EXPLOSIVE EVENTS ON A SUBARCSECOND SCALE IN IRIS OBSERVATIONS: A CASE STUDY

IRIS is a NASA Small Explorer mission, which was launched into a sun-synchronous orbit on 2013 June 27. It takes simultaneous spectra and images of the interface region (i.e., chromosphere and transition region) of the Sun with 033–04 spatial resolution and as low as 2 s cadence. IRIS observes the far-UV wavelength band in the spectral range from 1331.6 to 1358.4 Å that includes two bright C ii lines with a spectral sampling of 12.98 mÅ pixel⁻¹, and in the range 1380.6–1406.8 Å (containing several Si iv and O iv lines) at 12.72 mÅ pixel⁻¹. IRIS also operates in the near UV spectral range from 2782.6 to 2833.9 Å at 25.46 mÅ pixel⁻¹ recording two bright Mg ii lines. For comparison, the SUMER spectral sampling is around 44.0–45.0 mÅ pixel⁻¹ in the 660–1600 Å, i.e., IRIS has more than three times higher spectral resolution. The instrument is designed in such a way that a slit with a size 033 × 175'' guides the sunlight into a spectrograph while a reflective coating directs the sunlight outside of the slit into an imaging system making slit-jaw images (SJIs). Therefore, the slit position appears as an emission-blocked vertical line on the slit-jaw images (see the example in Figure 1). The slit can either stay in a fixed position to record sit-and-stare spectra or scan a large area with a selected step size. The slit-jaw images can be taken in four different channels, one centered at 1330 Å with a 40 Å bandpass recording chromospheric and strong continuum emission, one at 1400 Å again with a 40 Å bandpass imaging the lower transition region, one at 2796 Å with a 4 Å bandpass providing images of the upper chromosphere, and one at 2832 Å with a 4 Å bandpass for high-contrast photospheric imaging.

Zoom In Zoom Out Reset image size

The IRIS data set used in this study includes spectral observations taken in a sit-and-stare mode with 8 s exposure time and 9 s cadence, and slit-jaw images taken in the 1330 Å channel also at 9 s cadence. Four spectral windows were transferred to the ground containing two C ii lines (1334.5 Å and 1335.7 Å, log  T = 4.3 K), the Si iv 1402.8 Å line (log  T = 4.8 K), and the Mg ii k 2796.4 Å line (log  T = 4 K). IRIS was targeting an equatorial extension of a polar coronal hole across its boundaries (Figure 1). Figure 1 shows an overview of the coronal hole in the 211 Å channel of AIA and one of the SJ images. The coronal hole boundaries are defined as the region where the AIA 211 Å emission drops to half of that in the surrounding quiet Sun region. The spectral observations were taken in a sit-and-stare mode, but a compensation for the solar differential rotation was not applied to the data.

The present study uses IRIS level 2 data provided by the IRIS team. These are science products after dark current removal, flat-field, and geometric correction have been applied. We found that spiked pixels are still present in the spectral data. To flag these pixels, we first determined the maxima of the line profiles and the standard deviation (σ) of the maxima were calculated. The pixels with line profiles that have a maximum exceeding 3σ were flagged and excluded from any further analysis. A further wavelength correction for the orbital variation of the line position was required due to the temperature change of the detector and the spacecraft–Sun distance variation over the course of an orbit. To correct this, we used the standard program provided by the IRIS team (Tian et al. 2014).

The AIA observations used in this study were taken in the UV channels including the 1700 Å and 1600 Å at about 24 s cadence, and in the EUV 304 Å, 171 Å, 193 Å, 211 Å, 335 Å, 131 Å, and 94 Å passbands at about 12 s cadence. The spatial resolution of the AIA images is 12 (06 pixel⁻¹). The HMI longitudinal magnetograms analyzed here have a 45 s cadence and were taken from 17:40 UT to 19:17 UT (IRIS observed from 18:42 UT to 19:00 UT) in order to investigate the magnetic field evolution that precedes and follows the IRIS observations. The HMI pixel size is 0505 and a 1σ noise level is 10 G (Liu et al. 2012).

