INITIATION AND EARLY DEVELOPMENT OF THE 2008 APRIL 26 CORONAL MASS EJECTION

The radio images are provided by the multifrequency NRH at four frequencies (432, 327, 228, and 150.9 MHz; Kerdraon et al. 1997). The radio spectral data are obtained by the Nançay Decameter Array (DAM; Lecacheux 2000) in the frequency range 70–20 MHz and by the WIND/WAVES spectrometer (Bougeret et al. 1995) in the frequency range 13.8 MHz–20 kHz.

STEREO consists of two spacecraft drifting ahead of and behind the Earth in their orbit. The separation of the two spacecraft enables us to monitor CMEs from two different perspectives. On 2008 April 26, the A and B spacecraft were, respectively, ≈24° ahead and ≈−26° behind from the Earth. The Extreme Ultraviolet Imager (EUVI) is part of the Sun–Earth Connection Coronal and Heliospheric Investigation (SECCHI) instrument on board the STEREO spacecraft. EUVI observes the solar chromosphere and the lower corona at four frequencies ("wavelengths") with high temporal and spatial resolutions (Howard et al. 2008). We used only the 171 Å channel because of the high cadence (150 s) available during this event.

The emission of coronal loops is also analyzed using data from the SOHO/EUV Imaging Telescope (EIT; Delaboudinière et al. 1995). Full-Sun maps of the longitudinal magnetic field in the photosphere are provided by the Michelson Doppler Imager (MDI) on board SOHO (Scherrer et al. 1995).

The soft X-ray light curves at 1–8 Å are obtained from the Geostationary Operational Environmental Satellite (GOES) spacecraft. The X-ray emissions are imaged over a broad energy range by the Reuven Ramaty High-Energy Solar Spectroscopic Imager (RHESSI; Lin et al. 2002).

Figure 1 shows, in a broad frequency range, the evolution with time of the dynamic radio spectrum (70–0.05 MHz) and of the radio flux measured by the NRH at 432, 327, 236, and 150.9 MHz (no spectral observation is available at these frequencies). The bottom plot shows the time profile of the soft (1–8 Å; GOES) X-ray flux. This CME event is associated with a B 3.8 soft X-ray flare, i.e. a weak flare which has no detectable hard X-rays. The flare starts before 13:54 UT and peaks at 14:08 UT in soft X-rays.

Zoom In Zoom Out Reset image size

According to the evolution of the radio and X-ray emissions, we distinguish two main phases: the initiation and eruptive ones. The onset of the initiation phase is characterized by the occurrence of an interplanetary type III burst. During this phase a few radio sources, only detected at 432 MHz, are seen on the images (Figure 2). However, these sources are too weak to be seen on the total flux curve (Figure 1). Later on, the radio emission increases progressively at 432 and 327 MHz. The eruptive phase starts with an increase of the soft X-ray flux and of the radio flux in the four NRH channels. This increase is accompanied by a few interplanetary type III bursts that are also detected in the DAM frequency range. A strong sporadic activity starts in the DAM frequencies around 13:57 UT. This is followed by the buildup of a coronal type II burst and of a group of type III bursts (after ≈14:05 UT). We shall see in the next sub-sections that the spatial evolution of the radio sources leads us to distinguish two parts (I and II) in the initiation phase and three parts (I, II, and III) in the eruptive phase.

Zoom In Zoom Out Reset image size

The observations of STEREO-A and B are rotated toward the SOHO point of view before comparison with the EIT images and identifying the relative position of the radio sources. The EUVI image taken at 13:41:00 UT is the base of our base-difference images. During this CME/flare event, the MDI magnetogram nearest in time is at 14:24:00 UT. The temporal difference between MDI and NRH data introduces a spatial shift that has been corrected.

