The Astrophysical Journal, 975:33 (11pp), 2024 November 1 https://doi.org/10.

3847/1538-4357/ad7586
© 2024. The Author(s). Published by the American Astronomical Society.

On the Nature of Nonthermal Broadening of Spectral Lines Observed by IRIS

Kyuhyoun Cho1,2 , Bart De Pontieu2,3,4 , and Paola Testa5
1
Bay Area Environmental Research Institute, NASA Research Park, Moffett Field, CA 94035, USA
2
Lockheed Martin Solar & Astrophysics Laboratory, 3251 Hanover Street, Palo Alto, CA 94304, USA
3
Rosseland Centre for Solar Physics, University of Oslo, P.O. Box 1029 Blindern, NO-0315 Oslo, Norway
4
Institute of Theoretical Astrophysics, University of Oslo, P.O. Box 1029 Blindern, NO-0315 Oslo, Norway
5
Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02193, USA
Received 2024 May 24; revised 2024 August 26; accepted 2024 August 28; published 2024 October 23

Abstract
The origin of nonthermal broadening in solar spectra is one of the long-standing questions in solar physics. Various
processes have been invoked—including unresolved flows, waves, and turbulent processes—but definitive answers
are lacking. To investigate the physical processes responsible for nonthermal broadening, we examine its relation
with the angle between the magnetic field and the line of sight in three different closed-field regions above plage
regions at different locations on the solar disk. We obtained the nonthermal width of transition-region Si IV 1403 Å
spectra observed in active regions by the Interface Region Imaging Spectrograph, after subtraction of the thermal
and instrumental line broadening. To investigate the dependence of the measured broadening on the viewing angle
between the line of sight and magnetic field direction, we determined the magnetic field direction at transition-
region heights using nonlinear force-free extrapolations based on the observed photospheric vector magnetic field
taken by the Helioseismic and Magnetic Imager on board the Solar Dynamics Observatory. We found that the
nonthermal broadening shows a correlation with downward motion (redshifts) and alignment between the magnetic
field and the observer’s line-of-sight direction. Based on the observed correlations, we suggest that velocity
gradients within plasma flowing down along the magnetic field may lead to a significant portion of the observed
nonthermal broadening of transition-region spectral lines in closed fields above plage regions.
Unified Astronomy Thesaurus concepts: Solar transition region (1532); Solar physics (1476); Solar atmosphere
(1477); Solar ultraviolet emission (1533)

1. Introduction P. Testa et al. 2014, 2016, 2020; Juraj Lorincik et al. 2022;
K. Cho et al. 2023). The relation between the nonthermal
The nonthermal broadening of spectral lines is the amount of
velocity and magnetic field direction can reveal whether the
excess broadening in the observed solar spectra compared to the
nonthermal velocity is mainly associated with plasma motions
value expected based on the thermal properties of the emitting
along the magnetic field or perpendicular to the magnetic field.
ion (thermal broadening) and the instrumental properties
This is crucial for identifying the processes responsible for the
(instrumental broadening). The nonthermal broadening in
nonthermal broadening. In the case of MHD waves, for
transition-region lines is of high interest, because it may hold
example, if a region where the magnetic field is perpendicular
important clues to the heating mechanism that is active in the
to the LOS shows larger nonthermal broadening, this can
solar atmosphere, as it may provide an observational signature of
provide support for an interpretation in which transverse waves,
energy transfer from the lower solar atmosphere. After the early
such as Alfvénic waves, contribute significantly to the broad-
report by B. C. Boland et al. (1973), many previous studies have
ening. Conversely, a positive correlation of increased broad-
commonly found excess broadening of about a few tens of
ening for regions with magnetic fields parallel to the LOS
kilometers per second for transition-region lines. Several
direction can point to longitudinal waves, such as slow MHD
previous analyses of nonthermal broadening are reviewed in
waves, as a more likely process. Similarly, the superposition of
J. T. Mariska (1992) and G. Del Zanna & H. E. Mason (2018).
Nonthermal broadening can be interpreted as a cumulative unresolved field-aligned plasma motions (in a low-plasma-β
effect of multiple velocity components along the line-of-sight environment) also will be correlated with parallel LOS and
(LOS) direction within a single detector pixel. There are several magnetic fields.
For this reason, many previous studies have investigated the
candidates for explaining what kinds of physical processes
center-to-limb variation of the nonthermal broadening in solar
cause the multiple velocity components, e.g., acoustic or
transition-region lines emitted in the low corona. These studies
magnetohydrodynamic (MHD) waves (J. T. Mariska et al.
generally assume that the magnetic field is typically oriented
1978; K. G. McClements et al. 1991; K. P. Dere &
vertical to the solar surface, so the nonthermal broadening can
H. E. Mason 1993; A. A. van Ballegooijen et al. 2011;
be interpreted as motions parallel to the magnetic field when at
B. De Pontieu et al. 2015), unresolved turbulent motions
the solar disk center and perpendicular to the field when at the
(G. A. Doschek & U. Feldman 1977; D. I. Pontin et al. 2020),
limb. Most studies in the literature have reported that the
and magnetic reconnection events (I. M. Sarro et al. 1997;
broadening is not correlated with the distance from the disk
center (R. Roussel-Dupre et al. 1979; J. T. Mariska 1992;
Original content from this work may be used under the terms
of the Creative Commons Attribution 4.0 licence. Any further K. Cho et al. 2023). This finding has been interpreted as
distribution of this work must maintain attribution to the author(s) and the title being due to both transverse and longitudinal waves or field-
of the work, journal citation and DOI. aligned flows equally affecting the nonthermal broadening

