DYNAMICS OF ON-DISK PLUMES AS OBSERVED WITH THE INTERFACE REGION IMAGING SPECTROGRAPH, THE ATMOSPHERIC IMAGING ASSEMBLY, AND THE HELIOSEISMIC AND MAGNETIC IMAGER

Observational data were obtained from IRIS, AIA, and HMI instruments from 16:26 UT to 17:51 UT on 2014 July 12. We used AIA passbands at 193, 171, and 1600 Å and HMI line-of-sight (LOS) magnetograms. We used standard AIA prep routines to produce level 1.5 images. IRIS data are taken in sit-and-stare mode with the slit centered at 239'', −559'' pointing adjacent to a coronal hole. This dataset has three spectral lines, namely C ii 1335.71 Å, Si iv 1402.77 Å, and Mg ii k 2796 Å. We use Si iv 1402.77 Å for this study. Slit-jaw images (SJIs) were available only with the 1330 Å filter. We have used IRIS level 2 processed data which are corrected for dark current, flat field, and geometrical corrections etc. The cadence of 1330 Å SJIs and spectra was ∼5 s with a pixel size of 0166. AIA and HMI images were de-rotated before AIA 1600 Å images were co-aligned with IRIS SJIs 1330 Å using cross-correlation. Figure 1 shows the plume as seen in various AIA channels, IRIS 1330 Å SJI, and HMI LOS magnetogram. The magnetogram (Figure 1(b)) shows that the plume is dominated by negative-polarity magnetic field marked with red contours (−50 G). We made an Movie 1 (84 minutes, available online) with HMI LOS magnetogram images; it shows the evolution of negative and positive flux with time. In the movie, red and green contours represent magnetic field strengths of −50 and 20 G, respectively.

Figure 1 clearly shows that the IRIS slit is crossing the footpoints of the plume. We extrapolate the coronal magnetic field with a potential field approximation, using the LOS magnetic field observed by HMI as the boundary condition at the photosphere (Figure 2). We select a large field of view (FOV) around this plume structure to approximately satisfy the flux balance condition. This FOV magnetogram is projected to disk center correcting for the difference between normal and LOS values, and we compute the potential field in a 3D box (Gary 1989) encompassing the plume structure.

Zoom In Zoom Out Reset image size

We trace field lines having footpoints in strong magnetic patches ($| {B}_{z}| \;\gt \;$ 25 G). In Figures 2(a) and (b), they are overlaid on the boundary magnetogram along with contours (±25 G). Most of the field lines from the magnetic regions are open (yellow). The shapes of the two major open-field structures associated with the strong negative magnetic patches are consistent with the funnel structure of the plume originating from the network boundary, as revealed from the AIA images (Figures 1(c) and (d)). However, only the northernmost structure appears bright in the AIA images at the time of the observations. Note that the difference in the orientation is due to projection, which is corrected in this model. It is also noted that the plume originates from the network boundary, as revealed from IRIS SJIs (Figure 1(a)).

Apart from these open field lines, low-lying field lines (blue) connect the weak field regions in the immediate neighborhood of the main plume structure. Movie 1 shows the continuous emergence of positive flux followed by flux cancellation. Reconciling the model of Shibata et al. (1994), the open-dipped field lines at the edges of the plume, in a dynamical scenario, facilitate continuous reconnection with the neighboring closed field lines to supply heat and energy for sustaining the plume.

We have focused on the dynamics of two footpoints (negative polarity) as marked by a triangle and square (at "Y" positions −546'' and −564'' respectively) in Figure 1. We study the time evolution of this small region as seen in IRIS Si iv 1402.77 Å spectral profiles. At first, a single Gaussian fit is performed on the weighted averaged (three pixels along the slit and three in time) spectral profile. We average the spectral profiles to increase the signal-to-noise ratio, using error bars as calculated by iris_pixel_error.pro as weights. From the single Gaussian fit we derive peak intensity, Doppler shift, line width, and asymmetry coefficient (see Figure 3). We use the median of the centroid of the fitted Gaussians as the rest wavelength to estimate the Doppler velocity. The evolution of the spectral profile at these two positions as a function of time is shown in Movie 2 (available online). In this movie IRIS 1330 Å SJIs are also included to show the jet-like features at network boundaries. To analyze the non-Gaussian aspect of the spectral profiles, we use the coefficient of asymmetry defined in Dolla & Zhukov (2011). However, contrary to what was done in that work, we do not normalize by the error bars of spectral intensities, because we are interested here in quantifying the asymmetry and not in assessing its statistical significance. The coefficient of asymmetry therefore provides the percentage of area in the profile that deviates from the fitted single Gaussian, according to the pattern defined by the following formula:

