The Astronomical Journal, 163:164 (12pp), 2022 April https://doi.org/10.

3847/1538-3881/ac4e92
© 2022. The Author(s). Published by the American Astronomical Society.

Simultaneous Detection of Optical Flares of the Magnetically Active M-dwarf Wolf359

Han-Tang Lin1, Wen-Ping Chen1,2 , Jinzhong Liu3 , Xuan Zhang3 , Yu Zhang3 , Andrew Wang4 , Shiang-Yu Wang4 ,
Matthew J. Lehner4,5,6 , C. Y. Wen4, J. K. Guo1, Y. H. Chang1, M. H. Chang1, Anli Tsai1 , Chia-Lung Lin1 , C. Y. Hsu1, and
Wing Ip1,7
1
Graduate Institute of Astronomy, National Central University, 300 Zhongda Road, Zhongli, Taoyuan 32001, Taiwan; [email protected]
2
Department of Physics, National Central University, 300 Zhongda Road, Zhongli, Taoyuan 32001, Taiwan
3
Xinjiang Astronomical Observatory, Chinese Academy of Sciences, Peopleʼs Republic of China
4
Institute of Astronomy and Astrophysics, Academia Sinica, No. 1, Sec. 4, Roosevelt Road, Taipei 10617, Taiwan
5
Department of Physics and Astronomy, University of Pennsylvania, 209 South 33rd Street, Philadelphia, PA 19125, USA
6
Center for Astrophysics, Harvard & Smithsonian, 60 Garden Street, Cambridge, MA 02138, USA
7
Graduate Institute of Space Science, National Central University, 300 Zhongda Road, Zhongli, Taoyuan 32001, Taiwan
Received 2021 August 17; revised 2022 January 13; accepted 2022 January 13; published 2022 March 11

Abstract
We present detections of stellar flares of Wolf 359, an M6.5 dwarf in the solar neighborhood (2.41 pc) known to be
prone to flares due to surface magnetic activity. The observations were carried out from 2020 April 23 to 29 with a
1 m and a 0.5 m telescope separated by nearly 300 km in Xinjiang, China. In 27 hr of photometric monitoring, a
total of 13 optical flares were detected, each with a total energy of  5 × 1029 erg. The measured event rate of
about once every two hours is consistent with those reported previously in radio, X-ray, and optical wavelengths
for this star. One such flare, detected by both telescopes on April 26, was an energetic event with a released energy
of nearly 1033 erg. The two-telescope light curves of this major event sampled at different cadences and exposure
timings enabled us to better estimate the intrinsic flare profile, which reached a peak of up to 1.6 times the stellar
quiescent brightness, that otherwise would have been underestimated in the observed flare amplitudes of about 0.4
and 0.8, respectively, with single telescopes alone. The compromise between fast sampling so as to resolve a flare
profile versus a longer integration time for a higher photometric signal-to-noise ratio provides a useful guidance in
the experimental design of future flare observations.
Unified Astronomy Thesaurus concepts: Stellar flares (1603)
Supporting material: animation