The emission in the three IRIS spectral lines (Mg ii k, C ii, and Si iv) is shifted along the IRIS slit. To correct for this offset we used fiducial emission marks made by various phenomena along the slit. We found an offset between C ii and Si iv of 7 pixels (i.e., Si iv is 7 pixels lower than C ii) and Mg ii k is 4 pixels higher than C ii. The IRIS 1330 Å slit-jaw and the AIA 1600 Å images both have a strong continuum contribution which permits a straightforward alignment between both data sets. The AIA 1600 Å images were then aligned with the HMI data.

The IRIS spectral and imaging data were taken at the boundaries of an equatorial extension of a polar coronal hole (Figure 1). The spectral observations obtained in the Si iv line were first analyzed by applying a single Gaussian fit. Then the data set was inspected for strong blue and/or redshifted emission above 50 km s⁻¹. We identified one such event with strong non-Gaussian profiles (solar_y = 200''). Judging by its spectroscopic appearance, the event carries all the observational characteristics of the EE (see Section 1 for more details). The EE measures ≈15 along the IRIS slit. In the radiance images, produced from the sit-and-stare slit observations, it appears as a bright compact structure in the Mg ii k, C ii,  and Si iv lines (Figure 2). Note that because compensation for the solar differential rotation was not applied during the IRIS observations, at these heliographic coordinates the slit scans an area of 1'' in 460 s. For an event with a size of 15, it will take approximately 11 minutes to be scanned from west to east. In the SJ images, the EE appears as a compact bright-point-like structure, which is visible during the whole observing period, i.e., 18 minutes 46 s.

Zoom In Zoom Out Reset image size

In Figure 3 (top and bottom left) we show the radiance and the Doppler-shift images obtained from the Si iv line. The Doppler-velocity image was produced by applying a single Gaussian fit. Away from the EE, the line is dominated by noise due to the low count rate of Si iv but the emission in the EE has a good signal-to-noise ratio allowing reliable calculation from a single Gaussian fit. As we mentioned above the spectral information obtained during the event is both spatial and temporal, although the observations are in a sit-and-stare mode. While the EE moves under the IRIS slit, first a redshift dominated emission is registered for the time between 18:45:27 UT and 18:46:42 UT, i.e., during 75 s. The Sun would rotate under the IRIS slit for this period of time by approximately 016, i.e., ∼118 km. An example of the redshifted Si iv line profile is given in Figure 3 (top row, 2nd panel–"A"). Gradually blueshifted emission is seen to increase from 18:46:57 UT to 18:47:49 UT, (i.e., in 52 s) until the emission in both wings is almost equal (Figure 3, top row, 3nd panel–"B"). From 18:47:58 UT until 18:51:55 UT (237 s), the emission is blueshift dominated (Figure 3, top row, 4th panel–"C"). The phenomenon was scanned in 388 s, which means that the size of the EE in the Si iv line (i.e., transition region) is ≈084. Considering the projection angle, the EE should have a size of 094 on the solar surface.

Zoom In Zoom Out Reset image size

The Doppler-shift pattern seen in the single Gaussian images was further investigated using the red–blue (RB) asymmetry method. RB asymmetry is defined as the difference of the emission in the red and blue wing of a spectral line at the same Doppler velocity (or Doppler-velocity range) and is given by

$$\begin{equation*} {\rm RB}_{\Delta \lambda _1}=\int _{\lambda _0+\Delta \lambda _1-\delta \lambda _{\omega }}^{\lambda _0+\Delta \lambda _1+\delta \lambda _{\omega }}I_\lambda d\lambda - \int _{\lambda _0-\Delta \lambda _1-\delta \lambda _{\omega }}^{\lambda _0-\Delta \lambda _1+\delta \lambda _{\omega }}I_\lambda d\lambda, \end{equation*}$$

