On the relation between transition region network jets and coronal plumes

TL;DR

This study investigates the relationship between coronal plumes and network jets rooted in solar network lanes, revealing that regions with stronger magnetic convergence produce more dynamic jets and visible plumes.

Contribution

It introduces an automated method to track network jets and links magnetic convergence strength to jet dynamics and plume formation.

Findings

Coronal plumes are visible only in regions with stronger magnetic convergence.

Network jets are higher and faster in regions with visible plumes.

Stronger magnetic convergence correlates with more dynamic jets and plume formation.

Abstract

Both coronal plumes and network jets are rooted in network lanes. The relationship between the two, however, has yet to be addressed. For this purpose, we perform an observational analysis using images acquired with the Atmospheric Imaging Assembly (AIA) 171{\AA} passband to follow the evolution of coronal plumes, the observations taken by the Interface Region Imaging Spectrograph (IRIS) slit-jaw 1330{\AA} to study the network jets, and the line-of-sight magnetograms taken by the Helioseismic and Magnetic Imager (HMI) to overview the the photospheric magnetic features in the regions. Four regions in the network lanes are identified, and labeled ``R1--R4''. We find that coronal plumes are clearly seen only in ``R1''&''R2'' but not in ``R3''&``R4'', even though network jets abound in all these regions. Furthermore, while magnetic features in all these regions are dominated by positive polarity, they are more compact (suggesting stronger convergence) in ``R1''&``R2'' than that in ``R3''&``R4''. We develop an automated method to identify and track the network jets in the regions. We find that the network jets rooted in ``R1''&``R2'' are higher and faster than that in ``R3''&``R4'',indicating that network regions producing stronger coronal plumes also tend to produce more dynamic network jets. We suggest that the stronger convergence in ``R1''&``R2'' might provide a condition for faster shocks and/or more small-scale magnetic reconnection events that power more dynamic network jets and coronal plumes.

Full text

\setlastpage\inarticletrue{opening}

1 Introduction

\ilabel

sec:intro

Coronal plumes (or rays) are bright ray-like features in the corona that could extend to tens of solar radii (e.g. Deforest et al., 1997; DeForest, Plunkett, and Andrews, 2001). They were first found in polar coronal holes during the eclipse more than 100 years ago (see descriptions of the observations in van de Hulst, 1950). Coronal plumes have also identified in UV, EUV and X-ray while such observations are available in the space era. With Skylab observations at Mg ix, Wang and Sheeley (1995) first identified plume-like features in the low-latitude coronal holes, and they proposed that coronal plumes are not unique in the polar region but may occur in open field region at any latitudes on the Sun. This has been confirmed by observations from SOHO, TRACE, Hinode, STEREO and SDO (e.g., Del Zanna and Bromage, 1999; Del Zanna, Bromage, and Mason, 2003; Wang and Muglach, 2008; Feng et al., 2009; Tian et al., 2011a; Yang et al., 2011; DeForest et al., 2018). The electron density of coronal plumes is 3–8 times larger than that in the inter-plume region (Wilhelm et al., 2011, and references therein), and their lifetimes range from $\sim$20 hours (Lamy et al., 1997) to days (Withbroe, Feldman, and Ahluwalia, 1991; Young, Klimchuk, and Mason, 1999; DeForest, Lamy, and Llebaria, 2001). Since coronal plumes are always present in coronal holes where are the major source of fast solar wind (e.g. Xia, Marsch, and Curdt, 2003; Xia, Marsch, and Wilhelm, 2004; Cranmer, 2009; Madjarska et al., 2012; Fu et al., 2015), the connection between coronal plumes and the solar wind has been studied intensively. Some have indicated that they might feed sufficient plasma and energy into the fast solar wind (e.g. Velli, Habbal, and Esser, 1994; Gabriel, Bely-Dubau, and Lemaire, 2003; Gabriel et al., 2005; Tian et al., 2010, 2011a; Fu et al., 2014; Liu et al., 2015), although they might be not the major source (e.g. Wang, 1994; Habbal et al., 1995; Hassler et al., 1999; Patsourakos and Vial, 2000; Giordano et al., 2000; Wilhelm et al., 2000; Teriaca et al., 2003). More details about the history of the observations of coronal plumes could be found in recent reviews (Wilhelm et al., 2011; Poletto, 2015).

