VELOCITY AND MAGNETIC TRANSIENTS DRIVEN BY THE X2.2 WHITE-LIGHT FLARE OF 2011 FEBRUARY 15 IN NOAA 11158

AR NOAA 11158 appeared near the disk center at location S19W03 on 2011 February 12 in the rising phase of the current solar cycle 24. Subsequent to its birth on February 12, it evolved very rapidly and quickly developed from a simple β- to a very complex βγδ-configuration by 2011 February 15. (A detailed study of the surface and sub-surface evolution of this AR has been undertaken and results therefrom will be communicated in a subsequent paper.)

During its disk transit, NOAA 11158 produced several C-class and 5 M-class flares. It also unleashed the first X-class flare of the current solar cycle observed on 2011 February 15. This event was a two-ribbon WLF as seen in HMI WL images. The flare peaked at 01:56 UT, when the AR was centered around S20W12 and was registered at an X2.2 magnitude on the Richter scale of solar flares. In Figure 1 (the top two rows), development of the flare is shown using SOT Ca ii H filtergrams of the AR NOAA 11158 on 2011 February 15.

The co-aligned, extracted HMI and AIA images of AR NOAA 11158, obtained in different wavelengths during the flare, are shown in Figure 2. The top panel shows the AIA images taken at the rising phase of the flare around 01:46 UT in 1700, 304, and 193 Å (from left to right). These wavelengths correspond to different layers of the solar atmosphere, viz., temperature minimum region, chromosphere, and corona, respectively. The bottom panel shows the HMI WL image, magnetogram, and Dopplergram taken around the peak phase of the flare at 01:53 UT.

Zoom In Zoom Out Reset image size

Animations of HMI magnetograms and Dopplergrams showed "magnetic" and faint "Doppler" transient features (TFs) appearing near the umbral boundary of the main sunspot during the peak phase of the X2.2 flare. The dotted (dashed) curves labeled L1 and L2 represent the locations of the TFs during the peak phase of the flare, at 01:53 UT, when the features were prominently seen in the HMI magnetogram. These curves are drawn over the subsequent figures for reference. It is evident from the bottom panel that these transients appeared in rather narrow ridges along the umbral boundaries of the opposite polarity sunspots. The images in the top panel further illustrate that L1–L2 were associated with the flare ribbons observed in different AIA wavelengths.

Profiles of relative mean flare intensity ΔI/I with respect to the quiet Sun in different AIA wavelengths, obtained from the extracted and co-aligned images, are shown in Figure 3. GOES soft X-ray (SXR) profiles in the wavelength range 1.0–8.0 (0.4–5.0 Å) are shown by solid (dotted) curves. RHESSI HXR light curve (dashed) in the energy band 25–50 keV is also shown, which is scaled in the plot range as marked in the left-hand side ordinate. GOES 1.0–8.0 Å and GOES 0.4–5.0 Å emissions peaked at 01:56:49 and 01:55:50 UT, respectively, while the RHESSI HXR emission peaked a few minutes before, at 01:53:40.

Zoom In Zoom Out Reset image size

The relative flare intensity profiles obtained in different AIA wavelengths are also plotted in Figure 3, according to the scale given in the right-hand side ordinate. As evident, the flare intensities in the AIA wavelengths of 193, 304, and 1700 Å started rising around 01:46:53, 01:44:41, and 01:45:44 UT, and peaked at 01:58, 01:56, and 01:54 UT, respectively. The flare decayed in all the wavelengths by 02:40 UT, i.e., after 1 hr of the onset of the flare. It is to note that the AIA emissions, particularly in 304 Å, saturated during the peak phase of the flare. Hence it is difficult to precisely identify the flare maximum in these wavelengths. However, it is evident that the AIA intensity profiles showed a similar trend to the GOES SXR, while, the HXR profile was much narrower with steep rise and decay phases.