where λ₀ is the wavelength of the line center, I_λ is the intensity of the spectral line, Δλ₁ is the offset from the line center, and δλ_ω is the wavelength range over which the RB asymmetry is determined. This method has been widely used in optically thick spectral-line analysis (e.g., in Hα, Madjarska et al. 2009; Huang et al. 2014, and the references therein), and it was also introduced to emission lines (De Pontieu et al. 2009). A variant of this method that differs in the determination of the line center is discussed in Tian et al. (2011). In the present study, the line center is obtained from an average profile in the region indicated in Figure 2. In Figure 3, bottom row, we show images of the RB asymmetry in the Si iv line in three Doppler-shift ranges: 50–70 km s⁻¹, 90–110 km s⁻¹ , and 130–150 km s⁻¹. The temporal variations of the RB asymmetry obtained by averaging over the EE along the slit in the Si iv line are given in the top panel of Figure 6. We clearly see from both figures that the RB asymmetry of the EE is positive only in the lowest velocity range (50–70 km s⁻¹). The blueshift dominant emission is found in the central part of the EE. Again we have to keep in mind that the information is space-time combined and the excess up-flow may later be followed by a down-flow or vice-versa. As seen by Madjarska et al. (2009) in the case of a surge, simultaneous plasma up- and down-flow along adjacent field-lines would produce simultaneous blue and/or redshifted emission with the appearance along the spectrometer slit depending on the line of sight.

The observations show that the redshifted emission is weaker than the blueshifted emission, indicating that less plasma at the formation temperatures of the three lines falls back toward the chromosphere. One explanation could be that the plasma is ejected from the chromosphere and part of it is heated to transition region and/or coronal temperatures. After the plasma deposits the heat in the upper atmosphere, it falls back to the chromosphere at different speeds and over a longer period of time. Another possible interpretation is that the plasma is ejected along looped magnetic-field lines thus leaving the field-of-view of the spectrometer slit.

The event also appears as a compact bright structure in the Mg ii k and C ii radiance images (Figures 4 and 5) similar to Si iv. Both lines appear with a central absorption core surrounded by two emission peaks. The Mg ii k blue peak is sometimes referred to as k_2v and the red as k_2r, while the line core is k₃. Here, we will simply refer to k_2v and k_2r as blue and red wings. The line profiles of the two absorption lines at three sampling pixels ("A," "B," and "C" in bottom rows of Figures 4 and 5) are red-wing-dominated, with equal wings, and blue-wing-dominated respectively, i.e., the same as for the Si iv profiles. The RB asymmetry in the two chromospheric lines is derived as for the Si iv line and is shown in Figures 4 and 5. The RB temporal variations are given in Figure 6. Note that Doppler shifts in the wings of the Mg ii k line are reported to be a signature of flow velocities as found by Leenaarts et al. (2013a, 2013b) and Pereira et al. (2013). The Mg ii k and C ii RB asymmetry images show very similar to behavior as the Si iv asymmetry with the only difference in the Doppler velocity range 130–150  km s⁻¹. That may be due to a blend in the red wing of these two lines (Mg ii k is blended by an unidentified line, and C ii 1334.5 Å is blended by the blue wing of C ii 1335.7 Å). Our results demonstrate that above 50 km s⁻¹ during a very dynamic event the wings of the chromospheric lines studied here carry identical information as the optically thin Si iv line.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

To investigate in more detail the radiance behavior in the wings and centers of the three lines, we produced lightcurves in their blue and red wings and line centers (Figure 7). The response in the wings of C ii is almost identical to the Si iv wing emission with only a small delay of 9 s for the blue wing of C ii to reach peak emission (top panel in Figure 7). This indicates that the wings of the two lines form at a similar temperature. The behavior of the Mg ii k wings is similar to that of the C ii. However, the emission in the wings of Mg ii k in addition to the Doppler shift is possibly also affected by emission related to wing formation in the lower chromosphere. The above is not valid, however, for the line centers. The line center of C ii is clearly reversed. We obtained the lightcurves in the line center of the Mg ii k and C ii lines, summing the emission from −5 to 5 km s⁻¹. The line-center emission of these two lines show the same behavior which is very different from the emission in the center of the Si iv line (Figure 7, bottom panel). The only difference between C ii and Mg ii k is the later response of Mg ii k (18 s) when a sudden intensity increase at the edges of the brightening linked to the EE is observed. This clearly shows that the line centers of both chromospheric lines are emitted from plasma at similar temperatures.

Zoom In Zoom Out Reset image size

A line center reversal in the C ii has been seen in HRTS (0.05 Å spectral resolution) data but has never been reported in SUMER (0.045 Å) observations (Figures 9 and 10 in Avrett et al. 2013). We made an automatic scan of all the C ii data in the SUMER archive visually searching for C ii profiles showing explosive event line profiles. We found numerous examples of EEs where C ii shows a central reversal but only in stronger events. One of the reasons that a central reversal is now clearly seen in the IRIS data is apparently due to the IRIS higher spatial resolution. In SUMER, a small central reversal will not be visible but instead a flat profile peak will be registered.

