Small-Scale Upflows in a Coronal Hole – Tracked from the Photosphere to the Corona

Coronal transients are known as sources of coronal upflows. With the commissioning of Solar Orbiter, it became apparent that coronal small-scale features are even more frequent than previously estimated. It was found that even small coronal features seen by Solar Orbiter can produce visible upflows. Therefore, it is important to study the plasma flows on small scales better and understand their atmospheric driving mechanisms. In this article, we present the results from a two-week coordinated multi-spacecraft observation campaign with Hinode, IRIS, and the GREGOR telescope. We identify a small region of coronal upflows with Doppler velocities of up to 16.5 km s−1. The upflows are located north of a coronal bright point in a coronal hole. We study the corona, the transition region, the chromosphere and the photospheric magnetic field to find evidence of underlying mechanisms for the coronal upflow. We find a complex photospheric magnetic field with several small mixed polarities that are the footpoints of different loops. Flux emergence and cancellation are observed at the constantly changing footpoints of the coronal loops. Reconnection of loops can be identified as the driver of the coronal upflow. Furthermore, the impact of the coronal activity triggers plasma flows in the underlying layers. This work highlights that frequent small coronal features can cause considerable atmospheric response and ubiquitously produce plasma upflows that potentially feed into the solar wind.


Introduction
The solar atmosphere is the outer region of the Sun consisting of several layers: the photosphere, chromosphere, transition region, and corona.Each layer is characterised by different temperature-density regimes with different physical properties and phenomena.The photosphere has a temperature that decreases to about 4000 K and shows clear evidence of the convection that takes place below the surface.The next higher layer, the chromosphere, starts at what is called the temperature minimum at around 3800 K (Avrett, 2003) and increases from there again with height; it is followed by the thin transition region in which the temperature increases drastically up to several million K in the corona.The status of the transition region (e.g.Peter, 2001) as an individual narrow layer, however, is still a topic of Extended author information available on the last page of the article debate as the lower transition region is usually considered to be dominated by small-scale loops (Hansteen et al., 2014) and its upper part closer connected to the corona via conduction through coronal funnels (Gabriel, 1976).The solar corona itself has been an intense topic of research for several decades.Nonetheless, the cause of the rapid increase in temperature from thousands to millions of degrees is still not confirmed.This is known as the coronalheating problem.It was first noted in the 1940s and multiple discoveries have been made since then.Various heating mechanisms have been proposed since then and as of today, they might act simultaneously.The most favoured ideas are nanoflares (Parker, 1988) and Alfvén waves (e.g.Alfvén and Lindblad, 1947;Osterbrock, 1961).Nanoflares are produced by magnetic reconnection, which transfers energy from the magnetic field into the plasma as kinetic energy, which can be translated into heat.Alfvén waves, on the other hand, are oscillations in the plasma along the magnetic field.They originate in the photosphere and travel through the atmosphere into the corona where they release the stored energy.
In contrast to the lower layers, the solar corona is characterised by larger structures that can broadly be separated into three types: 1. coronal holes (CHs) are regions of lower emission in X-rays and extreme-ultraviolet (EUV), 2. active regions (ARs) are bright structures composed of complex loop systems, 3. the quiet Sun (QS) refers to regions that do not contain CHs or ARs.These three distinct types of regions are only directly visible in the solar corona; however, differences can be backtraced to lower regions.CHs and the QS are not easily distinguishable in lower temperatures, which is probably due to the comparable density of short, low, cool loops in both regions (Wiegelmann and Solanki, 2004).However, the emission from parts of the chromospheric Mg II line is lower in CHs than in the QS when comparing regions with similar magnetic-field strength (Kayshap et al., 2018).The properties of the chromospheric C II resonance line as a function of the magnetic flux |B| shows lower intensities for CHs than for the QS and larger redshifts in the QS for the smaller |B|, in contrast to larger redshifts for large |B| in CHs (Upendran and Tripathi, 2021).Upendran and Tripathi (2022) have shown that a correlation exists between the photospheric magnetic field |B| and flows of the Mg II emission line, which are stronger in CHs than in the QS.The correlation of the flux |B| and the transition-region Si IV emission line shows excess upflows in CHs and downflows in the QS.Transition-region upflows show a correlation with both up-and downflows in the chromosphere.Nonetheless, transition-region downflows only correlate with chromospheric downflows.