Coronal plumes are classified into two types, beam plumes and network plumes (Gabriel et al., 2009). Beam plumes have (quasi-)cylindrical shape with a base diameter of $\sim$30 Mm (e.g. Deforest et al., 1997), and their footpoints normally correspond to coronal bright points (e.g. Wang, 1998; Wang and Muglach, 2008; Pucci et al., 2014). Network plumes (also named “curtain plumes”) are made up of many faint structures that are aligned in a curtain shape (Gabriel, Bely-Dubau, and Lemaire, 2003). Network plumes are rooted in regions along the edge of supergranular boundaries that are corresponding to network lanes in the chromosphere (Gabriel et al., 2009; Rincon and Rieutord, 2018).

No matter which type coronal plumes belong to they are associated with bright network lanes, where the associated magnetic features are dominated by unipolarity (e.g. Newkirk and Harvey, 1968; Wang and Sheeley, 1995; Wang et al., 1997; Deforest et al., 1997; Wang and Muglach, 2008; Wang, Warren, and Muglach, 2016; Avallone et al., 2018). Recently, by investigating the evolution of the magnetic features associated with tens of coronal plumes as observed by HMI, Wang, Warren, and Muglach (2016) found that coronal plumes form where unipolar network elements inside coronal holes converge to form dense clumps and fade as the clumps disperse again. Similar results have been confirmed by Avallone et al. (2018), who also make a quantitative analysis and found that coronal plume appear when convergence of the associated magnetic flux surpasses a flux density of $\sim$200–600 Mx cm*-2*. Along with the dominant polarity, small bipolar features also frequently appear nearby and might be cancelling with the dominant unipolarity (e.g., Wang and Sheeley, 1995; Wang et al., 1997; Wang and Muglach, 2008). Based on the behaviours of magnetic features in the footpoints of the coronal plumes, Wang and Sheeley (1995) proposed a magnetic reconnection scenario, in which reconnection occurs between open fields (corresponding to the dominant unipolarity) and nearby small-scale loops (corresponding to the small bipolar features). Energy dissipation due to such magnetic reconnection process should take place in the base of corona (Wang, 1994). The released energy in the magnetic reconnection might conduct downward to the chromosphere and results in plasma evaporation that maintains the high density in the coronal plumes (Wang, 1998).

Many studies have focused on the connection between coronal plumes and other dynamic phenomena in the solar atmosphere. For coronal plumes with coronal bright points persisting at their bases, studies found that they appear several hours after the coronal bright points first appeared and fade away hours after the coronal bright points disappeared (Wang, 1998; Wang and Muglach, 2008; Pucci et al., 2014). While coronal jets are also closely linked to bipolar regions with one dominant polarity (e.g. Doschek et al., 2010; Huang et al., 2012; Sterling et al., 2015; Panesar et al., 2016; Panesar, Sterling, and Moore, 2018; Panesar et al., 2018), the connection between coronal jets and coronal plumes has also been investigated. Using white light observations during the eclipse data and EIT 195 Å data from SOHO, Lites et al. (1999) reported that the polar coronal plume was disturbed by the jet with a speed of $\sim$200 km s*-1* embedded therein. Raouafi et al. (2008) showed that over 90% of the 28 jets are associated with plumes, and they also found that 70% of those plume-related jets are followed by plume haze occurring minutes to hours later while the rest of the jets result in brightness enhancement in the pre-existed plumes. In more recent studies, Raouafi and Stenborg (2014) and Panesar et al. (2018) confirmed the close relationship between coronal plumes and coronal jets, and they further found that a large number of small jets and transient bright points frequently occurring in the bases of coronal plumes could be the main energy source for coronal plumes.