1. Introduction spectral types (brown dwarfs) have also shown surface activity
(e.g., Rutledge et al. 2000).
Solar flares are commonly observed surface phenomena,
Our target, Wolf 359 (GJ 406; CN Leo), at a heliocentric
attributable to acceleration of plasma during magnetic recon-
distance of 2.41 pc (Gaia Collaboration et al. 2021), is a known
nection near sunspots and active regions, that lead to sudden
eruptive-type red subdwarf (Kesseli et al. 2019). Previously, six
brightening observed in radio, optical, and X-ray wavelengths.
flare events were detected during 12.8 hr of monitoring in radio
While the detailed heating mechanism, i.e., how the magnetic
frequencies, equivalent to 47 events per 100 hr (Nelson et al.
energy is converted to gas kinetic energy is still unclear (Benz
1979). Extreme ultraviolet flare events have been reported to
& Güdel 2010), it is believed that conductive and radiative
occur at least daily (Audard et al. 2000), and major X-ray flares
processes are involved in the cooling phase. The solar flares are
have also been detected in this star (Liefke et al. 2010). Using
classified by the peak flux in soft X-rays, with the most
ground-based and Kepler/K2 observations combining long-
powerful being class X peaking at >0.1 erg s −1 cm−2. A
and short-cadence light curves of Wolf 359, Lin et al. (2021)
typical solar flare releases 1029–1032 erg. Rare, major flares,
derived a flare occurrence rate of once per day for events with a
which release a total energy more than ∼1033 erg, are linked to
total flare energies >1031 erg, and ten times per year for
coronal mass ejection events which influence space weather
superflares with released energies >1033 erg. Such an activity
and pose potential hazards to terrestrial environments.
Other stars, notably late-type dwarfs, being largely con- level is considered high even among known flaring M dwarfs.
Magnetic reconnection may not be limited to surface
vective are even more predisposed to flare activity and
extrusion as in the case of the Sun. The field lines may be
encompass a larger range of energy output, particularly the
linked to some intricate spin–orbit magnetospheric interaction
young ones with fast rotation rates (Feinstein et al. 2020).
with close-in companion stars, as in RS Canum Venaticorum or
Based on the K2 light curves of G- to M-type dwarfs, Lin et al.
BY Draconi type variables. It is known that newly born stars
(2019) conclude that later type stars have higher flare
may anchor their magnetic field to circumstellar disks
occurrence frequencies but generally with less energetic output.
(Feigelson & Montmerle 1999), with which the entwined field
M-dwarf flares with much shorter durations have been detected
lines are susceptible to reconnection and result in outbursts
in millimeter wavelengths (MacGregor et al. 2020). Later
(1036 erg) or in extended flaring loops (Hayashi et al. 1996;
Shibata & Magara 2011). Superflare events of red dwarfs are
Original content from this work may be used under the terms
of the Creative Commons Attribution 4.0 licence. Any further suspected to have similar interactions with orbiting giant
distribution of this work must maintain attribution to the author(s) and the title exoplanets (e.g., Klocová et al. 2017), though there is so far no
of the work, journal citation and DOI. definite supporting evidence. Wolf 359 is a fast rotator (Guinan

