Teh 2023 ApJ 959 68

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

The Astrophysical Journal, 959:68 (7pp), 2023 December 20 https://doi.org/10.

3847/1538-4357/ad09b4
© 2023. The Author(s). Published by the American Astronomical Society.

Kinetic and Shear Alfvén Waves in a Large Guide-field Reconnection at Earthʼs


Magnetopause
Wai-Leong Teh1 , Wenzhe Zhang2,3 , and Huishan Fu2,3
1
Space Science Centre, Institute of Climate Change, Universiti Kebangsaan Malaysia, Bangi, Malaysia
2
School of Space and Environment, Beihang University, Beijing, People’s Republic of China
3
Key Laboratory of Space Environment Monitoring and Information Processing, Ministry of Industry and Information Technology, Beijing, People’s Republic of
China
Received 2023 August 9; revised 2023 October 20; accepted 2023 November 3; published 2023 December 7

Abstract
A large guide-field (∼1.1Bo) reconnection X-line, observed by the Magnetospheric Multiscale spacecraft during an
outbound magnetopause crossing, is studied for Alfvén waves. Here Bo is the reconnecting field magnitude. The
current sheet thickness of the magnetopause was ∼2.6 ion inertial lengths (∼269 km), where field-aligned counter-
streaming electrons were observed and Hall electromagnetic fields were identified. A remarkable finding was that a
kinetic Alfvén wave (KAW) was seen in the magnetopause upstream region after a shear Alfvén wave (SAW) was
encountered in the magnetopause layer. The presence of both the SAW and KAW near the reconnection X-line is
for the first time reported. In the spacecraft frame of reference, the SAW has a dominant frequency at ∼0.74 Hz,
while the KAW has two dominant frequencies at ∼0.38 and ∼0.64 Hz. The wave energy for KAW and SAW was
mostly carried away from the reconnection site by the Poynting flux parallel to the magnetic field. The parallel
temperatures for ions and electrons were increased at KAW. The peaks of T∥/T⊥ for ions were located near the
wave peaks, while the ratio peaks for electrons were near the wave troughs. Our findings suggest that KAWS and
SAWs can be generated by asymmetric reconnection with a large guide field.
Unified Astronomy Thesaurus concepts: Alfvén waves (23); Space plasmas (1544)

1. Introduction Using kinetic particle-in-cell simulations, Shay et al. (2011)


found that the Hall magnetic field near the separatrices is
Alfvénic fluctuations are coupled fluctuations in the plasma
associated with KAWs. Gershman et al. (2017) confirmed from
velocity (V) and magnetic field (B), where changes in V and B
Magnetospheric Multiscale (MMS) observations that there is a
are governed by DV = aDB m0 r , with ρ being the
conservative energy exchange between ion-scale KAWs and
plasma mass density and α being constant. When ΔV and ΔB
particles in the reconnection exhaust at the magnetopause.
are correlated/anticorrelated, the transverse Alfvénic fluctua- These previous studies concluded that KAWs play an essential
tions are propagating antiparallel/parallel to the underlying role in the energy transfer process and in facilitating magnetic
magnetic field. A pure Alfvén wave has α being ±1. Super- reconnection.
position of different modes will result in deviations from the This paper aims to report MMS observations of a large
pure Alfvén waves. The Alfvénic fluctuations have been widely guide-field reconnection X-line at the dawn flank magnetopause
observed in various space plasma environments, for example, for Alfvén waves. A remarkable finding was that both the SAW
in the solar wind (e.g., Gosling et al. 2009; Paschmann et al. and KAW were encountered during the X-line crossing. This
2013), magnetosheath (e.g., Roberts et al. 2018), plasma sheet event was identified from a large data set (62 cases) of well-
boundary layer (e.g., Wygant et al. 2002), and ionosphere (e.g., determined rotational discontinuities by Haaland et al. (2020) at
Chaston et al. 2002). From a shear Alfvén wave (SAW), a the flank magnetopause. The multi-spacecraft timing method
kinetic Alfvén wave (KAW) is formed when the perpendicular (e.g., Vogt et al. 2011; Zhang et al. 2022) is implemented to
wavelength becomes comparable to the ion thermal gyroradius estimate wavevector properties (e.g., wave propagation direc-
(d = cs/Ωi). Here cs= kb T^ mi is the ion thermal speed and tion, phase speed, and wavenumber), using four-point magnetic
Ωi = eB/mi is the ion gyrofrequency. KAWs are obliquely field measurements.
propagating low-frequency waves, while SAWs can be parallel
and oblique ones.
KAWs have been considered as an energy source for auroral 2. MMS Observations and Analysis Results
particle acceleration (e.g., Lysak 2023). The electrons that
generate discrete aurora in the upper ionosphere are accelerated Around 23:43:20 UT, on 2015 December 15, the four MMS
by the parallel electric field, which is induced by KAWs from a spacecraft with a tetrahedron formation (Burch et al. 2016)
reconnection site in the magnetotail. It has been demonstrated made an outbound magnetopause crossing located around (8.9,
from Cluster observations that Poynting fluxes associated with −5.0, −0.7) RE (Earthʼs radius) in geocentric solar ecliptic
KAWs radiate obliquely outward from the reconnection (GSE) coordinates. The inter-spacecraft distances were small,
diffusion region in the magnetotail (Chaston et al. 2009). ranging between ∼24 and ∼14 km. In Figure 1, the data were
based on the MMS1 observations of magnetic fields from the
Original content from this work may be used under the terms
fluxgate magnetometer (FGM; Russell et al. 2016), electric
of the Creative Commons Attribution 4.0 licence. Any further fields from the electric double probes (EDP; Lindqvist et al.
distribution of this work must maintain attribution to the author(s) and the title 2016), and ion and electron plasma moments from the fast
of the work, journal citation and DOI. plasma investigation (FPI; Pollock et al. 2016), at time

