The Magnetic Environment of a Stealth Coronal Mass Ejection

AR 11165 has a small bipolar magnetic configuration. As detailed in O'Kane et al. (2019) AR 11165 began to emerge on 2011 February 25 at the polarity inversion line of a previously decayed AR. In fact, the polarity inversion line was the site of repeated flux emergence with further episodes of emergence observed over a time period of ∼44 hr starting at 10:00 UT on 2011 February 28. Figure 3 (bottom panel) shows the evolution of the radial component of the flux in the AR, with a 6 hr running average from 12 minute cadence data. The difference of approximately a factor of 2 between the negative and positive fluxes is likely due to the difficulty in distinguishing the negative flux emerging as part of AR 11165 from that of the preexisting negative flux, some of which has been captured by the method used.

Zoom In Zoom Out Reset image size

The MGN processed 304 Å AIA data show that AR 11165 has a bright core interwoven with dark absorbing plasma threads (see Figure 4). The 304 Å data also reveal three episodes of two-ribbon flare formation at the periphery of the AR bright core: two episodes occur after the flux emergence has ended (see Figure 3) but prior to the stealth CME and one occurs at the time of the stealth CME probable onset. The flare ribbons are faint and lack the classic hooks that would typically be observed in association with the eruption of a flux rope. However, given that stealth CMEs lack clear observational signatures due to their lower energy, weaker flare ribbons (or incomplete ribbons, because not all field lines undergoing reconnection have heated plasma at their footpoints) are to be expected. Then, the faint flare ribbons are in keeping with the stealth nature of the CME. Figure 4 shows that the flare ribbons are at least partially rooted in the dispersed magnetic field of the previously decayed AR, rather than involving the newly emerging flux. This is most apparent in the flare ribbons that are located on the positive polarity field and indicates that the associated reconnection process involves structures that are overlying the emerging bipole.

The first episode of flare ribbon formation begins on 2011 March 2 at ∼06:10 UT (Figure 4, panel 2). The ribbons remain illuminated for around 4 hours and are also temporally coincident with a reorganization of the corona as seen in AIA 193 Å data. Details of the coronal evolution are given in O'Kane et al. (2019) but in summary the lower altitude section of a loop structure apparently connected to the eastern ribbon dims and appears to expand at ∼06:10 UT on 2011 March 2. STEREO-A EUVI 195 Å data also show the motion of this loop structure on the eastern side. From this perspective a southward motion of the loop leg is also detected.

The second episode of two-ribbon flare formation begins on 2011 March 2 at ∼11:00 UT and the ribbons remain bright for several hours fading by ∼15:00 UT on 2011 March 2 (Figure 4, panel 3). As is the case with the first flare ribbon episode, the ribbons of the second episode form and are extended on the western side further to the north than the AR main positive polarity concentration. Both the first and the second ribbon episodes are temporally coincident with coronal brightenings observed in the AIA 335 Å wave band data at 06:30 UT and 15:45 UT on 2011 March 2. STEREO EUVI 195 Å data also show a brightening of the active region and the formation of new loops at 07:30 UT and 11:30 UT, coincident in time with the first and second episodes of ribbon formation respectively. However, these brightenings are not sufficient to produce flares that are detectable in the GOES soft X-ray light curve. No separation of the two ribbons is observed in either episode in keeping with the confined nature of the flares (i.e., no eruptions were observed). This, along with the location of the ribbons at the AR periphery, indicates that the reconnection site is located above its core field.

At ∼22:30 UT on 2011 March 2 the two ribbons activate for a third time; however, this time they separate away from the AR core (Figure 4, panel 4). The two ribbons are observed to separate from each other over an approximately 8 hour period. Interpreting these flare ribbons in the context of the standard two-ribbon flare model and subsequent studies (see Moore & Labonte 1980; Kitahara & Kurokawa 1990; Fletcher et al. 2004; Qiu 2009) would indicate that, during the first two episodes of two-ribbon formation, a reconfiguration of the coronal magnetic field above the newly emerged flux (AR core) occurs. This reconfiguration involves magnetic reconnection, which most probably played a key role in forming the structure that then went on to erupt during the third episode of ribbon occurrence. This finding adds to the growing body of work showing that eruptive structures can form during reconnection episodes in the corona that produce confined flares and that could have a flux rope configuration (Patsourakos et al. 2013; James et al. 2017). From the displacement of the flare ribbons with respect to a direction parallel to the polarity inversion line and the shear of the AR coronal loops we infer that its magnetic field was of positive chirality, in agreement with the positive magnetic shear found in coronal loops (see also Section 4.1 and Figure 9).

