Observation of the origin of CMEs in the low corona

Transcription

1 Astron. Astrophys. 355, (2000) ASTRONOMY AND ASTROPHYSICS Observation of the origin of CMEs in the low corona C. Delannée 1,, J.-P. Delaboudinière 1, and P. Lamy 2 1 Institut d Astrophysique Spatiale, Université Paris Sud CNRS, bat 121, Orsay, France 2 Laboratoire d Astronomie Spatiale, Traverse du Siphon, Les Trois Lucs, Marseille, France Received 9 December 1998 / Accepted 11 January 2000 Abstract. The aim of the main observing program with EIT on board SOHO, is to monitor the whole Sun surface in the Fe xii emission line at 195 Å, every 17 minutes. The very beginning of some CMEs can be observed. We interpret Fe xii images in conjunction with He ii,hα and coronagraph observations over a period of 6 days. We find that 7 prominences produced ejections. An active region produced 9 ejections. Five ejections are seen as dark bubbles propagating above the solar limb while 9 are seen as dimmings on the solar surface. The 3 other ejections are bright bubbles observed rising up above the limb. Thirteen of the 17 observed ejections are related to a CME. Two CMEs of the 15 CMEs observed with LASCO C2 are not related to Fe xii low corona events. Probably, these CMEs have their origin behind the limb. Prominences give rise to quite slow CMEs, km s 1 while fast CMEs, km s 1, originate close to active regions. We conclude that CMEs start in the low corona and that large scale coronal structures reconfigurations occur when these disturbances propagate outward. Key words: Sun: activity Sun: atmosphere Sun: corona Sun: particle emission Sun: prominences 1. Introduction Previously the coronal mass ejections were mainly observed by coronagraphs. As defined by Hundhausen (1993), coronal mass ejections are an observable change in coronal structure that (1) occurs on a time scale between a few minutes and several hours and (2) involves the appearance of a new, bright whitelight feature in the coronagraph field of view presumably associated with increased local density, which propagate outward. The Skylab, Solwind and SMM coronagraphs gave an extended view of CMEs. They were classified by visual shapes. Their velocity, mass, kinetic energy, angular width, and the angle of the trajectory relative to the solar equator plane, were systematically computed for about 3000 observed CMEs (Gosling et al., 1976, for the Skylab mission, Howard et al., 1986, for the Solwind mission and S t Cyr & Webb, 1991, Burkepile & S t Cyr, 1993 and Hundhausen, 1993, for the SMM mission). Now present at Goddard Space Flight Center, SoHo EAF, mail code 682.3, Greenbelt, MD USA Many authors (St Cyr & Webb, 1991, Kahler using SMM data and Geophysical data report) found that CMEs are initiated in association with active regions or prominences. The solar surface associated signatures are found to be flares and prominence ejections (Webb and Hundhausen, 1987). However, from these observations, only some prominence ejections or flares could be associated with CMEs due to the poor temporal resolution and temporal coverage between the observations of the corona and of the solar surface. In all the previous references, the analysis of the initiation of the CMEs were obtained doing comparison of statistics of the location of the CMEs and of the underlying activity. The main conclusion obtained on this basis and reported in Hundhausen (1993), is that CMEs are large scale perturbations of the coronal magnetic field. Dimmings and flares observed in X-rays are also associated to CMEs (Sterling & Hudson 1997, using the Soft X-ray Telescope on the Yohkoh satellite). Harrison (1992), observed 3 cases of CMEs associated with flares using SMM and GOES data and concluded that quite often a flare cannot be the source of a mass ejection. The Solar and Heliospheric Observatory (SOHO) provide the possibility to follow the CME from the low corona (below 1.14 R ) to the far corona (out to 30 R ). Dere et al. (1997a) analyzed a CME observed by the Extreme Ultra-Violet Imaging Telescope (EIT) on board SOHO on December, Images in the Fe xii line at 195 Å were obtained every 12 minutes. At this high cadence, one could observe the motion of a very faint wave produced above a prominence, seen on about 4 images. This feature was associated with a white light CME observed with LASCO C2. Since then, other events have been observed with the SOHO instruments and analyzed: February, by Gopalswamy et al. (1997) using LASCO, EIT, SXT and WIND data and Dere et al. (1997b) using EIT and LASCO data, April, by Thompson et al. (1997) using EIT data, May, by Thompson et al. (1998) using EIT and LASCO data, and by Aurass et al. (1999) using EIT, LASCO, GOES data, March, by Ciaravella et al. (1999) using EIT and UVCS data, etc. All these observations were obtained with a temporal resolution and temporal covering between the different instruments observations good enough to permit to follow the CME from the solar surface. However, Wiik et al. (1997) observed an eruptive prominence and its associated CME with two spectrographs (SUMER and CDS) and one coronagraph