The most striking feature of this phase is the presence of an interplanetary type III burst which starts at about 8 MHz (Figure 1). A burst source, observed by the NRH at 432 MHz, is located at the southern border of the negative magnetic polarity (Figure 2(a)). The time duration of this radio source, from 13:47:58 UT to 13:49:31 UT, defines the extent of the initiation phase I. The radio source is located above the northern end of bright EUV loops located at the south of the AR (Figure 2(b)). This base-difference image shows a slight increase of the loop brightness, in time coincidence with the radio source.

The EUVI base-difference images also show the presence of a bright front around a pair of dimmings, as well as brightenings around the PIL. These features are characteristics of CME initiation.

At 13:49:37 UT, the flux increases at 432 MHz by a factor up to ≈1.5 (Figure 1). This corresponds to a new radio source located over the PIL of the AR (Figure 3(a)) between the two dimmings (Figure 3(b)). Since this source is located in the internal part of the AR we call it IN. After about 50 s, a 327 MHz source is located at a similar location. During the period from 13:49:31 to 13:51:30 UT, the radio bursts are mainly present at these two high frequencies and the locations are almost stationary. This defines the initiation phase II.

Zoom In Zoom Out Reset image size

In EUV, this phase is seen mostly as a continuation of the previous one, so EUV observations have not sufficient cadence to respond to the change of location for this weak energy release (Figure 3(b)). Finally, in the interplanetary frequency range, this phase has no new radio emission.

This phase, from 13:51:30 UT to 13:57:00 UT, is characterized by an increase of the radio fluxes at the four NRH frequencies (Figure 1). Several type III bursts are detected in the radio spectrum at DAM and WAVES frequencies at the end of this phase. The distribution of the radio sources observed at the four NRH frequencies overlies the AR magnetic field and is aligned along a dominant east–west direction (Figure 4).

Zoom In Zoom Out Reset image size

More precisely, the mean location of the source at 432 MHz is the same as in the initiation phase II (source IN) and is near the soft X-ray source. Double sources are observed at both 327 MHz and 228 MHz (Figures 4(c) and (d)). One is close to source IN and the other one is located in the east part of the AR, so we called it the source E. With an Earth viewpoint, Figure 4(b) shows that most of the radio sources at 150.9 MHz are located, in projection, above the observed EUV bright front, and a few sources are located ahead of the front (which is moving eastward).

All together the radio sources are aligned along the direction that joins the dimming pair. The IN sources are observed at all NRH frequencies, except 150.9 MHz, and the source positions are not significantly dependent on the frequency. The E sources are observed at all NRH frequencies, except 432 MHz; their positions at 150.9 MHz are on average located eastward of the positions at 228 MHz. However, at both frequencies, their distributions overlap the sources at 327 MHz. Then the IN sources trace the internal energy release, while the E sources trace the erupting magnetic configuration.

In the time interval 13:57:00–14:04:30 UT, very few type III-like bursts are observed in the 70–20 MHz frequency range; these bursts are barely detected by the WIND/WAVES instrument. The most significant spectral evolution is the appearance of bursty activity of narrow bandwidth around 35 MHz (Figure 1). Details of this activity are shown on an expanded timescale in Figure 5. This figure displays the DAM spectrum, the corresponding integrated radio flux in the 30–50 MHz frequency range and the flux at 150 MHz. This phase is also characterized by the evolution of the radio sources observed by the NRH, as summarized in Figure 6.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Similarly to the previous two phases, there are almost stationary radio sources over the PIL located near the soft X-ray source (sources IN, Figure 6). However, at 228 and 150.9 MHz, two newer groups of sources appear as marked in Figure 6(b). One is located at the southeast of the AR (called S). The other one is located at the northeast of the AR (called N). We note that there was a spatial gap between the sources E of the previous phase and the new sources N at both 228 MHz and 150.9 MHz; hence, the sources N are newer ones.

Figure 7(a) shows that at 228 MHz, the brightness temperatures of both N and S sources are correlated when both of them are detected. The flux evolution has several main episodes of emission. The S sources are more intense than the N sources. This very variable radio emission is also present at 150.9 MHz during the entire time interval of the eruptive phase II.