1
The Astrophysical Journal, 975:33 (11pp), 2024 November 1 Cho, De Pontieu, & Testa

measuring the transverse component in weak-field regions

(M. G. Bobra et al. 2014). In addition, we need to consider the
difference of the magnetic field inclination at photospheric
heights and at transition-region heights, because the magnetic
fields in the higher atmosphere should be determined by the
configuration and proximity of neighboring magnetic pola-
rities, even if most photospheric magnetic fields depart from the
surface vertically. Therefore, it is worthwhile investigating the
relation between nonthermal broadening and magnetic field
inclination (with respect to the LOS) in strong enough magnetic
field regions at the height of the transition-region emission, i.e.,
a few thousand kilometers above the surface.
In this study, we investigate the relation of the nonthermal
broadening to the magnetic field inclination at transition-region
heights in closed-field regions above plage regions. The plage
regions are easily identified as bright regions in 1600 Å images
observed with the Atmospheric Imaging Assembly (AIA;
J. R. Lemen et al. 2012) on board SDO, which have a relatively
strong magnetic field strength of greater than 50 G. We
obtained Si IV 1403 Å spectra from IRIS (B. De Pontieu
Figure 1. An example of a magnetic field inclination map from HMI
observations. The white gray color (90°) indicates perpendicularity to the LOS et al. 2014) full-disk mosaic data and performed Gaussian
direction. fitting to determine several spectral parameters. A nonlinear
force-free field (NLFFF) extrapolation enables us to derive the
magnetic field vector at transition-region heights. Through the
(S. W. McIntosh et al. 2008; B. De Pontieu et al. 2015). comparison between the inclination angle of the magnetic field
Observations also show enhanced broadening near the solar and LOS, and the spectral parameters in the transition region,
limb (Y. K. Rao et al. 2022; M. Carlsson & B. De Pontieu we investigate the physical mechanisms driving nonthermal
2023), above the solar limb (J. T. Mariska et al. 1979; broadening in transition-region spectral lines.
S. W. McIntosh et al. 2008; M. Carlsson & B. De Pontieu
2023), and in the middle to upper corona (e.g., R. Esser et al. 2. Observation and Analysis
1999; Y. Zhu et al. 2024). These results are used to support the
IRIS takes full-disk mosaic data at a monthly cadence.6
claim that the transverse waves are a major source of the
Here, we use the data taken on 2013 October 21, when several
nonthermal broadening (M. Carlsson & B. De Pontieu 2023).
well-developed active regions were distributed across the solar
At the same time, P. Testa et al. 2016 found that a comparison
disk (see Figure 2), providing different viewing angles. The
of disk and limb Interface Region Imaging Spectrograph (IRIS)
full-disk mosaic consists of 185 different pointings, and the
observations of Fe XII in active region moss (i.e., in the
field of view (FOV) of each pointing is about 128″ × 180″,
transition region of hot loops) suggests the dominance of
with a spatial sampling of 2″ and 0.66″, respectively. The total
field-aligned motions, compatible with heating by magnetic
number of pixels is 10,496 × 548 along the scanning and slit
reconnection. The literature suggests that a mix of both
directions, and the total scanning time is about 14 hr. The IRIS
longitudinal and transverse effects appear likely to play a role.
mosaic data contain six spectral windows (Mg II h, Mg II k,
These effects may have a different relative importance for the
Si IV 1394, 1403 Å C II 1334, and 1335 Å). Among them, we
observed broadening, depending on the type of solar region
studied (quiet Sun or active region). selected the Si IV 1403 Å line to investigate the nonthermal
However, it is not clear that the magnetic field at transition- broadening in the transition region. The exposure time of the
region heights is predominantly vertical. This has been studied Si IV 1403 Å window was 2 s, with a spectral binning of 2.
for some numerical models of the quiet Sun (H. Peter et al. We fitted the Si IV 1403 Å spectra with a single Gaussian
2006), including the impact on the center-to-limb variation of function. To select spectra with sufficient statistics in the Si IV
line broadening, but for active regions, this has not been spectral emission line, to accurately derive its parameters, we
studied in detail. Yet, for such regions, it seems particularly only applied the single Gaussian fitting to the pixels where the
likely that the magnetic field direction is not always vertical at peak of the Si IV was higher than 4 DN s–1. The four parameters
transition-region heights, as it is dominated by the complex derived from the fitting are: Gaussian amplitude, Doppler shift,
magnetic field topology caused by large-scale neighboring Gaussian width (σ), and background level (see Figure 3).
regions (plage and sunspots) of opposite polarity. The Previous studies used a variety of definitions for spectral width,
Helioseismic and Magnetic Imager (HMI; J. Schou et al. including Gaussian width, FWHM, and 1/e width; the relation
2012) on board the Solar Dynamics Observatory (SDO; among these parameters is as follows:
W. D. Pesnell et al. 2012) provides full-disk photospheric
vector magnetograms at a 12 minutes cadence. The full-disk FWHM = 2 2 ln 2 s , w1 e = 2 s. (1 )
map of the inclination of the magnetic field with respect to the
LOS (which we hereafter call the inclination) generated from We chose the 1/e width for the nonthermal width definition, as
HMI data does not show a clear center-to-limb variation (see this width (when expressed as a velocity) can be interpreted as
Figure 1). This may be due to the photospheric magnetic field the most probable velocity in a statistical sense. The
containing considerable horizontal field strength in the quiet
Sun (D. Orozco Suárez 2012) or to the lack of sensitivity for 6
https://iris.lmsal.com/mosaic_allin1.html