2 726 C. Delannée et al.: Observation of the origin of CMEs in the low corona (LASCO C3) on board SOHO. They could not decide which of the prominence or the CME was ejected first. Dryer et al. (1998) observed a flare produced by the AR 7978 on July and the associated CME. They did not find any associated eruptive prominence. The relations between flares, prominence ejections and CMEs are difficult to establish observationally. Multiwavelength and high time cadence images are needed to follow the phenomena continuously from the photosphere to the corona. There exist many theoretical models for the initiation of CMEs developed numerically or analytically. Amari et al. (1996), Chen (1997), Gibson & Low (1998) described the relation between the eruption of a prominence and the production of a CME. In the models, the CME is a consequence of the prominence eruption. Antiochos et al. (1999) described a possible way to produce flares and an associated CME in a quadrupolar magnetic field region. The flare is the result of the reconnection as well as the CME. In this paper, we analyze chromospheric and low corona events, observed by EIT, related to CMEs observed by the Large Angle and Spectrometric Coronagraph (LASCO), in the period from 1 to 6 November We try to find the temporal and spatial continuity between the low corona events and the CMEs. In Sect. 2 the methods of observations are described. The morphologies of the events observed in the low corona and the chromosphere are described in Sect. 3. In Sect. 4, we discuss the relations between the low corona events and the CMEs. We conclude in Sect Observations The observations are obtained with two different kind of instruments. EUV band pass filters in the EIT telescope permit the observation of the low corona just above the solar surface, and LASCO C2 and C3 coronagraphs permit the observation of the high corona in white light EIT - low corona observations The Extreme ultra-violet Imaging Telescope (EIT) on board SOHO (Delaboudinière et al., 1995) observes the low corona, from 1 to 1.4 R in 4 band passes centered on: 304 Å which contains the He ii line emitted at typical temperatures of K and the Si xi line emitted at typical temperatures of 10 6 K; 171 Å which contains the Fe ix line emitted at typical temperatures of K, and Fe x line emitted at typical temperatures of K; 195 Å which contains the Fe xii line emitted at typical temperatures of K and 284 Å which contains the Fe xiv line emitted at typical temperatures of K. The spatial resolution is 2.62 arcsec per pixel. A special observing mode is used to detect the origin of the CMEs (CME watch) in which images are taken with the filter centered on 195 Å (Fe xii) every 17 minutes in average. This cadence is sufficient to observe dynamic events on an average number of 5 images per event. These images are processed as described in Moses et al., The intensities of the images is displayed in logarithmic scale. In order to detect faint events, time series of the difference between successive images (running difference images) are needed. This allows us to observe motions and make events evident, but with this method it is difficult to understand the exact shape evolution of the features. A dark feature may appear bright in the difference with a following image. The bright parts of this following image can also be bright in the difference image. So, the bright and dark parts of the previous and the following features can be confused. We use the difference images only to increase the sensitivity in detecting motions between successive images. Original images are used to compute the shape, the brightness and the location of the detected features. In addition, images in Hα from various ground based observatories (Hight Altitude Observatory, Ondřejov Observatory, Holloman AFB and Meudon Observatory), and images in He ii taken with the filter at 304 Å on EIT are used to analyze the events and their underlying features (sunspots and prominences/filaments) LASCO - observations from 2 to 5 R The Large Angle Spectrometric Coronagraph (LASCO) C2 on board SOHO (Brueckner et al., 1995), observes the corona from 2to5R. We use a filter centered on 5800±600 Å. The emission in this spectral band is due to Thomson scattering by free electrons and He I line emission. One image every hour in average is taken over the whole period of observations. This cadence is high enough to track CMEs on an average number of 8 images per event. The corona has two components: the F corona due to the diffraction by dust, and the K corona due to the scattering on electrons. The F corona is not affected by CMEs. To remove it, we use images of difference. An average image over one day is computed, removing particular images where data is missing or where events are appearing. The difference between each image of each day with this mean daily image is computed. This method is free of the problems appearing when the running difference method is used and is able to extract some very faint features. It is not possible to use this method in the case of EIT images, because too much temporal and spatial fluctuations are present on the EIT images so that the mean image over one day has no real value. The difference image method can introduce artifacts: features smaller than the characteristic thickness of the streamers cannot be resolved. Moreover, we have to take care that low time cadence of images can show structures which are apparently moving but which are in fact not related (see a sample on Fig. 1). We do not take into account these features. 3. Description of the events It is necessary to clarify the terminology we use. Burkepile and S t Cyr (1993) classified the CMEs observed in white light with a coronagraph in several types: Loop/Cavity, Cavity, Core, Mound, Blob, Jet or Tongue. The mound type was called by several authors halo (Howard, 1982, Vourlidas, 1997, etc), we prefer to use this denomination. We indicate in our paper if a

3 C. Delannée et al.: Observation of the origin of CMEs in the low corona 727 Fig. 1. Sample of artifact features produced by the difference image method we used for the analysis. The images obtained on 97/11/01 from 6:44 to 9:21 (4 images) show apparently moving bright and dark features. loop, a cavity, a core or a halo appears in the LASCO C2 field of view. Blob, jet, tongue are not seen during the period under consideration. The previous works showed that several dynamical structures are related to CME: dimmings of the solar surface and above the solar limb, EIT waves, flares, prominences eruptions. During the period under study, we analyzed 17 events which had similar shape and behavior than these well known structures. They were 9 dimmings, 3 bright bubbles which were related to a prominence eruption and 5 dark bubbles. On these 17 events, 13 were related to a CME. The observed events appear above active regions or above filaments or prominences. They have different shapes which we classify depending on their underlying structure and on their location on the disc or above the limb. All the observations are summarized in Table 1. The EIT events are presented in front of their corresponding LASCO events Active regions A single active region (AR 8100) produced ejections in the period of observations, from 97/11/01 to 97/11/06, during when the active region traveled from the central meridian to the west limb. The associated dynamical structures are: loops, jets (elongated features with short life time) (Fig. 2), flares (Fig. 3, 8), EIT waves and/or dimmings (Fig. 3, 4, 9). According to the previous studies of events observed in Fe xii with EIT (Dere et al., 1997, Gopalswamy et al. (1997), Dere et al. (1997b) and Thompson et al. (1998)) the dimmings are expected to be the very first signature of a CME originating in an active region. We observed 9 dimmings coming from AR 8100 in 5 days: on 97/11/02 at 21:08, on 97/11/03 at 5:03, at 9:12 and at 10:33, on 97/11/04 at 5:58, on 97/11/05 at 6:49 and at 11:41 and on 97/11/06 at 00:18 and at 12:01. Some can be produced with a very short time delay between two dimmings: about 1 hour. The last dimming is very large and very dynamic. After this, the active region remains quiet until it passes behind the limb. In the following we describe these dimmings. As their apparent morphology depends on their location, on the disc or above the limb, we describe them in two subsections corresponding each to their locations. The relation to flares and CMEs are also determined On the disc a) EIT events morphologies Five dimmings were produced close to AR 8100 when it passes the central meridian from 97/11/02 to 97/11/04. During this period the dimmings are always observed located on the disc only. On 97/11/03 at 9:12, a loop like structure which has its feet in the active region becomes very bright, and the region inside the loop becomes dark (Fig. 3). The dimming then expands northward with its border retaining the shape of a loop