The upflow in the solar atmosphere (e.g.Tian et al., 2021) is of interest since it may feed into the solar wind.The solar wind is a structured, continuous flow of charged particles (Parker, 1958).It consists of two regimes with speeds of ∼300 -500 km s −1 and ∼700 -800 km s −1 at a distance of 1 AU, which are known as the slow and fast solar wind.While the fast solar wind is known to originate from coronal holes, the slow solar wind during solar minimum is associated with the streamer-belt region (e.g.McComas et al., 2008).The detailed mechanisms creating the slow solar-wind outflows are not fully understood yet (e.g.Abbo et al., 2016).Analyses of coronal flows show upflows from inside of coronal holes and their boundaries, and at active-region edges (Kojima et al., 1999;Sakao et al., 2007;Harra et al., 2008).Hara et al. (2008) showed that outflows at the edges of an AR and a coronal hole are persistent and can have an additional velocity component of up to 100 km s −1 .Barczynski et al. (2021) recently identified three parallel mechanisms for upflows associated with AR borders: 1) reconnection of closed loops and open field lines in the lower corona or upper chromosphere, which implies the plasma upflow in the corona and downflow in the upper chromosphere; 2) reconnection of chromospheric small-scale loops and open magnetic fields that causes plasma upflow in the transition region and downflow in the lower chromosphere; and 3) expansion of magnetic-field lines allowing chromospheric plasma to go through the transition region into the corona due to wave propagation in the lower atmosphere.The solar wind has a high internal complexity at a distance of 36 -54 solar radii, as shown by Bale et al. (2019) with data from the Parker Solar Probe mission.This complexity cannot be achieved by a large and constant source and requires a large number of small and omnipresent sources of outflowing plasma.
Several coronal features have already been identified as sources of upflow plasma.Wellstudied transients with upflows are, for example, coronal jets (e.g.Raouafi et al., 2016) and bright points (e.g.Madjarska, 2019).Jets can come in different shapes and sizes and are known as sources of upflow.Recent observations of small-scale jets with the Solar Orbiter HRI EUV have shown speeds of 60 ± 8 km s −1 and the corresponding models have predicted Doppler velocities of up to −50 km s −1 (Panesar et al., 2023).A coronal bright point is formed out of small-scale coronal loops connecting opposite magnetic polarities.It can experience heating visible as coronal emission due to the emergence of photospheric flux.However, the chromospheric response can be delayed in time, which suggests that the heating takes place in the corona (Madjarska et al., 2021).Coronal loops in general are important as sources of upflows and EUV emission.The source of EUV emission in a coronal loop does not necessarily have to be coronal but can also be caused by chromospheric flows (Petralia et al., 2014).Furthermore, models have shown that kink oscillations in loops can heat their coronal counterparts and cause chromospheric evaporation (Guo et al., 2023).On the other hand, chromospheric plasma can be heated and transported into the corona via type-II spicules (De Pontieu et al., 2011).Similar to type-II spicules are network jets, which are likely their on-disk counterparts and might play an important link from the transition region to the corona (Tian et al., 2014).Small-scale jetting activity in coronal plumes is known as "jetlets" (Raouafi and Stenborg, 2014) and might play a crucial role in impacting the conditions of the solar atmosphere (Raouafi et al., 2023).Schwanitz et al. (2021) have demonstrated several cases of small-scale coronal features that do not have a strong intensity increase or obvious jet behaviour, but cause upflows with speeds of more than 6 km s −1 .A sub-sample of these features has shown X-ray characteristics of coronal jets, which were previously just regarded as X-ray brightenings (Sterling et al., 2022).In addition to their high appearance rate, all of these coronal transients share their relatively small sizes and short lifetimes in comparison to larger coronal features such as CHs and ARs.However, it is challenging to determine whether these upflows become outflows into the solar wind.
With the launch and commissioning of Solar Orbiter (Müller et al., 2020), new coronal features have been revealed and insights were brought into the mechanisms of the solar corona.Many of them are small-scale features that were not seen well enough with previous instruments and were revealed with the EUV Imager (EUI: Rochus et al., 2020).The first features to mention are small-scale EUV brightenings, first seen by Berghmans et al. (2021).They have sizes of 1000 -5000 km, lifetimes of 10 -200 s (Berghmans et al., 2021) and can reach heights above the photosphere of 1000 to 5000 km (Zhukov et al., 2021).They are most frequently formed over opposite magnetic fluxes and are often observed accompanied by flux cancellation (Panesar et al., 2021;Kahil et al., 2022).The additional value of highresolution data from the High Resolution Imager at 174 Å(HRI EUV ) for the study of coronal upflows was shown by Schwanitz et al. (2023), where two upflows could be associated with a small EUV brightening and loop brightenings.This demonstrates that features on various size scales have to be considered as potential sources of upflowing plasma.
These findings on small-scale EUV features and upflows show that their occurrence rate and potential impact on the corona and heliosphere might have been underestimated in previous studies.Previous projects on small-scale coronal upflow mostly focused on the coronal activity and the photospheric magnetic field.It became obvious that the coronal features and the corresponding magnetic activity are on the scale of a few arcseconds, which was not analysed by the previous studies.Furthermore, the chromosphere and transition region were not covered sufficiently.To study a coronal upflow, the coronal activity, the potential chromospheric activity, and small-scale photospheric magnetic fields, we designed a new observation campaign that we discuss in this article.Coronal upflows are best measured with the imaging spectrometer Hinode/EIS, which is complemented with imaging data from SDO/AIA.To study the chromospheric activity IRIS is used.SDO/HMI observes the photospheric magnetic field continuously on larger scales, while the GREGOR telescope is used to sequentially study the photospheric magnetic field on smaller scales.The combination of these five instruments gives good coverage from the photosphere up to the corona and allows us to trace back sources of coronal activity.