Recently, using the high resolution observations of the transition region from the Interface Region Imaging Spectrograph (IRIS), Tian et al. (2014) discovered that networks in the upper chromosphere and transition region of coronal holes are occupied by numerous small-scale jets (hereafter, network jets). Network jets having lifetimes of 20–80 s and speeds of 80–250 km s*-1* are prevalent in the network regions. They have been suggested to be powered by magnetic reconnection and many of them were found to be heated to at least 105 K. Network jets have also been found in quiet-sun region, which are slower and shorter than those in the coronal holes (Narang et al., 2016; Kayshap et al., 2018).

Since both coronal plumes and network jets are rooted in the networks and both of them might be powered by magnetic reconnection, One would naturally ask a question what the relationship between the two phenomena are. Targeting on this question, here we show a set of IRIS and SDO coordinated observations of a few network regions, where present different dynamics of coronal plumes and network jets. We develope an automatic method to identify and track jets in the network regions. With this method, we obtain birth rates, lifetimes, lengths and speeds of the network jets. We then compare these parameters depending on different part of the network regions.

The paper is organised as follows. The observations and methodology are described in Section \irefsec:obser. The results are shown in Section \irefsec:results and Section \irefsec:resstat. The discussion is given in Section \irefsect:discussion and the conclusions are given in Section \irefsect:conclusions.

2 Observations and Data Analysis

\ilabel

sec:obser

2.1 Details of Observations\ilabelsubsec:data

The data were taken on 2015 December 4 from 01:21 UT to 02:19 UT. They were collected by IRIS (De Pontieu et al., 2014) and the Atmospheric Imaging Assembly (AIA, Lemen et al., 2012) and the Helioseismic and Magnetic Imager (HMI, Schou et al., 2012) aboard the Solar Dynamics Observatory (SDO, Pesnell, Thompson, and Chamberlin, 2012).

IRIS was operated in a very large sit-and-stare mode and the IRIS slit-jaw (SJ) imager was observing only at 1330 Å passband with a cadence of 9 s. The spatial scale of the IRIS SJ images is 0.167*′′$\times$0.167′′. In order to reduce the telemetry load, the data had been binned by $2$ pixels$\times$2 pixels and thus the spatial scale of each grid of the data array is 0.334′′$\times$0.334′′*. The level 2 IRIS data are used and no further calibration is required.

The AIA and HMI data were downloaded from JSOC. The AIA data include that taken in 1600 Å and 171 Å passbands. The cadences of the AIA 1600 Å and 171 Å data are 24 s and 12 s, respectively. The pixel size of the AIA data is 0.6*′′$\times$0.6′′. The HMI line-of-sight magnetograms with a cadence of 45 s and a pixel size of 0.6′′$\times$0.6′′*are used. The AIA and HMI data are prepared with standard procedures provided by the instrument teams, and the level 1.5 data are analysed.

The images taken from different passbands are aligned using several referent features in the images. We first align IRIS 1330 Å images to AIA 1600 Å, and then HMI magnetograms to AIA 1600 Å. Although images from different passbands of AIA have been aligned each other by the data processing pipeline, we also check the alignment between 1600 Å and 171 Å using referent features and found an offset of 1*′′* at both Solar_X and Solar_Y directions.

In Figure \ireffigfov, we show the context of the region-of-interest taken with AIA, HMI and IRIS around 01:21 UT. The region is at the boundaries of an on-disk coronal hole (see Figure \ireffigfova). We can see that the region consists of a cluster of positive magnetic features (Figure \ireffigfovb), which are aligned in a typical magnetic structure of network regions (e.g. Xia, Marsch, and Curdt, 2003; Xia, Marsch, and Wilhelm, 2004). The dominant polarities are in line with the coronal hole seen in the AIA data. The network regions could be clearly seen in AIA 1600 Å and IRIS SJ 1330 Å images (see the bright lanes in Figures \ireffigfovc&d). Many jets rooted in the network lanes are found, and they can be clearly seen in the IRIS SJ 1330 Å images. In AIA 171 Å images (Figures \ireffigfova&e), we can see that a set of plume-like features are rooted in some of the region.

