observergets the impression of rotating twin dimming regions or dimmings that evolvefrom a ‘sheared’ to a more ‘potential’ configuration while they move away fromthe source region of the eruption, respectively. Once the stretched field lines ofthe overlying arcade reconnect, their upper parts supply the erupting flux rope
SOLA: mikl_v2_jul11.tex; 20 October 2018; 0:36; p. 18 igure 10.
Apparent migration of the CoGs of the localized dimming regions. Top row: sideview (a) and top view (b) before the eruption. The flux rope (helical structure) is rootedat both sides of the PIL (straight line). Close to the PIL the overlying arcade consists oflow-lying, strongly sheared loops, whereas at greater distances from the PIL the amount ofshearing decreases and the altitude of the loop tops increases. Bottom row: side view (c) andtop view (d) during the eruption. For clarity, only one field line of each field line triple from(a) and (b) is shown. The field lines of the overlying arcade are pushed upward by the eruptingflux rope. Shortly before the thus stretched field lines reconnect, dim patches appear near theirfootpoints. The dark, filled ellipses mark the positions of the CoGs of the coronal dimmings atdifferent times (black/gray: CoGs near currently/recently stretched lines). The arrows point inthe direction of the apparent motion of the CoGs. The white ellipses highlight the bright flareribbons (energy deposition sites) at the footpoints of the currently reconnected field lines. with new poloidal flux, while their lower parts form the postflare loops. Sincethe first appearing postflare loops are created from the low-lying, inner, andstrongly sheared loops, they are strongly sheared themselves, whereas the laterappearing postflare loops are created from the less and less sheared outer loops,and thus, have a more and more potential configuration.In analyzing the behavior of the relatively small twin dimmings DR1 and DR2on the solar surface we reach the conclusion that the observed motion patternreflects not only the eruption of the flux rope, but also the ensuing stretchingof the overlying arcade. The above-the-limb dimming region DR-L, however, is
SOLA: mikl_v2_jul11.tex; 20 October 2018; 0:36; p. 19 arger than DR1 and DR2 by a factor of 100 ( cf. top panel of Figure 5 andFigure 8(b)), and thus, rather represents the vast cross section of the coronalvolume, evacuated by the CME.Finally, the existence of both STA and STB/EUVI observations of the 13February 2009 CME event allowed us for the first time to compare directlytwo methods of deriving the CME velocity in the initial stage of the eruption.The velocity values obtained from above-the-limb observations of the leadingedge of the CME (CME tracking) and on-disk observations of the expandingglobal-scale dimmings were comparable (212 ±
10 km s − and 228 ±
36 km s − ,respectively). This demonstrates that global-scale, on-disk dimmings, originatingfrom the source region of the eruption, propagate with a speed similar to thatof the leaving CME front, and thus, supports the general view that global-scaleEUV dimmings are on-disk signatures of CME launches. Accordingly, they canbe used as a means for identifying CME source regions and estimating initialCME velocities, which is of particular interest in connection with Earth-directedhalo CMEs and space weather forecast. Acknowledgements
This study was supported by the Austrian Austrian Science Fund(FWF): P20145-N16.
Hinode is a Japanese mission developed and launched by ISAS/JAXA,with NAOJ as domestic partner and NASA and STFC (UK) as international partners. It isoperated by these agencies in co-operation with ESA and NSC (Norway). The SECCHI dataused here were produced by an international consortium of the
Naval Research Laboratory (USA),
Lockheed Martin Solar and Astrophysics Lab (USA),
NASA Goddard Space FlightCenter (USA),
Rutherford Appleton Laboratory (UK),
University of Birmingham (UK),
Max-Planck-Institut for Solar System Research (Germany),
Centre Spatiale de Liege (Belgium),
Institut d’Optique Theorique et Appliquee (France), and
Institut d’Astrophysique Spatiale (France).