4 728 C. Delannée et al.: Observation of the origin of CMEs in the low corona Table 1. Table summarizing all the events observed on EIT and LASCO, and showing their correspondences. The numbers in brackets indiquate the date at which the CME appeared. day appearance disappearance velocity (km s 1 ) EIT 1 15:42 19:43 28 NW to W prominence, bright bubble 2 12:20 13:40 35 NE AR, dark bubble 21:08 21:42 35 S to N AR, dimming 3 2:02 2:50 55 SW prominence, dark bubble 5:03 5: N 60 W 9:12 10: N 100 W from SW to NE and NW from SW to NE and NW 10:33 11:41 10 N 60 S from SW to NE and NW trajectory shape appearance disappearance 21:18 2:18 (2) velocity (km s 1 ) LASCO trajectory shape 120 W loop, cavity and core 12:36 11: SW loop, cavity 4:20 13:29 50 SW loop, cavity AR, dimming 6:10 9: N 225 S AR, dimming 11:11 14: N 114 S NW SW NW SW and and halo halo AR, dimming 11:44 16: N SW halo 4 2:21 3:59 20 SE prominence, V 4:40 5: SE loop, cavity shape 5:58 7:26 90 S to N AR, dimming 6:10 6: W halo 5:20 6:49 53 NW prominence, dark 6:44 7: NW core bubble 17:58 22: SW loop, cavity 19:28 00:49 (5) 3-15 SW prominence, bright bubble 19:37-7 SW prominence, wave like motion 5 6:49 7:43?? AR, large scale dark bubble, involve several active regions 11:41 12: from SW to NW AR, large scale dark bubble, involve several active regions 18:33 23:50 5 NW prominence dark bubble 6 00:18 3:10 10 SW AR, large scale dark bubble, involve several active regions 12:01 12:49 40 from SW to E, and W AR, large scale dark bubble, involve several active regions 00:18 (5) 6:10 (5) 85 W loop, cavity 7:28 8: W loop, cavity 12:10 13: W loop, cavity disturbed streamer 00:18 2: SW loop, cavity 12:10 13: W loop, cavity

5 C. Delannée et al.: Observation of the origin of CMEs in the low corona 729 Fig. 2. Loops, jet and streamer observation in the Fe xii emission line at 195 Å, on 97/11/03 at 11:24. Fig. 3. Flare and bright loop and the second dimming produced by AR 8100 in Fe xii on 97/11/03 at 09:12. They are both expanding on the solar disc. The bright loop is called an EIT wave in Thompson et al. (1998). with its feet in the active region (Fig. 4). During this expansion, some other loops which connect a northern active region (AR 8101) to AR 8100 change in brightness and in shape. Finally, the dimming slowly fades. All the dimmings observed on the disc, i.e. from 97/11/02 to 97/11/04, have a similar behavior. For each of these dimmings, a flare is detected by the GOES satellite: on 97/11/02 a C2.2 flare at 20:49, on 97/11/03 a C8.6 flare at 4:38, a M1.4 flare at 9:10 and a M4.2 flare at 10:29, on 97/11/04 a X2.1 flare at 5:58. In 195 Å the flares are a very bright compact region. On 97/11/04 at 5:58, the flare is so intense that it produces bleeding visible as a straight intense feature located at the western edge of the active region and instrumental scattered light visible as a diffuse feature of shape of an oblate disc of about 0.6*2.4 R which is an enhancement of about 6% in intensity. Nevertheless this event is very similar to the others. The flares are observed in the same image than the very beginning of the dimmings so it is not possible to decide what is produced first: the flare or the dimming. The bright edge of the dimmings is interpreted as a Moreton wave by Thompson et al. (1998), called in this article EIT waves. Moreton waves are theoretically described by Uchida (1974) as waves produced by an eruption and propagating on the photosphere and in the corona. Moreton waves are expanding but not always all around the center of the active region. From 97/11/02 to 97/11/04, the structures are always expanding from the South, where the AR 8100 is located, to the North. No part of these structures are

6 730 C. Delanne e et al.: Observation of the origin of CMEs in the low corona Fig. 4. The second dimming produced by the AR 8100 in Fe xii on 97/11/03 at 09:12. Images are running differences. The limb is shown by the black circle. The dimming region is enclosed by dashed lines. It is growing Northward. The points where the velocities were calculated are shown at the heads of the arrows in the fourth image. observed going southward. This is observed several times for the same active region on 97/11/02 at 21:08, on 97/11/03 at 5:03, at 9:12 and at 10:33, on 97/11/04 at 5:58. The first dimming (on 97/11/02 at 21:08) is very small. We follow the trajectories of the leading edge of the dimmings (Fig. 5). We did this estimation for two points of the structure of the dimming on 97/11/03. These points are shown at the heads of the arrows on the last image of Fig. 4. The estimated velocities are quite larger for the point going East Northward than for the one going West Northward. Moreover, the motion seems to be decelerated East Northward, and accelerated West Northward. The linear interpolation of the time distance diagram of the leading edge of the dimmings gives velocities, projected on the image surface, in the range km s 1. In fact it is very difficult to say that the so called EIT waves, i.e. the bright edge of the dimmings in Fe xii, are really waves as they could be the projection of the cavity and the surround bright loop of a CME rising in altitude and growing as described in the Gibson & Low (1998) process. The fact that the dimmings and their bright edges are observed going Northward and not Southward and that the NE part is going faster than the NW part can be interpreted either as a strong effect of the magnetic Fig. 5. Trajectories and velocities of two points of the dimming on 97/11/03 at 9:12. Diamonds correspond to the point going East northward, and the crosses to the one going West northward. The velocities are quite larger for the point going East northward than the one going West northward. field on the trajectory of the dimming or as the effect of the projection angle on the solar surface as a cavity grows and goes up, following the coronal magnetic field lines. The Hα observation from Meudon observatory shows that many filaments/prominences attached to the active region, appear and disappear during all the period, see Fig. 6. A sunspot appears in the active region on 97/11/03.