2.2 Network jet identification and tracking

\ilabel

subsec:aaptinj

Because network jets are extremely abundant in network regions, they demand an automatic method to be identified and tracked. Here we develop an automatic algorithm for network jet identification and tracking. In this method, if a feature is brighter than 2.5 times of background and extended for more than 4*′′* from the base near network lane, it is considered as a jet-like feature. An identified jet-like feature is then traced back and forth in time to obtain its evolution. The height of a jet-like feature is determined by the base and the point at its extending direction where its brightness drops below 2.5 times of the background. If a jet-like feature can be recognised in more than one frames and its heights are changing with time, it is considered as a network jet. By determining the heights of a jet changing with time, its speed could be obtained. More details of the algorithm could be found in the Appendix.

3 Dynamics in the network regions

\ilabel

sec:results

In Figure \ireffigfov and the associated animation, we show the evolution of the region seen in IRIS SJ 1330 Å and AIA 171 Å. In the IRIS SJ 1330 Å images, we can see that jets are abundant in the network lanes. The network jets are fine and dynamic, which are the typical characteristics of these phenomena (Tian et al., 2014). In regions “R1”–“R4”, the network jets ejected in directions almost parallel to each other, which allows our automatic algorithm to be used (see the Appendix for details). In AIA 171 Å images, it is clear that plume-like features are rooted in the regions of “R1” and “R2”, while they are hardly seen (or too weak to be seen) in the regions of “R3” and “R4”. It shows that the network regions of “R1” and “R2” are much brighter than “R3” and “R4”. In particular, the average irradiance of “R1”&“R2” seen in IRIS SJ 1330 Å is about 2–4 times of that in “R3”&“R4”.

As shown in Figure \ireffigfov and the animations, network jets rooted in the studied regions are very dynamic. In AIA 171 Å images, we can also observed many disturbances in the coronal plumes. We speculate that a number of network jets were resulting in disturbances in the coronal plumes. Because of the complex background in both transition region and coronal images, in most of the cases we cannot identify one-to-one relation between transition region activities and the coronal plume disturbances.

The HMI magnetograms show that the regions are dominated by positive polarities (Figure \ireffigfovb). In agreement with Wang, Warren, and Muglach (2016) and Avallone et al. (2018), the regions with clear coronal plumes (“R1”&“R2”) have magnetic features more compact than that in “R3”&“R4”. In most of the regions during the observing period, very few negative polarities were seen in the regions. This indicates that magnetic features with opposite polarity to the major one, if existed, should have sizes/lifetimes under the resolution or strengths under the resolving power (i.e. $\sim$10 Mx cm*-2* in the present case, Liu et al., 2012). To investigate this issue, one will need higher resolution data, which might be provided by the Goode Solar Telescope (GST, Goode et al., 2010; Cao et al., 2010), the forthcoming Daniel K. Inouye Solar Telescope (DKIST) and the planning Chinese Advanced Solar Observatories – Ground-based (ASO-G). Occasionally, we can also observe that small negative polarities appear nearby the dominant positive ones and immediately disappear within around 1 minute (i.e. close to the HMI temporal resolution), thus we cannot link such magnetic activities to any particular network jets.

We examine the magnetic topology of the region using full-disk potential field extrapolation provided by the PFSS package (Schrijver and De Rosa, 2003) in the solarsoft. In Figure \ireffigfield, we display the potential field extrapolation of the region based on observations taken on the days from December 1 to December 4, which are tracking the region from the east hemisphere to the west hemisphere. It is clear that the region of interest is dominated by open field lines, and the open fields could be seen at least three days before our observations. The open topology of magnetic field is in agreement with coronal hole observed in the AIA coronal passbands, and it provides a condition for birth of coronal plumes and network jets. As described above, however, some network lanes in this open field region are rich in coronal plumes but some others not while both are rich in network jets. In the following, we will compare the dynamics of the network jets in “R1”–“R4”, and investigate the possible relation between coronal plumes and network jets.