$$\begin{eqnarray}&&A=\displaystyle \frac{1}{{I}_{0}}\underset{k}{\displaystyle \sum }\epsilon (\lambda )\cdot {\rm{sgn}}({{\rm{\Delta }}}_{k}(\lambda ))\cdot {{\rm{\Delta }}}_{k}(\lambda ),\end{eqnarray} \tag{ 1 }$$

where λ is the wavelength, I₀ is the total intensity in the spectral line and ${\rm{\Delta }}(\lambda )={s}_{k}(\lambda )-{f}_{k}(\lambda )$ is the difference between the spectrum ${s}_{k}(\lambda )$ and the fitted Gaussian ${f}_{k}(\lambda )$, which are discretized on bin k. The contribution factor $\epsilon (\lambda )$ is defined as follows:

$$\begin{eqnarray}\epsilon (\lambda )=\left\{\begin{array}{ll}-1 & \mathrm{if}\;\lambda \in [{\lambda }_{0}-2\sigma ;{\lambda }_{0}-\sigma )\\ 1 & \mathrm{if}\;\lambda \in [{\lambda }_{0}-\sigma ;{\lambda }_{0})\\ -1 & \mathrm{if}\;\lambda \in ({\lambda }_{0};\;{\lambda }_{0}+\sigma ]\\ 1 & \mathrm{if}\;\lambda \in ({\lambda }_{0}+\sigma ;{\lambda }_{0}+2\sigma ]\\ 0 & \mathrm{otherwise},\end{array}\right.\end{eqnarray} \tag{ 2 }$$

where ${\lambda }_{0}$ is the center and σ the half-width at $1/\sqrt{e}$ of the fitted Gaussian. The sign of the coefficient of asymmetry indicates in which wing of the profile most imbalance is present (negative and positive on the blue and red wings, respectively). Another method of quantifying the profile asymmetry is the red–blue (RB) asymmetry method. The RB technique has been widely used in both optically thick (e.g., in Hα, Madjarska et al. 2009; Huang et al. 2014) and optically thin (e.g., De Pontieu et al. 2009; Tian et al. 2011b) spectral-line analysis.

Figure 4 shows spectral profiles at different instances (as marked in Figure 3) along the slit. We find that the spectral profiles are significantly non-Gaussian with two or more components present. We fit the spectral profiles at these instances with a single Gaussian (shown in orange) and double Gaussian (in green). Spectra averaged during the first 20 minutes of observations at the same "Y" positions are shown in blue. Both components of the double Gaussian are shown in gray. We observe both blueward and redward asymmetries at different instances (see Figure 4). The large asymmetry and double Gaussian behavior of the spectral profile indicates the presence of flows at these instances. Such behavior can also be attributed to waves (Verwichte et al. 2010) but we can discard this interpretation in our observations because the associated fluctuations in intensity are too large to correspond to linear waves only (see Figure 5). Non-Gaussian behavior is present on several occasions during the time of observation of IRIS at the positions marked by the triangle and square in Figure 1 (see Movie 2 online). IRIS 1330 Å SJIs show the presence of several small-scale jets throughout the time of observation at these two positions. Reconnection jets cause enhancements at the line wings, which lead to large asymmetries, larger line width, greater intensity, and large Doppler shift. The redshift in spectra could be a projection effect as the plume is close to the south pole. Besides this the redshifts may indeed imply that at least part of the material is falling back. This is consistent with Figure 5 where the brightenings are associated with strong blueshifts, with progressive evolution not only toward zero but also redshifts (i.e., velocities slightly positive). There is probably a combination of both blueshifts and redshifts during this recovery phase.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In Figure 5 we show the wavelet analysis of spectral intensity, Doppler shift, and line width at the plume footpoint marked by a triangle (see Figure 1). The missing values in the left panel of Figure 5 correspond to the positions where a single Gaussian could not be fitted due to low counts. The green stars in Figure 5 represent the instances we have chosen to show individual spectral profiles (see Figures 4(e)–(h)). In the right panel global wavelet power peaks at 15.6 and 7.1 minutes for intensity, 26.2 and 7.1 minutes for Doppler velocity, and 11 and 17 minutes for line width. A small peak $\;\sim 16$ minutes in global wavelet power of Doppler velocity is also present. The horizontal dashed line is the cutoff above which edge effects come into play; thus periods above the dashed line cannot be trusted. The dotted line marks the 99% significance level for a white noise process (Torrence & Compo 1998).