1
The Astronomical Journal, 163:164 (12pp), 2022 April Lin et al.

& Engle 2018; 2.72 day) and is known to host at least two profile, which would not have been possible otherwise with a
exoplanets (Tuomi et al. 2019). One of these, Wolf 359 c single data set.
(radius 0.1272 RJupiter) is hot and close in (0.018 au, orbital Images were processed by the standard procedure of bias,
period 2.88 day) suggestive of a possible spin–orbit tidal lock. dark, and flat-field corrections. Aperture photometry was then
In the flare star AD Leo, a periodicity of 2.23 day was inferred performed using SExtractor (Bertin & Arnouts 1996, 2010)
and attributed to stellar rotation (Hunt-Walker et al. 2012). with an adaptive aperture to measure the brightness of
However, Lin et al. (2021) found no flare timing in Wolf 359 with an aperture size of 4–8 pixels for the NOWT
synchronization with the planetary orbital phase in Wolf 359. images and of 8–9 pixels for the TAOS images. Figure 2 shows
Here we report on an optical monitoring campaign of an illustrative image taken by NOWT and by TAOS reported
Wolf 359. The star was monitored for one week in 2020 April here with the photometric aperture marked.
simultaneously with two telescopes. In addition to reaffirmation No standard star was observed and the brightness is
of the flare rate, with 13 events detected in 27 observing hours, referenced to that of the stellar quiescent state. In every case,
this paper focuses on one major flare observed by both two reference stars near Wolf 359 in the same image frame
telescopes, affording the possibility to derive the underlined were also measured to assess any variations due to the sky.
flare profile, whose amplitude would have been underestimated Figure 3 exhibits the NOWT light curves obtained during the
by any single light curve as a result of finite exposure time. campaign with the individual flare events marked. The major
While a flare is quantified by the total released energy event detected on 2020 April 26 by both NOWT and TAOS
(essentially scaled with the amplitude multiplied by the telescopes is presented separately in Figure 4. While the
duration), usually an event is recognized mainly by a brightness reference stars remained steady in brightness, Wolf 359
spike. Our study indicates how intrinsically moderate events experienced a brightening of ∼0.65 mag detected by NOWT,
could escape detection, and provides guidelines for proper and ∼0.38 mag detected by TAOS.
sampling specific to certain profiles in the experimental design. Table 1 summaries the photometric measurements used to
plot the light curves in Figure 3. Columns 1, 2, and 3 list,
respectively, the calendar date, telescope, and the Heliocentric
2. Observations and Data Analysis Julian Date (HJD) of the observation (middle of an exposure).
The remaining columns are magnitude and associated error of
2.1. CCD Imaging and Light-curve Extraction Wolf 359, and of the two reference stars.
The observations reported here were carried out from 2020 In a total of 27 data hours, 13 flare events were identified
visually in the light curves, in accord with the (non)variation of
April 23 to 29 simultaneously by the Nanshan One-meter
the reference stars at the same time. The flare parameters for
Wide-field Telescope (NOWT) in Xinjiang, and one of the
each event such as the peak amplitude, the rising and decay
TAOS telescopes, except for the night of April 24 for which the
timescales were derived, from which the integrated total energy
TAOS site was weathered out. The TAOS telescopes, each of
was computed.
f/2 50 cm, used to be installed at Lulin Observatory to catch
chance stellar occultation events by transneptunian objects 2.2. Flare Properties
(e.g., Alcock et al. 2003; Zhang et al. 2013). Two of the
original four TAOS telescopes were relocated in the spring of A flare profile is parameterized with (1) the epoch and
2020 to Qitai Station in Xinjiang, some 300 km from Nanshan, amplitude of the observed peak in the light curve, and relative
also operated by Xinjiang Astronomical Observatory. to the peak, (2) the rising timescale, and (3) the decline
The NOWT was equipped with an E2V back-illuminated timescale. Usually (2), signifying the energizing process is
CCD203-82 camera, with 12 micron pixels, spanning 1 13/ relatively fast, whereas (3), relevant to the cooling mechanism
pixel on the sky. For the data reported here, the NOWT drops off slower, typified by an exponential or a power-law
observed Wolf 359 in the R band for the first five nights, and in decay.
the V band for the rest two nights. The exposure time was 18 s, First, the observed light curve in unit of flux (count) f (t) is
with a dead time of approximately 12 s between exposures, subtracted of then divided by the detrended quiescent stellar
amounting to a cadence of ∼30 s. count (a linear fit to the flux away from an event) within the
The TAOS telescope was equipped with a Spectral spectral filter f0. To compute the released energy from such a
Instrument 800 camera with 13.5 micron pixels and a plate normalized light curve, Δf/f0 = ( f (t) − f0)/f0), we follow the
scale of 2 78/pixel. A custom-made filter was used which has equivalent duration method described by Gershberg (1972).
a flat-response in 500–700 nm approximately comparable to an The flare is approximated by a typical blackbody of effective
SDSS r ¢ filter. For the observing campaign of Wolf 359, the temperature Tflare of 9000 K (Mochnacki & Zirin 1980), thereby
exposure time was 45 s, with a dead time of ∼0.5 s. having a bolometric luminosity of
Wolf 359 has a high proper motion due to its proximity, with
μα = −3866.338 ± 0.081 mas yr−1 and L flare = sAflare Tflare
4
,
−1
μδ = −2699.215 ± 0.069 mas yr (Gaia Collaboration et al. where σ is the Stefan-Boltzmann constant, and Aflare is the flare
2021). Figure 1 displays its position in four epochs, three
area, related to the observed flare luminosity as
recorded by the Digital Sky Survey in years 1953, 1988, and
1995, whereas the last one was taken in 2020 reported in
this work.
¢ = Aflare
L flare ò Bl (Tflare) Rl dl.
No attempt was made to synchronize the shutter openings of
the two telescopes. The different cadences, hence sampling Here Bλ(T) is the Planck function and Rλ is the spectral
functions, of the two telescopes observing the same flare event response function for which only the filter response is
in turn provide the possibility to derive the underlined flare considered. The observed photospheric luminosity of a star of

2
The Astronomical Journal, 163:164 (12pp), 2022 April Lin et al.

Figure 1. The motion of Wolf 359. The first three images are from Digital Sky Survey taken in (a) 1953, (b) 1988, and (c) 1995, whereas the last image (d) was taken
in 2020 as a part of this work, all shown with J2000 coordinates.

Figure 2. Example images with Wolf 359 each marked with a red circle depicting the maximal photometric aperture size used in light-curve extraction: eight pixels for
NOWT (left image), and nine pixels for TAOS (right image).

