On the multi-threaded nature of solar spicules
aa r X i v : . [ a s t r o - ph . S R ] O c t D RAFT VERSION S EPTEMBER
11, 2018
Preprint typeset using L A TEX style emulateapj v. 5/2/11
ON THE MULTI-THREADED NATURE OF SOLAR SPICULES
H. S
KOGSRUD , L. R OUPPE VAN DER V OORT , AND
B. D E P ONTIEU
Institute of Theoretical Astrophysics, University of Oslo, P.O. Box 1029 Blindern, N-0315 Oslo, Norway and Lockheed Martin Solar & Astrophysics Lab, Org. A021S, Bldg. 252, 3251 Hanover Street Palo Alto, CA 94304 USA
Draft version September 11, 2018
ABSTRACTA dominant constituent in the dynamic chromosphere are spicules. Spicules at the limb appear as relativelysmall and dynamic jets that are observed to everywhere stick out. Many papers emphasize the important rolespicules might play in the energy and mass balance of the chromosphere and corona. However, many aspectsof spicules remain a mystery. In this Letter we shed more light on the multi-threaded nature of spicules andtheir torsional component. We use high spatial, spectral and temporal resolution observations from the Swedish1-m Solar Telescope in the H α spectral line. The data targets the limb and we extract spectra from spiculesfar out from the limb to reduce the line-of-sight superposition effect. We discover that many spicules displayvery asymmetric spectra with some even showing multiple peaks. To quantify this asymmetry we use a doubleGaussian fitting procedure and find an average velocity difference between the single Gaussian componentsto be between 20–30 km s − for a sample of 57 spicules. We observe that spicules show significant sub-structure where one spicule consists of many ’threads’. We interpret the asymmetric spectra as line-of-sightsuperposition of threads in one spicule and therefore have a measure for a perpendicular flow inside spiculeswhich will be important for future numerical model to reproduce. In addition we show examples of λ − x -slices perpendicular across spicules and find spectral tilts in individual threads providing further evidence forthe complex dynamical nature of spicules. INTRODUCTION
Spicules are relatively small and highly dynamic jets thatprotrude everywhere from the limb as observed in chromo-spheric spectral lines. Their nature is still a mystery owingto their dynamic nature and small size, ranging from somehundred kilometers all the way down to the resolution limitsof state of the art telescopes. Spicules have been extensivelystudied in the past and the earliest discovery of spicules datesback to 1877 by Secchi. The early spicule observations andmodels are reviewed by Beckers (1968) and Sterling (2000).Using Ca II H images from the then recently launchedHinode satellite (Kosugi et al. 2007; Tsuneta et al. 2008;Suematsu et al. 2008) De Pontieu et al. (2007b) found thatspicules come in two classes, type I and type II. The dis-tinction between type I and type II spicules is based on thestriking differences in dynamic behavior in the Ca II H 3968˚A passband of Hinode/SOT. Type II spicules are the domi-nant type in quiet-Sun and coronal holes regions while typeI spicules are almost exclusively found in active regions.Type I spicules show a slower apparent outward velocity (15-40 km s − ) and are seen to fall back toward the limb af-ter reaching maximum height, whereas type II spicules showfaster outward velocity (30-110 km s − ) and are observedto fade from the Ca II H passband at their maximum extentlikely due to heating. While it is unclear what drives type Ispicules, their dynamic behavior show similarities with thoseof dynamic fibrils in active region and some quiet Sun mottleswhich have been studied in detail by Hansteen et al. (2006),De Pontieu et al. (2007a), Rouppe van der Voort et al. (2007)and Mart´ınez-Sykora et al. (2009). Hints of sub-structurein spicules have been found in Hinode data, Suematsu et al.(2008) and Sterling et al. (2010) suggest that many spiculeshave two components.De Pontieu et al. (2011) show that brightenings appear inhotter passbands in the Atmospheric Imaging Assembly (AIA, Lemen et al. (2012)) after type II spicules disappearfrom H α , which strengthens their importance as energy me-diators in the outer atmosphere. It is however not clear towhat extend spicules reach coronal temperatures or how theirenergy is deposited in the corona.There are a very limited amount of numerical studies oftype II spicules compared to the wealth of observational datathat exists. Mart´ınez-Sykora et al. (2011) study in detail onespicule similar to a type II spicule that evolved naturally intheir large scale simulation. Type II spicules are generallynot present in large