EXTREMELY LARGE EUV LATE PHASE OF SOLAR FLARES

EVE observations measure the solar EUV differential irradiance from 0.1 to 105 nm with unprecedented spectral resolution of 0.1 nm, temporal cadence of 10 s, and accuracy of 20% (Woods et al. 2012). The EVE Level-2 processing produces a combined set of merged spectra provided in a pair of files: one file (EVS) contains the full spectra, and the other (EVL) contains the isolated lines from ionized solar elements and the bands simulated for other instruments such as AIA.

We used EVL data to study the EUV late phase of this flare event. We chose three lines with their central wavelength of 13.3 nm (Fe xx, ∼10 MK), 33.5 nm (Fe xvi, ∼2.5 MK), and 17.1 nm (Fe ix, ∼0.7 MK), corresponding to high temperature, warm temperature, and cool temperature, and plot their irradiance variability (normalized) as black, blue, and yellow curves in Figure 1, respectively. Similar to the GOESSXR profile, the profile of the high-temperature line 13.3 nm only had a single peak and peaked at almost the same time (∼13:29 UT) as in X-ray. However, in the profile of the warm 33.5 nm line, there was a distinct second peak (P$_{2}\;\sim$ 16:21 UT; blue vertical dotted line c in Figure 1) other than the peak in the main phase (P $_{1}\;\sim$13:32 UT); the second peak lagged about 170 minutes behind the first peak. The existence of this second peak confirmed that this flare did have a EUV late phase. Moreover, the peak value of the late phase is much larger than that in the main phase; the ratio between the second and the first for peaks is as high as 2.1.

The profile of the cool line 17.1 nm had two peaks: the main-phase peak at ∼13:32 UT and the late-phase peak at ∼17:37 UT, yellow vertical dotted line d in Figure 1. We note that, during the late phase, there were multiple peaks, and we chose the largest one representing the peak time. Obviously, there was a strong dimming in the cool line between the two phases. Immediately following the main-phase peak, the irradiance of the cool 17.1 nm line began to decrease until it reached its minimum around 15:36 UT. The flux intensity at the minimum was significantly lower than the pre-flare background level. The flux then had a slow recovery phase until 16:21 UT, which was followed by a faster ascending phase. Note that the transition time between the slow recovery to fast ascending coincides well with the peak time of the late phase in the 33.5 nm line. Based on the fact that this is a confined flare, it could be concluded that the dimming or depression in the cool line was mainly due to the thermal effect rather than the mass-loss effect.

The EVS spectral data product provided the spectrum irradiance variability of the flare within the wavelength range of 6–37 nm (see Figure 2(a)). The units for EVE spectrum irradiance were ${\rm watt}\;{{{\rm m}}^{-2}}\;{\rm n}{{{\rm m}}^{-2}}$. Integrating over this wavelength range, we obtained the total EUV radiative energy-loss rate (units ${\rm J}\;{{{\rm m}}^{-2}}\;{{{\rm s}}^{-2}}$) of EVE EUV spectrum at Earth (1 AU). We further converted the radiative energy-loss rate to energy loss at the Sun (${\rm ergs}\;{{{\rm s}}^{-1}}$) by multiplying a factor of $1.406\times {{10}^{30}}$ ($=\;{{10}^{7}}\cdot 2\cdot \pi \cdot {{(1\;{\rm AU})}^{2}}$), assuming a uniform angular distribution of the radiation (Woods et al. 2006). The same conversion method has been applied to GOES SXR observations, and the results are displayed in Figure 3.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

It was interesting to note that there were also two peaks in the time series of the radiative energy-loss rates of the EVE EUV spectrum, while only a single peak in the GOES SXR profile. The time of the first peak is 13:34 UT, lagging about 5 minutes behind the GOES X-ray peak (13:29 UT). This short lag is likely the consequence of the PFLs' cooling effect. The time of the second peak is 16:47 UT, later than the 33.5 nm late-phase peak and earlier than the 17.1 nm peak. Apparently, the second peak is larger than the first peak in energetics. From the energy release point of view, we defined these two peaks as the peaks of the main phase and the EUV late phase. The time separating the two phases is  about 15:03 UT (see Figure 3, vertical solid line). For calculating the flare energy, we subtracted the pre-flare value from the energy-loss rates (hereinafter we use 12:30 UT as the pre-flare time), then integrated them in the time range of both phases. As shown in Figure 3, the total radiative energy in GOES SXR 0.1–0.8 nm was about $5.26\times {{10}^{28}}\;{\rm erg}$, and the ratio of the main phase ($4.60\times {{10}^{28}}\;{\rm erg}$) to the late phase ($6.63\times {{10}^{27}}\;{\rm erg}$) is about seven to one. However, in the EVE EUV 6–37 nm range, the total flare radiative energy is about $1.03\times {{10}^{30}}\;{\rm erg}$, and the ratio between the main phase ($2.31\times {{10}^{29}}$) and the late phase ($7.96\times {{10}^{29}}$) is about 1 to 3.4. Obviously, the late phase had much more energy released than that in the main phase. It is thus called the event of the extremely large EUV late phase.