3
The Astronomical Journal, 163:164 (12pp), 2022 April Lin et al.

Figure 3. The NOWT light curves of Wolf 359 (in black) and of the two reference stars (in gray). Each flare event detected is marked by a vertical red line. The
observations from April 23 to 27 were taken in the R band, whereas on April 28 and 29 they were in V. The sudden flux drop on April 27 around HJD 2458967.2 was
due to weather conditions, manifest also in the reference light curves. The major event on 2020 April 26 detected also with TAOS was analyzed separately and not
shown here.

radius R* is then Finally the total flare energy is estimated from the bolometric
luminosity of the star multiplied by the equivalent duration
L*¢ = pR*2 ò Bl (T*) Rl dl. Df
¢ to L ¢ is the amplitude,
The ratio of L flare
E flare = aL* ò f0
dt, (1 )
*
Df L¢ where α is the constant accounting for the correction for the
¢ =
Cflare = flare ,
f0 L¢ blackbody assumption and the filter response. Taking the stellar
*
temperature as 2900 K (Fuhrmeister et al. 2005), α of 0.11 for
and the area of the flare becomes the standard Johnson–Cousins R filter, and 0.05 for the V filter,
ò Bl (T*) Rl dl Equation (1), adopting log L* L  = -2.95  0.05 (Pavlenko
¢
Aflare = pR 2 Cflare , et al. 2006), leads to the derivation of the flare energy, Eflare
*
ò Bl (Tflare) Rl dl released in an event. Note that this method assumes the flare
with the flare luminosity being computed as, temperature to be constant throughout the event, and the
behavior of the flare is the same in all spectral bands.
4
1 ⎛ Tflare ⎞ ¢ ò Bl (T*) Rl dl Table 2 summarizes the parameters of the 13 events
L flare = L* ⎜ ⎟Cflare .
4 ⎝ T* ⎠ ò Bl (Tflare) Rl dl including the superflare detected simultaneously by two
telescopes. For all events reported here, the rising time is less
than ∼30 s, i.e., shorter than the cadence of each of the

4
The Astronomical Journal, 163:164 (12pp), 2022 April Lin et al.

Figure 4. The major event detected on 2020 April 26 by both NOWT (in red) and TAOS (in blue). For each telescope, the measured instrumental magnitudes of
Wolf 359 and of a comparison star are displayed to validate the variability. Typical photometric errors are about 0.01 mag so smaller than the sizes of the symbols.

telescopes, so was not derived. This rising/heating timescale is released energy (in unit of energy) or relative to stellar
contrasted to those of several minutes among solar-type flares photospheric luminosity (in unit of power). A superflare of
(Yan et al. 2021). In our analysis, the light curve then takes a solar-type stars releases 1033–1038 erg (Schaefer et al. 2000),
straight line from the stellar quiescent state, i.e., one data point which given a typical duration of ∼30 minutes (Yan et al.
prior to the peak. The date/time refers to the middle of the 2021) amounts to a ratio to stellar luminosity
exposure within which a flare occurred. The duration of an 10-4 L   L flare  101 L . An M-dwarf flare, on the other
event is estimated from one data point prior to the peak to hand, gives out a total energy, 1031–1034 erg, with the fast
where the light curve falls below the uncertainty in Δf/f0, rotators liberating more, up to 1035 erg (Lin et al. 2021). For the
typically 0.002 for NOWT and 0.01 for TAOS. The events reported here, the most energetic one has E ∼ 3 × 1031
determination of duration time therefore is somewhat sub-
erg within ∼25 minutes, hence with L* = 1.1 × 10−3 Le for the
jective, but serves to gauge the relative time length of an event.
star, leading to L flare L* » 0.5%. The two-telescope event is
The events had energies ranging from ∼3 × 1029 to ∼3 × 1031
erg, lasting for a couple of minutes to over 20 minutes. Our powerful, having the rising and exponential decay timescales
campaign was not long enough to catch more powerful, both less than about 30 s, with a peak amplitude comparable to
presumably rarer events. the stellar flux, we refer this major event as a superflare.
We note that the term “superflare” is applicable to solar
events, but not well defined for stars, whether it refers to total

5
The Astronomical Journal, 163:164 (12pp), 2022 April Lin et al.

Table 1
Photometric Data for Wolf 359 and Reference Stars

Date Telescope Epoch mW a smW a m1 b sm1b m2c sm2 c

HJD mag mag mag mag mag mag
2020 April 23 NOWT 2458963.1156348 14.7537 0.0016 16.1537 0.0032 16.1891 0.0033
2458963.1169079 14.7529 0.0019 16.1874 0.0040 16.1823 0.0041
2458963.1181114 14.7546 0.0019 16.1804 0.0040 16.1860 0.0040
2458963.1184702 14.7455 0.0019 16.1843 0.0040 16.1740 0.0041
2458963.1188406 14.7506 0.0019 16.1736 0.0039 16.1807 0.0040
2458963.1192110 14.7538 0.0019 16.1769 0.0040 16.1817 0.0040
2458963.1195697 14.7526 0.0019 16.1965 0.0040 16.1828 0.0040
2458963.1199400 14.7478 0.0019 16.1916 0.0040 16.1856 0.0041
2458963.1203104 14.7497 0.0019 16.1996 0.0041 16.1905 0.0041

Notes. Table 1 is published in its entirety in the machine-readable format. A portion is shown here for guidance regarding its form and content.
a
The variable mW is the magnitude of Wolf 359.
b
The variable m1 is the magnitude of reference star 1.
c
The variable m2 is the magnitude of reference star 2.