It is quite evident from Figure 5 that the instances of the intensity peaks correspond to instances of large Doppler shifts and line widths and the flux cancellation of emerging positive flux. Movie 1 shows the evidence of a positive-polarity field around the dominant negative polarity. Cancellation of positive polarity at the edges of dominant negative polarity (region enclosed in a yellow box in movie 1) leads to the formation of reconnection jets that are characterized by intensity enhancement in spectra and apparent outflows in IRIS 1330 Å SJIs.

In order to compare dynamics at the transition region (as seen from IRIS) and the corona (as seen from AIA), we create smooth background-subtracted time–distance maps of AIA 171 Å and AIA 193 Å for the artificial slice 0 (see Figures 1(c) and (d)) co-spatial with the IRIS slit shown in green in Figure 1. We choose the width of the slice to be four pixels. We average along the width of the slice to increase signal to noise. Resulting time–distance maps are shown in Figure 6 (left panel). We observe quasi-periodic alternate bright and dark ridges. We fit the ridges with a straight line as done in Kiddie et al. (2012) and Krishna Prasad et al. (2012). We isolate a ridge and estimate the position of maximum intensity. We then fit the points of maximum intensity with a straight line, and thus the mean value of slope and error bars are estimated. The slope of the straight line gives an estimate of the speed of the PDs. Since we are interested in the average behavior of the PD, we take the mean of the speeds of all the ridges, and therefore some of the ridges in AIA 171 Å seem to be more inclined than overplotted straight lines. We find the average velocity of PDs is 51 ± 3 and 66 ± 8 km s⁻¹ in AIA 171 and 193 Å respectively.

Zoom In Zoom Out Reset image size

We carry out a similar analysis for two artificial slices marked as 1 and 2 in Figures 1(c) and (d) for AIA 171 and 193 Å, respectively. The slices are placed by looking at the direction of propagation of significant PDs in unsharp mask images. The widths of slices 1 and 2 are chosen to be 5 and 6 pixels respectively. Corresponding time–distance maps with best-fit overplotted ridges for slices 1 and 2 are shown in Figure 6 (middle and right panels respectively). We note that the velocity is higher in the hotter channel (AIA 193 Å) and the ratio between the velocity observed in AIA 193 and 171 Å is 1.29 ± 0.23, 1.23 ± 0.1, and 1.32 ± 0.07 (as compared to the theoretical value of 1.25 if the PDs are magnetoacoustic waves) for slices 0, 1, and 2 respectively.

We also estimate the velocities of PDs in slices 0, 1, and 2 using a cross-correlation method as done in Kiddie et al. (2012). We isolate individual ridges and estimate the time lag using cross-correlation between two positions. The velocity of the ridge is estimated by dividing the distance between two positions by the time lag. Thus, we compute the velocities of several ridges, and their mean value and standard deviation are estimated. Mean velocities and error bars (standard deviation for cross-correlation) using two different methods for slices 0, 1, and 2 are summarized in Table. 1.