An emission increase is clearly seen in the line center of Mg ii k and C ii at the west (the start of the scanning of the event as observed in a sit-and-stare mode) and east (end of the scanning) edge of the event. In Figure 4 (bottom row, first panel), we show the Mg ii k line-center radiance image where the two brightenings are seen. After careful investigation we found that this increase is not due to a line center emission increase but rather to a shift of the wings of both the C ii and Mg ii k lines toward the line center. On the west site of the event the intensity in the red wing increases and shifts toward the line center (shorter wavelength) while on the other edge (east or end of the scanning) of the event the intensity in the blue wing rises shifting again toward the line center (longer wavelength) (Figure 8). This behavior appears temporally and spatially where the Si iv line-wing-emission excess is observed, and it suggests that an intensity increase toward the line center is related to the plasma dynamics. Radiative transfer calculations are only available under the assumption of ionization equilibrium, therefore, these are not entirely valid for the atmospheric conditions in the presence of a highly energetic event (Leenaarts et al. 2013a, 2013b; Pereira et al. 2013). Doyle et al. (2013) showed that for dynamic bursts with a decay time of a few seconds, the Si iv line can be enhanced by a factor of two to four in the first fraction of a second with the peak in the line contribution function occurring initially at a higher electron temperature due to transient ionization compared to ionization equilibrium conditions.

Zoom In Zoom Out Reset image size

IRIS provides a unique opportunity to study solar phenomena in simultaneously taken spectroscopy and imaging data. Figure 10 displays the evolution of the event seen in the 1330 Å slit-jaw filter. At the beginning of the observations (18:42:08 UT), a brightening system with three bright cores can be seen in the area (see the arrows in the first panel of Figure 10). The EE is distinctly present in the AIA UV 1600 Å and 1700 Å channels with the same general shape. The three cores, however, are impossible to distinguish due to the lower (up to 3–4 times) spatial resolution of AIA. The event is seen in the AIA 1600 Å images from 17:54 UT to 19:19 UT, i.e., for 85 minutes.

The event was also investigated in the 304 Å, 171 Å, and 211 Å channels shown in Figure 11. In the AIA 304 Å and 171 Å images a dynamically evolving bright complex feature is present. During its evolution some loop pattern becomes apparent, but to clearly identify individual loops and to link to the underlying magnetic field configuration is close to impossible. One of the footpoints of the complex brightening is rooted in the three-core 1600 Å feature where the EE is also found to originate. The investigation of the associated HMI magnetograms shown in Figure 14 indicates that the EE is located above two opposite-polarity-canceling magnetic features (see the online animation with Figure 9). It also shows that the HMI resolution is too low to make any reliable field extrapolations for more detailed investigation of the magnetic field structures and associated small-scale dynamics. The bright core feature linked to the EE appears very dynamic in the IRIS 1330 Å images and a very close look reveals that continuously small-scale ejections take place during the entire IRIS observation period. Most of the ejections have a mini-jet-like shape. They propagate in random directions, and some of them (as the clear example shown with a black arrow in image at 18:56:20 UT of Figure 10) have a width of one pixel (i.e., 0166). Such fine scale jet structures have so far only being seen with ground-based telescopes, e.g., Ellerman bombs (e.g., Watanabe et al. 2011). Part of the ejected material appears to contract back to the source region toward the end of the 1330 Å image series a disk projection of what appears to be a small cloud-like feature is ejected after which the source region become weaker. To follow this evolution please view the online close-look SJ 1330 Å image animation (Figure 10). Chae et al. (1998b) found an association between EEs observed in Si iv 1402 Å and chromospheric upflows identified as a blueshifted Hα profile at 1'' spatial resolution. No jet features were observed but rather "dark dots" with a size of 2''–3'' and lifetime of 1–2 min were identified in the Dopplergram with an upflow velocity of 15–30 km s⁻¹. The authors suggested that the "chromospheric upflow events may be the manifestation of cool plasma material flowing into magnetically diffusive regions, while explosive events represent hot plasma material flowing out of the same regions" (see the abstract of the article).