Table 2 A more energetic flare conceivably would take a longer time

Flare Event Parameters to dissipate. Because a flare profile may be complex (e.g., more
Date/Time Amplitude Energy Time Constant Duration
than a simple exponential or power-law decay, or multiple
HJD−2458960 (Δf/f0) (erg) (s) (s) flares in the light-curve segment), we correlate the duration of a
flare with energy, exhibited also in Figure 5. A log–log linear
3.1974805 0.09 5.68 × 1030 61 698 0.49  0.08
fit gives tflare µ Eflare . Maehara et al. (2015) related the
4.1942563 0.05 1.48 × 1030 61 534
4.2042208 0.02 4.46 × 1030 457 1248
magnetic reconnection timescale to the Alvén time and derived
5.1751989 0.03 3.47 × 1030 15 712 analytically for solar-type stars, tflare µ Eflare
1 3
, to be comparable
5.1937739 0.09 4.72 × 1030 12 772 to their observed slope of 0.39 ± 0.03. Such a duration-energy
6.1336424 0.82 3.28 × 1031 23 1515 relation is also observed in the active dM 4e dwarf GJ 1243
6.1336110a 0.41 3.26 × 1031 43 1319 (Hawley et al. 2014; of a slope about 0.4, reading from their
7.1918455 0.08 6.64 × 1030 34 745 Figure 10), and in other M dwarfs (Raetz et al. 2020). Our slope
8.1159728 0.04 1.08 × 1030 83 350 is marginally steeper but, given the incompleteness for weaker
8.1413180 0.03 2.87 × 1029 35 126 events, should be overestimated. If so, this indicates a similar
8.1512477 0.05 8.30 × 1029 98 254
energizing mechanism in M dwarfs as in solar-type stars.
9.1471391 0.04 2.79 × 1029 23 126
9.1739656 0.17 3.18 × 1030 35 603
9.1912327 0.18 3.38 × 1030 47 539
4. Intrinsic Flare Profile Derived from Two-telescope Light
Curves
Notes.
a
Measured also by TAOS; all others by NOWT only. The simultaneous detection of the same event with different
sampling functions allows us to break the parameter degen-
eracy and thereby to infer the underlined flare profile. By
varying profile parameters, the light curve that would have
3. The Flare Rate of Wolf 359
resulted for a given sampling function of a telescope is
The flare rate we detected, 13 flares in 27 hr, equivalent to compared with what was actually observed. The profile that
about 48 events per 100 hr, is consistent with literature results yields the least deviation in the chi-squared sense in the two
(e.g., Nelson et al. 1979; 47/100 in radio wavelengths). In light curves, is considered the “best” solution, closer to the
terms of energetics, Figure 5 presents the cumulative frequency “truth” than any individual observed light curve alone.
distribution of the flares of Wolf 359 listed in Table 2. Fitted The normalized light curves by the two telescopes are
with a power law, log n µ b log E , where ν is the flare displayed in Figure 6. Note that NOWT detects a relative
frequency distribution with energy greater than E (Hunt-Walker amplitude of ∼0.8, whereas TAOS detects an amplitude of
et al. 2012), the index β = −0.87±0.19 if the data at the ∼0.4. There is a caveat that, because of the different filters
lowest-energy end, incomplete in our experiment, are excluded. used, f0 differs for each telescope. Conceivably, the star itself is
Even with our limited number of events, spanning two orders red in color, but the flare should be relatively blue, but here we
of magnitude in energy, the distribution seems more complex assume the same f0 to normalize each light curve. The peak of
than a single power law, a conclusion that has been drawn for the flare was sampled differently: TAOS detected it at HJD of
this star and for active M dwarfs in general (Lin et al. 2021). By 2458966.133611, whereas NOWT detected it at HJD of
and large, Wolf 359 produces a flare as powerful as 1030 erg 2458966.1336424, with an offset of about 2.7 s.
approximately once every three hours. While a fast rotating M The sampling effects on an intrinsically continuous function
dwarf like Wolf 359 tends to produce frequent and powerful include the integration time (averaging the signal), dead time
flares (Lin et al. 2019), it is not known whether the boosted (no signal), and the lapse, i.e., the offset time between the peak
magnetic activity of our target is linked to its own rapid spin relative to the sampling window. It is this offset and finite
alone or to the close-in planet with an enlarged emission sampling time that smear off the peak of the flare and distort the
volume. shape of a light curve, accounting for different flare statistics

