Multiwavelength High-resolution Observations of Chromospheric Swirls in the Quiet Sun

We present observations carried out between 08:07:24 and 09:05:46 UT on 2012 June 21, using the CRISP instrument on the Swedish 1 m Solar Telescope. The spatial resolution of these observations is 016 at 6302.0 Å and 020 at 8542 Å. The spatial sampling of the detector is 0059/pixel. The time cadence of the data is 8 s. The data were reconstructed using Multi-Object Multi-Frame Blind Deconvolution (MOMFBD; van Noort et al. 2005). The field of view (FOV) of 55'' ×55'' was centered in the quiet-Sun region, at solar-x = −31 and solar-y = 699. Using CRISP, we observed 11 line positions along the Hα line scan with increments of ±0.26 Å from the line center at 6563.0 Å up to ±1.29 Å. For the Ca ii line scan, we observed 19 line positions with increments of ±0.055 Å from the line center at 8542.0 Å up to ±0.495 Å. We obtained polarimetric observations in Fe 6302 Å at a single wavelength position at +0.043 Å from the line center. Figure 1 shows the SST FOV on an SDO-AIA 1600 channel full-disk image of the Sun. The context images for SST obtained in Hα at +1.29 Å from the line core, Ca ii 8542 Å at 0.495 Å from the line core, and Stokes-V of Fe 6302 Å are shown in subpanels (e)–(g). Reference images from SDO-AIA covering temperatures ranging between 4500 and 600,000 K are shown in subpanels (b)–(d) and (h).

Zoom In Zoom Out Reset image size

Chromospheric swirls are difficult to identify without clear, well-developed signatures in intensity. Therefore, we have adopted a set of visual criteria for selecting candidate swirls. First, we need to observe a circular pattern in intensity, at least at one time, in the line core of Ca ii 8542.11 Å and in its wings at ±0.11 Å, as well as in the Hα line core and 0.26 Å red wing. In Hα, swirls have a more fragmented appearance than in Ca ii. Second, we require that the swirl is associated with one or more photospheric magnetic concentrations observed in Fe 6302 Å. This emphasizes the connection between photospheric and chromospheric dynamics as suggested by Wedemeyer-Böhm et al. (2012).

In our data set, we have been able to visually identify 13 candidate swirls using these criteria. Their locations are indicated on an SST context image in Figure 1. Also, further basic details are provided for all swirls in the Appendix. The observed swirls have an average diameter of 15 and lifetime of 9–10 minutes, as seen in the Ca ii wings. The lifetime is estimated based on the duration of time over which the feature can be visually tracked. These signatures are consistent with previous reports by Wedemeyer-Böhm et al. (2012). The five most promising swirl candidates are shown at a given time in the various channels in Figure 2. As we shall see, these swirls also broadly satisfy the identification criteria for a chromospheric swirl set out by Wedemeyer et al. (2013). We have constructed Doppler difference images from intensity differences at ±3.5 km s⁻¹ from the Ca ii line core. The blue color implies upflow and the red downflow. During the appearance of swirls in the Ca ii line, we observe cospatial bright-intensity ring fragments in Hα. Such cospatial Ca ii–Hα swirl features have not been reported before.

Zoom In Zoom Out Reset image size

Under the second criterion, the population of swirls can be split into two categories: five associated with multiple MCs, and eight associated with a single MC. We have chosen to present the analysis of the clearest swirl from both categories. Swirl 1 is representative of a chromospheric swirl formed above multiple MCs and is similar to Swirls 3, 8, 9, and 10. Swirl 2 is representative of a chromospheric swirl formed above a single MC and is similar to Swirls 4, 5, 6, 7, 11, 12, and 13. The time evolution of Swirls 1 and 2 is shown in Figures 3 and 4, respectively. We shall first provide an overall description of the appearance and evolution of the two swirls in the chromospheric channels and the underlying MCs in the photospheric channel. Later we shall undertake an in-depth analysis.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Swirl 1: The evolution of Swirl 1 is shown as a set of time snapshots in the various channels in Figure 3. The swirl is visible in the chromosphere between 08:42 and 08:57 UT. At 08:46 UT, roughly 10 minutes after the appearance of the MC, a circular structure is observed first in the Ca ii +0.11 Å red wing, followed by the Ca ii line core and −0.11 Å blue wing a minute later. Around the same time, circular patterns are observed in the Hα core and ±0.26 Å wings, though partially obscured by chromospheric fibrils. The circular chromospheric pattern further develops into a swirl, as seen as a pattern of intensity enhancement that forms one or more spiral curves, or "arms." Within the curve, the intensity is reduced, which enhances the swirl pattern. It is observed to rotate around its axis, as well as expanding as it evolves. Also, in Hα there is a modulation of the overall swirl intensity with time. Overall, the swirl is visible for 15 minutes.

