PLASMA UPFLOWS AND MICROWAVE EMISSION IN HOT SUPRA-ARCADE STRUCTURE ASSOCIATED WITH AN M1.6 LIMB FLARE

The AIA instrument on board SDO has six narrowband EUV channels that are sensitive to different ionization states of iron (Boerner et al. 2012). The AIA produces high spatial resolution images (12) every 12 s. Figure 2 shows multi-thermal structures observed by the AIA at ∼00:20, during the decay phase of the flare. In two of the channels, 131 Å and 94 Å, the images clearly reveal a SA structure situated above the FA and extending to the high corona. For a flare, the AIA 131 Å channel contains a contribution from Fe xxi, formed at 11 MK, while the 94 Å channel contains lines from Fe xviii, formed at 7 MK (Lemen et al. 2012). At the same time, these channels also contain a significant response to cooler plasma temperatures of ∼1 MK. Therefore, one should be careful when estimating, even roughly, the temperature of the plasma using these two filters. Luckily, the AIA 171 Å and 193 Å bandpasses have temperature response curves which peak around the temperature range close to the lower temperature components of the 131 Å and 94 Å channels (see Figure 11). Since there is no emission consistent with the SA structure in these two channels (Figure 2), only the very hot plasma, over at least 7 MK, fills the SA structure. On the other hand, the FA are observed at all wavelengths, where they appear at growing heights as time progresses. The evolution of the flare in all AIA wavelengths is available as a mpeg movie in the electronic edition of the journal. In the movie, the AIA 131 Å and 94 Å channels show that the hot plasma extended toward the high corona. In the case of the 335 Å channel, the region of hot plasma appears dark due to an absence of 2.5 MK plasma, which is the peak temperature response of this channel. In the other three wavelengths, coronal loops around the flare expanded rapidly and swayed.

Figure 3 shows a sequence of snapshots taken by the AIA 131 Å channel. At the flare onset (Figure 3(b)), small-scale flare loops brighten, and the hot plasma above the flare loops stretches out toward the high corona. After the flare peak, the FA comes into view with rising motion, and the SA structure stops expanding and brightens gradually, with flows observed along two slits denoted by Flow A and Flow B in Figure 3(d). Figure 4 shows the position versus time stackplots, which we constructed by extracting ten-pixel wide slits of Flows A and B. These stackplots demonstrate that the flows of the hot plasma are upward and continuous at the particular positions during the decay phase. Anti-sunward flows in the SA structure are contrary to sunward SADs from McKenzie & Hudson (1999) and Gallagher et al. (2002) in their directionality. Also, it seems clear that it differs from the turbulent flows confirmed by McKenzie (2013). However, the upflows and SADs might have an identical mechanism, such as the magnetic reconnection process which can produce bidirectional outflows. Estimated flow speeds gradually decreased from 380 km s⁻¹ at 00:12 UT to 100 km s⁻¹ at 00:34 UT for Flow A and from 630 km s⁻¹ at 00:12 UT to 135 km s⁻¹ at 00:31 UT for Flow B. In the late decay phase (Figure 3(f)), the emission from the FA steeply decreases, while the SA structure shows continuous emission from the hot plasma.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Some of this hot plasma may be expelled out from the sun to the interplanetary space in the form of upflows and/or may cool down and fall back to the surface. Before the flare start time, SOHO/LASCO observations from CDAW data center⁵ revealed sunward flows of dark voids at around 23:16 UT on November 4. We note that these flows could interact with the SA structure but there is no evidence of such flows in the AIA data during the flare. On the other hand, Figure 5 shows that the bundle of coronal loops (indicated by two white arrows) at 193 Å connected to Flow A at 131 Å. Also, Flow B showed curved streaming as if it follows an existing field line in the SA structure (see the mpeg movie of the electronic edition of journal). Thus, we suggest that the SA structure consists of a thick coronal loop confining the hot plasma at the loop apex and large-sized loops passing the hot plasma to the other site. This hot plasma might be supplied from the SA structure itself or the lower part of the SA structure, most likely from the FA, which is the only remarkable emission source below the SA structure.