References
Aschwanden, M. J., Fletcher, L., Schrijver, C. J., Alexander, D.: 1999,
Astro-phys. J. , 880.Aschwanden, M. J., Nitta, N. V., Wuelser, J., Lemen, J. R., Sandman, A.,Vourlidas, A., Colaninno, R. C.: 2009a,
Astrophys. J. , 376.Aschwanden, M. J., Wuelser, J. P., Nitta, N. V., Lemen, J. R.: 2009b,
SolarPhys. , 3.Attrill, G. D. R., van Driel-Gesztelyi, L., D´emoulin, P., Zhukov, A. N., Steed,K., Harra, L. K., Mandrini, C. H., Linker, J.: 2008,
Solar Phys. , 349.Bewsher, D., Harrison, R. A., Brown, D. S.: 2008,
Astron. Astrophys. , 897.Cohen, O., Attrill, G. D. R., Manchester, W. B., Wills-Davey, M. J.: 2009,
Astrophys. J. , 587.Chen, J.: 1989,
Astrophys. J. , 453.
SOLA: mikl_v2_jul11.tex; 20 October 2018; 0:36; p. 20 hen, P.-F., Wu, S. T., Shibata, K., Fang, C.: 2002,
Astrophys. J. Lett. ,99.Culhane, J. L., Harra, L. K., James, A. M., Al-Janabi, K., Bradley, L. J.,Chaudry, R. A., et al. : 2007,
Solar Phys. , 19.Delann´ee, C.: 2000,
Astrophys. J. , 512.Dennis, B. R., Veronig, A., Schwartz, R. A., Sui, L., Tolbert, A. K., Zarro, D. M.,Rhessi Team: 2003,
Adv. Space Res. , 2459.Dennis, B. R., Zarro, D. M.: 1993, Solar Phys. , 177.Eyles, C. J., Harrison, R. A., Davis, C. J., Waltham, N. R., Shaughnessy, B. M.,Mapson-Menard, H. C. A., et al. : 2009,
Solar Phys. , 387.Fan, Y., Gibson, S. E.: 2004,
Astrophys. J. , 1123.Freeland, S. L., Handy, B. N.: 1998,
Solar Phys. , 497.Galvin, A. B., Kistler, L. M., Popecki, M. A., Farrugia, C. J., Simunac, K. D. C.,Ellis, L., et al. : 2008,
Space Sci. Rev. , 437.Harra, L. K., Hara, H., Imada, S., Young, P. R., Williams, D. R., Sterling, A. C.,Korendyke, C., Attrill, G. D. R.: 2007,
Publ. Astron. Soc. Japan , 801.Harra, L. K., Sterling, A. C.: 2001, Astrophys. J. Lett. , 215.Harrison, R. A., Bryans, P., Simnett, G. M., Lyons, M.: 2003,
Astron. Astrophys. , 1071.Harrison, R. A., Lyons, M.: 2000,
Astron. Astrophys. , 1097.Howard, R. A., Moses, J. D., Vourlidas, A., Newmark, J. S., Socker, D. G.,Plunkett, S. P., et al. : 2008,
Space Sci. Rev. , 67.Jin, M., Ding, M. D., Chen, P. F., Fang, C., Imada, S.: 2009,
Astrophys. J. ,27.Kaiser, M. L., Kucera, T. A., Davila, J. M., St. Cyr, O. C., Guhathakurta, M.,Christian, E.: 2008,
Space Sci. Rev. , 5.Kienreich, I. W., Temmer, M., Veronig, A. M.: 2009,
Astrophys. J. Lett. ,118.Kosugi, T., Matsuzaki, K., Sakao, T., Shimizu, T., Sone, Y., Tachikawa, S., et al. :2007,
Solar Phys. , 3.Luhmann, J. G., Curtis, D. W., Schroeder, P., McCauley, J., Lin, R. P., Larson,D. E., et al. : 2008,
Space Sci. Rev. , 117.