During the initial stage, the arms of this swirl are redshifted but the swirl center is blueshifted, as seen in the Doppler difference images in Figure 3. However, later around 08:48–08:51 UT, the Doppler signal changes sign, where the arms are now seen blueshifted and the swirl center is redshifted. We have emphasized this effect in Figure 3 by placing black crosses on the swirl centers at times of a change in Doppler sign. The chromospheric dynamics are analyzed in more detail in Sections 3.2 and 3.3.

In the photosphere, two MCs of the same polarity become visible in Fe i Stokes-V around 08:35 UT, one to the north (top MC) and one to the south (bottom MC). The bottom MC is cospatial with the chromospheric swirl, such that the swirl center tracks the MC position. The top MC moves south a distance of about 100 km toward the bottom MC. At the same time, the bottom MC has two components of motion, which may be interpreted as a rotation of the MC around an external center that itself is translating. Furthermore, the evolution of its shape is suggestive of rotation of the MC itself. These rotational motions, combined with the cospatiality, suggest that this MC lies at the footpoint of and is connected with the chromospheric swirl.

Swirl 2: the evolution of Swirl 2 is shown as a set of time snapshots in the various channels in Figure 4. The swirl is visible in the chromosphere between 08:18 UT and 08:34 UT. A circular pattern begins to appear around 08:18 UT in Ca ii. This pattern develops into a swirl similar to Swirl 1 from 08:26 UT. However, the circular pattern is not prominent in Hα. Similar to Swirl 1, Doppler difference images show a blueshifted center at the swirl with a redshifted arm, as well as a change of Doppler sign with time. Again, we have placed black crosses on the Doppler difference images to indicate a change in Doppler sign.

In the photosphere, a single MC is present that is cospatial with the swirl center. We track its motion from 08:16 UT. The MC initially has a complex morphology. However, around 08:25 UT it becomes circular and more compact. This can also clearly be seen in the Hα +1.29 Å red wing. This change in morphology proceeds within a minute of the formation of the swirl pattern, and therefore suggests a connection between the MC and the overlying swirl.

The timing, evolution, and location of photospheric MCs suggest that a vertical magnetic field structuring is supporting the swirl. We shall analyze in detail the photospheric motion of the MCs in Section 3.1. The intensity modulation of swirls seen in Hα has a periodicity similar to the acoustic oscillations present throughout the chromosphere. In Section 3.2 we shall analyze how the acoustic oscillations interact with the swirls and explore if the acoustic oscillations are modulated by the swirls. The main swirl features of rotating arms and expansion are analyzed in Section 3.3.

For this data set, we have photospheric polarimetry available in a single wavelength position at +0.043 Å from the center of the Fe i 6302 Å line. Due to the availability of a single wavelength position, we restrict our analysis to the tracking in the Stokes-V signal of the horizontal displacements of magnetic concentrations cospatial with swirls, to obtain the temporal evolution of their horizontal velocity vector, that is, their speed and direction. This is achieved by using the SWAMIS tracking code (DeForest et al. 2007). SWAMIS detects and identifies features in the images using a two-threshold signed discriminator combined with a topological pixel-grouping method. Detected features between two time steps are associated by maximizing feature intersection both forward and backward in time.