1
The Astrophysical Journal, 959:68 (7pp), 2023 December 20 Teh, Zhang, & Fu

Figure 1. (a)–(g) MMS1 observations of a large guide-field reconnection X-line during the magnetopause crossing enclosed by the vertical lines, with vectors shown
in LMN coordinate systems. (h)–(j) The rates of energy conversion and the parallel electric fields. The magenta dashed line in panel (a) denotes the ambient
reconnection guide field. The green line in panel (e) denotes the Hall electric fields evaluated in the X-line moving frame.

resolutions in burst modes. To remove high-frequency noises, outflows were, respectively, accompanied by BN > 0 and BN <
the electron plasma moments (ne, Te, and Ve) were smoothed 0, consistent with a two-dimensional reconnection scenario for
by a Savitzky–Golay filter (Press & Teukolsky 1990) with a the magnetopause. Using the measured electric field data
span width of 101 data points. The current density was between 23:43:13 and 23:43:24 UT, the estimated velocity of
calculated as j = qne(Vi − Ve) using the FPI data, where q is the the X-line in GSE was V0 = (−33.9, −5.8, −67.3) km s−1,
electron charge. In Figure 1(a), the two vertical dashed lines from the least squares method developed by Sonnerup &
denote the time interval (23:43:12 UT–23:43:28 UT) of the Hasegawa (2005). The normal magnetopause velocity was then
magnetopause crossing from the magnetospheric boundary calculated as V0 · N = −16.8 km s−1 and therefore the
layer (BL>0) to the magnetosheath (BL < 0). At the vertical thickness of the magnetopause current layer was |V0 · N|Δt =
lines, the magnetic fields in GSE were B1 = (1.1, −26.0, 35.0) ∼269 km (∼2.6 λi), where Δt = 16 s is the duration of the
nT and B2 = (−15.6, −41.7, −23.4) nT for the magnetospheric magnetopause crossing. Here λi = 102 km is the ion inertial
and magnetosheath sides, respectively. The magnetic shear length, based on the upstream (magnetosheath) values of
angle was ∼84° across the magnetopause. The LMN boundary n = 5 cm−3.
coordinates in Figure 1 were defined as M = (B1 × B2) × As indicated in Figure 1(e), the Hall electric fields evaluated
(B2–B1)/|(B1 × B2) × (B2–B1)| (the guide-field direction), in the X-line moving frame, (E + (Vi - V0 ) ´ B)N , were
N = B1 × B2/| B1 × B2| (the magnetopause normal), and pointing toward the magnetosheath side within the magneto-
L = M × N (the reconnecting field direction). The resulting pause, similar to the results shown in Figure 6(d) of Pritchett &
axes in GSE are L = (0.267, 0.250, 0.931), M = (0.170, 0.938, Mozer (2009) for asymmetric reconnection with a guide field.
−0.301), and N = (0.949, −0.238, −0.208). The reconnecting The Hall magnetic field, BM − BG, was positive for the
field magnitude B0 was estimated ∼31.4 nT, based on the quadrants (ViL < 0, BL>0) and (ViL>0, BL < 0), as shown in
average value of |B1 · L| and |B2 · L|. As indicated by the Figure 1(a). A remarkable feature was observed in the β > 1
magenta dashed line in Figure 1(a), the ambient reconnection region (β is the ratio of the ion thermal pressure to the magnetic
guide field across the magnetopause was estimated as BG = pressure; see Figure 1(f)), namely, that low-frequency waves
B1 · M = B2 · M = ∼−34.8 nT, corresponding to ∼1.1 B0. As were found in both the magnetic field and ion velocity
seen in Figure 1(b), reconnection outflows (>100 km s−1) were components in the L and N directions. These waves were also
observed in ViL and their signs were reversed from negative to seen at the current densities jL and jN. Later, it will be
positive during the magnetopause crossing, suggesting that a demonstrated that changes in BL and BN were negatively
magnetic X-line moved southward through the spacecraft. The correlated with those for the ion plasma velocity. Furthermore,
southward (V iL < 0) and northward (ViL > 0) reconnection low-frequency waves were also found in the ion and electron