Zoom In Zoom Out Reset image size

In Figure 7(b), the sources N1 and N2 at 150.9 MHz (red stars) move toward the northwest (indicated by an arrow). At 228 MHz, the sources N1 and N2 have the same projected positions as at 150.9 MHz. Simultaneously the sources S1 at 228 MHz (green square) and S2 at 228 and 150.9 MHz (green and red stars) move toward the southeast. This evolution trend continues at 150.9 MHz for N3 in the eruptive phase III (yellow stars).

The radio spectrum changes suddenly after 14:04:30 UT (Figure 1), as summarized below.

A coronal type II burst is present at DAM frequencies, while it is later not detected below 10 MHz (not shown). Two groups of type III bursts are observed in the DAM and WIND/WAVES data (Figure 1). They are also detected at 150.9 MHz. Their positions measured at 150.9 MHz coincide with the sources called N3 and S3–S3' in Figure 7(b). The direction finding capabilities of the WIND/WAVES receiver (Manning & Fainberg 1980) also show that the N3 bursts are observed in an azimuthal direction of the order of 5° and with an elevation angle measured from the ecliptic plane of about 5° (courtesy of S. Hoang 2010, private communication), so the IP type III bursts are coming from a solar source located to the northeast. A few bursts are observed at 150.9 MHz at 14:06:20 UT during about 15 s in the sources called S3 (Figure 7(b)). A new sporadic source S3^' at 150 MHz is detected from 14:06:51 UT to 14:07:16 UT. This time coincides precisely with the onset of the type II burst and the strong type III-like emission seen by WIND/WAVES. Note that S3^' is located in projection near the EUV bright/dark features in Figure 7. The comparison in time with STEREO-A running differences EUVI and COR1 images (Figure 8; see also Figure 5 of Cheng et al. 2010) strongly suggests that this type II burst is associated with the onset of the coronal wave near the southern flank of the CME. This is coherent with Cheng et al. (2010) who noted that, during the CME eruption, a streamer disturbance was triggered at the southern side as indicated by a black arrow in the running difference image (14:25–14:15 UT) shown in their Figure 5.

Zoom In Zoom Out Reset image size

ARs are formed by new magnetic flux coming from the convective zone. After a time period of a few days to a week, this magnetic flux starts to be dispersed by the convective motions. This flux dispersion leads to convergent flows toward the PIL, increasing the magnetic shear and forcing flux cancellation (e.g., van Driel-Gesztelyi & Culhane 2009, and references therein). From various observations, this flux cancellation is one of the primary features associated with flares and CMEs (Livi et al. 1989; Kosovichev et al. 2001; Somov et al. 2002; Lin et al. 2004; Wang 2006).

In the studied AR, Cheng et al. (2010) have found a continuous flux cancellation near the PIL lasting for several days. Such cancellation implies the buildup of a flux rope with coronal loops transformed by reconnection into S-shaped loops (e.g., Moore et al. 1995; Green & Kliem 2009). This also implies that the whole AR magnetic configuration increases in size (since the reconnection of sheared field lines removed one anchorage to the photosphere; see, e.g., the MHD simulations of Aulanier et al. 2010, and references therein). This expanding magnetic configuration is expected to reconnect with the surrounding magnetic field (e.g., see Baker et al. 2009, and references therein).

In order to determine the surrounding magnetic field, the potential field extrapolation of the longitudinal photospheric field from Mount Wilson Observatory is computed with a source surface at 2.5 R_☉ (Wang & Sheeley 1992). We find an open positive flux westward of the AR, as shown by the example of the enhanced blue field line in Figure 9(b). As the AR core magnetic field is expanding (as shown by the bright front and dimming appearance), a current sheet is necessarily formed in between the core field and the open flux passing on the side and above the AR because both magnetic fields have a large difference of orientation and they are pushed against each other. At some point of the evolution, when the current sheet is too extended, magnetic reconnection occurs as found in the MHD simulation studied by Masson (2010).