The light curves in different wavelengths follow the standard flare model for energy propagation at different heights in the solar atmosphere (Kane 1974). The general understanding of the flare physics is based on the concept of the reconnection of the magnetic field lines at coronal heights (Sturrock 1966; Hirayama 1974; Kopp & Pneuman 1976). Energy released during a flare covers a wide range of wavelengths; however, it is easier to observe in certain spectral lines as shown in Figure 3. During the pre-flare stage, where the release of the stored magnetic energy is triggered, the chromospheric flare ribbons are easily seen (see Figure 1, top). The increasing separation of the flare ribbon with time results from the reconnection of the magnetic field lines at successively increasing coronal heights. Further, the energetic particles released in the corona during a flare take increasingly longer time to propagate through the denser layers downward to the photosphere. This agrees with the observed flare light curves in 304 Å that started rising at 01:44:41 UT and thereafter in 1700 Å corresponding to the photosphere and the temperature minimum region. As the flare progressed, the temperature also increased making the flare ribbons visible in higher temperature plasmas in the coronal regions corresponding to the 193 Å emission.

The observed "magnetic" and "Doppler" TFs appeared during the peak phase of the X2.2 flare of 2011 February 15 (cf. Figure 2). Similar features were reported earlier during the more energetic X17/4B and X10/2B flares on 2003 October 28 and 29 (Maurya & Ambastha 2008, 2009, 2010a) and the X5.6 flare on 2001 April 6 (Qiu & Gary 2003). Venkatakrishnan et al. (2008) also reported cospatial Doppler ribbons associated with the Hα flare ribbons of the X17/4B flare. These TFs, usually located around the cooler umbral boundary of the sunspots, were found to be largely cospatial with the flare ribbons observed at different heights of the solar atmosphere, viz., the chromosphere, transition region, and corona (see Figures 1 and 2).

A time sequence of the consecutive difference images of the HMI magnetograms of AR NOAA 11158 in Figure 1 shows dark patches representing the magnetic transients along the curves L1 and L2. These TFs appeared and faded in a few minutes' period (01:50–02:02 UT) during the impulsive phase of the flare. From Figure 1, it is evident that the transients were cospatial with the observed Ca ii H line flare ribbons. However, while the Ca ii H flare ribbons separated away, the TFs remained nearly stationary with time.

Using an automated method described in Maurya & Ambastha (2010b), we determined that the Ca ii H flare ribbons separated out with an average velocity of 8 km s⁻¹. This is much smaller as compared to the earlier reported velocities in the range of 50–75 km s⁻¹ for the flare ribbons and the TFs observed during the X17/4B superflare of 2003 October 28 in NOAA 10486. In that event, the TFs rapidly moved away from the weak field regions of the PIL reaching a maximum separation of around 27.9 ± 0.4 Mm. Another part of the TF observed near the leading sunspot (i.e., strong magnetic field region), on the other hand, remained stationary (MA09). Similarly, the TFs associated with the X10/2B flare on 2003 October 29 in the same AR occurred in the umbral/penumbral region of the following sunspot. Those TFs also remained stationary as observed here in the case of the X2.2 flare on 2011 February 15.

3.2.1. Anomalous Reversal of Magnetic Polarity

We measured magnetic flux (MF) and DV in AR NOAA 11158 along a horizontal line PQ (Figure 4, top) drawn through the location "A" of the observed TF before and during the flare maximum in order to investigate their variations. The line PQ was selected by plotting the MF profiles along a horizontal raster moving from the bottom to the top of the selected magnetograms. The profiles of MF and DV along PQ are shown in Figure 4 (bottom), where the solid and dashed curves represent the MFs/velocities before (01:39:00 UT) and during (01:53:15 UT) the flare maximum, respectively. The MF profiles during the pre- and peak phases of the flare matched at all points along PQ except at the location "A" (see Figure 4(c)). It gives clear evidence of an abnormal sign reversal in MF polarity at "A" around the peak phase of the flare from the pre-flare phase. The DV profiles along PQ obtained during the pre- and peak phases of the flare, however, do not match so well as in the case of the MF profiles (see Figure 4(d)). This is mainly due to the effect of the solar p-mode oscillations. But, there is also a large increase in the DV during the peak phase of the flare at "A," i.e., the location of the sign reversal in the magnetic polarity.