Observation Campaign and Data Processing
The dataset used for this article was recorded on 10 May 2021.It was part of a coordinated two-week multi-instrument observation campaign conducted from 10 -14 May 2021 and 17 -21 May 2021.Its aim was to provide as complete as possible imaging and spectroscopic observations of the atmosphere to find the lower-atmospheric traces of small coronal features.It used two space-based satellites: Hinode and IRIS; and one ground-based instrument: GREGOR.Additionally, data from the SDO satellite are available.We will describe the instruments and their usage in the following subsections.

Solar Dynamics Observatory (SDO): AIA and HMI
The Solar Dynamics Observatory (SDO: Pesnell, Thompson, and Chamberlin, 2012) is a space-based NASA mission that was launched in 2010 to study the Sun.It is equipped with three different instruments: the Extreme Ultraviolet Variability Experiment (EVE: Woods et al., 2012), the Helioseismic and Magnetic Imager (HMI: Scherrer et al., 2012) and the Atmospheric Imaging Assembly (AIA: Lemen et al., 2012).We have worked with data from the latter two instruments.
We make use of the HMI line-of-sight magnetograms, which display the full-disk photospheric magnetic flux at a frequency of one image per 45 s.This is done by measuring the Fe I line at 6173 Å.The pixel scale of the instrument is 0.5 × 0.5 and the noise level is considered to be 7 G.
AIA is a full-disk imaging instrument with seven EUV wavebands covering temperatures from 20 000 to 20 000 000 K. It has a pixel size of 0.6 × 0.6 and cadence of 12 s.We have used pre-processed images from the Joint Science Operations Center (JSOC), which are rescaled level-1.5 data.
Due to the continuous full-disk observations of SDO instruments that cover several atmospheric temperatures, SDO serves well as a reference frame.Therefore, we have aligned all other data to SDO data.

GREGOR Telescope
The GREGOR telescope (Schmidt et al., 2012) is a solar telescope located at the Teide Observatory on the Izaña mountain in Tenerife.It had its first light in 2009 and was officially commissioned in 2012.It is a Gregorian telescope with a primary mirror of 1.5 m diameter.It is capable of observing visible and infrared wavelengths that are emitted in the photosphere and chromosphere.
For this article, we use the GREGOR Infrared Spectrograph (GRIS: Collados et al., 2012), which is a grating spectrograph.It has a slit length of about 60 and a sampling of 0.13 .The observation campaign ran for two weeks 10 -14 May 2021 and 17 -21 May 2021.However, due to local clouds and bushfires, observations were only possible on the 10th, 11th and 20th of May.All observations were taken in the morning due to the solar altitude and local air temperature.There are seven rasters on 10 May 2021 covering the interval 07:47 -09:12 UT.GRIS was operated with 100 raster steps, which leads to a field-of-view of 13 × 60 .The data are processed from raw data to final data products by the GRIS reduction code V8.1 This allows us to directly use the Stokes parameters I, Q, U, and V to characterise the electromagnetic radiation.We focus our analysis on the Si I absorption line at 10827 Å, which is a photospheric line and therefore allows direct comparisons to SDO/HMI.We use the relative circular polarisation Si I V/I that, in the weak-field approximation limit, scales linearly with the line-of-sight magnetic field (e.g.Stenflo, 2013).
We align the derived LOS magnetic data from GRIS with the LOS magnetic data from HMI.We first use the initial pointing coordinates from the observation to find the approximate position of the GRIS data.In a second step, we reduce the GRIS resolution to HMI resolution to be able to cross-correlate them.A rotation was not necessary.The derived offset is then applied to the GRIS data.