2
The Astrophysical Journal, 975:33 (11pp), 2024 November 1 Cho, De Pontieu, & Testa

Figure 2. (a) Si IV nonthermal velocity map from IRIS mosaic observations. (b) LOS magnetic field mosaic map from HMI observations. The blue numbers indicate
the locations of the numbered (NOAA) active regions. (c) AIA 1600 Å intensity mosaic map from AIA observations. (d) Observation time for the IRIS mosaic data.
The red boxes indicate the selected rectangular areas or regions of interest. The white color indicates regions that lack data (e.g., because of data dropouts or because of
gaps between different mosaic pointings). Note that the selected regions span across different pointings, and especially the time difference between two successive
horizontal sequences is usually greater than 1 hr.

nonthermal broadening is obtained by subtracting 1/e instru- ±30 km s−1, to exclude poor fits for extremely noisy cases. We
mental broadening (winst = 0.023 Å; B. De Pontieu et al. 2014) also generated the total intensity map of Si IV 1403 Å through
and 1/e thermal broadening, which corresponds to the Si IV integration of the spectra over a wavelength range of
peak ionization temperature of log T[K] ∼ 4.8 (wth = 0.011 Å; 1402.77 Å ± 1 Å.
B. De Pontieu et al. 2014), using the following relation: To investigate the environments of our region of interest
(ROI), we need to make mosaic maps from different
observational data, analogous to the IRIS mosaic map. We
wnth = w12 e - winst
2
- wth2 . (2 ) generated AIA 1600 Å and HMI LOS magnetogram mosaic
maps. Because of the long scan time for IRIS mosaic maps, the
We assumed that the thermal broadening does not significantly spatial (heliographic x and y) position in these maps
vary with ionization temperature, because the response function corresponds to different observing times. We selected the
of an Si IV ion has only a single narrow peak in the ionization closest (in time) observational data of AIA and HMI for every
equilibrium (see Figure 9 in H. Peter et al. 2022). So we fixed IRIS slit position and filled their respective mosaic maps using
the thermal broadening regardless of the temperature variation. 2D interpolation to allow pixel-by-pixel comparison.
The nonthermal broadening and the Doppler shift are We selected three rectangular regions of interest for this
converted to velocity units (Doppler velocity and nonthermal study. The red boxes in Figure 2 represent the selected areas.
velocity). We only focus on the pixels that have a nonthermal They are located at different positions on the solar disk and are
velocity less than 40 km s−1 and a Doppler velocity between parts of NOAA active regions 11874, 11877, and 11875,