Zoom In Zoom Out Reset image size

What processes caused the observed abnormal MF sign reversal and the DV enhancement? Are they real changes? Or, is it related to the line-profile change (Ding et al. 2002; Qiu & Gary 2003) or some other process such as the impact of a shock wave (Zharkova & Zharkov 2007) on the Fe i line-production region? More recently, Kosovichev (2011) reported "sunquake" sources in the regions of TFs and suggested that these arise due to thermal and hydrodynamic effects of high-energy particles heating the lower atmosphere. Apart from the observed transients in MF and DVs, an increase in the continuum intensity was seen in the form of WLF ribbons. Notably, the flares studied in MA09 were also WL events during which similar TFs were detected.

One may attribute the observed sign reversals to the expected change in the spectral line profile from absorption to emission occurring during these large WLFs. However, the non-LTE calculations by Ding et al. (2002) were carried out for the spectral line Ni i 6768 Å, used in GONG and MDI instruments for magnetic and Doppler measurements. They showed that this line, formed in thermally stable regions, can turn into emission only by a large increase of electron density and not by heating of the atmosphere by any other means. They suggested that such non-thermal effects are most pronounced in a cool atmosphere where the continuum is maintained at a low intensity level. The observed flare-associated sign reversal in MF reported here also occurred in the cooler umbral boundary of the sunspot in agreement with Ding et al. (2002). However, there are also exceptions as reported in the case of the X17/4B flare on 2003 October 28 in NOAA 10486, where the TFs formed in the weak magnetic field regions around the PIL.

If the line formation model given by Ding et al. (2002) for Ni i were to be applicable also to the Fe i line used in the HMI measurement, one would suggest that the observed sign reversal is due to the change in the line profile. Observational evidence of the Ni i line-profile change was not available for the WLFs previously studied in MA09 due to the lack of spectral data from MDI and GONG. Therefore, our interpretation of the origin of the transients was limited to their spatiotemporal association with WLF kernels and non-thermal HXR sources. However, as the required spectral data are now available from the HMI for the X2.2 flare studied here, we can look for the observational evidence of any flare-related changes in the Fe i line profiles at the location of the TFs. The results are discussed in Section 3.4.

3.2.2. Spacetime Maps of Magnetic Flux and Doppler Velocity

In order to study the temporal variations of the TFs, we constructed spacetime maps of MF and DV in the AR along the line PQ as shown in Figure 4(top). We added the corresponding magnetic and DV values along PQ during the period 01:38:15–02:15:45 UT on 2011 February 15. We also constructed a similar spacetime map for the HMI WL images.

The spacetime maps thus constructed are shown in Figure 5(top). Here, the x- and y-axes represent time and longitudinal positions along PQ, respectively. As observed during the period 1:47–2:08 UT, these maps show a clear signature of the magnetic and Doppler transients in the box drawn along the time line EF at the longitude 195''. MF and DV values along the line EF are plotted as dashed-dotted curves in the corresponding bottom panels. The solid (dotted) curves show the integrated GOES-15 SXR flux in the wavelength range 1.0–8.0 (0.5–4.0) Å, while the dashed curves represent the WLF intensity scaled on the left-hand side of the y-axis.

Zoom In Zoom Out Reset image size

It is evident from Figure 5(bottom) that MF and WLF began rising at 01:47 UT with a delay of 1 (2) minutes from the rise of the GOES flux in the wavelength range 1.0–8.0(0.4–5.0) Å. However, MF reached its peak much faster at 01:53:15 UT as compared to various other flare intensities: WLF at 01:54:00 UT and GOES SXR 0.4–5.0 (1.0–8.0) Å at 01:55:50 (01:56:49) UT. Similarly, the MF and WLF profiles decayed much faster than the GOES flux. The DV profile also peaked simultaneously with the WLF and MF profiles. Its profile returned back to the original level at 2:00 UT, i.e., about 10 minutes before the MF. From these time profiles at "A" (Figure 4, top), we thus infer that the observed transients and WLF had similar temporal profiles, reaching their peaks 2–3 minutes prior to the peaks for the GOES fluxes. Also, the DV profile was comparatively much sharper than the MF profile.