The Interface Region Imaging Spectrograph (IRIS)
The Interface Region Imaging Spectrograph The observation mode during the campaign was "Large coarse 64-step raster 126x120 64 s".The IRIS data in this article were taken on 10 May 2021 between 07:02 UT and 09:57 UT with a total of 16 rasters.Each raster has a field-of-view of 126 × 120 with 64 steps and an exposure time of 8 s per step.This means each raster takes about 10 min.The spatial pixel size along the y-direction is 1/3 and along the x-direction 2 .The focus was on observing the spectra of non-coronal lines.The particular rasters have taken spectra both in near-and far-ultraviolet.The SJIs of the campaign were taken with a cadence of 37 s in all four wavebands.We used the level 2 SJIs to track intensity changes in the photosphere, chromosphere, and transition region.The raster data were used to calculate the Doppler velocities of plasma at different heights.We analysed the Mg II k emission line at 2796 Å for the chromosphere and the Si IV emission line at 1403 Å for the lower transition region.The other lines of this study are C II 1336 Å, Fe XII 1349 Å, O I 1356 Å and Si IV 1394 Å.These lines were not used due to their weak intensities and low signal-to-noise ratio.The Mg II k 2796 Å line is characterised by two peaks (k 2V , k 2R ) and a core between them (k 3 ), which provides different diagnostics at different atmospheric heights (Leenaarts et al., 2013a,b).We extracted the spectral position (λ k 2V , λ k 2R , λ k 3 ) and intensity (I k 2V , I k 2R , I k 3 ) for each feature by using a double-Gaussian approach, where each Gaussian covers a peak and gives its intensity and spectral position.The core features were calculated by extrapolating the data between the two peaks with the combined Gaussians.The minimum of this function gives I k 3 and its location λ k 3 .We have calculated the peak separation velocity c with λ k 0 being the wavelength at rest.This Doppler velocity represents mid-chromospheric heights at around 1200 -1550 km.To calculate upper-chromospheric velocities at around 2200 km, we have used c.The Si IV 1403 Å is an unblended line formed in the transition region at around 65 000 K. It can be fitted by a single Gaussian that was then used to calculate intensities and Doppler velocities.

Hinode EUV Imaging Spectrometer (EIS)
The Hinode spacecraft (Kosugi et al., 2008) is a Japan Aerospace Exploration Agency (JAXA) mission launched in 2006.It carries three different instruments on board: the Solar Optical Telescope (SOT: Tsuneta et al., 2008), the X-ray Telescope (XRT: Golub et al., 2008) and the Extreme-Ultraviolet Imaging Spectrometer (EIS: Culhane et al., 2007).We used only EIS because the SOT observation ran before our feature of interest is seen and XRT did not observe in general.EIS is an EUV spectrometer, which measures in two frequency bands 170 -212 Å and 246 -292 Å that cover emission lines with formation temperatures from 50 000 to 20 000 000 K.
The line with the highest sensitivity in the low-frequency waveband is the Fe XII, which is emitted at coronal temperatures of around 1.5 million K. Its peak emission and especially its spectral shift, which is used to calculate Doppler velocities, have been proven as useful observables for the quiet Sun and coronal features such as jets (e.g.Kamio et al., 2007;Moreno-Insertis, Galsgaard, and Ugarte-Urra, 2008), CBPs (e.g.Tian et al., 2008;Pérez-Suárez et al., 2008) or small-scale upflows (Schwanitz et al., 2021(Schwanitz et al., , 2023)).
The three rasters used in this article were taken on 10 May 2021 with starting times at 05:37 UT, 07:15 UT, and 08:53 UT.They had a field-of-view (FOV) of 120 × 512 , with 60 pointing positions of the slit, which resulted in a pixel size of 2 in solar-X and 1 in solar-Y.The exposure time at each position is one minute.
The data was processed with the Python library EIS Python Analysis code (EISPAC: Weberg et al., 2023).We used a single Gaussian to fit the Fe XII emission line at 195.12 Å.This was done in a spectral window for the centroid from 195.08 Å to 195.16 Å.A double-Gaussian approach, which incorporates the blending of the weaker Fe XII line at 195.18 Å (Young et al., 2009), did not converge for most cases due to the low intensity in the quiet Sun and coronal holes.We extracted the peak heights from these Gaussian fits to obtain the intensity and the spectral shift of the centroids to calculate the plasma Doppler velocities.The theoretical line at rest was used as the reference wavelength to calculate these Doppler velocities.The derived Doppler maps were then corrected for the orbital effects of the spacecraft by assuming a median zero shift per column.
The Hinode/EIS Fe XII 195.18 Å data was aligned with the SDO/AIA 193 Å waveband, which covers the Fe XII line at 195.18 Å as well as the hotter Fe XXIV line, which is emitted at 20 000 000 K. However, we do not expect such high temperatures.To align these two datasets, we lowered the spatial resolution of the SDO/AIA data to match that of Hinode/EIS data.We then used a cross-calibration technique to calculate the spatial shift between the two data sets and applied the derived shift to the Hinode/EIS data.