Zoom In Zoom Out Reset image size

NoRH provides microwave imaging data at two frequencies, 17 and 34 GHz, with a spatial resolution of 10'' and 5'', respectively. We synthesized microwave images with a cadence of 10 s. To improve the signal-to-noise ratio of the 34 GHz images, 30 images are averaged over 5 minutes during the decay phase. The second panel of Figure 1 shows the spatially integrated flux at 17 and 34 GHz over the flare region exhibited in Figure 6. The impulsive phase of the flare shows a drastic increase and decrease in flux, which peaked at 23:57 UT. Then, a gradual decrease starts around 00:00 UT (vertical dashed line in the second panel of Figure 1).

Zoom In Zoom Out Reset image size

Figure 6 shows sequential maps of brightness temperature (T_B) at 17 GHz with 34 GHz T_B contours over plotted. Figures 6(a) and (b) are taken at the onset and the peak of the flare, and Figures 6(c)–(f) are obtained during the decay phase. At the peak of the flare, the bright NoRH source corresponds to the flare loops seen by AIA, which is revealed as the FA during the decay phase, and the position of maximum brightness is coincident with the top of the flare loops. Simultaneously, a weak but clear blob-like structure appears at a greater height, above the bright loop-top source. During the decay phase, the blob-like structure stretches out and fills the region of the SA structure shown in the AIA 131 Å image. The T_B of the SA structure increases, while the brightness of the flaring source decreases (Figures 6(c)–(f)). This can be seen more clearly in Figure 7, which shows the time profile of brightness temperature at 17 and 34 GHz for two regions, the top of the FA and the SA structure, which are denoted by white boxes in Figure 6(f). During the decay phase from 00:00 UT, T_B in the FA exhibits a decreasing pattern, while T_B in the SA clearly increases up to 1.9 × 10⁴ K at 17 GHz and 5 × 10³ K at 34 GHz. For the 34 GHz emission, the SA structure was seen just after the middle of the decay phase, ∼00:20 UT.

Zoom In Zoom Out Reset image size

NoRH observations provide an opportunity to derive a spectral index, α (log [Flux₃₄/Flux₁₇]/log [34 GHz/17 GHz]), for spatially resolved regions. Gyrosynchrotron emission leads to α < 0 and thermal free–free emission leads to α ∼ 0 in the high-frequency regime (see Dulk 1985). Figure 8 shows time variation of index α derived from the SA and the FA. For the SA, the α index is valid only after 00:20 UT when the SA emission is observed at 34 GHz. The index α is around –1 at the flare peak time, while it goes to ∼0 during the decay phase. It is apparent that the radio emission comes from gyrosynchrotron during the impulsive phase, while it comes from thermal free–free and optically thin plasma during the decay phase. This is reconfirmed from microwave spectra in the bottom panel of Figure 1, which are taken by NoRP (Nakajima et al. 1985) multiple channels at 1, 2, 3.75, 9.4, and 17 GHz. At the flare peak time, the spectrum exhibits a turnover frequency between 9.4 and 17 GHz (solid line), and in the early decay phase, 00:00 UT, the spectrum becomes flat, e.g., α ∼ 0 (dashed line).

Zoom In Zoom Out Reset image size

With regards to the microwave emission of the SA structure, Nakajima & Yokoyama (2002) presented microwave observations of a collimated ejection which occurred above a flare loop and lasted for several minutes. The authors suggested it was caused by an injection of high-energy electrons, which may have been generated by the magnetic reconnection of a jet. This is the only example that clearly shows microwave emissions extended above the flare loops and off-limb of the sun similar to our event. However, our event shows an increase of microwave emissions over a long time (∼1 hr), while Nakajima & Yokoyama (2002) studied short lasting structure, and thus they differ in kind.

RHESSI observes X-ray and γ-ray photons in the energy range 3 keV to 17 MeV. The instrument is capable of both (separate and combined) imaging and spectroscopy with a spatial resolution of 23 and an energy-dependent spectral resolution of ∼1–10 keV. Unfortunately, there is no data available during the impulsive phase of the flare due to satellite night. During the decay phase, low-energy photons were observed in the 6–20 keV energy range (third panel of Figure 1), while there is no counts above background at energies greater than 20 keV.