Following the analysis of Stangalini et al. (2017), we use empirical mode decomposition (EMD; Huang et al. 1998) to separate the high-frequency horizontal fluctuations from the low-frequency part of motions of magnetic elements detected in Fe i Stokes-V. This allows us to study the low-frequency motions of MCs and protect our analysis from spurious results associated with rapid remnant seeing-induced motions. From the practical point of view, this is done by combining only those initial mode functions obtained from the EMD that are representatives of the low-frequency oscillations in the signal, while ignoring the remaining ones that contain high-frequency information and, generally, measurement noise. In what follows, we present the results of this analysis case by case for the MCs associated with the swirls.

Swirl 1: while there are two photospheric magnetic elements in the FOV of Swirl 1, only the bottom one appears cospatial with the swirl, as seen in the chromospheric lines (see Figure 3). Figure 5 (upper panel) shows the low-frequency evolution of the velocity vector obtained from tracking the feature across the solar surface, in polar coordinates. For most of the lifetime of the structure, both the magnitude and direction of the velocity vector evolve slowly. A coherent change of direction is indicative of rotation of the magnetic element around an external center. The direction angle does not necessarily have to complete 360° as this motion may be superimposed on a linear translation of this external center. Here, the velocity vector covers a range of direction angles of about 200°. The rotation has a period of about 120 s. We also identify times at which the velocity vector direction reverses. This may suggest the appearance of new impulses in the photospheric plasma flows. We conclude that the MC rotates around an external point, which itself is undergoing a slow linear translation (Stangalini et al. 2017 dubbed such motion to be rotational), and is not consistent with a random process. This motion seems consistent with a magnetic element moving in a larger photospheric vortex flow field.

Zoom In Zoom Out Reset image size

For the sake of completeness, we also show in Figure 5 the power spectra of the fluctuations of the two components of the horizontal velocity vector (i.e., v_x and v_y). Both components are well correlated and display several harmonics up to ∼40 mHz or, equivalently, down to a period of 25 s. It is worth noting that the maximum power is located at ∼30 mHz (33 s).

Swirl 2: the same analysis is made of the single magnetic element associated with Swirl 2 and is shown in Figure 6. As for the previous case, the evolution of the direction of the velocity vector is not random. However, the velocity vector covers a narrower range of direction angles of about 150°. Also, there are instants at which the motion reverses. Hence the motion may also be consistent with a linearly polarized and transverse oscillatory movement with a periodicity around 100–120 s superimposed on top of a translation. The amplitude of the velocity vector is smaller than in the previous case. This is also seen in the power spectra. It is not possible to unambiguously identify harmonics in the spectra. Similarly to Swirl 1, the power is also concentrated in the same spectral region up to 40 mHz (the Nyquist frequency is around 80 mHz).

Zoom In Zoom Out Reset image size

The analysis of Swirls 3, 4, and 5 shows that, at least in some cases, there are two magnetic structures rotating around one another. This is for instance somewhat clear in Swirl 1, showing the temporal evolution of the photospheric polarization signals. However, a detailed study where the signals in the photosphere spatially resolve this rotation is needed.

The high-β chromosphere oscillates with a natural period of 3 minutes (e.g., Fleck & Schmitz 1991; Hansteen 1997). In order to investigate the association of acoustic oscillations with swirls, we analyze time series constructed from the average intensity across the FOV in the various channels, using the continuous wavelet transform (CWT; Torrence & Compo 1998) with a Morlet mother wavelet. A time series covers the evolution of the chromospheric plasma before, during, and after the swirl. Different wavelengths across a single spectral line, and between lines, probe physical conditions at different atmospheric heights. Therefore, we use cross correlations between CWTs at different wavelengths to assess physical couplings of plasma in the swirls at different heights (Grinsted et al. 2004). We define the cross-wavelet transform (XWT) of two time series f(t) and g(t) as

$$\begin{eqnarray}&&\mathrm{XWT}{(f,g)=\mathrm{CWT}(f)\mathrm{CWT}(g)}^{* },\end{eqnarray} \tag{ 1 }$$

where * indicates the complex conjugate. The XWT reveals time intervals and period ranges where the two time series show a common high power (significance) and hence a strong correlation. Significance in the XWTs of two time series is shown using vectors that indicate at each time and period the strength and direction (phase relation) of the correlation. The lag of time series f with respect to time series g is thus represented by the direction (angle). If f is in-phase with g (0°), the correlation vector points to the right. If f is in antiphase with g (180°), the correlation vector points to the left. For a 90° phase lag of f with respect to g, the correlation vector points down. Similarly, for a −90° phase lag (a 90° lead) of f with respect to g, the correlation vector points up.