Zoom In Zoom Out Reset image size

In the top panel of Figure 12, we present the lightcurves in the three spectral lines together with the lightcurve of the event in the SJ images (see the remarks in Figure 10). The lightcurves of the emission in the AIA EUV channels are shown in the panels below and are obtained from the boxed region overplotted on the first column in Figure 11. The region selected for producing the lightcurves is larger (765 × 735) in comparison to the event because we need to account for the movement of the feature and for the activity in its close proximity (perhaps triggered by the EE), see Figure 10. Thus the lightcurves are formed partially by the EE source region and also by the loops rooted in it. The lightcurves are smoothed by three frames. We need to point out that the dip in the SJ lightcurve during the peak of the event seen in the spectral lines is purely instrumental (see Section 2 for more details). The slit is obscuring the event during this period of time and because of the small size of the event at least one-third of its emission is blocked.

Zoom In Zoom Out Reset image size

The lightcurves in Figure 12 show two periods of brightening increase. The first lasts for ∼8 minutes but is actually composed of several intensity peaks each with duration between 90 and 120 s. The second brightening has the same duration as the individual spikes of the first brightening. These intensity variations are clearly linked to the small-scale ejections and brightness increase in the source region. The first intensity increase starts in the AIA 304 Å channel at ∼18:44 UT and ends at ∼18:53 UT. The start in the SJ images cannot be defined because of the spectroscopic observations. In the 171 Å channel the flaring of the EE begins later, i.e., ∼18:46 UT, and stops earlier at ∼18:52 UT. The event is first seen in the AIA 193 Å channel at the same time as in AIA 304 Å. In AIA 131 Å, it appears to be delayed by one minute with respect to the 304 Å and 193 Å channels but this can also be due to the weak signal in this channel. The start of the brightening in the 211 Å channel is impossible to determine.

At 18:55 UT, the EE starts to flare up again (only seen in the imaging data) and reaches maximum in the IRIS 1330 Å slit-jaw images at 18:56:01 UT. During this stage, the event shows a clear jet-like structure seen in IRIS 1330 Å SJ images (black arrow in image at 18:56:20 UT of Figure 10), and can also be followed in the AIA UV/EUV channels (arrows in the 3rd column of Figure 11). The time of the emission maxima are 18:56:55 UT in 304 Å, 18:57:47 UT in 171 Å (44 s later with respect to AIA 304 Å), 18:57:42 UT in 193 Å (102 s), 18:57:47 UT in 211 Å (102 s), and 18:58:32 UT for 131 Å (152 s). Because of the 12 s cadence of the AIA data, the 171 Å, 193 Å, and 211 Å actually respond simultaneously. Only the 131 Å channel is delayed with respect to the other coronal temperature channels by ≈50 s. The emission recorded in the 131 Å channel is known to be dominated by Fe viii and several transition region lines. Therefore, the later response can be related to a cooling rather than heating during the EE. The 171 Å channel is dominated by Fe ix but if a feature at low temperature is observed, the recorded emission will be at temperatures log T (K) < 5.7 (Vanninathan et al. 2012; Brooks et al. 2011). Although the AIA 193 Å channel is dominated by three Fe xii lines, it has a significant transition region emission contribution from non-identified lines (Del Zanna et al. 2011). The AIA 211 Å channel is also known to record transition region emission, a lot coming from Fe viii lines. However, 50% of the lines emitting in this channel remain unidentified (Del Zanna et al. 2011). Only in active regions we do have a strong contribution from Fe xiv. To conclude, the AIA channels do not give a straight answer on the temperature of the observed explosive event.

The clear intensity increase in AIA 171 Å, 193 Å, 211 Å, and 131 Å suggests that during the explosive event, plasma was ejected reaching temperatures to which all of these channels are sensitive. To determine this temperature, we used the emission measure (EM) Loci method (see, e.g., Winebarger et al. 2011; Alexander et al. 2013, etc.). The EM loci curves are constructed from the channel response functions calculated using the method described in Del Zanna et al. (2011) and the emission at around 18:58 UT. The resulting curves are shown in Figure 13. The EM loci suggests that log T = 5.36 ± 0.06 (K) is the most probable temperature of the event. The second loci-curve clustering at 6.18 ± 0.1 K is less plausible following the discussion above on the temperature response of the AIA EUV channels. An EUV active-region jet was reported by Chae (2003) in TRACE 1600 Å (transition region and continuum emission) and 171 Å. The jet, which was much larger than the jets reported here, had a temperature of 2–3 × 10⁵ K. The temperature of the jet was obtained from a TRACE filter ratio method.