RHESSI X-ray images reveal two coronal soft X-ray sources in the 8–20 keV energy range: a compact source near the limb (FA), located at the top of the flare loops; and a faint extended source high in the corona (SA), which is cospatial with the hot SA structure (Figures 9 and 10). The extended SA structure, however, is significantly weaker than the compact source. Due to the limitation of RHESSI's dynamic range, it is difficult to image the SA source in the presence of the bright compact source using the traditional CLEAN algorithm (Hurford et al. 2002). However, using the two-step CLEAN method described in Krucker et al. (2011), we are able to recover the SA source. Applying this method, the resulting compact and SA sources are reconstructed with spatial resolutions of ∼12'' and ∼35'' respectively. The Figure 9 shows both sources for two time intervals, 00:14:08–00:18:40 UT (red contours) and 00:41:30–00:44:30 UT (blue contours). The compact source contour values are 70%, 80%, and 90% of the compact source maximum intensity, while the extended source contours are 3.3%, 3.5%, and 3.7% and 27%, 28%, 29% of the compact source maximum intensity for two time intervals, respectively.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Gallagher et al. (2002) has shown a similar examination of a SA structure using RHESSI data. During the decay phase of their event, a thermal source appeared high in the corona at an energy range of 6–25 keV, and the altitude of the source centroid moved to greater heights with a velocity of 9.9 km s⁻¹. In our case, detailed imaging spectroscopy to recover the temperature and emission measure of the SA source is not possible, due to the low counting statistics, but RHESSI spectroscopy indicates that X-ray emission of the flare, which contains both the SA and FA source, is certainly thermal. The centroid of the SA source shifts around 10'' during the decay phase with a speed of around 38 km s⁻¹ (Figure 9) but, unfortunately, the nature of the two-step CLEAN does not allow us to determine if this motion is real. Indeed, Gallagher et al. (2002) found hot thermal structure high in the corona and FA below it in EUV observation. However, during the decay phase, one X-ray source appeared just as high in corona. Thus, we presume that, in our event, the FA which has a predominant X-ray source might play a certain role for the formation of the SA as an energy source.

To derive a plasma temperature, we used the two GOES X-ray channels (1–8 Å and 0.5–4 Å) for the FA source and AIA 131 Å and 94 Å images for the SA source (see Section 2.1). We have used observations from different instruments for each source due to the saturation effects on the FA region in AIA 131 Å images. However, since the FA region includes the flare loops, which are the main source of soft X-ray, it is reasonable to determine T of the FA using the ratio of the two GOES X-ray channels even though they are full-sun observations. Assuming an isothermal plasma, the plasma temperatures T for the FA and SA have both been estimated by the filter ratio method. The ratio R is given as

$$\begin{equation} R=\frac{{\rm DN}_{\lambda _1}}{{\rm DN}_{\lambda _2}} = \frac{{\rm Response}_{\lambda _1}}{{\rm Response}_{\lambda _2}}, \end{equation} \tag{ 1 }$$

where DN is the digital number converted from observed photon counts. The temperature response function of each filter is derived using the coronal abundances in CHIANTI 6.0.1 (Dere et al. 1997, 2009). The temperature of SA was estimated using the filter ratio method for the AIA 131 Å and 94 Å channels (see Section 2.1). The top panel of Figure 11 shows the temperature response curves for six AIA channels (obtained using aia_get_response.pro in solarsoft). In order to use this pair of AIA data, there are two important sources of error to be considered (see Boerner et al. 2012): 25% uncertainty on the instrument calibration and deficiencies of the cooler lines on 131 and 94 Å in the CHIANTI atomic physics databases. We found that the estimated uncertainty on the temperature caused by instrument calibration is not significant, under 9%. Since the SA structure appears only in the AIA 131 Å and 94 Å, it is reasonable to suggest that the structure consists of predominantly hot plasma. The high definition of the SA structure revealed in AIA 131 Å images supports the suggestion that the majority of the emission comes from the thermal plasma within the temperature response of AIA 131 Å. We estimate the temperature of the SA to be in the range log T = 6.7–7.2 (dashed vertical lines in Figure 11), which avoids overlap with the temperature response range of the AIA 335 Å channel.