Table 1.  Velocity of PDs Using Two Different Methods of Ridge Fitting

Slice	Channel	Velocity (km ${{\rm{s}}}^{-1}$)
		Using Best-fit Straight Line	Using Correlation
0	AIA 171 Å	51 ± 3	61 ± 12
	AIA 193 Å	66 ± 8	81 ± 11
1	AIA 171 Å	59 ± 3	60 ± 14
	AIA 193 Å	72.5 ± 1	80 ± 18
2	AIA 171 Å	72 ± 5	63 ± 10
	AIA 193 Å	95 ± 3	83 ± 25

Download table as:  ASCIITypeset image

Since the velocity is higher in the hotter channel (AIA 193 Å) using two different methods, we believe that these are propagating slow magnetoacoustic waves. We carry out a wavelet analysis at positions marked with dashed lines in Figure 6 (left panel). We find that there exist at least two dominant significant periodicities close to the footpoint of the plume (see global wavelet power plot in Figure 7). We choose the significance level to be 99% for a white noise process. We find that the periodicities in intensity in AIA 171 and 193 Å at slice 0 are similar to those in IRIS Si iv 1402.77 Å peak intensity (see Figure 5 top panel). We also find that the periodicity of 7.2 minutes is present in AIA 171 Å, AIA 193 Å, and IRIS spectral intensity, which suggests that the quasi-periodic reconnection jet outflows could be responsible for quasi-periodic PDs seen in AIA 171 and 193 Å.

Zoom In Zoom Out Reset image size

For slices 1 and 2 the rows marked with dashed lines in Figure 6 are used in the wavelet analysis (Figure 7). We find the two dominant periodicities of 13.2 and 7.2 minutes (24.3 and 7.2 minutes) and 17.1 and 6.6 minutes (26.4 and 6.6 minutes) in AIA 171 Å (193 Å) for slices 1 and 2, respectively (see Table 2).

We also carry out the wavelet analysis at other positions (10 and 15 Mm in the x–t map). The results are summarized in Table 2.

Since the periodicity of ∼7 minutes is present in slices 0, 1, and 2, we create time–distance maps using Fourier filtered images. We perform a Fourier transform at each pixel location. Fourier power is then multiplied by a Gaussian peaking at 7 minutes with a FWHM of 1 minute and the inverse Fourier transform is applied to get the reconstructed light curve at the same pixel location. Fourier filtered movies of AIA 171 and 193 Å are available online as Movies 3 and 4, respectively. Time–distance maps for slices 1 and 2 using Fourier filtered images are shown in Figure 9. We find that the periodicity of ∼7 minutes is significant in AIA 193 Å and mostly toward the end of observation (from 60 to 80 minutes in Figure 9 lower right panel) at the position of slice 2. This is also evident from wavelet plots as shown in Figure 8 where we note that a periodicity of ∼7 minutes is present at the position of slice 2 at the start and toward the end of the observation from 60 to 80 minutes (see Figure 8(d)). This fact is also supported by Fourier filtered movie 4 where we see the significant PDs propagating outwards at the position of slice 2 toward the end of observation. A Similar behavior is found in AIA 171 Å wavelet plots in slices 1 and 2 but the presence of the 7 minutes periodicity at later times is most prominent in AIA 193 Å at the position of slice 2. This indicates that these PDs are triggered by some drivers at specific instances.

Zoom In Zoom Out Reset image size

Since significant PDs are observed at later times at the position of slice 2, we fit the significant ridges with a straight line between 60 and 80 minutes in AIA 171 and 193 Å Fourier filtered time–distance maps for slit 2 as shown in Figure 9 right panel. The method of fitting is the same as explained in Section 2.3. We estimate the velocity to be 64 ± 3 km ${{\rm{s}}}^{-1}$ (73 ± 3 km ${{\rm{s}}}^{-1}$) and 55 ± 3 km ${{\rm{s}}}^{-1}$ (79 ± 2 km ${{\rm{s}}}^{-1}$) in AIA 171 Å (193 Å) for slices 1 and 2 respectively. We find that PDs of 7 minutes period are propagating with larger velocity in the hotter channel (AIA 193 Å), which further supports the fact that these could be propagating slow magnetoacoustic waves. However, we would like to point out that the PDs of 7 minutes period are not so clearly observed in AIA 171 Å and we see uneven ridges that could affect the velocity estimation. At the position of slice 1 (Figure 9 left panel) PDs of 7 minutes period are not clearly seen maybe because the period of 7 minutes is not the dominant one in the time series as shown in Figure 8(a). Thus we could fit only two ridges, one at 15 minutes and another at 65 minutes after the start of observation. We fit the corresponding ridges in AIA 193 Å and find that in slice 1, too, the velocity is higher in the hotter channel.