Zoom In Zoom Out Reset image size

From the observed photospheric evolution (flux cancellation; Cheng et al. 2010), the computed coronal magnetic field and the present theoretical knowledge on magnetic reconnection, we deduce the following theoretical scenario, illustrated in Figure 9(a). As the AR core field grows in size, the magnetic connections located at the external border of the core, i.e. between the positive (C) and negative (B) polarities, reconnect with the open field lines anchored at the positive polarity (A) outside the AR. An example of these pre-reconnection field lines is shown in Figure 9(a) with blue field lines (a closed one connecting C to B and an open one anchored at A). A pair of reconnected field lines is shown in red.

Next we confront the above theoretical expectations to the coronal observations which trace the consequences of energy input in both sets of reconnected field lines. As expected, the open reconnected field lines have no counterpart in EUV because the volume of the associated magnetic flux tubes is too large to permit a significant increase of plasma density after the heat input (since the EUV emission depends on the square of the density). However, the close reconnected field lines are seen in EUV as a faint loop (Figure 2(b)). The loop is formed by both evaporation and heating of chromospheric plasma to coronal temperatures. Moreover, a source at 432 MHz is cospatial with the northern footpoint of the faint EUV loop (Figure 2(b)). This source is probably a consequence of the non-thermal electrons accelerated during the reconnection and propagating downward. The non-thermal electrons traveling upward along the newer open field lines produce the type III burst when the conditions for radiation emission are met. It starts at about 8 MHz, which corresponds typically to a radial distance of 2–3 solar radii (Leblanc et al. 1998).

In summary, the magnetic reconnection scenario, expected theoretically, is supported by the observed evidence of the energy input in the coronal plasma: a radio source on the footpoint of an EUV loop and interplanetary type III bursts.

This reconnection reduces the downward magnetic tension, so it decreases the stabilizing effect of the upper magnetic arcade located over the AR. This is expected to favor the later eruption of the core magnetic field located below the arcade. This is similar to the breakout model proposed by Antiochos et al. (1999) in a closed quadrupolar configuration, but here it involves open fields for the larger-scale connectivity region. However, from the observations we cannot tell if this process contributes only marginally or if this process is rather a key process in the initiation of the eruption.

Magnetic flux cancellation was detected at the PIL of the AR during a few days before this CME (Cheng et al. 2010). The coronal activity associated with this slow reconnection is typically difficult to detect. However, Cheng et al. (2010) reported several jets and activity in the filament located over the PIL, up to two days before the CME. They also reported EUV brightenings 1 hr before the eruption starts. Just at the beginning of initiation phase II, a radio source, called IN, is observed at 432 MHz over the PIL, where flux cancellation was observed. However, this new radio source indicates a location in the low corona rather than in the photosphere, which implies that a newer reconnection occurred. Indeed, such reconnection is expected to start behind an erupting flux rope since a current sheet is formed above the PIL in MHD models (e.g., van Ballegooijen & Martens 1989; Forbes & Isenberg 1991; Amari et al. 2000). Such reconnection continues the transformation of the sheared arcade to twisted field lines (as also present in the pre-eruptive phase due to flux cancellation). Then, the magnetic flux involved in the flux rope is expected to significantly grow during the initiation phase (as well as later on).

After about 50 s, the radio burst at 327 MHz appears at the same location. The shifting of the radio burst from high frequency to lower frequency implies a lower plasma density for a radio emission occurring at the plasma frequency (or its harmonic). Since the density is typically decreasing with height, this is an indication that the reconnection site is moving to greater heights as expected in the above MHD erupting models, and in agreement with the observed growing of flare loops (formed below the reconnection site).