Results
We start our analysis by searching a small-scale coronal upflow region in the Fe XII emission line.We searched for features that are similar to those found by Schwanitz et al. (2021), have a diameter on the order of 1000 km, and show coronal upflows that are stronger than −6 km s −1 .The features should be observed by as many different instruments as possible.We found such an upflow region on 10 May 2021. Figure 1 shows the region of interest highlighted by a red box.It is located at the boundary between the quiet Sun and a southern coronal hole, which is magnetically mostly negative.The smaller blue boundary highlights the region, where we have measured a coronal upflow, it is explained in more detail in Section 3.1.We note that the upflow region is located north of a small bright point and south of a bay in the CH, which extends towards the west and south.The region east of the upflow region is part of the quiet Sun with smaller loop structures.

Coronal Upflow
The coronal upflows in this article are identified in the Fe XII emission line.Figure 2 shows the results from the three rasters taken on 10 May 2021 from 06:29 to 09:51 UT.Each column represents the result from one raster, the top row shows the peak intensity and the bottom row the Doppler velocities of the Gaussian Fe XII emission line.All six panels are, additionally to the spatial coordinates solar-X and solar-Y in arcsec, labelled with the scanning time for the position of the slit on the x-axis.This allows for easier comparisons of Hinode/EIS raster data with the other instruments.Vertical stripes in the rasters are missing data.Furthermore, we want to note the existence of artefacts, which are expected in regions of low signal in particular CHs.The most prominent ones are three horizontal lines of redshifts at solar-X −769 , −789 , and −809 .They can be confirmed as non-physical since they are only present in regions of low intensity and appear in all rasters at the same slit position, which indicates an instrumental effect, probably warm pixels.The columns in Figures 3, 4, 5, and 6 also display the data that is the closest in time to the rasters in Figure 2.
The intensity of the Fe XII does not show strong changes between the three different rasters over the course of roughly 3.5 h in Figure 2. The largest change can be seen in the reduction of the intensity of the bright loop structure at around [130 , −780 ], which is In contrast to that, the Doppler-velocity maps show distinct changes.In general, the flows in the region are characterised by blueshifts in the southern region of the CH and redshifts at the CH boundaries in the north.However, these regions are characterised by a low signal-to-noise ratio and single pixels should not be overinterpreted.The small brightening at [180-195 , −840 ] is redshifted such as the central bright point BP1 where the downflows seem to decrease over time.The region of interest (blue contour) is directly north of the bright point structure and south of a CH bay.In the first raster (left column), this region shows only weak upflows of mostly 1 -4 km s −1 with a maximum of 6.7 km s −1 .In the second raster (central column), we can see strong compact upflows at [160 , −805 ] with a maximum of 16.5 km s −1 , which is on the top-left boundary of the bright point BP1.One raster later (right column), we can see more elongated upflows over the whole top edge of the BP1.Since the upflow in the third raster is slightly more westwards, it is not clear whether it is the same upflow as in the preceding raster or a new one.The strongest measured Doppler velocity for the upflow region in this raster is 15.3 km s −1 .The drawn contours in this and all following plots are a combination of the upflows measured in the second and third Doppler maps.
The identified upflow region is rather compact and clearly separated from surrounding upflows.The fact that the existing upflows appear within 1 h and 40 min and then change rather drastically suggests a coronal transient.However, no clear coronal feature can be seen in the Fe XII emission in the rasters.This motivates a more detailed study of imaging data in the next step.