3
The Astrophysical Journal, 975:33 (11pp), 2024 November 1 Cho, De Pontieu, & Testa

Cartesian coordinates, the result may have errors if we directly

use the simply cropped observational data, because: (1) the
physical size of a pixel depends on the location, due to the
projection effect; and (2) the coordinates of the observed vector
magnetogram differ from those of the local frame on the solar
surface, except for a viewing geometry at the disk center. To
minimize the projection effect, we converted the heliocentric
Cartesian coordinates to spherical coordinates with the solar
center as the origin and the solar radius as the distance. Then
we made an equidistant grid, with its center located at the
center of the ROI and a grid size of about 0°. 03. The latter
corresponds to the HMI pixel size (0 5 ; 360 km at disk
center). We selected a grid area that is sufficiently large to
cover the whole active region that contains the rectangular area
(Figure 5(a)). It seems that polarities outside of our grid area
may affect magnetic fields near the edge of the active region.
Although we think that our main interest, the variation of the
inclination near the inner edge region, is not significantly
affected by the coverage of the grid. To calculate the NLFFF in
a local frame, we converted the magnetic field components
from the observer’s frame to a local frame that has solar east,
north, and the normal of the solar surface as coordinate axes,
Figure 3. An example of an observed spectrum and a single Gaussian fit result. then created input data using 2D interpolation (the underlying
The location of the sampled spectrum is shown as the circles in Figure 6. The
black histogram and gray dashed line indicate the observed spectrum and the
image in Figure 5(b)).
rest wavelength of Si IV (1402.77 Å). The thick red solid line shows the result We obtained the 3D vector magnetic field through NLFFF
of a single Gaussian fitting. The red vertical line and red horizontal arrow show extrapolation (Figure 5(b)). The calculated vector magnetic
the Doppler shift (23.3 km s−1) and 1/e width (31.1 km s−1). The green line fields were converted to the observer’s coordinates again
shows a spectrum with a thermal width corresponding to log T[K] = 4.8, the (Figure 5(c)). We took the fields at 2200 km (;7 pixels) above
IRIS instrumental width, and a zero nonthermal width for comparison.
the photosphere as the transition-region height and calculated
the angle i between the extrapolated magnetic field direction at
respectively (see Figure 2(b)). We considered each active
the transition region and the LOS direction (Figure 5(d)). We
region as a simple magnetic bipole and centered the rectangular
used the cos i value for the analysis. The value cos i = 1
area on the farther polarity region from the solar disk center
indicates the magnetic field is perfectly aligned with the LOS
between the two polarities. This choice is a way of maximizing
the magnetic field inclination effect (Figure 4). If two polarities direction. Finally, we generated a mosaic map of the magnetic
that are offset from the disk center are connected with potential- field inclination at the transition region to allow pixel-by-pixel
like magnetic fields, the inner edge (i.e., closer to the solar disk comparison with other data. There are various different NLFFF
center) of the farther polarity region has parallel magnetic fields magnetic field extrapolation methods (see M. L. De Rosa et al.
to the LOS direction at transition-region heights. On the 2009). We expect that the result will not be very sensitive to the
contrary, the magnetic fields at the outer edge (i.e., closer to the selection of the method if we use the same photospheric vector
solar limb) of the farther polarity region are perpendicular to magnetic field as the boundary condition, because our target
the LOS direction. This effect is similar to the asymmetry of a height, the transition region, is close to the input boundary.
sunspot’s penumbra near the limb. The penumbra toward the For more accurate comparisons, we used two masks to define
center of the solar disk is narrower than that toward the solar the ROIs in the rectangular areas. First, we confined our ROIs
limb. This is generally a geometric projection effect caused by to the strong-magnetic-field region. This mask is important for
the combination of the Wilson depression and the magnetic two reasons: (1) it ensures that only regions in which the
field inclination (A. Wilson & N. Maskelyne 1774). Mean- measurement error in the observed HMI vector magnetogram is
while, the inclination effect illustrated in Figure 4 is not clear in less significant are included; and (2) it is more advantageous to
the disk center region, so we did not include AR 11873 in our study the effect of the magnetic field on nonthermal broadening
analysis. than using weaker-magnetic-field regions. We perform a
This regional dependence of the magnetic field inclination in Gaussian smoothing in spatial dimensions with a σ of 5 pixels,
our simple cartoon geometry can be validated using an NLFFF then take regions where the smoothed photospheric LOS
magnetic field extrapolation. We use the NLFFF package7 in magnetic field strength is stronger than ±50 G, depending on
IDL SolarSoft for the NLFFF magnetic field extrapolation the sign of the main polarity. Second, we adopted a temporal
(M. S. Wheatland et al. 2000). For the extrapolation, the masking. Since the NLFFF was calculated for the three specific
photospheric vector magnetic field at a specific moment is times for each rectangular area, the observational data should
required as a boundary condition. We selected three different be compared with the regions that were observed nearest in
observation times when the IRIS slit passes through the center time, in order to reduce errors caused by temporal evolution or
of the three rectangular areas and took the temporally closest solar rotation. We chose a region that was observed within
HMI vector magnetograms as the boundary conditions. Since ±720 s, which corresponds to the time cadence of the HMI
the local magnetic field extrapolation generally uses the vector magnetogram, from the observing time of the input data
for the NLFFF calculations. Since the selected rectangular
7
https://sprg.ssl.berkeley.edu/~jimm/fff/optimization_fff.html areas span across different horizontal sequences of the IRIS

4
The Astrophysical Journal, 975:33 (11pp), 2024 November 1 Cho, De Pontieu, & Testa

Figure 4. A schematic illustration of the magnetic field configuration and its relation to the LOS direction at transition-region heights. The definition of the inclination
i is described in several cases. If we assume that the active region located far from the disk center has a simple bipole configuration, the inclinations of the magnetic
fields at the inner edge (i.e., toward disk center) of the farther polarity region are close to parallel to the LOS direction (the red circle). In contrast, the magnetic field
inclination at the outer edge of the active region (i.e., toward the limb) is close to perpendicular to the LOS direction (the blue circle).

mosaic (see Figure 2(d)), it is inevitable that the temporal fields are closer to parallel to the LOS in the inner edge of the
masking removes some of the lower parts of the rectangular farther polarity region, in agreement with the expectation from
areas. The final mask combines both the strong-field mask and our cartoon (see Figure 4).
the temporal mask. The most important finding is that the distribution of the
Si IV nonthermal velocity also shows a similar pattern as the
3. Results magnetic field inclination (Figure 6(e)). The averaged non-
thermal velocity in the ROI is about 20 km s−1 (green color).
Figure 6 shows the comprehensive information for region 1.
One can find that the distribution of the nonthermal velocities is
It demonstrates that the region with a strong LOS field from the
not uniform, but the pixels with higher nonthermal velocity are
field strength masking (the blue contour in Figure 6(a)) matches
the bright plage in the AIA 1600 Å intensity map (Figure 6(b)) concentrated in the region toward the disk center. When
and bright Si IV intensity region (Figure 6(c)). The AIA 1600 Å comparing with the magnetic field inclination map, this region
intensity in the ROI does not show a specific regional of enhanced broadening appears to be associated with the
distribution. The magnetic field inclination, however, shows region where the magnetic field is aligned with the LOS
substantial variation. Figure 6(d) clearly shows that the direction. To further investigate and quantify this, we analyzed
magnetic field at transition-region heights in the northeastern a scatter plot (pixel-by-pixel) between the two parameters. The
part of the plage region is parallel to the LOS direction (red 2D histogram in Figure 6(g) does indeed show an association
area), while the southwestern part is not (white or light purple between nonthermal velocity and the cosine of the magnetic
area). Particularly, the northeastern direction is the direction to field inclination. They have a positive linear correlation, with a
the solar disk center (the white arrow in Figure 6(d)). Pearson correlation coefficient of 0.46. The linear fitting result
Considering that we chose the farther (from the disk center) demonstrates that the region with a magnetic field parallel to
polarity region as our ROI, the northeastern part becomes the the LOS shows nonthermal velocities that are about
inner edge of the farther polarity region. Thus, the magnetic 12.7 km s−1 higher than in the region with a magnetic field