4 Statistical analysis of the network jets

\ilabel

sec:resstat

In “R1”–“R4”, we identify 1293 network jets, of which 619 located in regions “R1” and “R2” and the rest 674 located in regions “R3” and “R4”. Based on their coronal plume activities, in the statistics we group network jets from “R1” and “R2” as one category, and “R3” and “R4” as another category. The birthrates of network jets in these two types of regions are $7.8\times 10^{-16}$ m*-2* s*-1* (“R1”& “R2”) and $4.7\times 10^{-16}$ m*-2* s*-1* (“R3”& “R4”), respectively. Please note that these birthrates are the bottom limits because we only take into account the network jets that allow speed calculations (see the Appendix for detail). In Figure \ireffigstatres, we give the statistical analysis of three parameters (lifetime, height and speed) of the network jets in the two categories of regions.

In the regions rich in coronal plumes (i.e. “R1”&”R2”), the average lifetimes, heights and speeds of all 619 network jets are 45.6 s with a standard deviation ($1\sigma$) of 35 s, 8.1*′′with $1\sigma$ of 1.6′′* and 131 km s*-1* with $1\sigma$ of 64 km s*-1*, respectively. In the regions poor in coronal plumes (i.e. “R3”&”R4”), the average lifetimes, heights and speeds of all 674 network jets are 50.2 s with $1\sigma$ of 35.4 s, 5.5*′′with $1\sigma$ of 1.9′′* and 89 km s*-1* with $1\sigma$ of 45 km s*-1*, respectively. In average, the network jets from regions rich in coronal plumes are higher and faster than that from the regions poor in coronal plumes. However, we also see that the distributions of each parameter from the two kinds of regions are largely overlapped. The obtained values of heights and speeds are spreading in the ranges consistent with the measurements in the previous studies (Tian et al., 2014; Narang et al., 2016; Kayshap et al., 2018), but the mean and most probable values measured here are smaller. This could be resulted from the line-of-sight effect since the region studied here is closer to the disk center. Although the obtained lifetimes are generally in agreement with the previous studies (Tian et al., 2014; Narang et al., 2016; Kayshap et al., 2018), there is a large portion showing lifetime of 18 s (i.e. two frames in the time series). This kind of short-lifetime jets were not included in the studies of Tian et al. (2014) and Narang et al. (2016) due to the temporal resolutions of their data, but they appear to be common (see e.g. De Pontieu, Martínez-Sykora, and Chintzoglou, 2017; Martínez-Sykora et al., 2017).

In the speed regime, the histogram show two peaks at $\sim 50$ km s*-1* and $\sim 110$ km s*-1* with a clear division at $\sim 90$ km s*-1*. Although with less samples, such speed distribution has also been seen in that given in Tian et al. (2014), but with different peak speeds at $\sim 110$ km s*-1* and $\sim 160$ km s*-1* and a division at $\sim 140$ km s*-1*. Such difference between their results and ours can be understood if we take into account the line-of-sight effect. The questions are whether such two-peak distribution is universal and whether this two-peak distribution indicates different species of network jets. These could be an interesting topic for future studies using higher cadence data and a larger number of samples.

The distribution histograms of the lifetimes, heights and speeds of the network jets are overlapped with those of spicules that also originated from network regions (see e.g. Sterling, 2000; Xia et al., 2005; de Pontieu et al., 2007; Zhang et al., 2012; Pereira, Pontieu, and Carlsson, 2012; Pereira et al., 2014). We speculate that a part of the network jets identified here are responses of spicules in the IRIS SJ 1330 Å passband. To compare with the speed histogram of spicules (de Pontieu et al., 2007; Pereira, Pontieu, and Carlsson, 2012), that of the network jets much bias toward more than 100 km s*-1*.