2
The Astrophysical Journal, 959:68 (7pp), 2023 December 20 Teh, Zhang, & Fu

correlated with δVa in between the gray lines, but not found
in between the green lines. The δVe perturbations were present
at all three components, where the regression slopes were −0.6,
−1.3, and −0.6 for the L, M, and N directions, respectively. In
the next section, it will be shown that the first waveform is of
an SAW and the other is of a KAW. Figure 3(g) compares the
Poynting flux magnitude, |S|, with the parallel one, |S∥|, and
indicates that the wave Poynting flux parallel to the magnetic
field carries most of the energy, expected for SAW and KAW.
The ratios Te∥/Te⊥ and T∥/T⊥ show that the parallel
temperature for ions and electrons was increased at KAW
(Figure 3(h)). It is noteworthy that the peaks of T∥/T⊥ were
located near the wave peaks denoted by the red dashed lines,
while the peaks of Te∥/Te⊥ were near the wave troughs.

3. Wavevector Analysis
Figure 4(a) shows for the four MMS spacecraft the δBL of
the first waveform in the time interval between 23:43:15 and
23:43:21 UT, where the high-frequency fluctuations in δBL
have been removed by the Savitzky–Golay filter with a span
width of 91 data points. One can see that the crossing times of
Figure 2. (a) Electron PAD for the energy range of 0.2–2.0 keV. (b)–(d) The the wavefront for the four spacecraft were different. Figure 4(b)
overlaid plots of the magnetic field and the ion and electron velocity (−Vi and shows the power spectral density (PSD) of δBL based on
−Ve) for each component in LMN coordinate systems. The vertical lines mark MMS1. A single spectral peak, denoted by the red dashed line,
the duration of the magnetopause crossing. was present at fm = ∼0.74 Hz, smaller than the proton
gyrofrequency fci (green line), in the spacecraft frame. Using a
bandpass filter at 0.70–0.78 Hz, a single-frequency wave can
velocities in the low β region (β < 1). The rates of energy be approximately derived. Therefore, the wavevector, k, and
conversion, j · E, Ve · (j × B), and j · E¢, are shown in the wave phase speed, vph, can be estimated by the multi-
Figures 1(h) and (i), where E¢ = E + Ve ´ B. The dot product k
spacecraft timing method, where m = ( v ) = åa4 = 1 ka tga and
of j · E¢ can be written as j · E − Ve · (j × B), where Ve · (j × B) ph