By subtracting the spectrum irradiance at the pre-flare time (hereinafter we use 12:30 UT as the pre-flare time), we derived the EVE EUV spectrum variabilities of different phases (panels (b)–(d) in Figure 2; we use 5 minute averages here), which were main-phase peak (∼13:29 UT), coronal dimming (∼15:36 UT), and late-phase peak (∼16:21 UT), respectively. From these variability profiles, we identified 19 different spectral lines with significant variations (${\Delta }I\geqslant 20$, units: ${{10}^{-6}}\;{\rm W}/{{{\rm m}}^{2}}/{\rm nm}$) and listed their properties in Table 1 (based on the Chianti Atomic Database; Dere et al. 1997; Landi et al. 2013). The table indicates that only the cool corona line, Fe ix 17.1nm, shows the dimming. Except for the choromosphere line He ii 30.4 nm and warm corona line Fe xvi 33.5 nm, which had larger enhanced radiation in all phases, other lines had enhanced radiation either in the main phase (hotter lines with shorter wavelengths, <15 nm) or in the late phase (cooler lines with longer wavelengthes, >15 nm).

Table 1.  The Properties of Lines in Figure 2

Ion	λ	Tmax	Main Phase	Dimming	Late Phase
	(nm)	$({{{\rm log} }_{10}})$	${\Delta }I\ ({{10}^{-6}}\;{\rm W}\;{{{\rm m}}^{-2}}\;{\rm n}{{{\rm m}}^{-2}}$)
Fe xviii	9.4	6.9	28	24	...
Fe xix	10.8	7.0	30	...	...
Fe xxi	11.8	7.1	32	...	...
Fe xx	12.2	7.1	22	...	...
Fe xxi	12.9	7.1	42	...	...
Fe xx	13.3	7.1	84	...	...
Fe xxii	13.6	7.1	24	...	...
Fe ix	17.1	5.9	−29	−40	...
Fe x	17.5	6.1	...	...	25
Fe xi	18.0	6.2	...	...	24
Fe xii	19.5	6.2	...	...	22
Fe xiii	20.3	6.3	...	...	21
Fe xiv	21.1	6.3	...	...	20
He ii	25.6	4.9	...	...	25
Fe xv	28.4	6.4	...	...	73
He ii	30.4	4.9	96	111	170
Fe xvi	33.5	6.8	43	59	68
Fe xvi	36.1	6.8	...	20	36
Mg ix	36.8	6.0	...	...	24

Download table as:  ASCIITypeset image

The six lines Fe xx 13.3 nm, Fe xviii 9.4 nm, Fe xvi 33.5 nm, Fe xiv 21.1 nm, Fe xii 19.5 nm, and Fe ix 17.1 nm in Table 1 correspond to the six coronal passbands of AIA (Fe xx 131 Å, Fe xviii 94 Å, Fe xvi 335 Å, Fe xiv 211 Å, Fe xii 193 Å , and Fe ix 171 Å). Thus, we further inspected the AIA image sequences to study the morphology and dynamic process of the flare event.

AIA takes high-resolution images (4096 × 4096 pixels, 06 pixel size, and 15 spatial resolution) of the whole Sun with a high cadence of 12 s. Furthermore, the AIA instrument has 10 passbands sensitive to different temperatures, thus allowing the investigation of the thermal property of flare structures. Six of the ten passbands are sensitive to coronal temperatures as mentioned above, and we only use the images from these six passbands in this paper.

From the AIA images, we found that the flare event occurred at NOAA AR 11121, centered at the heliographic coordinates 20° S and 75° E. Therefore, it was located at the southeast limb allowing us to better study its structure along the radial direction in the lower corona. To eliminate the interference on flux calculation from other ARs, we cut a sub-region of the AIA images containing the AR 11121 and focused on these sub-images in different wavebands. By summing all the pixel values in this sub-region and plotting its variation against time (Figure 5(a)), we got very similar profiles as the EVE observations (Figure 1), which confirmed that emissions for both the main phase and the late phase were mainly originated from this sub-region.

Zoom In Zoom Out Reset image size

An interesting high-lying hot-blob-like feature was found in these sub-region images. As the bright PFL arcade rose and reached a certain height, the hot blob-like structure appeared above, then kept expanding and ascending until it was stopped by the "cold" overlying loops (see the online AIA animation). This blob-like structure was diagnosed as hot plasma (5 ∼ 18 MK) since it was brightening only in hot passbands, while the PFLs were bright in all passbands (see Figures 4(b)–(d)). As the structure faded in the hot passbands, certain similar structures began to appear in the warm passband 335 Å, coinciding with the late phase. We plotted the region variabilities of this hot structure along with the PFLs in Figures 5(b) and (c), corresponding to the white solid box and dotted box, respectively, in Figure 4(b). Based on these variability profiles, we confirmed that the EUV emissions for main phase were mainly from the PFLs, while the EUV late-phase emissions were mainly from the area of the hot structure.