scale simulations of the solar atmosphere,even though observations indicate that type II spicules are thedominant constituent in the chromosphere and corona inter-face region.The motion of type II spicules is found to consistof a combination of tree motions: Field-aligned flows,transverse swaying of the central axis and torsional ro-tation around the central axis. This is observed fromspicules off-limb (De Pontieu et al. 2007a; Pereira et al. 2012;De Pontieu et al. 2012) and from the disk-counterpart ofspicules (rapid blue-shifted excursion) (Langangen et al.2008; Rouppe van der Voort et al. 2009; Sekse et al. 2012,2013a,b).The aim of this paper is to study the multi-threaded natureof spicules in high resolution images and the torsional compo-nent in more detail by searching for clean spectra of spiculesfar off-limb. Targeting off-limb is especially important be-cause the spectra are not as optically thick compared to on-disk and therefore are easier to interpret. Given the enormousline-of-sight effects close to the limb we need to go to greatheights to reduce superposition. OBSERVATIONS
We analyze a high quality dataset obtained with the CRISPinstrument (Scharmer et al. 2008) at the Swedish 1-m SolarTelescope (SST, Scharmer et al. (2003a)). CRISP is an imag-ing spectro-polarimeter using dual Fabry-P´erot etalons as de-scribed by Scharmer (2006) and it works as a tunable narrow-band filter that can rapidly scan spectral lines isolated by aprefilter. One camera is situated just after the prefilter calledthe wide band camera (WB) and two more cameras are placedafter the dual etalons. In H α the full width at half maximumof the transmission profile for CRISP is 6.6 pm and 0.49 nmfor the H α prefilter.The data analyzed was from 27 June 2010 between11:43:12 − ′′ inside from the limb. The pixel sizeis 0.059 ′′ and the entire field of view (FOV) covers about57 × ′′ of the Sun. Both observations from STEREO(Howard et al. 2008) from the same day and AIA a few dayslater show that the solar region below the off-limb part is mag-netically less active than the enhanced network in the centerof the FOV. The size of the FOV region off-limb is approxi-mately Mm .The observational sequence scanned the H α line at 41equidistant positions symmetrically around line-center. Thescan started at − m ˚A with steps of m ˚A, equivalent toa Doppler offset of − km s − in steps of km s − . Thecadence of the series is 9 s and 281 scans were recorded.The data targeted a region of plage just inside the south-eastern limb of the Sun. A sunspot was situated close to, butoutside the FOV.The observations were post-processed with the MOMFBD(van Noort et al. 2005) procedure. The MOMFBD procedureassures that the NB images are all aligned to the WB imageand therefore to each other, thereby achieving a high degreeof spectral integrity in the spectral scan. The WB and NBobjects are given equal weight in the restoration process. Thisprocess may leave residual seeing artifacts due to changes inthe differential seeing. This is compensated for by addinganother object to the restoration following Henriques (2012).It is important to note that off-limb there is reduced signalin the wide-band channel. This makes it difficult to computethe wavefront deformation and makes the restoration less suc-cessful. However, the seeing in this dataset is excellent andvery stable and it was verified that spurious signals from spa-tial misalignment due to the sequential nature of the acquisi-tion method are negligible.We used CRISPEX (Vissers & Rouppe van der Voort 2012)to explore and analyze the observations. ANALYSIS
When we look at the data it is apparent that many spiculesdo not evolve independently, but rather evolve as small groupsof spicules in a collective fashion. The spicules in the groupsappear and disappear nearly at the same time and have simi-lar apparent motion. Two examples are shown in Fig. 1. Inthe first example two spicules have an apparent outward mo-tion. After 81 s (not shown in the figure, see movie in theonline material) the rightmost spicule splits into two spicules.After an additional 27 s the leftmost spicule splits into twospicules. In the second example a larger ensemble of spiculeshave appeared and appears to move outward. The very brightspicule to the left at t = 27 s appear to be made up of twospicules and at later times they appear to separate. In Fig. 2we show three examples of spicules merging or splitting upover a short period of time. In the first and third example thesplitting appears along nearly the entire length of the spiculeat the same time indicating that the relative motion of the in- Figure 2.