is the rate of work done by the Lorentz force in the ideal MHD. |k| = 1. Here κα is the reciprocal vector for the spacecraft
Therefore, the nonzero j · E¢ indicates the nonideal MHD tetrahedron (Chanteur 1998) and tγα = tα − tγ is the crossing
energy transfer between electromagnetic fields and plasmas time difference between two spacecraft, where tγ is at the
(Zenitani et al. 2012). One can find in Figure 1(j) that reference spacecraft, which is MMS1 in this study. The time
significant parallel electric fields were present during the delays tγα were determined by the maximum value of the cross-
energy conversion process. correlation function. The timing analysis results for the first
waveform were vph = ∣ m ∣ = 281 km s−1 and k = vphm =
1
In Figure 2 each component of the magnetic field vector in
LMN coordinates was plotted on top of the corresponding (−0.491, −0.816, 0.304) (GSE). A wave normal angle θBk of
component of -Vi and -Ve, together with a color map of the ∼18° formed between k and the magnetic field and the ratio of
electron pitch angle distribution (PAD) for energy range k⊥ to k∥ was ∼0.3. Since the magnetic X-line is moving
0.2–2.0 keV. It is seen that changes in B were negatively southward (V0z <0) and kz > 0, the wave is thus propagating
correlated with Vi and Ve. Field-aligned counter-streaming away from the diffusion region. The wavenumber, k, can then
be roughly estimated as 2 v m = 1.644 × 10−2 km−1, which
f
electrons were detected within the magnetopause, which is
ph
considered one of the signatures for the reconnection ion gives k⊥d = ∼0.2. Here d = ∼41 km is the ion thermal
diffusion region (e.g., Teh et al. 2012). Changes in BL and BN gyroradius based on upstream values of T⊥ = 410 eV and
were mostly coupled with -ViL and -ViN within the magneto- B = 50 nT. In the ion velocity frame of reference, the wave
pause, while in the upstream region, the magnetic field v ·k
frequency was estimated as fps = fm - 2p = ∼0.41
variations were coupled with both the ion and electron
velocities for all three components. Hz = ∼0.118 fci, where v = (−115.8, −119.8, −95.5)
By calculating the consecutive differences in Vi (δVi), the (km s−1) is the average ion plasma velocity. In summary, the
wave structures of Vi were manifested and then compared with overall result suggests that the first waveform is of an SAW, in
k
δVa = δ(B/ m0 mp n ), as shown in Figures 3(a)–(c). Here Va is terms of the parallel wave propagation, k^ <1, and k⊥d <1.

the ion Alfvén velocity, mp is the proton mass, and n is the Figure 4(c) shows the δBL of the second waveform for the
plasma density. For the time interval enclosed by the green four spacecraft in the time interval between 23:43:27 and
lines in Figures 3(a)–(c), the δVi perturbations were negatively 23:43:46 UT. In Figure 4(d), two spectral peaks with
correlated with δVa and dominant at the L and N directions. comparable PSD were present at fm1 = ∼0.38 Hz (black
The slope of a regression line was ∼−1.3 between δViL and dashed line) and fm2 = ∼0.64 Hz (red dashed line) in the
δVaL and ∼−1.6 between δViN and δVaN, while these two spacecraft frame. A single-frequency wave can approximately
slopes became −0.3 and −0.7 for the time interval enclosed by be extracted for fm1 and fm2, by using a bandpass filter at
the gray lines. Comparisons of δVe with δVa in Figures 3(d)–(f) 0.35–0.40 Hz and 0.6–0.7 Hz, respectively. For fm1,
revealed that the δVe perturbations were also negatively vph1 = 120 km s−1 and k1 = (−0.272, −0.222, 0.936) (GSE),

3
The Astrophysical Journal, 959:68 (7pp), 2023 December 20 Teh, Zhang, & Fu

Figure 3. (a)–(f) Wave structures for ion and electron plasma velocities, and magnetic field. (g) Comparison of the Poynting flux |S| to |S∥|. (h) The ratio of parallel to
perpendicular temperature for ions (blue) and electrons (red). The first and second waveforms are enclosed by the green and gray lines, respectively. The red dashed
lines denote the wave peaks.