Corona and Transition Region
To find potential causes of the coronal upflow we make use of the imaging data from SDO/AIA.The changes in the corona in the 193 Å filter are seen in the video AIA_193_nor-mal_50.mp4.We have highlighted two regions of high activity R1 and R2 in Figure 3.The video shows the previously described bright point BP1 at [160-175 , −815 ] with several substructures.First, we note that the upflow region north of BP1 has a lower intensity than its surroundings and it seems to be encircled by the BP1 in the south and a faint loop in the north.The BP1 itself seems to consist of parallel loops with a west-east orientation.Within the upflow region, several smaller features appear: a jet at 06:29 UT in R1, a brightening at 06:50 UT at [175 , −800 ] travelling eastwards ending in R1, and a strong loop brightening at 07:20 UT in R1.A permanent rearrangement of the BP1 loops and a decrease in intensity takes place over the time period 05:00 -11:00 UT.In addition to the small features in the upflow region, several further brightenings occur in the BP1 and its vicinity.In general, BP1 shows a high level of activity and restructuring.The same region in the AIA 171 Å filter is shown in the video AIA_171_normal_50.mp4, which also shows strong activity in   at 06:46 -06:52 UT, 4 min earlier than in the corona.The eastern part of the upflow region is highly active between 06:54 -07:45 UT, which is just a few minutes before the upflow is observed in that region by Hinode/EIS.The western part of the upflow region, which is covered in the third Hinode/EIS raster, is surrounded by increased TR activity around 09:00 -10:45 UT.
To highlight the activity levels, we have calculated standard deviation maps for the SDO/AIA 193 Å, 171 Å and 304 Å wavebands.They are displayed in Figure 3.Each column corresponds to the time of the upflow in Figure 2 ±30 min.The top row shows the 304 Å waveband, where we can see that activity within the upflow region is relatively low except in the small point in the south of R1.This changes towards the next raster where we have the strongest activity right in the eastern part of the upflow region at [162 , −803 ] covering most of R1.This point exceeds all activity levels in the BP1 and is comparable to the activity in the bright loop region at [130 , −775 ].In the last raster, the activity in 304 Å was mostly around the upflow region.This behaviour is comparable to the activities in 171 Å (central row) and 193 Å (bottom row).The largest difference in these is the prominence of the coronal loops, which are best seen in 193 Å and show the highest activity in the central raster.It can be clearly seen that we have a long loop (labelled with "L1" in Figure 3), which spans from [160 , −806 ] to [180 , −802 ] and a short active loop (labelled with "L2" in Figure 3) from [160 , −800 ] to [168 , −798 ].Both loops are rooted in R1, while L1 ends in R2, L2 ends in between.The footpoint of L2 in R1 is further north.To better understand the regions of high activity and the two loop structures, we analyse those regions in the lower atmosphere in Section 3.3 and their magnetic configurations and footpoints in Section 3.4.

Chromosphere and Transition Region Analysis Using IRIS
The IRIS slit-jaw images (SJIs) are imaging data that help us to track the activity to the lower TR, the chromosphere and the photosphere.Figure 4 shows IRIS SJIs, where each column represents the same time as the Hinode/EIS rasters in Figure 2. The white region on the right shows missing data that is outside of the field-of-view.Pixels hit by cosmic rays appear as white dots.The rows from top to bottom go lower in the atmosphere and to lower temperatures, respectively: 1400 Å (lower transition region at 80 000 K), 1330 Å (upper chromosphere at 30 000 K), 2796 Å (upper chromosphere at 15 000 K) and 2832 Å (upper photosphere at 6000 K).
The 1400 Å filter is dominated by emission from the Si IV line and provides a good continuation from the SDO/AIA 304 Å filter.We first note that we have a region of several intense points in the eastern part of the upflow region at [158 , −803 ] in the first image at 07:06:56 UT.This corresponds to the previously identified R1.A second stronger region with increased intensity is at [178 , −798 ], which is northwest of the upflow region, previously identified as R2.We see that those two regions are present throughout the whole observation time and change in their structure and intensity.However, they are always more prominent than the region of the coronal BP1, which extends to about −820 in solar-Y.A similar structure can be seen in the 1330 Å filter, which is dominated by the C II line.The two bright regions R1 in the east of the upflow and R2 north-west of it are very prominent.However, they are now not only made up of individual bright points but are more extended, which corresponds to the presence of spicules.This makes it less distinctive from neighbouring regions.The third row shows data from the 2796 Å filter, which is dominated by the Mg II k line.We can again identify the two regions R1 at [158 , −803 ] and R2 at [178 , −798 ] in the image at 06:07:00 UT.However, they are less intense in comparison to other regions.In particular, R2 at [178 , −798 ] looks more washed out.The bottom row shows the 2832 Å filter with the Mg II wing showing the photosphere, which shows individual bright pores in the two described regions but that are not different from other ones.
The IRIS SJIs show high levels of activity in the TR and upper chromosphere and no significant activity below.This could be explained by an energy release and material injection in that region, which could be triggered locally or in the corona.The smaller activity in lower regions could then be a secondary effect of its downfall.Smaller brightenings could be the result of interacting chromospheric loops.