5
The Astrophysical Journal, 975:33 (11pp), 2024 November 1 Cho, De Pontieu, & Testa

Figure 5. An example of how we obtain an inclination map for region 3 in Figure 2. (a) An equidistant grid is superimposed on the HMI LOS magnetic field map. The
yellow solid lines indicate every 50 grid points. The observing time of the HMI data is shown at the top of the panel. (b) The NLFFF extrapolated magnetic fields
(green lines) in the local frame. The underlying image shows the equidistant grid data of the radial magnetic field in the photosphere. (c) The NLFFF extrapolated
magnetic fields (green lines) superimposed on the AIA 171 map. The observing time of the AIA data is shown at the top of the panel. Considering that the coronal
loops in the 171 Å image outline the coronal magnetic field lines, one can consider the goodness of the magnetic field extrapolation. (d) The cosine of the magnetic
field inclination at transition-region heights (about 2200 km above the photosphere). A value of 1 indicates magnetic field lines that are parallel with the LOS.

perpendicular to the LOS, i.e., the broadening is enhanced by a positive values, indicating that downflows are ubiquitous at
factor of 2.1. transition-region heights in strong-field regions above plage.
The distribution of the Doppler velocity is also well The other ROIs also show similar results (Figures 7 and 8).
associated with the cosine of the magnetic field inclination Although their locations on the solar disk and viewing angles
(Figure 6(f)). Relatively strong downflows (∼20 km s−1) were are different, they commonly exhibit, in the inner edge of the
observed in the inner edge region. The 2D histogram between farther (from disk center) polarity, the following: a magnetic
the cosine of the magnetic field inclination and the Doppler field parallel with the LOS, enhanced Si IV nonthermal
velocity similarly shows a positive correlation, with a Pearson velocity, and relatively strong downflow. The positive correla-
correlation coefficient of 0.48 (Figure 6(h)). As both non- tion between the parameters can also be found in the 2D
thermal and Doppler velocities show similar behavior with histograms. The coefficients obtained from the 2D histograms
respect to the cosine of the magnetic field inclination, they also are summarized in Table 1. The quantities show some regional
have a good correlation with one another. Figure 6(i) exhibits a variations. For example, in the case of the relation between the
positive correlation between them, with a Pearson correlation cosine of the magnetic field inclination and the nonthermal
coefficient of 0.47. This latter result is consistent with a velocity, the Pearson correlation coefficients (r) and the
previous report (A. Ghosh et al. 2021). It is noteworthy that the gradients of the linear fitting (c1) decrease in the order of
Doppler velocities measured from the Si IV line generally show regions 1, 2, and 3. This may be related to the complexity of the

6
The Astrophysical Journal, 975:33 (11pp), 2024 November 1 Cho, De Pontieu, & Testa

Figure 6. Observational data and 2D histogram for region 1: (a) LOS magnetic field strength map; (b) AIA 1600 Å intensity map; (c) Si IV intensity map; (d) map of
the cosine of the inclination between the magnetic field at the transition region and the LOS, where the red regions, or cos i  1, represent regions where the magnetic
field is parallel to the LOS direction; (e) Si IV nonthermal velocity map; (f) Si IV Doppler velocity map; (g) 2D histogram between the cosine of the magnetic field
inclination at the transition region and the Si IV nonthermal velocity; (h) 2D histogram between the cosine of the magnetic field inclination at the transition region and
the Si IV Doppler velocity; and (i) 2D histogram between the Si IV Doppler velocity and Si IV nonthermal velocity. The blue contours and purple contours in (a)–(c)
indicate the magnetic field strength masking and temporal masking, respectively. The unshaded areas in (d)–(f) indicate the major part of the intersection between the
two maskings, the ROI. All analyses were conducted with the data within this unshaded area. The white arrows in (d)–(f) show the direction to the solar disk center.
The linear fitting results are shown with the blue lines, and their coefficients and the Pearson correlation coefficients are exhibited in the upper left corner of every 2D
histogram plot. The circles in (c), (e), and (f) indicate the positions of the sampled spectra in Figure 3.

magnetic field configuration, i.e., deviation from the simple boundary of the FOV in the velocity map in Figure 7(f) and
bipolar configuration assumed in our cartoon (Figure 4). the distinct coefficient c0 in Figure 7(h).
Another notable difference is that the constant of the linear
fitting between the magnetic field inclination and the Doppler
velocity has a fairly large negative value in region 2 4. Conclusion and Discussion
(c0 = −8.2). We found that this was not real but was caused We have investigated the relation between the cosine of the
by an error in the wavelength calibration. Because the strong magnetic field inclination, with respect to the LOS, and the
noise contaminates the reference line (O I 1355.5977 Å), the observed Si IV nonthermal velocity. Our investigation is
wavelength calibration has an error and the Si IV line looks motivated by the fact that the inclination of the magnetic field
blueshifted. It is responsible for the discontinuity at the with respect to the LOS is expected to have a different effect on

7
The Astrophysical Journal, 975:33 (11pp), 2024 November 1 Cho, De Pontieu, & Testa

Figure 7. The same as Figure 6, but for region 2. The noisy patterns in the IRIS data were caused by cosmic-ray impacts on the detector, when IRIS passed through the
South Atlantic Anomaly.