= ∼5.6. For fm2, vph2 = 170 km s−1 (Figure 5(b)) and KAW (Figure 5(d)). In Figure 5, the cross
k^
with θBk = ∼80o and k
and k2 = (−0.739, −0.230, 0.633) (GSE), with θBk = ∼67° and triangle symbols denote the previous analysis results for
k SAW and KAW. A good agreement is found at the cross
and k^ = ∼2.3. The resulting wavenumbers for fm1 and fm2 are
 symbols, where at the same wave frequency the result of the
k1 = 1.962 × 10−2 km−1 and k2 = 2.362 × 10−2 km−1, giving cross symbol is close to the one with a high wave power, but a
k⊥d = 0.8 and 0.9, respectively. The wave frequencies in the large discrepancy at the triangle symbol. This discrepancy
ion velocity frame, i.e., v = (−56.8, −72.1, 48.5) (km s−1), could be attributed to the multiple wave branches in KAW.
were fps1 = ∼0.13 Hz = ∼0.029 fci and fps2 = ∼0.30
Hz = ∼0.066 fci. Note that the normalized frequency in the 4. Summary and Discussion
plasma frame in Figure 4(d) was calculated using the k2 result.
Apparently, the second waveform is different than the SAW A large guide-field (∼1.1Bo) reconnection X-line, observed
seen in the magnetopause layer. It is suggested that the second by the MMS spacecraft during the outbound magnetopause
waveform is of a KAW, in terms of the perpendicular crossing, has been examined for Alfvén waves. The current
propagation, k⊥/k∥ > 1, and k⊥d ∼ 1. Since k1z and k2z are sheet thickness of the magnetopause was ∼2.6 ion inertial
directed northward, the KAW is thus propagating away from lengths (∼269 km), where the field-aligned counter-streaming
the reconnection site. electrons were detected and the Hall electromagnetic fields
Additionally, the wave dispersion relation was examined for were identified. An SAW was present in the magnetopause
SAW and KAW, using the multi-spacecraft timing method layer. Remarkably, a KAW was seen in the magnetopause
developed by Zhang et al. (2022), where the phase speed and upstream region after an SAW was encountered. Low-
wavevector were calculated at each time for each single- frequency magnetic field fluctuations were negatively corre-
frequency wave in the range from zero to the Nyquist lated with the changes in the ion and electron velocities at
frequency. In Figures 5(a) and (c), the dispersion relations KAW, but only in the ion velocity at SAW. The wavevector k
are shown for θBk = 0°–30° (SAW) and θBk = 60°–90° was predicted by the multi-spacecraft timing method using
(KAW), where the color code denotes the wave power P( f, k) four-point magnetic field measurements. The timing analysis
k
derived from a wavelet transform and the red dashed line results of k^ , k⊥d, and the wave propagation direction were

denotes the local Alfvén speed. As compared to SAW, the satisfied for both the KAW and SAW. In the spacecraft frame,
KAW has a much wider k range at the same frequency, the wave frequency was ∼0.74 Hz for SAW and ∼0.38 Hz/
suggesting the presence of multiple wave branches in the ∼0.64 Hz for KAW. The parallel Poynting flux carried most of
KAW. The frequency versus k⊥d is also shown for SAW the wave energy in the KAW and SAW, away from the

4
The Astrophysical Journal, 959:68 (7pp), 2023 December 20 Teh, Zhang, & Fu

Figure 4. The consecutive difference in BL at the four MMS spacecraft for (a) SAW and (c) KAW. The PSD of δBL based on MMS1 is displayed for (b) SAW and (d)
KAW, where the red and black dashed lines denote the spectral peaks, and the proton gyrofrequency is at the green line. The two values of the normalized frequency
correspond to 1 Hz and 10 Hz in the spacecraft frame.

reconnection site. The parallel temperature for ions and The flow speed of the plasma frame is distinct between SAW
electrons was increased at KAW and the peaks of T∥/T⊥ for and KAW, where the SAW was observed in the faster plasma
ions were located near the wave peaks, while the ratio peaks for frame. In the spacecraft frame, the wave frequency (∼0.74 Hz)
electrons were near the wave troughs. of the SAW is fairly close to that of the KAW at ∼0.64 Hz, and
It is unexpected that the perturbed fields of the SAW are not their Doppler shifts (v · k ) are fairly consistent (∼0.34 Hz).
correlated with the electron velocity variations. During the Since the SAW was in the faster plasma frame as compared to
SAW time interval, strong field-aligned, counter-streaming the KAW, the wavelength of the SAW is expected to be larger
electrons were present (Figure 2(a)). This will lead to a non- than that of the KAW.
Maxwellian distribution in the phase space density. Moreover, The magnetic energy of the Hall magnetic fields is believed
the calculated electron bulk velocity is significantly smaller to be propagating away from the reconnection site by Alfvén
than the electron thermal speed (∼4400 km s−1). One may waves. It has been demonstrated in simulations that the Hall
speculate that the electron bulk velocity may not be accurately magnetic field structures are carried away by KAWs (Shay
estimated. Thus, this result needs to be interpreted with caution. et al. 2011). From a 2D kinetic simulation of magnetotail