The eruptive phase I is characterized by the appearance of new radio sources E at 327 MHz, 228 MHz, and 150.9 MHz (Figure 4), while the source IN was continuously present at 432, 327, and 228 MHz. This phase is also characterized by a peak in the acceleration of the CME leading edge (see Figure 7 of Cheng et al. 2010) and by the development of strong twin dimmings.

In another CME, observed on 2002 June 2, Pick et al. (2005) concluded that both observed radio sources are due to the accelerated particles in or near the reconnecting current sheet located below the erupting flux rope for a CME which was a limb event. We propose the same mechanism for the present event. A part of these particles is expected to be accelerated downward creating the IN radio source, while the other part is injected upward, at the border of the flux rope, creating the eastward radio sources E (Figure 4). However, as described below, the geometry of the observations is different than in the 2002 June 2 event.

The difference in the perspective is illustrated in Figures 10(a) and (b) by the based-difference images from STEREO-A at 13:58:30 UT and from STEREO-B at 13:59:03 UT (we select the beginning of eruptive phase II simply because the asymmetry is even more visible as the CME develops). The double ribbons, the dimmings, and the bright front are present in both images, but their structures in STEREO-A were much more asymmetrical than in STEREO-B. Wood & Howard (2009) concluded that STEREO-A observed this CME from the side and STEREO-B observed it almost along its ejecting direction (Figure 10(c)). The NRH on the Earth observed this CME with a perspective in between the STEREO spacecraft, so the radio sources viewed in projection on the disk are shifted toward the east direction proportionally to their coronal heights.

Zoom In Zoom Out Reset image size

The radio source IN, observed typically at higher frequency, is expected to be at lower height, and so moderately shifted by the side view from the Earth. This source is indeed always located close to the soft X-ray source and to the two eruptive flare ribbons (Figures 4 and 10). Both series of impulsive bursts and a smooth component are distinguished (Figure 11), then we interpret this radio source as the emission of particles accelerated downward from the reconnection region (or possibly from the shock associated with a downward jet), and we locate it above the observed flare loops. Note that the brightness temperature of the smooth component is smaller than 2 × 10⁶ K, so the emitting mechanism can be thermal or non-thermal within flare loops.

Zoom In Zoom Out Reset image size

The radio source E is the counterpart of the radio source IN, as we interpret this radio source as the emission of particles accelerated upward. This interpretation implies that, during the impulsive phase I, the magnetic structure is small enough so that the radio emission level is around (or even above) the top of the erupting structure. Then, the radio source is observed up to the CME bright front (Figure 4(b)).

At 150.9 MHz, part of the sources E is located in front of the bright EUV front observed by STEREO-B (Figure 4(b)). The difference of perspective has been corrected, and we also overplotted the radio data on SOHO/EIT data (as in Figure 7(b)) to check that the easternmost sources E are truly in front of the bright EUV front. So it is plausible that part of sources E at 150.9 MHz is due to the reconnection of the erupting magnetic configuration with the surrounding magnetic field, in particular open magnetic field lines since several type III bursts are detected at lower frequencies (DAM and WIND/WAVES, Figure 1). The electrons accelerated along large or open field lines can excite radio sources located significantly ahead of the front of the CME.

We also note that there is a region in which the projected E source positions, when also detected at 327 MHz, are not located eastward as the radio frequency decreases; they are observed at three frequencies (327, 228, and 150.9 MHz) along approximately the same line of sight and they are located close to the PIL. These observations may suggest that reconnection is driving non-thermal electrons in a comparable way to in source IN (i.e., that reconnection occurs at different locations along the PIL). Figure 4 provides some support to this suggestion since most of the 327 MHz sources are indeed located close to the PIL. A few other ones are located near the edge of the flux rope.