SOLA: mikl_v2_jul11.tex; 20 October 2018; 0:36; p. 21 andrini, C. H., Nakwacki, M. S., Attrill, G., van Driel-Gesztelyi, L., D´emoulin,P., Dasso, S., Elliott, H.: 2007,
Solar Phys. , 25.Mariˇci´c, D., Vrˇsnak, B., Stanger, A. L., Veronig, A. M., Temmer, M., Roˇsa, D.:2007,
Solar Phys. , 99.McIntosh, S. W., Leamon, R. J., Davey, A. R., Wills-Davey, M. J.: 2007,
Astrophys. J. , 1653.M¨ostl, C., Rollett, T., Lugaz, N., Farrugia, C. J., Davies, J. A., Temmer, M., et al. : 2007,
Astrophys. J. , accepted.Moreton, G. E., Ramsey, H. E.: 1960,
Publ. Astron. Soc. Pacific , 357.Moses, D., Clette, F., Delaboudini`ere, J.-P., Artzner, G. E., Bougnet, M.,Brunaud, J., et al. : 1997, Solar Phys. , 571.Neupert, W. M.: 1968,
Astrophys. J. Lett. , 59.Patsourakos, S., Vourlidas, A.: 2009,
Astrophys. J. Lett. , 182.Reinard, A. A., Biesecker, D. A.: 2008,
Astrophys. J. , 576.Sterling, A. C., Hudson, H. S.: 1997,
Astrophys. J. Lett. , 55.Temmer, M., Veronig, A. M., Kontar, E. P., Krucker, S., Vrˇsnak, B.: 2010,
Astrophys. J. , 1410.Temmer, M., Veronig, A. M., Vrˇsnak, B., Ryb´ak, J., G¨om¨ory, P., Stoiser, S.,Mariˇci´c, D.: 2008,
Astrophys. J. Lett. , 95.Thompson, B. J., Cliver, E. W., Nitta, N., Delann´ee, C., Delaboudini`ere, J.:2000,
Geophys. Res. Lett. , 1431.Thompson, B. J., Plunkett, S. P., Gurman, J. B., Newmark, J. S., St. Cyr, O. C.,Michels, D. J.: 1998, Geophys. Res. Lett. , 2465.T¨or¨ok, T., Kliem, B.: 2003, Astron. Astrophys. , 1043.Veronig, A. M., Brown, J. C., Dennis, B. R., Schwartz, R. A., Sui, L., Tolbert,A. K.: 2005,
Astrophys. J. , 482.Vrˇsnak, B., Mariˇci´c, D., Stanger, A. L., Veronig, A. M., Temmer, M., Roˇsa, D.:2007,
Solar Phys. , 85.Webb, D. F., Lepping, R. P., Burlaga, L. F., DeForest, C. E., Larson, D. E.,Martin, S. F., Plunkett, S. P., Rust, D. M.: 2000,
J. Geophys. Res. ,27251.Wuelser, J., Lemen, J. R., Tarbell, T. D., Wolfson, C. J., Cannon, J. C., Car-penter, B. A., et al. : 2004, In: Fineschi, S., Gummin, M. A. (eds.),
Telescopesand Instrumentation for Solar Astrophysics , Proc. SPIE , 111.
SOLA: mikl_v2_jul11.tex; 20 October 2018; 0:36; p. 22 arro, D. M., Sterling, A. C., Thompson, B. J., Hudson, H. S., Nitta, N.: 1999,
Astrophys. J. Lett. , 139.Zhang, J., Dere, K. P., Howard, R. A., Kundu, M. R., White, S. M.: 2001,
Astrophys. J. , 452.Zhang, T. L., Baumjohann, W., Delva, M., Auster, H.-U., Balogh, A., Russell,C. T.; 2006,
Planet. Space Sci. , 1336.Zhukov, A. N., Auch`ere, F.: 2004, Astron. Astrophys. , 705., 705.