Zoom In Zoom Out Reset image size

The white box shown in Figure 10 indicates the SA source area where we estimated the plasma temperature. The DN for the SA area was averaged and a background was subtracted. Considering that the SA location is in the corona, the background DN was found at a diffuse and stable emission region existing in the off-limb corona. We note that diffraction patterns, observed in AIA 131 Å (see Figure 3(c) in particular), caused by the bright flare loops contribute to the flux in the SA region. The measured contribution of diffraction patterns was found to be around 11% of the total DN of the SA area, so it is insignificant for these estimations (C. L. Raftery 2012, private communication). The bottom panel of Figure 11 shows the DN ratios of AIA 131 Å to 94 Å for the SA area (blue line). It is superposed on the ratio of temperature response curves of these two filters (black line). Using the GOES widget, we derived the temperature for the FL (White et al. 2005). Figure 12 shows the time variation of estimated plasma temperature for both regions, the FA (dotted line) and the SA (solid line). The FA temperature peaks at 18 MK at 23:56 UT and quickly decreases to 14 MK within 4 minutes. After that, it gradually decreases down to 8 MK. On the other hand, the SA temperature shows a steep increase around 10 MK until 23:57 UT but, during the decay phase, it seems to increase steadily until 00:10 UT and then decrease to 8 MK at 00:50 UT. Since the steady increase of the temperature takes place within given error bars, it may not be a real increase. However, it sustains its high temperature above 10 MK until 00:16 UT.

Zoom In Zoom Out Reset image size

The microwave emission, unlike EUV and X-rays, is little sensitive to the variation of the plasma temperature (see Equations (2) and (3)). Thus, it is appropriate to derive the emission measure of the plasma using the microwave emission during the phase when the plasma shows a clear temperature gradient. From the brightness temperature observed by NoRH at 17 GHz, we have derived the emission measure (EM) and electron number density (N_e) of the FA and the SA, during the decay phase, for the period from 00:00 UT to 00:50. Based on the observations, we assume that the microwave comes from the optically thin (τ_ν ≪ 1) and thermal free–free emission of plasma. Thus, the relation between brightness temperature (T_B) at specific radio frequency (ν) and plasma temperature (T) is described as (Dulk 1985):

$$\begin{equation} T_B \sim T {\tau }_{\nu }, \end{equation} \tag{ 2 }$$

where τ_ν is

$$\begin{equation} {\tau }_{\nu } = \frac{9.786 \times {10}^{-3}}{\nu ^2} \ln \Lambda \ \frac{{\rm EM}}{T^{3/2}}, \end{equation} \tag{ 3 }$$

where the coulomb logarithm ln Λ = ln [4.7 × 10¹⁰(T/ν)]. T is a plasma temperature calculated in Section 3.1 and T_B is obtained at ν = 17 GHz. The density of each sources can be derived from EM,

$$\begin{equation} N_e=\left (\frac{{\rm EM}}{z_{{\rm meas}}f}\right)^{1/2}. \end{equation} \tag{ 4 }$$

where z_meas is the averaged depth of the line of sight and f is the line of sight filling factor, f = z_real/z_meas (Dere 1982). The line of sight depth of the soft X-ray sources is assumed to be equal to the source width, z_FA ≈ 23 Mm and z_SA ≈ 40 Mm, respectively. It is difficult to determine the filling factors of observed loops or structures since present observations cannot resolve whether the plasma are filled uniformly or sparsely with separate multiple loops/structures. Dere et al. (1979), Dere (1982), and Tripathi et al. (2009) compared the density determined from the emission measure within the volume of interest with the density determined from two density sensitive spectral lines, establishing the filling factor within the source. The filling factor near unity was found by Dere et al. (1979) for solar flares and by Dere (1982) for two active region coronal loops. On the other hand, there seems to be evidence of filling factors as low as 5 × 10⁻² (Cargill & Klimchuk 1997). Recently, Tripathi et al. (2009) found that the filling factor increased with projected height of the hot coronal loops and was close to unity just beyond 40 Mm. In this analysis, we assume that the inner part of the FL and the SA is filled with plasma, and the electron number density along the line of sight is uniform, e.g., f ∼ 1.