6
The Astronomical Journal, 163:164 (12pp), 2022 April Lin et al.

Figure 5. (a) Cumulative frequency (or occurrence timescale) distribution and (b) duration vs. flare energy of the Wolf 359 flares reported in this work. The dashed
line in (a) is not meant for a fit but for illustration only because the data seem more complex than a linear fit.

between the long-cadence versus short-cadence K2 light curves computed light curve according to a specific sampling function
(e.g., Raetz et al. 2020; Lin et al. 2021). A grid of event was compared with the actual observed one (“observed” minus
parameters were used to compute the simulated light curves “computed”, “O − C”) to evaluate the chi-squared value (sum
with steps of 0.01 in the peak amplitude and 1 s in decay of (O − C)/O).
timescale, chosen as the appropriate step parameters with Figure 7 presents the best-fit results, whose parameters are
extensive simulations. For the decay portion, different models summarized in Table 3. The double-exponential model, adding
were exercised: a single-exponential function ( A e-t t1), a one more degree of freedom in the fitting, gives an over-all
double-exponential function ( A e-t t1 + B e-t t2 ), and a better account than the single-exponential function of the
power-law falloff (A t− γ), where A and B are amplitudes, t is fading part of the light curves, judged by the residual χ2. This
time, τ1 and τ2 are correspondent exponential time constants, is consistent with the time-resolved flares of another eruptive
and γ is the power-law exponent index. In each case, the red dwarf, GJ 1243 (Davenport et al. 2014), and supports the

7
The Astronomical Journal, 163:164 (12pp), 2022 April Lin et al.

Figure 6. As in Figure 4 for the superflare observed in 2020 April 26, but now each light curve has been detrended and rescaled relative to the quiescent stellar
brightness, in flux/count unit instead of magnitude. The horizontal segments mark the sampling function for NOWT (in red) and for TAOS (in blue), with the filled
parts for exposure and open (or a dark bar for TAOS) for dead time intervals.

Figure 7. The observed (filled symbols, NOWT in red and TAOS in blue) vs. computed (open symbols) light curves for (a) a single-exponential, (b) a double-
exponential, (c) a power-law decay function with an instantaneous impulse rise, and (d) a power-law decay function with a finite-time rise (16 s). For each model, the
analytic function is represented as a black curve, and the residuals (O − C, the observed minus the computed) are also shown. For (c) the peak amplitude is ΔF/
F0 = 2.92, while in (d) it is ΔF/F0 = 1.63.

Table 3
Best-fit Model Parameters

Model A B τ1 τ2 γ Energy (erg) χ2

One-exponential 1.05 32 5.97 × 1030 0.014
Two-exponential 0.94 0.11 24 234 8.41 × 1030 0.007
Power law 2.92 0.7 1.88 × 1031 0.011
Power law 1.63 0.6 2.06 × 1031 0.004

8
The Astronomical Journal, 163:164 (12pp), 2022 April Lin et al.

Figure 8. Effect of observing cadence of a continuous (a) power-law, or (b) a double-exponential function.

notion of possibly more than one cooling mechanism (radiative 5. Implication for Flare Observations
and conductive; Benz & Güdel 2010).
A flare event, detected either visually or by an algorithm
For the power-law model, we present two cases, one with an
(e.g., to recognize in a light curve an abrupt rise followed by a
instantaneous impulse rise, and the other with a finite-time rise. few data points above quiescence) is characterized by the
The latter is more realistic, but our data could not constrain the amplitude and duration, from which the total energy is derive.
rising timescale. Therefore an exponential rise of 16 s is The fact that the true superflare event reaches to at least 1.6
adopted here as an example, for which a peak of 1.63 times of times of stellar brightness, while the observed light curves
the stellar brightness is required to fit the data, whereas for an peak, respectively, at 0.8 and 0.4 times, as demonstrated in this
instantaneous rise, the peak would have been 2.92 then falling work, manifests how the sampling function affects the
off faster (with a slightly larger value of γ). In general a power- amplitude of an observed flare. The experimental design to
law decay requires a higher peak than an exponential model, detect sporadic stellar flares hence pertains to an integration
leading to an elevated total flare energy. In our data, the time as short as possible so as to resolve the flare profile given
impulse plus the rapid decay portion of the light curve spans no the kind of flare events targeted for detection, while
more than a few data points. This means that a higher time commensurate with a sufficient signal-to-noise ratio. Figure 8
resolution is needed than reported here in order to distinguish plots how cadence affects the detected amplitude of the
one cooling function from another. superflare event reported here, for which the peak of 1.6 times