Finally, the radio emissions indicate an intensification of magnetic reconnection above the PIL and below the erupting flux rope. Since part of the magnetic connections to the dense photosphere is cut during the process, it implies a fast growth of the magnetic configuration which is indeed observed as the expansion of the bright front and the intensification of the twin dimmings. The bright front is located inside field lines significantly inclined with respect to the local vertical since with STEREO-A the western part of the bright front is intense and thin (viewed edge-on), while the eastern part of the bright front is diffuse and extended (viewed from the side, Figure 10(a)). Such difference is not intrinsic to the erupting configuration since the eastern and western parts of the bright front have similar width when viewed from STEREO-B (Figure 10(b)). Such inclination, about 50°, is a consequence of the lateral expansion of the erupting field.

Both the lateral and vertical expansions also imply an important increase of the volume of the configuration, and so a decrease of the density and the intensification of twin dimmings (this does not require a true opening of the magnetic field lines). Since the plasma density is higher at low height, as stratified by the solar gravity, this dimming formation corresponds dominantly to a decrease of density at low height, so the twin dimmings mainly trace the locations of the flux rope footpoints.

The eruptive phase II is characterized by the occurrence of new sources N and S (Figure 6), and the flux measured at 150 MHz which appears to present a variability similar to that of the flux measured around 35 MHz (Figure 5). The flux at 228 MHz of the sources N and S is correlated during some time intervals, indicating that they have a common source of energetic electrons (Figure 7).

We suggest that the physical origin of sources N and S is due to reconnection of the erupting flux rope with the surrounding closed magnetic field. This interpretation is consistent with the observations, as follows. First, in the full-Sun potential extrapolation, large-scale field lines are present over the eastern part of the AR (orange lines in Figure 9(b)). Second, during the time period when the flux of sources N and S is correlated the radio emissions are mainly limited to the DAM frequency range, so the accelerated electrons are located in closed loops. The bursty DAM activity could be the signature of the reconnection region.

The interaction of the CME with the surrounding field is pronounced in the eruptive phase III since a series of type III bursts is observed (Figure 1). Moreover, the combination of radio and STEREO images shows that the type II burst has its origin near the southern neighboring of the flux rope and of the CME flank. This is coherent with the results of former studies which showed the link between the fast CME development and the origin of coronal shocks in the low corona. These studies established identifications of shocks propagating at CME flanks with streamers deflected when the shock impinges on them (Vourlidas et al. 2003; Yan et al. 2006; Lario & Pick 2007). Type III bursts observed at the same time originate near the compression region built up between the expanding CME and the neighboring open field line region (Pick et al. 2005; Yan et al. 2006).

Next, we investigate the global evolution of the flux rope from the corona to interplanetary space. The orientation of the PIL in the central part of the AR (Figure 2(a)) and the flare ribbons (Figure 10) are all inclined to the ecliptic plane by about 40°. However, the locations of the twin dimmings indicate that the flux rope has footpoints in an east–west direction. The radio sources also define approximately the direction of the flux rope (Figure 4(b)). From the comparison of the synthetic images built from density models with coronagraph observations by the STEREO spacecraft, Thernisien et al. (2009) and Wood & Howard (2009) concluded that the flux rope was tilted in the counterclockwise direction by 13° and 20°, respectively, to the ecliptic plane. This is a small rotation, at most 20°, between the low corona and interplanetary space.

The inverse S-shaped coronal loops before the eruption, the observed magnetic shear in the flare loops (Figure 4 in Cheng et al. 2010), as well as the separation of the flare ribbons along the PIL (Figure 10) all indicate a negative magnetic helicity (Démoulin & Pariat 2009). From MHD simulations (Török & Kliem 2003), as well as from observations (Green et al. 2007), a counterclockwise rotation is expected in the eruption of a flux rope with negative magnetic helicity. It is a consequence of a magnetic torque present in the erupting configuration, when the magnetic forces are strong enough, so most of the rotation is expected in the low corona. Then, the observed weak counterclockwise rotation of the flux rope is most probably a simple consequence of the eruption and it is plausibly present in the low corona in agreement with the present observational constraints.