Three examples of spicules splitting and merging in a short periodof time. The images are from H α line center and are radial filtered to makethe spicules stand out more. The cutouts are fixed in space. h is the intensity, µ is the mean of the distribution and σ isthe standard deviation of the distribution. The number in thesubscript refers to the two Gaussians. This function contains7 free parameters: b, h , µ , σ , h , µ and σ .The asymmetry analysis involves computing the asymmetry of a line profile, which we define for a discrete intensity lineprofile I ( λ ) to be: a ( i ) = P λ ′ − i ∆ λ − ∆ λ ′ λ = λ ′ − i ∆ λ I ( λ ) − P λ ′ + i ∆ λ +∆ λ ′ λ = λ ′ + i ∆ λ I ( λ ) I max − I min , (2)with i ∈ [0 , , , . . . ] . λ ′ is the location of the peak of theline profile, ∆ λ ′ and ∆ λ are user defined intervals (both setto 10 km s − ) and I max and I min are the maximum and min-imum value of the line profile respectively. The location ofmaximum asymmetry is found from the peak of | a ( i ) | .The procedure first counts the number of peaks in the lineprofile. If there are two peaks in the line profile µ and µ are placed at one peak each. If there only is one peak theasymmetry analysis is performed and the level of asymme-try is compared to a threshold. The threshold was set to . because at less asymmetry the double Gaussian fit would of-ten fail. If sufficient asymmetry is present we place one µ at the peak of the line profile and the other at the location ofmaximum asymmetry. If the level of asymmetry is below thethreshold then a single Gaussian function is fitted instead. The µ ’s are allowed to have a small “wiggle room”, ± km s − ,to achieve best fit. This effectively reduces the number of freeparameters down to 5.For the asymmetric single-peaked line profiles we performthe fit 50 times while varying the starting estimate for the µ placed at the location of peak asymmetry. We do this becausethe solution from the fitting procedure is at times dependenton the initial guesses for the parameters with that µ beinga more sensitive parameter. All unique solutions the proce-dure converges to are stored and near identical solutions aregrouped as one and the number of groups are counted. If thereis more than one group, i.e. the solution to the fitting is am-biguous, the line profile in question is discarded from furtheranalysis.To quantify the asymmetric line profiles we selected 91spicules by eye in moments with excellent seeing, in simi-lar fashion as for Fig. 3. The line profiles constructed at eachheight were an average of the two closest pixels perpendicularto the spicule axis and the pixel on the spicule axis. DoubleGaussian fit was performed for each line profile and from thiswe computed the average velocity difference in each spicule,shown in Fig. 4. At each height the level of asymmetry ofthe line profile was assessed. Only heights where the level ofasymmetry was above the threshold were included in the aver-age. In the histogram 57 spicules are included. The remainingspicules either did not display sufficient asymmetry (19), thefitting result was ambiguous (10), or a combination of the two(3).Further complexity in spicules is apparent when we look at λ − x slices across spicules, some examples in Fig. 5. The toppanel shows the spatial position of the slits with correspond-ing λ − x slices in the remaining panels. For most examples astriking feature is the tilt of the spectral line as function of per-pendicular distance across the spicules, most clearly seen inslit 8. A tilt in a λ − x slice indicates a velocity gradient acrossthe feature and may be interpreted as a sign of rotational mo-tion. The central spicule in slit 4 is even more complex. Itshows two peaks in the red and blue wings indicating an over-lap of two threads and both show different spectral tilts. Forslits no. 3 and 5 there are no clear velocity gradients acrossthe spicules. DISCUSSION
Figure 3.
Spectral line profile as function of height for 8 spicules. The upper panel shows a cutout of the full FOV of line center H α with the spicules studiedmarked by red dash-dot lines. The image is radial filtered to make the spicules stand out more clearly. The inclined gray line towards the lower right is theposition of the photospheric limb. In the remaining panels the spectral line profiles are shown for 5 different heights with increments of 25%. The black line isfrom the position closest to the limb while red is from the position furthest away from the limb. Each line profile is an average of three line profiles perpendicularto the spicule. The normalization factor is individual for each spicule. Figure 4.
Histogram showing the average velocity difference between thetwo single Gaussian components for 57 spicules.