9
The Astronomical Journal, 163:164 (12pp), 2022 April Lin et al.

Figure 9. The computed light curves of the intrinsic flare profile with a peak of 1.6 times of the stellar brightness (Figure 7(d)) sampled at a cadence of 20 s (in blue),
2 minutes (in orange), and 20 minutes (in green); (a) with a zero phase lag, or (b) with a 0.5 phase lag.

of stellar brightness would be degraded quickly; e.g., with a 30 Wolf 359, but serves to demonstrate vividly the essence of fast
s integration the detected peak drops to less than 70%. This sampling. The lesson is while the total energy of a flare can be
applies only to the specific event detected on 2020 April 26 for reasonably estimated with a single data set, different samplings

10
The Astronomical Journal, 163:164 (12pp), 2022 April Lin et al.

Figure 10. An animation illustrating the effect of discrete sampling of a continuous flare profile. The animation runs from a cadence of 1–30 s to sample the profile in
Figure 9, i.e., of a peak of 1.63 with a phase shift of 0.5. The run time duration of the animation is 15 s.
(An animation of this figure is available.)

of a flare is necessary to derive the true profile in order to data, the peak and energy are underestimated as the result of
distinguish the heating and cooling processes. One improve- sampling by finite integration time with a phase lapse. A major
ment of our experiment, other than with larger telescopes to flare was detected simultaneously by two telescopes on 2020
afford faster cadences, is to measure the same event at different April 26, for which the underlying flare profile is estimated.
passbands, or better yet with spectroscopy, thereby diagnosing The profile parameters are model dependent, but the “true” flare
the temperature variation during the flare. amplitude might reach as high as 1.6 times of the quiescent
Stellar flare activity may be elevated if the field lines have an stellar flux, whereas the two telescopes detected a peak level of
external source to anchor to, be it a circumstellar disk, a 0.8 and 0.4, respectively, with the total released energy nearly
companion, or an exoplanet, increasing the magnetic filling four times as large.
factors hence the emitting volume than by surface starspot pairs
or coronal loops (Benz & Güdel 2010). In stars like Wolf 359 We thank the referee for constructive comments to improve
there may well be a combination of solar-type surface flares the quality of the paper. We are grateful to Xinjiang
plus inflated star-planet events. Long-term high-cadenced Astronomical Observatory for support in installation and
monitoring observations are called for to derive any possible operations of the TAOS telescopes in Qitai. The XAO authors
rotation or orbital periodicity. acknowledge the financial aid from National Natural Science
As in the case of Kepler/K2, the Transiting Exoplanet Foundation of China under grant U2031204. The relocation of
Survey Satellite (TESS; Ricker et al. 2015) provides data as the two TAOS telescopes was funded by the grant MOST 105-
useful for stellar flare research as in the primary science in 2119-M-008-028-MY3. The work at NCU is financially
exoplanets, particularly if complemented with ground-based supported in part by the grant MOST 109-2112-M-008-
observations of high sampling rates (e.g., Howard et al. 2020). 015-MY3.
TESS are monitoring some bright stars of three different
cadences: 20 s, 2 minutes, and 10 minutes. Figure 9 illustrates ORCID iDs
how these three sampling rates would have detected the April Wen-Ping Chen https://orcid.org/0000-0003-0262-272X
26 event. Here a peak amplitude of 1.6 times of the stellar Jinzhong Liu https://orcid.org/0000-0002-7420-6744
brightness is adopted (see Figure 7(d)) with a zero phase lag, Xuan Zhang https://orcid.org/0000-0002-5750-8177
i.e., with the peak coinciding with the start of the sampling Yu Zhang https://orcid.org/0000-0001-7134-2874
window, versus with a 0.5 phase lag. One sees that for this Andrew Wang https://orcid.org/0000-0003-4708-5964
particular flare only the shortest (20 s) cadence, similar to the Shiang-Yu Wang https://orcid.org/0000-0001-6491-1901
data reported here, can resolve the profile, with a phase- Matthew J. Lehner https://orcid.org/0000-0003-4077-0985
dependent amplitude of 0.6 or 0.8, respectively, but neither the Anli Tsai https://orcid.org/0000-0002-3211-4219
2 minutes (for selected targets) nor the 10 minutes (for the Chia-Lung Lin https://orcid.org/0000-0001-5989-7594
whole frame) cadence can. Figure 10 presents an animation of
discrete sampling of a continuous flare profile shown in References
Figure 9.
In summary, our photometric monitoring of the red dwarf Alcock, C., Dave, R., Giammarco, J., et al. 2003, EM&P, 92, 459
Wolf 359 in 2020 April detected, in 27 data hours, 13 flare Audard, M., Güdel, M., Drake, J. J., & Kashyap, V. L. 2000, ApJ, 541, 396
Benz, A. O., & Güdel, M. 2010, ARA&A, 48, 241
events with released energy in the range ≈3 × 1029–3 × 1031 Bertin, E., & Arnouts, S. 1996, A&AS, 117, 393
erg, consistent with the flare occurrence rate for this star Bertin, E., & Arnouts, S. 2010, SExtractor: Source Extractor, Astrophysics
reported previously in the literature. For any single-telescope Source Code Library, ascl:1010.064