To study the dynamics of this flare event, two slices were cut from the AIA difference images (hot 131 Å, warm 335 Å, and cold 171 Å) across the hot structure in two directions, the radial direction (slice 1) and transverse direction (slice 2, both slices were indicated with arrows in Figure 4(a)). We stacked these slices along time as the slice-time plots in Figure 6. The appearance and ascending process of the hot structure were only obvious in the hot 131 Å plot (Figure 6(a)), while the PFLs brightened in all the plots. The hot structure appeared above the PFLs and maintained a certain distance between them. The separation and sharp boundaries of the two structures allowed us to identify their evolving leading edges (red triangles for the PFLs and blue diamonds for the hot structure in Figure 6(a)).

Zoom In Zoom Out Reset image size

Based on the height-time plots of the leading edges (Figure 7(a)), we further derived their ascending velocities and accelerations and plotted them with uncertainties in Figures 7(b) and (c). From these plots, we found that the hot structure appeared around 13:20 UT (long after the flare onset time, 12:43 UT) at the height of about 50 Mm, while the PFLs reached its maximum height of about 30 Mm. Then it accelerated to its maximum velocity around 150 km s⁻¹ at the time of 13:28 UT (solid vertical line in Figure 7), which was almost the same as the GOES SXR peak time (13:29 UT), considering the uncertainties. This result suggests that the drive source for the ascending is related to the flare reconnection.

Zoom In Zoom Out Reset image size

The overlying loop arcade was obvious in the cool 171 Å plot (Figure 6(c)). As above, we identified its leading edge and plotted the height-time and variations of velocities and accelerations in Figures 7(a), (d), and (e), respectively. From these plots, we found that the overlying loop arcade accelerated with the rising of the PFLs and the hot structure until it reached its maximum velocity around 18 km s⁻¹ at 13:34 UT, which was also the time when the rising of the hot structure stopped. Hence, there was a possible interaction between the overlying loop arcade and the rising hot structure, or, to put it another way, the overlying loop arcade prevented the hot structure from further eruption. Such failed eruption is consistent with the fact that there was no CME associated with the flare.

The Extreme-Ultraviolet Imager (EUVI; Wuelser et al. 2004) on board Solar TErrestrial Relations Observatory (STEREO; Davilla et al. 1996) also takes EUV images of the whole Sun. Thanks to the unique positions of STEREO A and B (see Figure 8(a)), the images from EUVI-B could provide a new perspective of the flare event. Although EUVI has multiple passbands, we use the 195 Å images, which have a cadence of 5 minutes and are comparable in temperature sensitivity with the EVE 19.5 nm and AIA 193 Å observations.

During the flare time, AR 11121 was located at the center of the EUVI-B 195 Å images. As we did to AIA images, a sub-region including the AR was cut out from the images, then we summed the pixel values and plotted the region variability profile (greed solid curve in Figure 8(b)). Figure 8(b) also gave the profiles of EVE 195 Å line irradiance variability (black solid curve) and AIA 193 Å region variability (green dotted curve). The consistence of the three profiles indicates that the three different instruments were observing the same feature.

There were three obvious peaks in all the three profiles, which were the main-phase peak (T $_{c}\;\sim$ 13:31 UT) and the first (${{T}_{d}}\;\sim$ 16:46 UT) and second (${{T}_{e}}\;\sim$ 17:21 UT) late-phase peaks, respectively. By inspecting the image sequences of AIA 193 Å and EUVI-B 195 Å (see the online STEREO animation), we found that the three peaks corresponded to the brightening of three loop systems. For better illustration, the snapshots of both the AIA 193 Å and EUVI-B 195 Å sub-images at three peak times were plotted in Figures 8(c)–(e) and 8(c')–(e'). The loop system responsible for the first peak was the PFLs, which were compact and south–north oriented (the red and blue pluses in Figure 8(c') represented their two footpoints, which were more obvious in the animation). The loop system for the second peak, or the late-phase peak, was semicircular. This loop arcade had a different orientation (east–west; Figure 8(d')) from that of the PFLs. It was more like the "late-phase loop arcade" discussed by Liu et al. (2013) since it had one footpoint (the red asterisks) near the PFLs, while another footpoint (blue asterisks) was located at a remote site.

The third loop system was a sigmoid-like structure as seen in the EUVI-B images (Figure 8(e')). It had the same orientation (east–west; footpoints: the blue and red diamonds) as the second loop system, but it was wider and higher based on Figures 8(d) and (e). The red curves in Figure 8(e) were the contours of the hot structure at its maximum height in the AIA 131 Å image (Figure 4(b)). The fact that the red curves could wrap the loop system very well suggests that the loop system might be the cool remnant of the hot structure. Thus, together, the late-phase loop arcade and the remainder of the failed-erupting hot structure were responsible for the extremely large EUV late phase.