STEREO-A COR1 data show that prior to the stealth CME two other small and faint Earth-directed eruptions occur. The first being initially seen in the COR1 field of view at 06:00 UT and the second at 11:30 UT on 2011 March 2. However, these eruptions originate from the northern hemisphere and not from AR 11165. Their small and faint structure makes determining their specific source region challenging. A small eruption is observed from AR11167, which is likely the source of the CME observed at 06:00 UT. Meanwhile, AIA and EUVI data show ongoing dimming on the southern side of NOAA ARs 11163 and 11164 between 00:00 UT and 08:00 UT on 2011 March 2, which may be associated with the CME observed in COR1 at 11:30 UT.

The lack of strong plasma emission from the erupting structure in EUV wave bands (and hence the classification of this event as a stealth CME) prevents a detailed analysis of the CME initiation and rise profile in EUV imaging data. However, a bright concave-up structure, that is best seen in STEREO-B EUVI 171 Å data, is observed at 00:14 UT (Figure 5, panel (e)), 1.75 hr after the third episode of flare ribbon formation. The 2 hr cadence of EUVI 171 Å data prevents a good analysis of the temporal evolution of this structure but it is seen again in the following image at 02:14 UT on 2011 March 3 (Figure 5, panel (f)). The concave-up structure observed in CMEs is interpreted as indicating the underside of a flux rope.

Zoom In Zoom Out Reset image size

Likewise, in the coronagraph data (COR1), there is no discernible structure at the leading edge of the stealth CME, only a concave-up structure at its trailing edge. The concave-up structure is followed in STEREO-B COR1 in order to determine the speed and acceleration of the eruption. However, it is not possible to confidently track the same plasma structure between EUVI data and coronagraph data due to the differing emission processes, so the kinematics of the CME are calculated only from STEREO-B COR data with a note that the EUVI-B concave-up structure is measured to be at a height of 1.54 R_⊙ from Sun center in the plane of the sky at 00:14 UT and 1.67 R_⊙ at 02:14 UT on 2011 March 3 (Figure 5, panels (e) and (f)), giving a very approximate plane-of-sky rise speed of 17 km s⁻¹. These heights are consistent with the underside of the stealth CME when it is first detected in STEREO-B COR1 data early on 2011 March 3 (see Figure 6).

The top panel of Figure 6 shows the height-time variation of the erupting CME using data from STEREO-B COR1 along a radial slice at 120^◦ measured in the clockwise direction from solar north. Figure 6 panel (b) shows the identified evolution of the underside of the stealth CME fitted using a Savitsky–Golay bootstrapping technique (see Byrne et al. 2013). This approach enables the point-to-point variation of the speed and acceleration of the CME to be estimated as shown in panels (c) and (d) of Figure 6. The initial measured speed of the white light CME, as determined from coronagraph data, at 04:23 UT on 3 March was found to be ∼31 km s⁻¹, with an acceleration of ∼102 m s⁻², consistent with the speed found from EUVI-B. From the kinematic data and the AIA 304 Å flare ribbon formation we conclude that the stealth CME initiation began in the time period between 22:30 UT on 2011 March 2 (the time in which the third episode of flare ribbons begin) to 00:14 UT on 2011 March 3 (the time in which the concave-up structure is first observed in EUVI-B).

Zoom In Zoom Out Reset image size

The evolution of the corona at a larger scale, encompassing AR 11165 and the complex formed by ARs 11163 and 11164 to the north, is studied using SDO/AIA images and STEREO/EUVI data. As seen in STEREO EUVI 195 Å wave band data from both Ahead and Behind spacecraft, transequatorial loops exist between these ARs in the northern and southern hemispheres. STEREO-A shows that these loops are visible by 18:30 UT on 2011 March 2, while they are observed by 19:55 UT in STEREO-B. From the Earth perspective, the transequatorial loops are observed in AIA 193 Å data around 19:40 UT on 2011 March 2 (Figure 7, left panel). These observations collectively show the presence of these large scale loops before the onset of the stealth CME but after the time period in which the coronal field above AR 11165 was reconfigured via magnetic reconnection as evidenced by the two-ribbon flares. We have searched for transequatorial loops in soft X-ray imaging data using Hinode/XRT; however, there is a gap in the full-disk XRT data between 2 March at ≈06:20 UT and 2011 March 3 at ≈06:20 UT.