the observed nonthermal velocity, depending on the main There is a possibility that our assumption of a fixed height
physical mechanism causing the nonthermal velocity. To find for the Si IV line formation (when estimating the magnetic field
clues to the origins of transition-region nonthermal broadening, inclination at transition-region height) introduces some uncer-
we compared the Si IV nonthermal velocity obtained from IRIS tainty. However, we find similar correlations when we assume
observations with the magnetic field inclination at transition- a fixed transition-region height of 4 Mm instead of 2 Mm,
region heights from NLFFF extrapolations based on photospheric which indicates that our assumption of a fixed height does not
vector field measurements with HMI. We discovered that higher significantly impact our conclusions.
values of the Si IV nonthermal velocity are preferentially Our findings imply that, in active regions, plasma motions
observed in regions where the magnetic field appears to be well along the magnetic field are a key contributing ingredient for
aligned with the LOS direction. Additionally, relatively strong causing enhanced nonthermal broadening in the transition
downflows also have a similar tendency as the nonthermal region. These findings are in agreement with the analysis of the
velocity. Our findings do not appear to depend on the relative hotter transition-region plasma observed with IRIS in Fe XII
location on the solar disk or different active regions. Our results emission (P. Testa et al. 2016), which also hinted at a
hold true for region 2, where the data are significantly affected by dominance of field-aligned flows (although they did not
noise introduced by cosmic rays. estimate a pixel-by-pixel inclination of the magnetic field with

8
The Astrophysical Journal, 975:33 (11pp), 2024 November 1 Cho, De Pontieu, & Testa

Figure 8. The same as Figure 6, but for region 3.

Table 1
Coefficientsa from the Relation between Parameters and Information about the Active Regions

Target Binc versus vnthb Binc versus vDopc vDop versus vnthd Magnetic Classificatione m = cos q f
c1 c0 r c1 c0 r c1 c0 r
Region 1 12.7 11.3 0.46 15.4 −2.2 0.48 0.4 16.4 0.47 β 0.57
Region 2 6.6 14.5 0.26 12.8 −8.2 0.37 0.3 18.6 0.31 β/βγ 0.78
Region 3 4.6 17.0 0.21 12.3 −2.0 0.42 0.2 18.6 0.32 βγ/βγδ 0.44

Notes.
a
The linear fit coefficients in a form of y = c1x + c0. The Pearson correlation coefficient is shown as r.
b
Panel (g) in Figures 6, 7, and 8.
c
Panel (h) in Figures 6, 7, and 8.
d
Panel (i) in Figures 6, 7, and 8.
e
http://helio.mssl.ucl.ac.uk/helio-vo/solar_activity/arstats-archive.
f
Based on the center of the rectangular region.