2.3.1. Jet-like Features in SJI 1300 Å

In this subsection we study the dynamic properties of the observed jet-like features. To determine the apparent speed of jets seen at the network boundary we place nine artificial box slices five pixels wide in SJI 1330 Å as shown in Figure 1(a). Slices are placed based on the direction of propagation of significant jets as seen in SJI 1330 Å in Movie 2 (available online). We average along the width of each box slice to increase signal to noise. Corresponding time–distance maps are shown in Figure 10. In the time–distance maps we fit significant ridges with straight lines. We find that several jets are preceded by brightenings seen in the time–distance maps, which suggests that small-scale reconnections could be triggering these jet-like features. We identify 62 jet-like features and fit them with dashed green curves as shown in Figure 10. The slope of the overplotted green dashed line gives an estimate of the apparent speed of the jet projected in the plane of sky. The velocity distribution of these jet-like features is shown in Figure 10 (last panel). We find that the distribution peaks around 10 km s⁻¹, which is similar to what has been reported by Shibata et al. (2007). Apart from outflows we see several downflows in the time–distance maps, which suggests that a certain amount of the jet material is falling back (see slit 4 in Figure 10).

Zoom In Zoom Out Reset image size

2.3.2. Correspondence between Jet-like Features in SJI 1300 Å and PDs in AIA 171 and 193 Å

In this subsection we explore whether the jet-like features can be responsible for the generation of the PDs.

To understand the source of PDs generated in the corona, we compare the spectral intensity of IRIS slit spectra at the footpoint (marked by the triangle in Figure 1) and the intensity of PDs observed in the time–distance map of slit 0 (see Figure 6) at 2 Mm, which coincides with the position of the triangle marked in Figure 1, for AIA 171 Å and AIA 193 Å. The left panel of Figure 11 compares the peak intensity of IRIS slit spectra (top left), PDs (along slit 0) observed in AIA 171 Å (middle left), and PDs (along slit 0) observed in AIA 193 Å (bottom left). The dotted–dashed line in black represents the peak spectral intensity while the green line represents the corresponding intensity peaks in AIA 171 and 193 Å. We find that IRIS spectral intensity peaks precede the AIA intensity peaks and we estimate a lag of 24–84 s with a mean lag of ∼60 s.

Zoom In Zoom Out Reset image size

Some of the peaks in spectral intensity do not correspond to sharp peaks in AIA 171 and 193 Å, maybe because the jet is aligned sideways, and do not move along the slice 0 placed co-spatially with the IRIS slit. We also compare the light curves at 0.5 Mm at box slice 8 placed in SJI 1330 Å (see Figure 1(a)) and 3 Mm at box slice 2 placed in AIA 171 and 193 Å (see Figures 1(c) and (d)). These two positions are also co-spatial.

From Figure 11, we note that there is fairly good one-to-one correspondence between jet-like features in slice 8 in IRIS SJI and PDs in slice 2 in AIA 171 and 193 Å. The dotted–dashed line represents the peak in IRIS SJI intensity. We also note that there is no significant lag except at the last ridge (marked by the dotted–dashed green line) where the lag is ∼120 s. We carry out a similar analysis for co-spatial positions at slice 3 placed in SJI 1330 Å and at slice 1 placed in AIA 171 and 193 Å. We found a lag of 84 s at two positions marked with green dotted–dashed lines.

It should be noted that the time lag between peaks in IRIS SJI and AIA is not uniform. Apart from the fact that some jets are preceded by the brightenings in x–t maps, at some positions we may not capture the exact time when a jet starts in IRIS SJI (see slice 8 in Figure 10 between 40 and 80 minutes). What we know is when a jet appears at the position of slices placed in IRIS SJI. It is also possible that when jets appear in a given slice, they have already covered some distance in the vertical direction (projection effect) and have generated the PDs visible in coronal channels. Thus they may appear co-temporal with PDs observed in AIA 171 and 193 Å.

These results allow us to propose that jet-like features in the transition region may cause PDs as observed in the corona.