7 C. Delannée et al.: Observation of the origin of CMEs in the low corona 731 Fig. 6. The active region AR 8100 was observed during the whole period in Hα at the Meudon Observatory. Many filaments associated to the active region appeared and disappeared. A sunspot appeared on 97/11/03. b) Related CMEs The related CMEs are halos, see Fig. 7 for 3 samples of the 4 halos which appeared. All the halos present a bright and diffuse front. They are very faint. They appear on 97/11/03 at 6:10, at 11:11 and at 11:44, on 97/11/04 at 6:10 corresponding respectively to the dimmings on 97/11/03 at 5:03, at 9:12 and at 10:33, and on 97/11/04 at 6:10. The last images showing the events on EIT are obtained after the first image showing the corresponding CME on LASCO C2 because the effect of the event on the solar disc last longer than the passage of the CME front. They propagate South Westward and North Westward. The propagation is radial. As it is very difficult to correct any angle projection on the trajectory of the CMEs, we computed all the velocities in the plane of the image. The derived velocities are in the range km s Above the limb a) EIT events morphologies The morphologies of the events produced from the AR 8100 on 97/11/05 and 97/11/06 are quite different as they are visible partly on the disc and above the limb. They occurred on 97/11/05 at 6:49 and at 11:41 and on 97/11/06 at 00:18 and at 12:01. They are associated with 4 flares: on 97/11/05 at 6:41 a C7.0 flare, at 11:36 a C4.7 flare and at 23:35 a C1.2 flare and on 97/11/06 at 11:55 a X9.4 flare (Fig. 8). This last flare produces instrumental scattered light which appears as a bright straight feature and a diffuse feature of shape of a square triangle of 2.3*2.7 R which is an enhancement of about 4% in intensity produced. Due to this scattered light it is difficult to observe the underlying features on the first image of the ejection on EIT. The events produced on 97/11/05 and on 97/11/06, present dimmings surrounded by a bright edge, i.e. EIT wave. Both structures, dimmings and bright edges, are observed on and above the solar disc, except for the event produced on 97/11/05 at 00:18 which does not present dimmings nor bright edge on the disc. A propagation from the South to the North is visible for the dimming and its bright edge produced on 97/11/05 at 11:41. The other events which occurred on 97/11/05 at 6:49 and on 97/11/06 at 12:01 are much faster than the ones which occurred from 97/11/02 to 97/11/04. So, the propagation from the South to the North of the dimmings and their bright edges are not visible, see Fig. 9 for a sample. However, from Maia et al. (1999), we believe that there is a propagation of the dimmings and their bright edges from the South to the North. In these cases, the region affected by the ejection has almost its full extension on the first image of each ejection. However, the dimmings slowly expand from the West to the East at velocities in the range km s 1, on the image surface. In the 4 cases studied in this section, the region above the West limb is affected and presents a strong decrease in intensity and a change in the coronal structures shapes. These dimmings fade with time in the same manner as the part of the dimmings on the disc does. Finally let us mention an aborted ejection. A dark bubble is observed in the North East on 97/11/02 at 12:20 above an active region (AR 8104). The coronal structures visible as loops or beginning of streamers above the active region change in shape very dynamically. The loops open and the streamer lines first diverge to let the dark bubble pass then return back to their initial positions. This dark bubble is observed going up and down, which means that it is not expelled. b) Related CMEs The associated CMEs present a bright and compact loop with an underlying cavity, see Fig. 10 for a sample. They appear on 97/11/05 at 7:28 and at 12:10 and on 97/11/06 at 12:10 (Fig. 10) corresponding respectively to the dimmings occurring on 97/11/05 at 6:49 and at 11:41 and on 97/11/06 at 00:18 and at 12:01. The loop and the cavity propagate Westward in the underlying streamer. The projected velocities of these CMEs are quite large, in the range km s 1. Due to the high velocity of the last CME its bright loop is deformed in a mushroom shape (Fig. 10) Filaments or prominences On the disc Unfortunately no filaments erupt in the studied period. However filament eruptions can usually be observed with EIT through the

8 732 C. Delannée et al.: Observation of the origin of CMEs in the low corona Fig halo CMEs appeared on 97/11/03 at 6:10, at 11:11, and at 11:44. The halos are enclosed by the dashed lines. Fig. 8. A flare observed in Fe xii produced by AR 8100 on 97/11/06 at 12:01. There is instrumental scattered light in the CCD due to the high intensity produced at this time.