Zoom In Zoom Out Reset image size

After the stealth CME, the transequatorial loops are still observed, albeit at a higher altitude indicating that a reconfiguration occurred, perhaps due to reconnection driven by the expanding structure of the stealth CME. STEREO-A EUVI data indicate that these higher altitude loops can be seen by 08:25 UT on 3 March and in STEREO-B EUVI data by 19:55 UT on the same day. The loops are persistent and continue to be observed into 2011 March 4 at which time STEREO-B data show that the top of the transequatorial loops reach a height of ≈234 Mm above the photosphere (1.34 R_⊙ from Sun center, see Figure 7). The loops are not clear in AIA at this point, presumably due to the low plasma density of the loops against a higher density background.

Figure 8, two left panels, shows the photospheric magnetic field configuration and the coronal emission structures of the northern hemisphere active region complex formed by ARs 11163 and 11164. These ARs exhibit sheared loops as can be seen by the way they cross our approximated polarity inversion lines (PILs, depicted as red dashed lines). In particular, the PIL in AR 11164 is quite complex and we only trace the main one. The global coronal structure of this AR displays roughly an inverse S-shape. The PIL of AR 11163 is much simpler and the loop inclinations to the PIL marginally also indicate a negative shear (see, e.g., Palmerio et al. 2017).

Zoom In Zoom Out Reset image size

Figure 8, two right-hand-side panels, show that AR 11165 has positively sheared loops as is expected from the hemispheric helicity trend for ARs in the southern hemisphere (Pevtsov & Balasubramaniam 2003). To confirm the sign of the shear, we have extrapolated the line-of-sight magnetic field of AR 11165 into the corona, using the HMI magnetogram under the linear force-free field (LFFF) approximation (${\rm{\nabla }}\vec{B}=\alpha \vec{B}$, with α constant, Mandrini et al. 1996; Démoulin et al. 1997). An example of the model is shown in Figure 9. The value of the free parameter of the model, α, is set to best match the observed loops at the time of the magnetogram used for the extrapolation, following the procedure discussed by Green et al. (2002). The best-matching value is positive, α = 6.3 × 10⁻³ Mm⁻¹. Our model also carries out a transformation of coordinates from the local AR frame to the observed one (see Démoulin et al. 1997) so that our computed field lines can be compared to the AIA observed loops (as shown in Figure 9).

Zoom In Zoom Out Reset image size

Since the northern complex of ARs and the stealth CME source region are connected by transequatorial loops (see Section 3.4), we now investigate the magnetic connectivity between them. Because of the presence of moderate opposite magnetic shear in the three ARs, we extrapolate the same HMI photospheric magnetogram used to model the coronal field of AR 11165 (Figure 9) under a potential field approach, using an area that encompasses both the northern and southern regions as the boundary condition. A Cartesian model at such a large size-scale is disputable; however, its use allows us to consider a single magnetogram including the three ARs with a higher spatial resolution (i.e., our nonuniform grid size corresponds to HMI spatial resolution at the center of our computational box) and closer in time to the analyzed events, reflecting more accurately the photospheric conditions. The results found in this section will be verified in Section 4.3 using a global model in spherical coordinates.

We searched for topological structures in the extrapolated model and found the presence of a magnetic null point to the northwest of AR 11165 (see Figure 10). The local field connectivity around a magnetic null can be described using the linear term of the Taylor expansion of the field around such a point (see Démoulin et al. 1994; Mandrini et al. 2014, 2015, and references therein). From the diagonalization of the Jacobian matrix of the field, we find three eigenvectors and the corresponding eigenvalues, which add up to zero to satisfy the field divergence-free condition. The eigenvalues are real for coronal conditions (a force-free field). A positive null point has two positive eigenvalues and conversely for a negative null. Figure 10 illustrates the location of a negative null point found at a height of 234 Mm above the photosphere (1.34 R_⊙ from Sun center); this is indicated by the intersection of three segments that correspond to the directions of the three eigenvectors of the Jacobian matrix. These segments are color coded to indicate the magnitude of the corresponding eigenvalues. For a negative null, dark blue (light blue) corresponds to the highest (lowest) negative eigenvalue in the null fan plane and red to the null spine eigenvalue.