We studied H α spicules in an excellent SST off-limb timeseries and often observe group behavior: multiple spicules actas ’threads’ in a larger structure, with each individual threaddisplaying a complex dynamic nature. The overall apparent motion is synchronized with the other threads in the struc-ture. We find that spectral line profiles of isolated spicules farabove the limb are often asymmetric, sometimes with cleardouble peaks, and are best fitted by a double Gaussian lineprofile. The scenario that is emerging from these observationsis that spicules are fundamentally multi-threaded with strongperpendicular flows inside ’one’ spicule. This multi-threadednature is compatible with a whole volume undergoing spicu-lar motions, with only a few threads visible at one time, andapparently harbouring turbulent perpendicular flows. The per-pendicular flows peak at around 20–30 km s − which is com-patible with previous reports of Alfvenic waves. All of thisprovide constraints on models for the formation and evolutionof spicules.Suematsu et al. (2008) suggest that more than 50% ofspicules appear as double-threaded structures in the Ca II3968 ˚A passband of Hinode. In our observations, we actu-ally find that we often see more than two threads. They reportseparation of the order of a few tens of an arc-second of the in-dividual threads while we see synchronous behavior at larger Figure 5. λ − x slices across many spicules. The upper panel shows 8 artificial slits as red lines overplotted on a cutout from the full FOV in line center H α . Theinclined gray line towards the lower right is the position of the photospheric limb and the image is radial filtered. The remaining panels show the λ − x slices ininverted color table with the distance measure proceeding from left to right along the slits in the upper panel. spatial extents at times exceeding 2 arc-seconds.We regard chance alignment of unrelated overlappingspicules as an unlikely cause of our findings: we selectspicules far from the limb to avoid line-of-sight superposi-tion and we often find near constant spectral asymmetry asfunction of height along the spicule.Which type of spicules are we analyzing? The classifica-tion of spicules into type I and type II are traditionally madein the Ca II 3968 ˚A passband. Here we analyze H α data andit is unclear how these types of spicules manifest in this di-agnostic - we are unaware of any detailed comparison of H α and Ca H spicules that addresses this question in the literature.McIntosh et al. (2008) and Pereira et al. (2013) show imagesof off-limb spicules in both Ca II and H α at the same timewhich show that the appearance is similar in the two spectrallines. We therefore assume the diagnostics are comparable forthe task of classifying spicules. De Pontieu et al. (2007a) andPereira et al. (2012) indicate that type I spicules are almost ex-clusively present in active regions. The target of our data setis magnetically much less active compared to active regionsand we observe very little downward motion of the spicules.We therefore speculate that we observe type II spicules. A possible explanation for the velocity gradientacross spicules can be inferred from the work ofvan Ballegooijen et al. (2011). These authors study theheating of the corona caused by Alfvenic turbulence. Theydeveloped a numerical model where upward propagatingwaves are created from flux-tube foot-point motions. Due tointeraction between downward propagating reflected waves,turbulence is created in the flux-tube in the photosphereand chromosphere. Our results show that the line-of-sightvelocity across spicules can be complex, not necessarilya constant gradient across the spicule. It is tempting tospeculate whether we observe the Alfvenic turbulence pro-posed by van Ballegooijen et al. (2011) but the speculation islimited by our spatial resolution ( ∼ km) compared to theless-than 100 km spatial scales in the model. The complexitywe sometimes see in our λ − x slices can be an indication ofhigher rotational modes than the fundamental mode whichmay provide support for this theory.Shelyag et al. (2011) use numerical simulations to showthat vorticity in the photosphere can be generated as a resultof interaction between granular flows and the magnetic fieldin the intergranular lanes. The simulation does not include thechromosphere and it is therefore difficult to say if the vorticitywe observe is connected to the process they study.What causes the multi-threaded nature of spicules? A re-cent paper by Antolin et al. (2014) shows that the observedfine strand-like structure of coronal loops can be generated bya combination of the line-of-sight angle and vortices gener-ated by the Kelvin-Helmholtz instability. Transverse MHDwaves generate the instability which creates the vortices. Theauthors show intensity images in the hot Fe IX 171 ˚A pass-band, and while not directly comparable to H α the similar-ities are striking. Further modeling is required to establishwhether the mechanism can explain the multi-threaded natureof spicules we see in H α .The research leading to these results has received fund-ing from both the Research Council of Norway and the Eu-ropean Research Council under the European Union’s Sev-enth Framework Programme (FP7/2007-2013) / ERC grantagreement nr. 291058. The Swedish 1-m Solar Telescope isoperated on the island of La Palma by the Institute for So-lar Physics of Stockholm University in the Spanish Obser-vatorio del Roque de los Muchachos of the Instituto de As-trof´ısica de Canarias. B.D.P. is supported by NASA contractNNG09FA40C (IRIS), and NASA grants NNX11AN98G andNNM12AB40P..The research leading to these results has received fund-ing from both the Research Council of Norway and the Eu-ropean Research Council under the European Union’s Sev-enth Framework Programme (FP7/2007-2013) / ERC grantagreement nr. 291058. The Swedish 1-m Solar Telescope isoperated on the island of La Palma by the Institute for So-lar Physics of Stockholm University in the Spanish Obser-vatorio del Roque de los Muchachos of the Instituto de As-trof´ısica de Canarias. B.D.P. is supported by NASA contractNNG09FA40C (IRIS), and NASA grants NNX11AN98G andNNM12AB40P.