9 C. Delannée et al.: Observation of the origin of CMEs in the low corona 733 Fig. 9. The large scale dimming of about R /3 observed in Fe xii, on 97/11/06 at 12:01. Images are images of difference with the image preceding the event taken at 11:41. The dimming is propagating from the East to the West. No propagation from the South to the North is visible even if the radio emission shows one (Maia et al., 1999). Fig. 10. Large mushroom CME appeared on 97/11/06 at 12:10, related to the dimming produced near AR 8100 when it was close to the limb. Due to the high velocity its bright loop is deformed into a mushroom shape. 195 Å filter. They become very bright before they leave. After an eruption, the filament channel becomes very dark Above the limb EIT events morphologies and related CMEs Prominences are dark in 195 Å. This is due to their cold material which absorbs the Fe xii hot line (Kucera et al., 1998). The above coronal structures are the same as the ones produced above an active region: loops and feet of streamers. We observe 7 ejections related to prominence motions. They are of two kinds: bright bubbles (2) and dark bubbles (5). The two bright bubbles appeared on 97/11/01 at 15:42 (Fig. 11) and on 97/11/04 at 19:28 (Fig. 13). In Fig. 11 a prominence is observed erupting. A dark feature is present in Fe xii at the feet of the prominence. Higher in altitude and more to the South than this dark feature is a bright structure which corresponds to the southern extension of the Hα prominence. The bright structure in Fe xii is composed of bubbles which move southward at velocities of about 28 km s 1. After this event, the prominence has totally disappeared. This eruption is related to a big CME presenting a loop, a cavity and a core which appeared in the LASCO C2 field of view at 21:18 (see Fig. 12). The CME is lying in the streamer and takes place around the equator. Its velocity is about 120 km s 1. In Fig. 13, a prominence is evolving going up above the limb on 97/11/04 at 19:28. Near the limb in Fe xii, a dark thin and twisted structure is observed. This structure corresponds to the

10 734 C. Delannée et al.: Observation of the origin of CMEs in the low corona Fig. 11. Bright bubble appeared as a prominence is erupting on 97/11/01 at 15:42. On the first line are the Fe xii observations, on the second one are the running difference images of the Fe xii observations and on the third line are the Hα observation of the region from the High Altitude Observatory, and He ii observation of the region before the ejection observed with EIT. The bright structure seen in Fe xii is composed of bubbles which correspond to the southern extension of the Hα prominence. A CME is observed related to this eruption. Fig. 12. Image at 23:17 of the loop, cavity and core of the CME which appeared on 97/11/01 at 21:18. This CME corresponds to the prominence eruption which occurred on 97/11/01 at 15:42. feet and the inner part of the prominence. At 19:37 a small bright bubble is observed in Fe xii moving upward. Notice that the bright Fe xii bubble seems to be the top of the He ii prominence but the exact overlying of the two images of the same structure is not possible due to the time delay ( 6 hours) between the He ii image and the nearest Fe xii image. The bright Fe xii bubble is apparently detached from the underlying prominence but in He ii, the prominence remains connected to the chromosphere until its top has disappeared from the EIT field of view. The motion of the bright Fe xii bubble is accelerated from 3 to 15 km s 1. The day after the prominence has totally disappeared. This prominence eruption observed with EIT is apparently not related to any CME observed with LASCO C2. The probable reason is that another prominence erupted at 19:37 and the two CMEs are impossible to distinguish from each other. A dark and bright feature observed in Fe xii on 97/11/04 is shown in Fig. 14. A prominence located in the South West is moving upward very slowly (7 km s 1 ). The prominence is observed in Fe xii as dark and bright almost straight features parallel to the solar limb (similar as a planar wave). We expect that the dark part of this feature corresponds to the prominence core and the bright parts to a region which is heated by the upward motion of the prominence. There is no possibility to make the exact overlying of the structures observed in Fe xii and in He ii as the images are obtained with to much time delay

11 C. Delannée et al.: Observation of the origin of CMEs in the low corona 735 Fig. 13. Bright bubble appearing in Fe xii above a prominence on 97/11/04 at 19:37. On the first and on the second lines are the Fe xii observations. Near the limb, a dark thin and twisted structure is observed. This structure corresponds to the feet and the inner part of the prominence as shown by the images in Hα from the Holloman AFB, and He ii from EIT on the third line. Thereafter the prominence was no longer visible. Fig. 14. The prominence eruption on 97/11/04 at 19:37, produced slow wave like motion of dark and bright features in the Fe xii observations. The first image is the Fe xii observation, the second image is the running difference image for a better look at this structure. He ii observations of the region with EIT are presented below. ( 6 hours). Nevertheless, Gopalswamy and Hanaoka (1998) observed that the prominence core produce a cavity in SXR and a bright feature at the interface of the prominence and the corona. In our case, the slow motion of the prominence lasted during 4 hours from 15:42 until 19:37. At 19:37, the prominence has disappeared. There is no possibility to derive the acceleration of the prominence as just before its disappearance the prominence is in a slow motion and then suddenly disappeared from the field of view. This third prominence erupted from the South West at 19:37 and is related to a CME presenting a loop and a cavity appearing on 97/11/05 at 00:18 in the LASCO C2 field of view (see Fig. 15). The trajectory follows the streamer. The loop is lying across the equator. The velocity is about 85 km s 1. Fig. 15. A CME appeared on 97/11/05 at 00:18 in the form of a loop with a cavity after the prominence eruption on 97/11/04 at 19:37. The trajectory follows the streamer. In these first three cases, the motion of the prominences is observed but not the one of any feature produced above the prominence. The structures described in the following are produced above the prominences but are not the prominences themselves. The structures are dark bubbles. Four dark bubbles appeared in the studied period of observations: on 97/11/03 at 2:02, on 97/11/04 at 2:21, on 97/11/04 at 5:20 and on 97/11/05 at 18:33. In the Fig. 16 is shown a V shape structure. In the South East, a prominence is slowly moving upward in Fe xii, on 97/11/04 between 00:15 and 2:21. At 2:21, the coronal structures which are the feet of a streamer begin to diverge. A decrease in intensity occurs between them. This decrease is moving radially at a velocity of about 20 km s 1. At 5:00, the coronal structures come back to their initial position. The V shape dark bubble is related to a CME observed with LASCO C2 which presents a loop and a cavity at 4:40 (Fig. 17). This CME appears before a