Zoom In Zoom Out Reset image size

We trace sets of field lines in the neighborhood of the null point, with starting positions in the direction of the three eigenvectors, to explore the different connectivity domains. This connectivity is illustrated in both panels of Figure 10. The left panel corresponds to the observer's point of view along the Sun–Earth line, while the right panel shows the location of the null point as viewed from the solar south. The field lines in both panels have been drawn in blue and red to indicate what we envisioned could be the pre-reconnection set (blue field lines) and the post-reconnection set (red field lines), following the evolution of the emission structures and transequatorial loops discussed in Sections 3.2 and 3.4. The illustrated connectivity shows that the positive polarity of AR 11165 is connected to the dispersed negative polarity of AR 11163 and its negative polarity to the positive polarity of AR 11164. The footpoints of the field lines in AR 11165 are located in the vicinity of the twin EUV dimmings observed in association with the stealth CME, as shown in Figure 8 of O'Kane et al. (2019). The location of the dimming regions, their evolution over a 9 hr period and their relationship to the CMEs are discussed in detail in that article.

The magnetic null point found in the local Cartesian model could play a crucial role in facilitating the initiation of the stealth CME; therefore, we have searched for its presence in a global potential field source surface (PFSS) model, which could also give us clues about the location of "open" field lines, their shape, and their possibility of channeling the stealth CME and influencing its propagation direction into the interplanetary (IP) medium. Because of the relevance of the role of this null point, we have checked the robustness of its presence using two different approaches to compute the PFSS global model.

Both PFSS models use as their lower boundary condition the HMI magnetic field synoptic map from Carrington rotation (CR) 2107, that ran from February 16 to 2011 March 16. HMI synoptic maps are computed from line-of-sight magnetograms by combining central meridian data from 20 magnetograms collected over a 4 hr interval. A synoptic map is made with the magnetograms collected over a full solar rotation (≈27.7 day interval) with 3600 × 1440 steps in longitude and sine latitude. Both models assume that the magnetic field becomes purely radial at a height, called the source surface, which is set to the value 2.5 R_⊙. Full details concerning the construction of synoptic maps can be found on the HMI website (http://jsoc.stanford.edu/jsocwiki/SynopticMaps).

Our first PFSS modeling approach uses the Finite Difference Iterative Potential-Field Solver (FDIPS) code described by Tóth et al. (2011). The FDIPS code, which is freely available from the Center for Space Environment Modeling (CSEM) at the University of Michigan (http://csem.engin.umich.edu/tools/FDIPS), makes use of an iterative finite-difference method to solve the Laplace equation for the magnetic field. The spatial resolution of this particular model is 1° in longitude (360 longitudinal grid points), 0.11 in the sine of latitude (180 latitudinal grid points) and 0.01 R_⊙ in the radial direction. We searched for a null point using a method similar to that discussed in the previous section but using a spherical geometry for the coronal field. Figure 11, left panel, shows the location of a null point and the connectivity of the field when the Sun is rotated so that AR 11165 is facing the observer. This null point is located at a height of ≈118 Mm, lower than in the Cartesian model, but the connectivity clearly resembles the one of that model (Figure 10).

Zoom In Zoom Out Reset image size

Our second approach uses the PFSS solver in Python (pfsspy; Yeates 2018; Stansby 2019). In this case we resampled the original field spatial resolution to a 540 and 270 pixels in longitude and latitude, respectively. This solver also uses finite differences, and follows the method in van Ballegooijen et al. (2000) to effectively compute discrete spherical harmonics global functions. We also found a magnetic null point at a similar location and with a similar connectivity in its neighborhood as before, but at a height of ≈104 Mm. Therefore, though there are differences that come from the different approaches used, and different boundary conditions between the local and global models, the presence of this null point between the AR complex in the north and the stealth CME source region is consistent. Finally, both PFSS models suggest that there are "open" field lines above the stealth CME source region that are inclined to the south (Figure 11, mainly the right panel).

Zoom In Zoom Out Reset image size

The inclination of the "open" field lines as well as the reaction of the surrounding potential field, present over AR 11165, may influence the CME propagation direction (e.g., Cremades et al. 2006; Gopalswamy et al. 2009). The erupting magnetic field is both compressing and bending the surrounding magnetic field. This induces a reactive force with both a gradient of magnetic pressure and a magnetic tension, with both forces proportional to B². B² is seen to be higher above the two northern hemisphere ARs, lower in the region of the null point, and increases again over AR 11165 (see Figure 11). The CME is therefore likely to be influenced by B² present in the surrounding potential field, in agreement with previous studies (Gui et al. 2011; Kay et al. 2013; Sieyra et al. 2001, and references therein) and to be deflected southward contrary to the general equatorial deflection of many CMEs (Cremades & Bothmer 2004). The presence of open field lines, deflected toward the south, is also expected to deflect the CME along them (Mäkelä et al. 2013; Wang et al. 2020, and references therein).