Swirl 1: we create time series from the Swirl 1 FOV in the various channels for the time interval from 08:21 to 09:05 UT. This covers a time span of 20 minutes before the appearance of the swirl and 7 minutes after the swirl is no longer clearly visible. Their CWTs and XWTs are shown in Figure 7. CWT wavelets corresponding to the various time series are shown in Figure 7. It shows that wave power is enhanced during the swirl. The power of the acoustic oscillations is significant (90%) for a period of 150 s (∼7 mHz frequency) in both Ca ii and Hα. In the absence of swirls, the acoustic oscillations in the chromosphere remain at a constant period of about 180 s. In the presence of the swirl, the power in the oscillation increases, especially in Hα, and the period shortens. The drop in oscillation power immediately before and after the swirl is indicative of a fast change in periodicity. We can conclude that the presence of Swirl 1 modulates the natural chromospheric oscillation frequency.

Zoom In Zoom Out Reset image size

The XWTs are shown in Figure 7. In Ca ii we note that in terms of phase lag the blue wing leads the line core by 90°, which in turn leads the red wing by 90°. In this scenario, we expect the XWT of the two Ca ii wing time series to be in antiphase. However, we find a 90° phase relation instead. We note that the XWT for those two time series has low power, so its phase relation is in doubt. Also, though the period is moderated, the XWT phase relations themselves are not changed by the appearance of the swirl. In Hα, the phase relations are more complex. We look at the phase relations for the first and second halves of the swirl time interval. For the first time half, the blue wing leads the line core by 90°, which in turn slightly leads the red wing. In the second half, roles are reversed. The blue wing has a tendency to lag the line core, which in turn slightly lags behind the red wing.

Swirl 2: we create time series from the Swirl 2 FOV in the various channels, for the time interval from 08:12 to 09:05 UT. The swirl starts at 08:23 UT and lasts for 16 minutes. Their CWTs and XWTs are shown in Figure 8. In Ca ii we detect an oscillation with a period between 230 and 300 s, with a tendency of the period to increase from 230 s at the beginning to 300 s near the end of the swirl. Before and after the swirl, the oscillation period is about 300 s. The presence of this swirl seems to reduce the oscillation period, but it is less pronounced than in Swirl 1. In Hα we only clearly detect an oscillation during the time of the swirl in the line core. Compared with Ca ii, the maximum oscillating power is located between 150 and 180 s. We are possibly detecting different acoustic modes that are locally dominant at different heights in the atmosphere.

Zoom In Zoom Out Reset image size

The XWT for Ca ii shows in terms of phase correlation that the line core leads the red wing, which in turn leads the blue wing. However, note that the swirl signature in the line position −0.11 Å is not clear. Because in Hα the signal is only clear in the line core, the XWT for this line is inconclusive.

We have also examined the wavelet and cross-wavelet transforms for line position time series for Swirls 3, 4, and 5. The results are less clear with reduced significance. Swirls 3 and 4 have oscillation periods around 200 s and no clear moderation in oscillation period during the swirl. Swirl 5, like Swirl 2, shows an oscillation with a periodicity between 200 and 300 s that has a trend toward larger period during the swirl. In Hα this periodicity, as well as some evidence for a 120 s period, is seen. In Ca ii, we find a similar correlation phase in the XWTs as for Swirl 1 for all three swirls. In Hα, Swirls 4 and 5 do not show anything significant, but Swirl 5 shows a phase pattern similar to Swirl 1.