12 736 C. Delannée et al.: Observation of the origin of CMEs in the low corona Fig. 16. A V shape structure was observed in Fe xii on 97/11/04 at 2:21. First line shows the Fe xii observations, second line shows the running difference images of the Fe xii observations, third line is the Hα, at the Ondřejov Observatory, and He ii observations of the region. The prominence is slowly rising and the dark bubble is related to a CME. The prominence lasted many hours after this event, but its disappearance did not give rise to any event observed with EIT or LASCO. halo CME produced by the AR 8100 at 5:58 but the differences in shapes, in locations and in velocities of the two CMEs permit us to distinguish the two phenomena. The loop is not regular and seems to present a twist. The two feet of the loop in the LASCO C2 field of view are distinct and the legs of the loop are crossing each other near their feet. So, the V shape ejection seen with EIT becomes a twisted loop and a faint cavity in the LASCO C2 field of view. As the loop presents a large almost circular head, it is maybe seen edge-on. The trajectory of the loop is laid in the streamer. The velocity is about 120 km s 1. After this event the underlying prominence remains visible in He ii and in Hα for 8 hours expanding upward but without any visible ejection. At 7:18 on 97/11/05 the prominence has disappeared in all lines but no eruption was visible neither with EIT in any wavelength nor with LASCO C2. This disappearance seems to be a so called thermal one. This prominence is maybe high enough to be heated and become invisible in chromospheric and coronal lines. In Fig. 18 is shown a dark bubble produced above a prominence on 97/11/04 at 5:20 in Fe xii. A very faint filament is present very close to the North West limb, and a very small point above the limb is visible in Hα on 97/11/04 at 17:11. In Fe xii a dark point moves on the solar disc toward the limb on 97/11/04 from 3:59 to 5:20. The underlying prominence is marginally visible in Hα and in He ii but the presence of the dark point moving at the solar limb just before the rise of the dark bubble is typical of the motion of a prominence, as described in Dere et al. (1997) for the CME which occurred on December 23, At 5:20, a dark bubble raises. The dark bubble is surrounded by a thin bright loop. The loop and the dark bubble are expanding in the corona in altitude and in longitude. The expansion in altitude occurs at velocities of about 53 km s 1. This ejected dark bubble is related to a CME which is mixed with a halo CME (see Fig. 19). The velocity is about 320 km s 1 but it is difficult to distinguish the two different events. On 97/11/03 at 2:02, a dark bubble raises above a prominence in the South West observed in Hα and in He ii (Fig. 20). The event is related to a CME which is mixed with a halo CME but a bright core moving more slowly than the halo remain when the halo disappears from the field of view. We therefore relate this core to the expelled CME. Its velocity is about 50 km s 1. After the ejection of the dark bubble the involved prominence remains and develops until 97/11/04 at 19:37 when it erupted and produced the wave like motion described above (see also

13 C. Delannée et al.: Observation of the origin of CMEs in the low corona 737 Fig. 17. A CME on the East part appeared on 97/11/04 at 4:40. It was produced by the prominence giving a V shape structure in Fe xii (195 Å) in the EIT field of view. In the North there is a strong leg of the CME and in the South a weak leg. The head of the CME is a circular structure. The trajectory of this CME is Southward and radial. Fig. 14). So, in the case of this prominence, during the period of its slow growing it produces a dark bubble which is observed going up. Then a day after, the prominence is fully developed and erupted very suddenly. The end of the life of this prominence is very different than the one which is present in the South East and produces a dark V shape bubble on 97/11/04 at 2:21, it develops fully over a day and slowly thermally disappears without any visible ejection. The dark bubble produced on 97/11/05 at 18:33 is not possible to relate to a CME. The reason is that the two CMEs produced by AR 8100 on 97/11/05 at 6:49 and at 11:41 and the CME produced by the prominence ejected on 97/11/04 at 19:37 are so large that the streamer is disturbed for a long time and the new CME is not visible in such a noisy surrounding. All of these dark and bright bubbles are observed going upward and growing. The propagation of the structures is radial on EIT in 2 cases only: the wave shape event (Fig. 14) and the V shape event (Fig. 16). The V shape structure is laid in the radial coronal structures. In the other cases, the propagation is not radial and follows the underlying streamer. The velocities of the ejections are quite low, in the range km s 1 as seen in projection by EIT. The prominences which produced dark bubbles (on 97/11/03 at 2:02 (Fig. 20), on 97/11/04 at 5:20 (Fig. 18), on 97/11/05 at 18:33) are not observed erupting at all. In fact it is very difficult to say if a part of the prominences erupted or not because the prominences can be seen edge-on and some structures can overlay in the line of sight. If one of its parts disappears, the prominences may remain in the same shape. In the two cases of bright bubbles and the one where the prominence had a wave like motion, the prominences disappear in He ii and in Hα. The CMEs related to the ejections presented in this Sect. (3.2.2) are typical ones. They present a bright loop, and an