Behaviour of the Magnetic Field in the Upflow Region
To better understand the chromospheric and coronal activity, it is crucial to analyse the photospheric magnetic configuration.The advantage of this study is the availability of photospheric magnetic-field data from GREGOR. Figure 5 shows the LOS magnetic field recorded by SDO/HMI and GREGOR/GRIS.The background of each figure shows the SDO/HMI LOS magnetograms, seven of which have vertical stripes that show the LOS magnetic field derived from the V/I Stokes parameters for the Si I line, measured by the GREGOR/GRIS instrument.They have higher resolution than SDO/HMI and allow a better understanding of the small-scale structures.The top-left and bottom-right panels represent the times of the first and third Hinode/EIS rasters, and the seven other rasters correspond to the second Hinode/EIS raster.The coronal upflow region north of BP1 is highlighted by the same blue contour as before.
We note two prominent polarities in each image: a positive one (red) in the east part of the upflow region and a negative (blue) north-west of the upflow.These correspond to the regions R1 and R2, which were previously seen as regions of high activity in the chromosphere and corona.They, furthermore, form the footpoints of the coronal loops that we see in SDO/AIA 193 Å.Additional negative polarities are seen in all frames below the upflow region, where the coronal bright point is located at solar-X = 815-818 .We first start our analysis with the two pure HMI images (top-left and bottom-right panels), which correspond to the first and third rasters.In addition to the two opposite polarities in R1 and R2, we further notice a smaller negative flux next to the positive R1 in the first image at 06:08:08 UT.This flux seems to be gone in the last timeframe at 09:21:35 UT, while the positive flux is split up into two smaller patches.However, when we look at the negative flux in R2 in the last frame, we see that a small positive flux has emerged west of it.Those two examples of opposite polarities already give strong indications for constant flux emergence and flux cancellation at small scales in the two regions of strong activity R1 and R2.
When we now look at magnetic activity on even smaller scales by using GREGOR/GRIS data, we see permanent flux emergence and cancellation.In particular, the five images, those in the central row and the first two in the bottom row, are of high interest since they cover the R1 region.We first see at 08:15:35 UT how the positive flux is surrounded by at least five patches of smaller opposite flux.Ten minutes later we only note two distinct negative patches, which means that the other ones have either been cancelled by the stronger positive flux or merged.When we now go 10 min further to 08:35:19 UT, the negative fluxes in R1 are almost gone and the positive flux has become stronger.Only 11 min later, at 08:46:51, we see two new negative flux regions, which last very briefly since they are already gone in the next raster at 09:02:52 UT.This permanent activity of flux cancellation and the emergence of new small polarities is not visible in SDO/HMI data.However, its detection is crucial to understand the footpoints of the coronal loops and the high chromospheric activity.

Flows
After obtaining a complete picture of the activity at different atmospheric heights, the next step is to analyse the related flows in different layers and at different temperatures.Figure 6 shows column-wise the Doppler-velocity maps at three times.The first three rows are rasters taken by IRIS and the fourth row the initial three Hinode/EIS rasters.The IRIS rasters are the closest in time to the respective Hinode/EIS raster.
The top row shows the Mg II k2 Doppler velocities that are measured in the midchromosphere.We see that the two regions R1 and R2, the magnetic footpoints of the coronal loop L1, both show compact upflows in all three rasters.A small downflow region can be seen in the north of R1, which corresponds to the footpoint of L2.Both regions are mostly surrounded by downflows.The region of the coronal upflow between R1 and R2 shows downflows as well.
The next higher emission is the Mg II k3, which is emitted in the upper chromosphere.It is displayed in the second row.The situation here is less distinct with respect to the footpoints and upflow regions.While the footpoint R1 first shows mostly upflows, it also has downflows in the second and third rasters.The other footpoint R2 first has both up-and downflows in the first, only upflows in the second and downflows in the third raster.The region between the two footpoint regions predominantly consists of upflows.
An alternating composition can also be seen in the Si IV line, which is displayed in the third row and represents TR temperatures.We note the existence of three artefacts that appear as horizontal lines of either pure up-or downflows at solar-Y −788 , −816 and −843 in all three rasters.The first raster shows both up-and downflows in the R1 footpoint, their location coincides with the location of the two opposite polarities described in Section 3.4 at that point.The second footpoint R2 clearly consists of pure downflows.This changes in the second raster, where R1 now has only downflows and R2 only upflows and changes again in the third raster where R1 shows upflows and R2 mostly downflows.
The bottom row displays the flows in the Fe XII coronal emission line as in Figure 2. We note that R1, R2 and the interspace always show upflows, but they are only stronger than −6 km s −1 in the second raster above R1 and between R1 and R2 in the third raster.