Figure 13 shows the derived EM (top) and N_e (middle) for the FA and the SA with time. EM and N_e of the FA decrease constantly from 1.0 × 10³⁰ to 0.2 × 10³⁰ cm⁻⁵ and 2.0 × 10¹⁰ to 1.0 × 10¹⁰ cm⁻³, respectively. Contrary to this, EM and N_e of the SA increase from 0.2 × 10²⁹ to 1.0 × 10²⁹ cm⁻⁵ and from 2.0 × 10⁹ to 5.0 × 10⁹ cm⁻³, respectively. The decreasing rate of N_e in the FA was 2.7 × 10⁶ cm⁻³ s⁻¹ throughout. On the other hand, the increasing rate of N_e of the SA was 1 × 10⁶ cm⁻³ s⁻¹ around the start of the decay phase and then gradually decreases to 1 × 10⁵ cm⁻³ s⁻¹ at the end of the flare. Based on the assumed z and calculated source area A, the total number of electrons decreased by 2.4 × 10³⁴ s⁻¹ for the FA and increased by 6.3 × 10³⁴ s⁻¹ around the start of the decay phase and 0.6 × 10³⁴ s⁻¹ at the end of the flare for the SA. Taking into account the increasing N_e and T in the SA, one can deduce that there is an injection of hot plasma into the SA from below.

Zoom In Zoom Out Reset image size

For consistency, we compare our estimates of temperature and emission measure with observations from RHESSI. A spatially integrated spectrum for the time interval 00:10:00 to 00:13:30 UT, taken immediately after spacecraft night, reveals a thermal spectrum with a temperature of 14 MK and an emission measure of 0.142 × 10⁴⁹cm⁻³ for the combined FA and SA source contributions. However, due to low counting statistics, it is not possible to perform imaging spectroscopy to determine independent values of temperature and emission measure of the SA source. Nevertheless, we attempt to compare the flux ratio of the SA and FA that we would expect to see from RHESSI imaging observations given the temperature and emission measure estimates found using AIA and GOES. Using values obtained from AIA and GOES, and assuming isothermal free–free emission, we calculated the flux ratio of the FA and the SA sources, i.e., F_FA: F_SA, that we would expect to observe in the soft X-ray range observed by RHESSI (blue lines, Figure 14). This was done using the solarsoft routine f_vth.pro, which determines the expected radiation as a function of energy. The function contains a contribution from the optically thin thermal free–free bremsstrahlung and free-bound continuum, and line emission, primarily from highly ionized iron (seen in RHESSI spectra at 6.7 keV and 8 keV), obtained using CHIANTI. The continuum emission takes the form I() ≈ (EM/T^1/2)exp (− /k_BT), where is photon energy and k_B is the Boltzmann constant. We then compared these values with those found from actual observations (red lines, Figure 14). We chose to make this comparison at 8 keV, where there were enough soft X-ray counts to reconstruct a RHESSI image over a narrow energy range. The calculations were done using the integrated time intervals required for RHESSI imaging. In general, the results are consistent with RHESSI observations. However, we note that the main source of error in these measurements comes from determining the source area of the SA, which is particularly difficult given its diffuse nature and intensity with respect to the compact source. We expect our source area to be accurate to around a factor of two; however, we note that it is possible that the source contains small- scale structure that cannot be reconstructed using this method.

Zoom In Zoom Out Reset image size

The thermal energy of the emitting plasma in the volume (V_meas), which is determined by a source area of A multiplied by the assumed line of sight depth of z_meas (see Section 3.2), is given by

$$\begin{equation} E_{{\rm th}}= 3k_BT \sqrt{{\rm EM} A V_{\rm meas}f}. \end{equation} \tag{ 5 }$$

The filling factor is assumed to be unity (see Section 3.2). The bottom panel of Figure 13 shows the estimated thermal energy for both sources. E_th of the FA gradually decreases from 1.5 × 10³⁰ erg at 00:00 UT to 0.4 × 10³⁰ erg at 00:50 UT, while E_th of the SA increases from 0.5 × 10³⁰ erg at 00:00 UT to 1.0 × 10³⁰ erg at 00:16 UT and then sustains this value until 00:50 UT. To determine a time scale for the energy loss, cooling processes by radiation and conduction are considered. Using an estimate of loop radius from AIA 131 Å images, L = 29 Mm for the FL and 89 Mm for the SA, we found that the predominant cooling process is conduction and the estimated cooling time scale ($\tau _c=4 \times 10^{-10}N_eL^2T_e^{-5/2}$) is 230 s for the FA and 390 s for the SA. However, considering the decreasing rate of the estimated temperature during the decay phase, the cooling time of the top of the FA down to 1 MK is 6400 s. Furthermore, the SA structure shows not cooling but heating for a while in the early part of the decay phase. Thus, it clearly shows that the significant energy injection took place in the FA and the SA during the flare decay phase.