5 Discussions

\ilabel

sect:discussion Since both network jets and coronal network plumes originate in network regions, the connection between them should be answered. In the present study, we made a statistical analysis of a few parameters of network jets in a few network regions (“R1”–“R4”), in which coronal plumes are clearly seen in “R1” and “R2” but almost invisible in “R3” and “R4”. We found that the network jets in “R1” and “R2” are statistically higher and faster than that in “R3” and “R4”. These observational results indicate more dynamic and energetic nature of network jets in the regions where coronal plumes are clearly seen. It suggests that the network regions of “R1” and “R2” could provide a condition in favor of both more energetic network jets and stronger coronal plumes.

While the unipolar features in “R1” and “R2” are more compact (implying a higher degree of convergence), it provides stronger compression in the regions. The compression in the footpoints of the magnetic flux tubes might produce shocks that could feed mass and energy to the higher solar atmosphere. The shocks have to be strong and dense enough to power plasma to produce coronal plumes in the region. In this scenario, one would expect that the network jets are also powered by shocks and therefore many network jets should directly feed mass and energy to coronal plumes. This will require a further study using data targeting on regions with cleaner background (e.g. polar regions), which allows identify one-to-one relation between network jets and coronal plumes. The higher degree of compression might also produce more complicate shearing motions in the footpoints of magnetic flux tubes rooted in the region. The complex shearing motions could generate a complexity of magnetic topology above the photosphere. Such complexity of magnetic topology includes magnetic braids (Parker, 1983b, a). As a result of magnetic braiding, magnetic reconnections occur and could heat the plasma to coronal temperature and accelerate the plasma to more than a hundred kilometer per second (Cirtain et al., 2013; Huang et al., 2018).

If there were mixed polarities in the region, the higher degree of convergence might directly drive more magnetic reconnection to occur and thus feed more mass and energy to higher solar atmosphere (He et al., 2010). In the present cases, although there are opposite polarities found nearby the dominant polarity, they are too rare to agree with the occurrence of the network jets. However, it is possible that many opposite polarities cannot be resolved with the current data, and higher resolution data provided by GST and DKIST might shed light on this problem.

In coronal plumes, propagating disturbances are usually seen (see e.g. DeForest and Gurman, 1998; Marsh et al., 2003; De Pontieu et al., 2007; Wang et al., 2009; De Pontieu and McIntosh, 2010; Tian et al., 2011b, a, 2012, etc.). It has been reported that spicules (De Pontieu et al., 2011; Pant et al., 2015; Jiao et al., 2015; Samanta, Pant, and Banerjee, 2015; Jiao et al., 2016) and/or shocks (Hou et al., 2018) could be the possible sources of the propagating disturbances in the coronal plumes. In the present study, the observations show many propagating disturbances along the coronal plumes. We can only speculate from the animation that some of the network jets should have directly linked to the propagating disturbances in the coronal plumes. However, because of the complex background emission, we were not able to identify one-to-one corresponding between such disturbances and network jets. Whether network jets can directly trigger propagating disturbances in the coronal plumes remains open.

6 Conclusions

\ilabel

sect:conclusions In the present study, we studied activities in four regions of network lanes, in which coronal plumes (viewed in AIA 171 Å passband) are clearly seen rooting in two (“R1”& “R2”) of the regions but are very fade in the other two (“R3”& “R4”). In all these regions, network jets seen in IRIS SJ 1330 Å passband are abundant. Positive polarities are dominant in these regions and negative polarities could only be seen occasionally. The positive polarities are more compact in “R1”& “R2”, suggesting higher degree of convergence. The average irradiance of “R1”& “R2” seen in IRIS SJ 1330 Å is 2–4 times of that in “R3”& “R4”.