If the MF and DV transients occurred due to line-profile changes (Qiu & Gary 2003), then they both are expected to exhibit similar temporal evolution. Similarity of the transient and WLF profiles suggests that the thermal heating process may have caused these transients. The time delay between the TFs and GOES SXR peaks indicates the travel time that the energetic charged particles released in the corona during the flare took to reach the photosphere and to produce the TFs and subsequent thermal heating giving rise to the SXR flux. We further discuss the spatial and temporal variations of TFs along the curves L1 and L2, and their association with WLF, Ca ii H, and HXR kernels in the following section.

In order to identify the temporal and spatial association of RHESSI HXR sources and flare ribbons with the TFs, we overlaid the 25–50 keV RHESSI HXR contours on the consecutive magnetogram and Dopplergram difference images (Figure 6). The transients are seen in the respective background images in the time period 1:49:57–02:01:12 UT. The HXR contours (dash-dotted) were drawn at 40%, 60%, and 80% levels of the maxima. These features were observed around the magnetic polarity inversion line (PIL); not drawn here to avoid overcrowding. The Ca ii H flare ribbons (solid contours) and the transients, i.e., L1 and L2, appear to be well correlated.

Zoom In Zoom Out Reset image size

The HXR sources evolved both spatially and temporally along L1 and L2 as shown in panels (a)–(d). The source region of HXR moved toward the west, parallel to the magnetic neutral line during the period when the TFs were seen. The observed motion of HXR sources is attributed to the temporal movement of the reconnection site as reported earlier by Masuda et al. (2001), Bogachev et al. (2005), and Zhou et al. (2008). From Figure 6, it is evident that the HXR sources were not very well correlated with the TFs during the entire peak phase of the flare as compared to the Ca ii H ribbons. Therefore, we infer that HXR sources alone were not the cause or effect of the TFs. This inference agrees with our earlier results obtained for the X10/2B flare on 2003 October 29 (MA09).

Although the WLF ribbons were faintly observable in the HMI continuum animations, they are rather difficult to identify even in the difference images. However, from the animations, the WLF ribbons appeared to be correlated with the TFs. This is further corroborated by the WLF intensity, magnetic and DV profiles of Figure 5(bottom).

Qiu & Gary (2003) reported that (1) the apparent sign reversal occurs in cool, strong-field (>1000 G) regions within sunspot umbrae, (2) locations of the anomaly are exactly co-aligned with the thick-target HXR sources, and (3) the transient reversal flux is temporally correlated with the HXR flux. Based on these results, they proposed that at the flare kernels the Ni i absorption line profile is either temporarily turned into emission or significantly broadened with a strong central reversal due to the non-thermal beam impact on the umbral atmosphere. However, our results for this X2.2 flare are not in full conformity with Qiu & Gary (2003) as we did not find a very good association of HXR and TFs. Therefore, the TFs observed during the flare on 2011 February 15 and the X10/2B flare on 2003 October 29, reported in MA09, are not fully explained only by the non-thermal process suggested in Qiu & Gary (2003).

Earlier studies based on numerical models (Machado et al. 1980; Vernazza et al. 1981; Ding & Fang 1989; Ding et al. 2002; Qiu & Gary 2003) have shown that the flare-associated transients may be attributed to a change in the spectral line profile. However, due to the non-availability of observational spectral data, the transients could be examined using only the imaged data of the flares at various wavelengths, and inferences were drawn based on this indirect method. We are now able to address this issue using SDO/HMI spectral observations. The HMI observes the Sun in two circular polarizations at six wavelength positions (±34.4, ±103.2, and ±172.0 mÅ) of the Fe i spectral line. Using these observations one can construct the line profile for any desired location of the full disk Sun. The spectroscopic imaging of HMI observations is described in a recent paper by Martínez Oliveros et al. (2011).