9
The Astrophysical Journal, 975:33 (11pp), 2024 November 1 Cho, De Pontieu, & Testa

respect to the line of sight, but instead assumed a dominance of support the nanoflare occurrence, the characteristics of the
the magnetic field orientation perpendicular to the surface, as in observable parameters are roughly similar to the simulated
previous works). We find that regions where the magnetic field results.
is parallel to the LOS have a nonthermal broadening of More generally, the well-known temperature dependence of
maximum 2.1 times higher than regions where the magnetic the nonthermal velocity supports our notion. The nonthermal
field is perpendicular. In the latter regions, however, the velocity has a peak around 3 × 105 K and decreases for
broadening is not zero. It thus appears that there are (at least) increasing and decreasing temperatures (J. T. Mariska 1992;
two different mechanisms at play. The mechanism we focus on J. Chae et al. 1998). This signifies that, in closed-field regions
here is the newly discovered field-aligned mechanism. on the solar disk, the nonthermal velocity is greatest at the
Which kind of physical mechanism can explain the observed transition-region temperature rather than at chromospheric or
relationship for these regions where the field is parallel to the coronal temperatures (D. H. Brooks & H. P. Warren 2016). The
LOS? Previous studies have traditionally suggested that MHD density in the solar atmosphere is expected to change most
waves (e.g., Alfvénic waves or slow MHD waves) may be the dramatically in the transition region, so it is the region where
dominant mechanism behind enhanced nonthermal velocities, as steep density gradients, and thus strong velocity gradients, are
mentioned in Section 1. It seems clear that Alfvénic waves are most easily expected to occur. This also matches with our
not the major source responsible for the inclination dependence, scenario. One caveat is that most previous studies of
at least in our cases. If the Alfvénic waves were the major nonthermal line broadening did not distinguish between
contributor to the nonthermal velocity in the observed regions, regions where the LOS is aligned or perpendicular with the
we should detect an enhancement of nonthermal broadening in magnetic field. Since it seems clear that there are at least two
regions where the magnetic fields are perpendicular to the LOS mechanisms at work, the selection of observations within the
direction, since Alfvénic waves are transverse waves. This is the samples studied can have a significant effect on the interpreta-
opposite to what we see in our observations. Slow MHD waves tion of the results.
could, in principle, be a candidate mechanism, since these would We note that many ions formed in the transition region and
show a positive correlation between nonthermal velocity and the their associated spectral line emission are affected by none-
cosine of the inclination angle between the magnetic field and quilibrium ionization effects. Such effects have been shown to
LOS. However, if slow-mode waves were the (only) dominant significantly increase the temperature (and thus height) range
mechanism, one would expect alternating blue- and redshifts, so over which the line is formed (K. Olluri et al. 2015). This could
that an IRIS Doppler velocity map should be dominated, contribute to enhanced broadening as well.
statistically, by a mix of blue- and redshifts. This is not the In our observations, there are some differences between the
case in our observations, which are dominated by the well-known observed regions. With respect to the Pearson correlation
pervasive redshifts in the transition region (J. T. Mariska 1992; coefficients, region 1 shows a tighter linear correlation than
J. Chae et al. 1998; H. Peter & P. G. Judge 1999; V. H. Hansteen region 3. We believe that the simplicity of the magnetic field
et al. 2010; H. Skogsrud et al. 2016). In other words, if slow- configuration may play a role. For the NLFFF method, a simple
mode waves are to contribute to the nonthermal broadening, any configuration of magnetic poles as a boundary condition leads
explanation also needs to simultaneously explain the pervasive to a more realistic field extrapolation. We can see that region 1
redshifts. It cannot be excluded that a mechanism causing is approximately close to a simple isolated bipole, so the
pervasive redshifts and a separate mechanism (like slow-mode NLFFF method can provide a more realistic estimate of the 3D
waves or shock waves; B. De Pontieu et al. 2015) are both acting field configuration. On the contrary, region 2 looks more
together in combination at the same time. complex and affected by another neighboring active region
An alternative explanation is that the nonthermal velocity is (AR 11879), and region 3 is even more complex, as evidenced
caused by a velocity gradient (within the height range over by the occurrence of 10 C-class flares in this region on the
which the transition-region spectral lines form) associated with observed date (see the magnetic classification in Table 1).
the downward field-aligned motions that we observe. It is clear Given the methodology used (NLFFF), it is thus perhaps not
that we are observing falling material, from the observed surprising that region 1 shows a higher correlation coefficient
redshifts. Since we are observing regions with low plasma β, than the other regions. We have already mentioned that region
the plasma motions we observe are moving along the field. One 2 was observed by IRIS when the spacecraft passed through the
potential scenario is that given the density stratification in the South Atlantic Anomaly. This leads to more cosmic-ray hits
atmosphere, as plasma falls down into the lower atmosphere, it from the impacts of energetic particles on the detector, causing
encounters increasing density, and eventually will slow down noise in the obtained spectral parameters (See Figures 7(c), (e),
and stop falling at some height. During this process, a and (f)). This can introduce noise in the observed correlations.
deceleration phase is inevitable, and this could lead to a In summary, our results indicate that, in strong plage regions,
change of velocity within the height range over which the there is a significant component to the nonthermal broadening
spectral line is formed. This change in velocity could then lead that is field-aligned and associated with strong downflows. If
to enhanced line broadening. This scenario is consistent with we assume that the result from region 1 is an ideal case, there is
what we found in the observational data. There are several still a considerable nonthermal velocity (∼11.3 km s−1) in the
previous studies supporting our explanation. S. Patsourakos & region where the LOS is perpendicular to the magnetic field
J. A. Klimchuk (2006) simulated the responses of several (Figure 6(g)). This is not negligible, comparing with the field-
spectra when nanoflares occur and reported that the spectra that aligned component (∼12.7 km s−1), and its origin is not well
formed just below 1 MK show weak redshift with a few tens of understood. We also see an enhancement of nonthermal
kilometers per second and broadening due to the mild draining velocity at the solar limb (see Figure 2(a)). As discussed,
motion. Even though these predictions do not extend to the low this may be a contribution from transverse waves, as pointed
temperatures we observe here, and our result does not directly out by several previous studies (J. T. Mariska et al. 1978;