14 738 C. Delannée et al.: Observation of the origin of CMEs in the low corona Fig. 18. A dark bubble surrounded by a thin bright loop appeared above a prominence on 97/11/04 at 5:20. On the first line are the Fe xii observations and on the second line are the running difference images of the Fe xii observations, the limb is shown by the dark arc of a circle. On the third line are the Hα observation of the region from the High Altitude Observatory, and the He ii observation of the region obtained with EIT. The prominence remains after the ejection as observed in Hα. underlying cavity followed in one case by a bright core. They propagate in the underlying streamer. Their velocities are in the range km s 1 on the image surface. 4. Discussion Only 4 of the 17 events seen on EIT have no LASCO counterpart. The first of the 4 event not related to a CME is a dark bubble produced above AR 8104 on 97/11/02 at 12:20 which is not expelled. The second one is the very small and very faint dimming produced on the solar disc near AR 8100 on 97/11/02 at 21:08. It would correspond to a halo CME but as halos are in general very faint, this ejection is probably too faint to be detected by LASCO C2. The third and the fourth not related events are produced on 97/11/04 at 19:37 and on 97/11/05 at 18:33, just after a very large event. The streamer remains very turbulent after these large ejections, so these two last EIT events are probably mixed with other features. Two LASCO events do not present any corresponding event on EIT. In fact this is a very low number of events because 50% of halo CMEs and some loop/cavity CMEs should have their origin behind the limb. It is possible to detect a few ejections coming from behind the limb with EIT. These ejections are visible if the densities can produce enough intensity to be detected by the CCD. The intensity on EIT varies as the electron density square and the one on LASCO varies as the electron density. If the internal density dilutes with the altitude as the coronal mean electron density does, then the intensity of the events varies as on LASCO. If an event is formed behind the limb and goes up radially, the projection on the sky plane will make it much fainter in comparison to the surrounding R 3 on EIT, and as R 1.5 corona in the EIT field of view than in the LASCO field of view. So, we should see much less events with EIT than with LASCO (about 50%). This statistic is done on a short period of 6 days, during which a quite high number of events are observed (17). During this period an active region observed on the disc is particularly active and produced many halo CMEs. Looking at the opposite side of the Sun which is visible about 15 days before and 15 days after the period of observation we see that there is a very low probability that such an active region is present behind the limb at the time of the observations. This can explain why we observe the origin of the CME in 88% of cases instead of 50% as expected. Our method of observation of CMEs leads us to find more CMEs than the LASCO team does (Table 2, ftp://lasco6.nascom.nasa.gov/pub/lasco/status/version2 - CME Lists/ CME List.txt). Particularly they found one less halo CMEs (on 97/11/03) because they are produced with a very short time delay (1 hour) and they can be mixed together when the images are treated with a method which does not reveal faint features. Thus, in the present study, the comparison with the low corona events leads to a better view of the higher corona events and helps to understand which features belong to different events. The different CMEs appearing on a LASCO C2 image can be separated by EIT if it can separate the sources in time and in location on the solar surface. The sketch presented in Fig. 21 resumes the observations obtained with EIT and LASCO in the cases of the dark and bright bubbles, i.e. in the cases of ejections which occurred at the limb. The dark bubbles observed in Fe xii with EIT above the limb seem to be similar to the cavities observed with LASCO C2. The dark bubbles which occurred on 97/11/04 at 5:20 and

15 C. Delannée et al.: Observation of the origin of CMEs in the low corona 739 EIT event CME Sun EIT streamer C2 occultor LASCO Fig. 21. Sketch representing the leading edge of EIT dark bubble appearing above the limb (it follows the underlying streamer) and the following CME as it is observed in the LASCO C2 field of view. Fig. 19. A CME associated with a dark bubble produced above a prominence on 97/11/04 at 5:20 appeared on 97/11/04 at 6:44 in the LASCO C2 field of view. A Halo CME related to a flare in AR 8100 on 97/11/04 at 5:58 appeared on 97/11/04 at 6:10 in the LASCO C2 field of view and is present on the same images. Fig. 20. A dark bubble appearing above a prominence on 97/11/03 at 2:02. The first three images are the running difference images of the Fe xii observations. The limb is shown by the dark arc of a circle. The dark bubble is enclosed by the dashed lines. On the last image is the He ii observation of the region obtained with the filter at 304 Å of EIT. on 97/11/05 at 18:33 are surrounded by a bright very thin loop, which can be the same object observed as the bright loop in the LASCO C2 field of view. However, the exact variations of the CME shapes are difficult to follow because there is a spatial gap in the field of view between the EIT and LASCO C2 instruments of about 0.5 R. The trajectory of the dark bubbles ejected near the limb are not radial and sometimes not straight on EIT. They go towards the equator. The northern part of the ejection seems deflected in the underlying streamer toward the equator and the southern part inflates towards the equator (Fig. 11, 18, and 20). The trajectories become equatorial and radial in most cases in the LASCO field of view (cf. the respective CME in LASCO C2 in Fig. 12, and 19 of the events cited above). When the ejections appear in the LASCO C2 field of view the loops are lying across the equator. It is possible as in Maia et al. (1999) to conclude that the loops in LASCO C2 are the result of reconnections that occurred in the LASCO C1 field of view. On the other hand the inflation of low loops in the active regions or over the prominences may push the overlying transequatorial loops connecting the two solar hemispheres leading to the appearance of the transequatorial bright loops in LASCO C2. The LASCO C1 coronagraph observations are obtained in Fe x and in Fe xv, from 1.1 to 3 R, but the Fabry-Perrot spectrometer used is very narrow, so no material with velocity higher than 20 km s 1 along the line of sight can be observed. The differences in shape of the ejections when observed are difficult to interpret because of the two different kinds of light emission corresponding each to different kinds of material. The loops appearing in the EIT events are very thin and sharp. They are difficult to observe. In the LASCO C2 field of view the loops