We have already shown the strong spatial association of magnetic transients with the WLF kernels. Figure 7(top) shows the consecutive difference magnetogram and WL images of the AR NOAA 11158 during the peak phase (01:53:15 UT) of the X2.2 flare. We selected a quiet (P₁) and a transient (P₂) location as marked by the arrows. Figures 7(c) and (d) show the line profiles at these two locations. The solid (dashed) curve represents the line profile during the pre- (peak) phase of the flare. These curves were obtained from a four-term Gaussian fitting (except for the peak profile at P₂ where such a fitting was not possible). It is evident that the line profile turned to emission at the wavelength Δλ = −34.4 mÅ and became broader during the peak phase of the flare as compared to the pre-flare phase. This change in the line-profile shape is expected to be the main reason for the observed flare-associated sign reversal of MF and the enhancement of DV at location P₂ during the peak phase of the flare. No such change was found at P₁, taken as a reference location far away from the flare.

Zoom In Zoom Out Reset image size

From the HMI spectral line-profile data, we now attempt to explain the reason for these transients. In the SDO/HMI measurements, magnetogram and Dopplergram maps of the full disk Sun are computed from the phase of the Fourier transform (first component) of the line-profile values. The phase is determined for the left-circular polarization (LCP) and right-circular polarization (RCP) components independently, and the Dopplergrams and magnetograms are constructed from the mean and difference, respectively (Schou et al. 2012). However, the standard procedure assumes only a moderate line shift due to DV and Zeeman splitting. The regions having large changes in the line-profile shape, e.g., line reversal, and large shifts would not be covered by the algorithm, resulting in artifacts in these measurements. For this large X2.2 flare, the spectral line core at the locations of the transient evidently turned from absorption to emission. Therefore, it is difficult to get "correct" values of MF and DV using the above algorithm.

In the following, we have used a two-point algorithm to illustrate the result of the observed spectral line reversal at the transient locations. The MF and DV values can be computed from the line intensities in the LCP and RCP as follows: let I^{LCP, RCP}_i, i = 0, ..., 5 be the line intensities in the two circular polarizations, LCP and RCP, at the six wavelength positions, Δλ_i = 68.8 i − 172.0, centered around the Fe i spectral line. Then the mean intensities in the blue and red wings of the line can be written as

$$I_b^{\rm LCP, RCP}=\frac{1}{3}\sum _{i=0}^2{I_i^{\rm LCP, RCP}}$$

and

$$I_r^{\rm LCP, RCP}=\frac{1}{3}\sum _{i=3}^5{I_i^{\rm LCP, RCP}},$$

respectively. The parameters sensitive to MF (say, ψ_B) and DV (say, ψ_V) fields can be written as

$$\psi _B=\frac{1}{2}\left[\left(I_b^{\rm RCP}+I_r^{\rm LCP}\right)-\left(I_b^{\rm LCP}+I_r^{\rm RCP}\right)\right]$$

and

$$\psi _V=\frac{1}{2}\left[\left(I_b^{\rm LCP}+I_b^{\rm RCP}\right)-\left(I_r^{\rm LCP}+I_r^{\rm RCP}\right)\right],$$

respectively.

The parameters ψ_B and ψ_V for any location of the AR NOAA 11158 can thus be computed using the 12 line intensity observations at six wavelength positions in the two circular polarizations. However, it is important to note that as such, these parameters do not represent the magnetic and DV fields. They are defined here only to examine the observed magnetic polarity sign reversal and DV enhancement at the transients' locations (see Figures 4(c) and (d)). Furthermore, this is used to illustrate the observed artifacts in magnetic and DV measurements resulting from the standard procedure adopted in SDO/HMI.