5
The Astrophysical Journal, 959:68 (7pp), 2023 December 20 Teh, Zhang, & Fu

Figure 5. The dispersion relation and frequency vs. k⊥d for (a), (b) SAW and (c), (d) KAW. The color code is the wave power derived from a wavelet transform. The
red dashed line in (a) and (c) denotes the local Alfvén speed at 221 and 394 km s−1, respectively. The cross and triangle results are obtained from a different method.

reconnection with zero guide field, Gurram et al. (2021) ORCID iDs
showed that magnetic reconnection can generate both SAWs
Wai-Leong Teh https://orcid.org/0000-0002-7114-1543
and KAWs when an ion-to-electron mass ratio of 400 and open
Wenzhe Zhang https://orcid.org/0000-0003-4364-3252
boundary conditions were implemented and that the SAWs can
Huishan Fu https://orcid.org/0000-0002-4701-7219
become the main carrier of wave energy during reconnection.
In this studied event, both SAW and KAW were detected near
the reconnection X-line, suggesting that KAWS and SAWs can References
be generated by asymmetric reconnection with a large
guide field. Burch, J. L., Moore, T. E., Torbert, R. B., & Giles, B. L. 2016, SSRv, 199, 5
Chanteur, G. 1998, ISSIR, 1, 349
Chaston, C. C., Bonnell, J. W., Peticolas, L. M., et al. 2002, GeoRL, 29, 1535
Chaston, C. C., Johnson, J. R., Wilber, M., et al. 2009, PhRvL, 102, 015001
Acknowledgments Gershman, D. J., F-Viñas, A., Dorelli, J. C., et al. 2017, NatCo, 8, 14719
Gosling, J. T., McComas, D. J., Roberts, D. A., & Skoug, R. M. 2009, ApJL,
This work was supported by the grants of Universiti 695, L213
Kebangsaan Malaysia (GP-2021-K020730 and GP-K020730). Gurram, H., Egedal, J., & Daughton, W. 2021, GeoRL, 48, e94201
W.L.T. thanks Stein Haaland for providing a list of Haaland, S., Paschmann, G., Øieroset, M., et al. 2020, JGRA, 125, e27623
magnetopause crossings by MMS. Special thanks to the Lindqvist, P.-A., Olsson, G., Torbert, R. B., et al. 2016, SSRv, 119, 137
Lysak, R. L. 2023, RvMPP, 7, 6
dedicated efforts of the entire MMS mission team for data Paschmann, G., Haaland, S., Sonnerup, B. U. Ö., & Knetter, T. 2013, AnGeo,
access. MMS data can be downloaded at the NASA CDAWeb 31, 871
at http://cdaweb.gsfc.nasa.gov/. Pollock, C., Moore, T., Jacques, A., et al. 2016, SSRv, 199, 331

6
The Astrophysical Journal, 959:68 (7pp), 2023 December 20 Teh, Zhang, & Fu

Press, W. H., & Teukolsky, S. A. 1990, ComPh, 4, 669 Sonnerup, B. U. Ö., & Hasegawa, H. 2005, JGRA, 110, A06208
Pritchett, P. L., & Mozer, F. S. 2009, JGRA, 114, A11210 Teh, W.-L., Nakamura, R., & Fujimoto, M. 2012, JGRA, 117, A12225
Roberts, O. W., Toledo-Redondo, S., Perrone, D., et al. 2018, GeoRL, 45, 7974 Wygant, J. R., Keiling, A., Cattell, C. A., et al. 2002, JGRA, 107, 1201
Russell, C. T., Anderson, B. J., Baumjohann, W., et al. 2016, SSRv, 199, 189 Vogt, J., Haaland, S., & Paschmann, G. 2011, AnGeo, 29, 2239
Shay, M. A., Drake, J. F., Eastwood, J. P., & Phan, T. D. 2011, PhRvL, 107, Zenitani, S., Shinohara, I., & Nagai, T. 2012, GeoRL, 39, L11102
065001 Zhang, W. Z., Fu, H. S., Cao, J. B., et al. 2022, ApJ, 936, 176

You might also like