16 740 C. Delannée et al.: Observation of the origin of CMEs in the low corona Table 2. list of CMEs as observed by the LASCO team, given at ftp://lasco6.nascom.nasa.gov/pub/lasco/status/version2 CME Lists/ CME List.txt. date appearance time (UT) central (deg) PA angular width (deg) speed s 1 ) (km comments 01 20: Concentric/overlapping structured loops 02 11: Faint mound; some structure internally; falls apart 03 05: Second faint mound w/bright structured knot near equator 03 11: Superposed on earlier mound CME; late structure 04 04: Loop w/structure; dimples noticeably in C :10 Halo Mini-snowstorm; daily in W&S; faint in N 05 00: Multi-part; part of previous halo?; later stages faster 06 4: Bright non-radial wave appears 06 12: X9 flare; Snowstorm LASCO EIT C2 occultor Sun EIT event streamer CME Fig. 22. Sketch representing the EIT dimmings appearing on the disc produced from the active region. With EIT, only the northern part of the event is observed where there is enough density integrated on the line of sight. On LASCO, this northern part is either occulted or too diluted when it emerges from the occulter to be observed. are very large, intense and diffuse. The dark bubbles are much larger than the cavities are, compared to their loops in EIT and in LASCO respectively. These differences in shapes are due to the differences in emission processes of the light received in the LASCO C2 and EIT detectors. EIT detects the Fe xii emission line. The processes of the emission lines are sensitive to the electron density at the temperature of the ion formation. So, the lack of density produces a large cavity, and the density enhancement at the edge of the expanding loop produces light only if the electrons have the appropriate temperature. On the opposite, LASCO C2 detects the Thomson scattered light. The edges of the loops seem to be very dense and large as the loops are very bright and large. The edges of the loops are diffuse due to the smooth decrease of the density in the loops towards their edges. The cavity is a lack of density but even a low density can produce light at these wavelengths. The prominence ejections observed as bright bubbles do not show the dark bubble formation. This is maybe due to the fact that when the prominence begins to erupt it is almost at the limit of the EIT field of view. So, the dark bubble should be very faint or outside the field of view of EIT. However, the prominence seems to appear in the LASCO C2 field of view as the core below the cavity (cf. Fig. 11 for the EIT observations and Fig. 12 for the LASCO C2 observation of the same prominence eruption). The sketch in Fig. 22 summarizes the observations of the dimmings produced on the disc and their related halos. The dimmings emitted on the disc from AR 8100 become halo CMEs while the ones emitted near the limb are CMEs with a loop and a cavity. If the expansions of the dimmings are observable in the NS direction, then they are always from the South to the North. No part of any dimmings are observed in the South. However, in LASCO C2 the CMEs appear as a halo expanding mainly towards the South or a loop over the equator. It is possible that the differences of the location and the propagation of the same ejection observed with the two instruments are mainly due to the differences in the emission of the detected light. If the dimmings are the same phenomenon as the cavities observed in LASCO C2 but observed under different angles, the shapes of the dimmings

17 C. Delannée et al.: Observation of the origin of CMEs in the low corona 741 Fig. 23. On the West, a circular bright bubble appeared on 97/11/04 at 17:58, and went Southward with an angle of 43 degree. It followed the streamer. are nearly cones. In the northern part of the cone, we will find more integrated density in the line of sight due to the projection angle of the cone on the solar surface. In the South, the density would be integrated on the very thin layer of the edge of the cone. Moreover, it is possible that the inflation of the dark bubbles toward the equator as we observed, can produce more intense compression of the material at the edge of the dimmings closest to the equator. So, the northern edge of the dimmings is brighter than the southern edge, so much that the northern edge is the only observable. For the dimmings produced on the disc, only the southern part of the corresponding halo CME is observed because the northern part is either occulted or too diluted to be observed when it emerges from the occulter. Only, three CMEs have trajectories with some angle to the equatorial plane: on 97/11/02 at 12:36, 35 degree South West, on 97/11/04 at 4:40, 33 degree South East, and at 17:58, 43 degree South West (Fig. 23). These trajectories are always straight and follow underlying streamers. They are initiated by EIT events going in the same directions. The prominences give slow CMEs: in the range 50 to 120 km s 1. Knowing the velocities of the CMEs is a quite tough problem. The CME produced by the prominence on 97/11/04 at 5:20, has velocities of about 320 km s 1, but in this case a fast CME produced by an active region is present. They are probably mixed together. The active region AR 8100 produced quite fast CMEs: km s 1. All the CMEs produced when AR 8100 was near the central meridian, i.e. 97/11/02 04, are of the halo type, so a projection angle effect on the trajectory mixed with the growth of the structure can give such high velocities. Finally, the computed velocity of the CME produced on 97/11/06 at 12:01 (1000 km s 1 ) is highly uncertain, as it was computed with only 2 points of measure of the trajectory. The LASCO team gives a velocity of 1560 km s 1 and Maia et al. (1999) gives a velocity of about 2000 km s 1 for this CME. 5. Conclusions We observed 17 events with EIT and 15 CMEs with LASCO C2. Thirteen ejections observed in the EIT field of view are followed in the LASCO C2 field of view. Four EIT events are not possible to relate with a CME. We resolve this problem for each of these events in the discussion. Two CMEs are not possible to relate with a EIT event, this is explained by the fact that EIT cannot observe the events located behind the limb. The ejections as observed by EIT in Fe xii emission line are 9 dimmings produced near an active region, 3 bright bubbles at the leading edge of an ejecting prominence and 5 dark bubbles, sometimes surrounded by a bright thin loop, produced above prominences or active regions. The prominences can produce ejections without being observed erupting themselves. The ejections follow in some few samples the radial direction. In the other cases, as shown in the sketch in Fig. 21, they are deflected by the surrounding streamer and they inflate towards the equator. They appear to be around the equator in LASCO C2 field of view. This can be due to the destabilization of the large magnetic field lines which expand to follow the lower expansion or due to the reconnection of the lower expanding magnetic loops with the higher magnetic loops. The sketch in Fig. 22 summarizes the observations of the dimmings and their related CMEs. A dimming propagating toward the equator is related to a halo propagating in the inverse