The values of the parameters ψ_B and ψ_V at location P₂ during the pre- (peak) phase of the flare were found to be −0.034(−0.008) and 0.036(0.054), respectively. It is evident from these computed values that the parameter ψ_B tends toward a positive value during the peak phase of the flare as compared to that in the pre-flare phase. This indicates the reason for the observed sign reversal in the magnetic polarity at location P₂ or at A (see Figure 4(c)). Similarly, the velocity parameter ψ_V increased during the peak phase of the flare, indicating the large enhancement in the HMI DV at P₂ or at A (see Figure 4(d)).

Figure 8 shows the temporal variation in the normalized line intensities at the six wavelength positions ±34.4, ±103.2, and ±172.0 mÅ, where panels (a) and (b) correspond to locations P₁ and P₂, respectively. Panel (a) shows that there is no significant change in the line-profile shapes at P₁ from the pre- to peak phases of the flare. The bottom panel (c) shows the corresponding variation in the RHESSI HXR flux (in the energy range of 25–50 keV), WL, photospheric MF (B_p), and DV (V_p). The parameters B_p, V_p, and WL are normalized to the plot range at the left ordinate axis. The RHESSI HXR fluxes are shown with respect to the right ordinate axis. The shaded area shows the time range of the transients. It is evident from panel (b) that the change in the line profile at P₂ is temporally associated with the transients. This is also the period of the peak phase of the flare as seen from panel (c). The spectral line intensity near the core (i.e., at wavelength ±34.4 mÅ) increased toward the continuum. The intensity enhancement was found to be stronger at the blue wing as compared to that in the red wing. This asymmetry resulted in the enhancement of DV in the same sign (Figure 4(d)) as explained earlier.

Zoom In Zoom Out Reset image size

From the spectral data available during this flare, it is thus evident that line-profile changes occurred during the impulsive phase of the flare. The estimated MF and DV values, as obtained by HMI algorithms, were affected by these changes during the impulsive phase of the flare, resulting in the observed TFs in magneto(Doppler)grams. It is important to note that a purely line-of-sight magnetic field would result in a spectral line with a missing pi-component. Also, the horizontal and vertical gradients in the magnetic field can cause asymmetric line profiles. The way to resolve these ambiguities would be to not just look at the LCP and RCP profiles, but all the stokes profiles, i.e., I±U and I±Q. Such profiles are now available from the SDO/HMI. We obtained the required data corresponding to this flare and aligned the Stokes vectors in a similar manner as was done for other data sets of this study. Figure 9 shows the consecutive difference maps of the Stokes parameters (I, Q, U, and V) constructed for the AR NOAA 11158 at the six wavelengths during the peak phase of the flare. Strong transient signals are seen in all these Stokes maps, which are particularly more prominent around the Fe i line center (i.e., at −34.4 and 34.4 mÅ).

Zoom In Zoom Out Reset image size

We have further analyzed the line profiles at several flare and quiet locations of the AR NOAA 11158 for examining the transient changes in the Stokes profiles. These profiles I±U and I±Q at transient location P₂ are plotted in Figure 10. Solid (dotted) curves correspond to the profiles during the pre- (peak) phase of the flare. There is a strong reversal at the wavelength −34.4 mÅ away from the line center, similar to the profile derived from the continuum as shown in Figure 7(d). Here it is important to note that the magnitude of Stokes I is much larger at all wavelengths compared to the magnitude of Stokes Q and U. Hence, the differences seen in the computed Stokes profiles I±Q and I±U are small.

Zoom In Zoom Out Reset image size

The observed line-profile changes could be related to the thermodynamic change occurring during the flare peak time as the height of the line formation can drastically change due to the moderate perturbation of temperature and density. This can be attributed to a large increase of electron density as shown by the non-LTE calculations of Ding et al. (2002) for the Ni i 6768 Å line. They suggested that the non-thermal excitation and ionization by the penetrating electrons generate a higher electron density that enhances the continuum opacity, thereby pushing the formation height of the line upward. The precipitation of electrons and deposition of energy in the chromosphere and the enhanced radiation in the hydrogen Paschen continuum give rise to the line source function, leading to an increase of the line core emission relative to the far wing and continuum.