Zoom In Zoom Out Reset image size

Having now the great opportunity provided by IRIS to observe spectroscopically the Sun at subarcsecond resolution, we face the challenge of having both incompatible coronal imaging and also magnetic field data. Nevertheless, we analyzed in detail the only available data from HMI. After the magnetic field associated with the explosive event was identified, we tracked its evolution starting an hour before the IRIS observations. A selection of HMI images is shown in Figure 14 and an animation in the online journal (Figure 9). At the start of the selected data set (i.e., 17:40 UT), a bipolar region is found in the area of the EE with one negative polarity fragment and a few positive ones. The distance between the two closest negative and positive fragments is ∼2''. The positive and negative fragments are moving toward each other while at the same time the positive flux increases. which is either due to flux emergence or convergence (the spatial scale and sensitivity of HMI does not permit a judgment as to which one of the two mechanisms is at work). Magnetic flux cancellation associated with EEs has been previously reported by Chae et al. (1998a) and Muglach (2008). In both studies, however, only around 63% and 19% (103 out of 165 and 7 out of 37), of the EEs resulted from flux cancelation, respectively. As in the present case, instrumental limitations could be the reason for this result and possibly larger or even all EEs are associated with magnetic flux cancellation. Thus studies using higher magnetic field resolution and sensitivity data are required, as will be discussed further below. Jet-like phenomena of various sizes and temperatures have been related to magnetic flux cancellation (e.g., Chae et al. 1999; Chae 2003; Chifor et al. 2008; Madjarska et al. 2012; Huang et al. 2012; Adams et al. 2014). The authors of all these studies suggest magnetic reconnection as the most probable driving mechanism.

Zoom In Zoom Out Reset image size

To evaluate the magnetic activity during the EE, we produced the temporal variations of the total positive and negative magnetic flux from the boxed region in Figure 14. A limited field of view was used around the region of the EE. The temporal variations shown in Figure 15 reveal that a flux increase starts prior to the event in both the positive (from ∼18:27 UT until 18:43 UT) and the negative (from 18:18 UT until 18:25 UT) polarities. The increase of the positive flux is at a rate of 2.7 × 10¹⁵ Mx s⁻¹, while the increase of the negative is 1.6 × 10¹⁵ Mx s⁻¹. At the beginning of the IRIS observations at 18:42 UT, the distance between the negative and positive magnetic fragments is within one HMI pixel when the magnetic-field cancellation was already ongoing. The cancellation rate is 4.7 × 10¹⁴ Mx s⁻¹ for the positive flux (linear fit to the curve between 18:39 UT and 19:17 UT), and 5.9 × 10¹⁴ Mx s⁻¹ for the negative flux (linear fit to the curve between 18:25 UT and 18:59 UT). We should stress here that these estimations are only a low limit because of instrumental limitations. The cancellation rates are close to those found for an X-ray jet (Huang et al. 2012), and much smaller than those recorded during a GOES C4.3 flare (Huang et al. 2014). As discussed in Huang et al. (2012), this flux cancellation rate strongly suggests that the event observed here results from an impulsive energy release most possibly magnetic reconnection. To compare, Chae et al. (2002) found 3.6 × 10¹⁴ Mx s⁻¹ and 9.7 × 10¹⁴ Mx s⁻¹ of flux cancellation rate for two cancellation sites, which the authors suggest to be consistent with a Sweet–Parker magnetic reconnection model. In the present case, the combined changes of the locations and the flux of different patches will result in reconfiguration of the field line connectivity that can only take place through magnetic reconnection. To fully understand the dynamical evolution of this event, reliable 3D models of the magnetic field are required. With the small size of this event, which represents only a few HMI pixels, a representation of the small-scale magnetic field is not possible to obtain by any extrapolation models. Without this, a detailed understanding of the underlying mechanism that drives this event is very challenging.

Zoom In Zoom Out Reset image size