Furthermore, considering Ca ii, the time series for the acoustic oscillation for all swirls (where significant) show the same periodicity in the core and in the wings. If the change in intensity in the various line positions is due to Doppler velocity, then a sinusoidal profile of the acoustic velocity would lead to a doubling of the oscillation frequency in the line-core time series (per cycle in the velocity time profile, a blue or redshifted antinode only occurs once, but a node occurs twice). We do not see this. This suggests that the velocity time profile has an asymmetric shape. The line core of the Ca ii line is formed at about 1500 km or less. This is above the height where three-minute acoustic waves are expected to start to form shocks and show asymmetric velocity profiles (Carlsson & Stein 1995, 1997). Also, the XWT phase relations between the core and wings in Ca ii can be reproduced in a simple model that uses synthetic time series evaluated at the core and wings of a Gaussian spectral absorption line that is Doppler-shifted by an asymmetric velocity time profile. For some of the swirls, the oscillation correlation phases in Hα are influenced by the swirl itself. Detailed forward-modeling of the interaction of a swirl and acoustic oscillations would be required to elucidate this further.

The results from wavelet analysis suggest that we most likely see a modification of the acoustic oscillations due to the magnetic structure propagating from the photosphere to the chromosphere. In the swirl there may be temperature increments, which are most likely related to the intensity enhancements observed in the fragments of the swirl in the Hα line core. A temperature increase leads to an increase in sound speed and hence a decrease in period at the same wavelength. A change in periods from 300 to 150 s represents a temperature increase by a factor of 2.

The motion that evolves in the swirl from the photosphere to the chromosphere is observed to rotate and expand. We investigate the dynamics within a swirl by introducing a polar coordinate system (r, φ) centered on the swirl. We visually track the swirl center in time in the Ca ii red wing and fit a polynomial to the center x and y coordinates to obtain a smooth center track. As before, we show our analysis for Swirl 1 and Swirl 2. Figures 9 and 10 show the Ca ii line core and ±0.11 Å. The signature of the swirl is obvious in Ca ii but much less clear in the Hα red wing and has been omitted. The images are true-type where the R, G, and B channels correspond to the red wing, line core, and blue wing, respectively. There is a clear periodicity in intensity over time across the whole swirls that is linked to previously mentioned chromospheric acoustic oscillations. The phase relations of the oscillations seen in various line positions, which were revealed in Section 3.2, are apparent here as well. Note that, to enhance the detailed dynamics within the swirl from the overlying acoustic oscillation, we rescale at each time the intensity with the average intensity across the whole FOV of the swirl. We cannot eliminate the acoustic signal completely, but differences in phase at different locations of the swirl become apparent in the time–distance plots.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Swirl 1: Figure 9 shows polar plots corresponding to Swirl 1 in Ca ii. In the integrated intensity plot calculated within radius r₀, we see continuous intensity enhancements related to the swirl. These have a period of ≈200 s, as observed in Section 3.2. These patterns of low and high intensities are also seen in the radial time–distance plots showing that there is a continuous average radial speed of 10 km s⁻¹ associated with the swirl. Furthermore, the angular time–distance plots reveal clear signatures of rotation, which become more apparent at larger radius. For Swirl 1, there is some suggestion it is rotating in a counterclockwise direction at a rate of about 0.04 rad s⁻¹ (160 s periodicity), while the outer region rotates clockwise at the same rate.

Swirl 2: Figure 10 shows polar plots corresponding to Swirl 2 in Ca ii 8542 Å. As in Swirl 1, Swirl 2 forms a similar radial time–distance plot that shows a continuous radial signal propagation at an average speed of 20 km s⁻¹. A counterclockwise and clockwise rotation of 0.06 rad s⁻¹ (105 s periodicity) has been superimposed as a solid and a dashed line.

In addition, Swirl 4 and Swirl 5 show radial and angular patterns, whereas Swirl 3 does not show any clear radial or angular patterns. Swirl 4 shows a counterclockwise rotation at the rate of 0.045 rad s⁻¹ near the center, slowing down to 0.03 rad s⁻¹ near the arms. Swirl 5 shows a counterclockwise rotation of 0.03 rad s⁻¹ in the outer regions. The swirl rotation is not dissimilar to the periodicity of the motion of the underlying photospheric MCs.