We developed an automatic method to identify and trace network jets in these regions. With the method, we identified and traced 619 network jets in “R1”& “R2” and 674 in “R3”& “R4”. With these samples, we carried out statistical analyses of lifetimes, heights and speeds of the network jets in “R1”& “R2” and “R3”& “R4”, respectively. The lifetimes, heights and speeds of the network jets in “R1”& “R2” are 45.6 s with a standard deviation ($1\sigma$) of 35 s, 8.1*′′with $1\sigma$ of 1.6′′* and 131 km s*-1* with $1\sigma$ of 64 km s*-1*, respectively. In “R3”&“R4”, the average lifetimes, heights and speeds of the network jets are 50.2 s with $1\sigma$ of 35.4 s, 5.5*′′with $1\sigma$ of 1.9′′* and 89 km s*-1* with $1\sigma$ of 45 km s*-1*, respectively. These results show that the network jets are in-average higher and faster (i.e. more dynamic) in the regions with visible coronal plumes than that without clear coronal plumes. We suggest that the convergence motions in the base of the network regions could build up energy and the energy could be released to the higher solar atmosphere though shocks and/or small-scale (under the current resolution) magnetic reconnections.

Acknowledgments

Acknowledgments: The research is supported by National Natural Science Foundation of China (U1831112, 41627806,41604147, 41474150, 41404135). Z.H. thanks the Young Scholar Program of Shandong University, Weihai (2017WHWLJH07). We acknowledge Dr. Hui Tian, Dr. Tanmoy Samanta and Prof. Rob Rutten for fruitful discussion. IRIS is a NASA small explorer mission developed and operated by LMSAL with mission operations executed at NASA Ames Research center and major contributions to downlink communications funded by ESA and the Norwegian Space Centre. Courtesy of NASA/SDO, the AIA and HMI teams and JSOC.

7 Appendix: Network jet identification and tracking algorithm

The network jets are identified and traced on the IRIS SJ 1330 Å images by the automatic method. The algorithm of the method includes a few steps as described in the follow.

(a) Region selection and base definition: In order to simplify the tracking procedures, we analyze the regions “R1”–“R4”, where the bright elements in the network lanes (viewed in IRIS SJ 1330 Å) are almost aligned and the network jets are ejected in directions almost parallel to each other. The bases of the network jets in each region are determined to be the edge of the bright network lanes. Because the network lanes are very dynamic with many transient bright dots, we make an artificial image that is sum of all the IRIS SJ 1330Å images taken in the observing period of time, and the edge of the bright network lanes are defined based on the artificial image.

(b) Jet-like feature identification in an image frame: In each region, starting from the bases of the network jets, we select four slices that extend (almost) perpendicular to the network jets and have 1*′′* separation between each two neighbours (see Figure \ireffigmethod1a). The variation of the SJ 1330 Å radiance along each slice is then obtain. The local peaks as defined in Huang et al. (2017) are identified on the variation curves (see Figure \ireffigmethod1b). Such a local peak indicates a local brightening. In order to avoid the noise effect, the local peaks with radiances less than 2.5 times of the background are excluded. If a local peak of a slice and the other local peak of the neighbouring slice appear at the slice positions with difference less than 1*′′*, they are considered to be results of the same bright feature. If a bright feature as defined by local peaks is found in all the four slices, it is considered to be a jet-like feature, and its location and extending direction are determined using the positions of the corresponding local peaks identified in the four slices.

(c) Length determination for the jet-like features: For each identified jet-like feature, a slice along its extending direction is made and the radiance variation along the slice is obtained. Along the slice, the farthest extending point of the jet-like feature is defined to be the location where the radiance drops below 2.5 times of the background (See Figure \ireffigmethod1c).

(d) Jet-like feature tracking and definition of network jet and its lifetime and speed: Steps (b) and (c) is run for each imaging frame. If a jet-like features found in one frame and the other jet-like feature found in the other close-in-time frame locate at the same position, they are considered to be the same jet-like feature that is evolving in time. If a jet-like feature is found in more than one image frame and less than 27 image frames (i.e. lifetime less than 4 minutes), it is considered to be a network jet. The heights of the network jet seen at different times are given by the lengths of the jet-like features. By following the heights of a network jet at different times, its speed could be obtained using a linear fit in the height vs. time space (see Figure \ireffigmethod1d). Please note that for each jet only its maximum height is used in the statistics (i.e. Figure \ireffigstatresb).