Conclusion
The datasets used are complex and probe different parts of the solar atmosphere.In order to summarise the results found, Table 1 shows the times of the flows measured by Hinode/EIS and brightenings in the chromosphere, transition region, and corona seen by SDO/AIA.As can be seen, we first observe activity in R2, which is followed by flux emergence and increased activity in R1 before the coronal upflow is measured there.Later, the activity is again evident in R2, which is towards the region of the second coronal upflow.
We have summarised our findings and their interpretation in Figure 7.The initial configuration starts with a longer loop that connects a positive polarity in R1 and a negative one in R2.This assembles the two permanent fluxes and the loop L1, which are present throughout the observation.Another negative polarity is present in the region R1 and forms two shorter loops that are not visible in imaging data.A third small loop is present with footpoints between the two regions.We can see the footpoints of L2 for about one hour.This time period shows high activity in the chromosphere and corona with several brightenings and at least one jet.Two regions of reconnection are highlighted by green crosses.This activity is due to two interacting loops and the large number of opposite polarities in R1 forming smaller loops.Outside of the investigated regions, coronal holes of negative polarity are present.They have a coronal upflow and a chromospheric downflow.This is the situation displayed in the top part of Figure 7.The next step is displayed in the central frame.After several loop brightenings along the loops, the longer loop L1 has reconnected and realigned to a farther positive polarity.The newly visible loop L2 has formed from the reconnection of smaller loops.This causes the coronal upflow, which we measure over R2 and explains why we see upflows only in the coronal part of the loop and downflows in the transition region and chromosphere.We note that the cartoon only displays loops that are clearly visible or those for which we have clear evidence because of the photospheric polarities that are present.We assume that other faint loops have existed before or after and some might have submerged.The eastern footpoint of L1 could be reconfigured and change place due to submergence or flux cancellation, which we strongly see over R1 and R2 in the photospheric magnetic field and as activity in the lower chromosphere.The lower panel of Figure 7 shows the final situation in which we see later the second upflow, which is located between the footpoints R1 and R2 and more towards R1.At this time we note stronger chromospheric and coronal activity around R2 and reduced activity around R1.This might be because the two smaller loops in the previous situation, which are closer to R1, submerge or are cancelled out later during the increase of the strong positive polarity that cancels the smaller negative polarities.We, therefore, see the corresponding coronal upflow later.Afterwards, both regions R1 and R2 show less activity.

Discussion
In this article, we have presented a coronal upflow with a maximum Doppler velocity of 16.5 km s −1 , which is caused by the restructuring of the magnetic flux and reconnection of small coronal loops.They are located at the edge of a coronal bright point in a coronal hole.The longer loop L1 connects two major opposite polarities and the second shorter-lived one is between smaller polarities of which one is close to the footpoint of L1.The permanent changes at their footpoints due to flux emergence cause instabilities and reconnection of the smaller loops, which is accompanied by several coronal brightenings.
The connecting loop between the two coronal holes with the same polarity and upflows show similarities to the structure of a pseudo-streamer (Wang, Sheeley, and Rich, 2007) in which a double arcade is surrounded by coronal holes of the same sign polarity.However, the much larger pseudo-streamers also show higher flow speeds of the order of hundreds of km s −1 .The total configuration might be similar though, however, on much smaller scales.

Figure 1
Figure 1 The left panel shows the region of interest in SDO/AIA 193 Å.The right panel shows the region of interest in SDO/HMI LOS magnetic field.The blue contour highlights the Hinode/EIS upflow and the red rectangle corresponds to the field-of-view used for other plots.

Figure 2
Figure 2 Top row: Three Hinode/EIS rasters showing the intensity of the Fe XII emission line.Bottom row: Same rasters, but showing the Doppler velocities.The x-axis additionally displays the raster time of the respective slit position.The colour bar shows the velocity range, where blue is blueshifted and red is redshifted.

Figure 3
Figure 3 SDO/AIA standard deviation maps, each column represents one time, which is the closest to the Hinode/EIS rasters in Figure 1.

Figure 4
Figure 4 IRIS slit-jaw images, each column represents the time, which is the closest to the Hinode/EIS rasters in Figure 1.

Figure 5
Figure 5 Overlay of SDO/HMI LOS magnetograms (fainter, larger FOV) and GREGOR Si I V/I data (more intense columns).

Figure 6
Figure 6 IRIS raster data showing upflows and downflows at different heights.Each column represents the time, which is the closest to the Hinode/EIS rasters in Figure 1.

Figure 7
Figure 7 Three cartoons display the photospheric field and magnetic loops.The shown polarities (red-positive, blue-negative) are based on the ones seen in HMI and GREGOR data.The top panel roughly displays the situation at 06:00 -07:30 UT, the central panel at 07:30 -08:30 UT, and the bottom panel at 08:30 -10:00.The open negative CH field lines are on the sides of the loop structures.Two green crosses in the top panel show reconnection between loops.

Table 1
Temporal evolution of relevant features in the region of interest.