10
The Astrophysical Journal, 975:33 (11pp), 2024 November 1 Cho, De Pontieu, & Testa

B. De Pontieu et al. 2015; Y. K. Rao et al. 2022; M. Carlsson & References

B. De Pontieu 2023). Our study has focused only on plage
regions with strong magnetic fields. The nonthermal velocity in Bobra, M. G., Sun, X., Hoeksema, J. T., et al. 2014, SoPh, 289, 3549
Boland, B. C., Engstrom, S. F. T., Jones, B. B., & Wilson, R. 1973, A&A,
weak-field regions may be affected by different mechanisms, as 22, 161
examined by A. Ghosh et al. (2021). For example, it is not clear Brooks, D. H., & Warren, H. P. 2016, ApJ, 820, 63
whether the magnetic field is strong enough to guide the plasma Carlsson, M., & De Pontieu, B. 2023, ApJ, 959, 87
as well in weak-field regions. In addition, explosive events are Chae, J., Schühle, U., & Lemaire, P. 1998, ApJ, 505, 957
(more) common in weak-field regions and likely are a Cho, K., Testa, P., De Pontieu, B., & Polito, V. 2023, ApJ, 945, 143
De Pontieu, B., McIntosh, S., Martinez-Sykora, J., Peter, H., &
significant contributor to the nonthermal velocity. We can Pereira, T. M. D. 2015, ApJL, 799, L12
even find signatures of explosive events in our data: a De Pontieu, B., Title, A. M., Lemen, J. R., et al. 2014, SoPh, 289, 2733
brightening is associated with high nonthermal velocity at De Rosa, M. L., Schrijver, C. J., Barnes, G., et al. 2009, ApJ, 696, 1780
(−350″, 30″) in region 3 (Figures 8(b), (c), and (e)). Prevalent Del Zanna, G., & Mason, H. E. 2018, LRSP, 15, 5
Dere, K. P., & Mason, H. E. 1993, SoPh, 144, 217
small events under the current detection limit may contribute to Doschek, G. A., & Feldman, U. 1977, ApJL, 212, L143
the nonthermal broadening in the same manner. Esser, R., Fineschi, S., Dobrzycka, D., et al. 1999, ApJL, 510, L63
It thus seems likely that multiple processes cause nonthermal Ghosh, A., Tripathi, D., & Klimchuk, J. A. 2021, ApJ, 913, 151
broadening of spectral lines. The relative importance of these Hansteen, V. H., Hara, H., De Pontieu, B., & Carlsson, M. 2010, ApJ,
718, 1070
processes may vary depending on the type of region. It is clear Lemen, J. R., Title, A. M., Akin, D. J., et al. 2012, SoPh, 275, 17
that revealing the origin of nonthermal broadening will give us Lorincik, J., Polito, V., De Pontieu, B., Yu, S., & Freij, N. 2022, FrASS,
valuable clues toward understanding the small-scale phenom- 9, 334
ena and energy transport in the solar atmosphere. Mariska, J. T. 1992, The Solar Transition Region (Cambridge: Cambridge
Univ. Press)
Mariska, J. T., Feldman, U., & Doschek, G. A. 1978, ApJ, 226, 698
Acknowledgments Mariska, J. T., Feldman, U., & Doschek, G. A. 1979, A&A, 73, 361
We gratefully acknowledge support from the NASA contract McClements, K. G., Harrison, R. A., & Alexander, D. 1991, SoPh, 131, 41
McIntosh, S. W., De Pontieu, B., & Tarbell, T. D. 2008, ApJL, 673, L219
NNG09FA40C (IRIS). P.T. was supported for this work by Olluri, K., Gudiksen, B. V., Hansteen, V. H., & De Pontieu, B. 2015, ApJ,
contract 8100002705 (IRIS) to the Smithsonian Astrophysical 802, 5
Observatory and by NASA grant 80NSSC20K1272. We Orozco Suárez, D. 2012, in ASP Conf. Ser. 454, Hinode-3: The 3rd Hinode
gratefully acknowledge helpful discussions regarding the Science Meeting, ed. T. Sekii, T. Watanabe, & T. Sakurai (San Francisco,
CA: ASP), 37
interpretation of the results with Viggo Hansteen and Juan Patsourakos, S., & Klimchuk, J. A. 2006, ApJ, 647, 1452
Martinez Sykora. We greatly appreciate Alberto Sainz Dalda Pesnell, W. D., Thompson, B. J., & Chamberlin, P. C. 2012, SoPh, 275, 3
and Marc DeRosa, who helped with the use of the IRIS mosaic Peter, H., Gudiksen, B. V., & Nordlund, Å . 2006, ApJ, 638, 1086
data and the NLFFF extrapolation, respectively. This research Peter, H., & Judge, P. G. 1999, ApJ, 522, 1148
has made use of NASA’s Astrophysics Data System and of the Peter, H., Chitta, L. P., Chen, F., et al. 2022, ApJ, 933, 153
Pontin, D. I., Peter, H., & Chitta, L. P. 2020, A&A, 639, A21
SolarSoft package for IDL. IRIS is a NASA small explorer Rao, Y. K., Del Zanna, G., & Mason, H. E. 2022, MNRAS, 511, 1383
mission developed and operated by LMSAL with mission Roussel-Dupre, R., Francis, M. H., & Billings, D. E. 1979, MNRAS, 187, 9
operations executed at NASA Ames Research Center and Sarro, I. M., de Sterk, H., Erdélyi, R., Montesinos, B., & Doyle, J. G. 1997, in
major contributions to downlink communications funded by ESA Special Publication, Vol. 404, Fifth SOHO Workshop: The Corona
and Solar Wind Near Minimum Activity, ed. A. Wilson (Noordwijk: ESA
ESA and the Norwegian Space Agency (NOSA). Resources Publications), 657
supporting this work were provided by the NASA High-End Schou, J., Scherrer, P. H., Bush, R. I., et al. 2012, SoPh, 275, 229
Computing (HEC) Program through the NASA Advanced Skogsrud, H., Rouppe van der Voort, L., & De Pontieu, B. 2016, ApJ, 817, 124
Supercomputing (NAS) Division at Ames Research Center. Testa, P., De Pontieu, B., & Hansteen, V. 2016, ApJ, 827, 99
Testa, P., Polito, V., & De Pontieu, B. 2020, ApJ, 889, 124
Testa, P., De Pontieu, B., Allred, J., et al. 2014, Sci, 346, 1255724
ORCID iDs van Ballegooijen, A. A., Asgari-Targhi, M., Cranmer, S. R., & DeLuca, E. E.
Kyuhyoun Cho https://orcid.org/0000-0001-7460-725X 2011, ApJ, 736, 3
Wheatland, M. S., Sturrock, P. A., & Roumeliotis, G. 2000, ApJ, 540, 1150
Bart De Pontieu https://orcid.org/0000-0002-8370-952X Wilson, A., & Maskelyne, N. 1774, RSPT, 64, 1
Paola Testa https://orcid.org/0000-0002-0405-0668 Zhu, Y., Habbal, S. R., Ding, A., et al. 2024, ApJ, 966, 122

11