11
The Astronomical Journal, 163:164 (12pp), 2022 April Lin et al.

Davenport, J. R. A., Hawley, S. L., Hebb, L., et al. 2014, ApJ, 797, 122 Lin, C. L., Ip, W. H., Hou, W. C., Huang, L. C., & Chang, H. Y. 2019, ApJ,
Feigelson, E. D., & Montmerle, T. 1999, ARA&A, 37, 363 873, 97
Feinstein, A. D., Montet, B. T., Ansdell, M., et al. 2020, AJ, 160, 219 Lin, C.-L., Chen, W.-P., Ip, W.-H., et al. 2021, AJ, 162, 11
Fuhrmeister, B., Schmitt, J. H. M. M., & Hauschildt, P. H. 2005, A&A, MacGregor, A. M., Osten, R. A., & Hughes, A. M. 2020, ApJ, 891, 80
439, 1137 Maehara, H., Shibayama, T., Notsu, Y., et al. 2015, EP&S, 67, 59
Gaia Collaboration, Brown, A. G. A., Vallenari, A., et al. 2021, A&A, 649, A1 Mochnacki, S. W., & Zirin, H. 1980, ApJL, 239, L27
Gershberg, R. E. 1972, Ap&SS, 19, 75 Nelson, G. J., Robinson, R. D., Slee, O. B., et al. 1979, MNRAS, 187, 405
Guinan, E. F., & Engle, S. G. 2018, RNAAS, 2, 1 Pavlenko, Y. V., Jones, H. R. A., Lyubchik, Y., Tennyson, J., & Pinfield, D. J.
Hawley, S. L., Davenport, J. R. A., Kowalski, A. F., et al. 2014, ApJ, 797, 121 2006, A&A, 447, 709
Hayashi, M. R., Shibata, K., & Matsumoto, R. 1996, ApJL, 468, L37 Raetz, S., Stelzer, B., Damasso, M., & Scholz, A. 2020, A&A, 637, A22
Howard, W. S., Corbett, H., Law, N. M., et al. 2020, ApJ, 895, 140 Ricker, G. R., Winn, J. N., Vanderspek, R., et al. 2015, JATIS, 1, 014003
Hunt-Walker, N. M., Hilton, E. J., Kowalski, A. F., Hawley, S. L., & Rutledge, R. E., Basri, G., Martín, E. L., & Bildsten, L. 2000, ApJL, 538, L141
Matthews, J. M. 2012, PASP, 124, 545 Schaefer, B. E., King, J. R., & Deliyannis, C. P. 2000, ApJ, 529, 1026
Kesseli, A. Y., Kirkpatrick, J. D., Fajardo-Acosta, S. B., et al. 2019, AJ, Shibata, K., & Magara, T. 2011, LRSP, 8, 6
157, 63 Tuomi, M., Jones, H. R. A., Butler, R. P., et al. 2019, arXiv:1906.04644
Klocová, T., Czesla, S., Khalafinejad, S., Wolter, U., & Schmitt, J. H. M. M. Yan, Y., He, H., Li, C., et al. 2021, MNRAS, 505, L79
2017, A&A, 607, A66 Zhang, Z. W., Lehner, M. J., Wang, J. H., et al. 2013, AJ, 146, 14
Liefke, C., Fuhrmeister, B., & Schmitt, J. H. M. M. 2010, A&A, 514, A94

12