ANDICAM I- and J-band monitoring of bright inner Galactic late-type stars

Abstract

Time-series photometry in the I and J bands of 57 inner Galactic late-type stars, highly probable red supergiant (RSG) stars, is presented here. 38% of the sample presents significant photometric variations. The variations in the I and J bands appear to be correlated, with ΔI ∝ ΔJ × 2.2, ΔI variations ranging from 0.04–1.08 mag, and ΔJ variations from 0.03–0.52 mag. New short periods (<1000 d) could be estimated for eight stars and range from 167–433 d. This work confirms that the sample is not contaminated by large-amplitude asymptotic giant branch stars. Furthermore, despite the large errors in distance, the period–luminosity diagram suggests that the sample is populating the same sequence as the known Galactic RSGs.

1 Introduction

Red supergiant (RSG) stars are late-type stars burning helium in the central core and with initial masses larger than 8–9 M_⊙. They are intrinsically bright at infrared wavelengths and, in principle, they can be detected at great distances even in the most obscured regions of the Galaxy. Unfortunately, the resemblance between RSGs and asymptotic giant branch (AGB) stars and the lack of precise distances complicate their detection.

In the last decade, the astronomical community has made a great effort to conduct medium- and high-resolution spectroscopic studies of RSGs. Nowadays, metallicity and temperature can be directly inferred from iron lines (e.g., Taniguchi et al. 2021), and some infrared lines have been found with strengths that correlate with the stellar luminosity (e.g., Messineo et al. 2021). The spectroscopic future looks promising with millions of spectra to be released by the Gaia, LAMOST, GALAH, and 4MOST surveys (e.g., de Jong et al. 2012; Gaia Collaboration 2021; Wu et al. 2021; Sharma et al. 2022). This will allow us to greatly improve the census of Galactic RSGs and to better learn how to classify them well and at minimum cost.

For improved Galactic distances, one must wait for the new releases of Gaia parallaxes and the pulsational periods of long-period variable stars. Periods for millions of stars will soon be available from the Gaia survey and the forthcoming LSST survey (Ivezić et al. 2019). Periods may yield distance estimates via a period–luminosity relation.

It has long been established that there is a correlation between the length of the period and the stellar luminosity of variable late-type stars. That such a relation also exists for variable RSG stars was noticed by Glass (1979) by analyzing seven RSGs in the LMC. The work in the LMC by Feast et al. (1980) confirmed it with an analysis of 24 RSGs. The RSG absolute K magnitude (M_K) vs. period (Per) relation falls above that of AGB stars, and their K-band amplitudes are typically smaller than 0.25 mag, while in AGB stars amplitudes range from 0.5–1.0 mag (Wood et al. 1983). Starting around the end of the 1990s, the description of long-period variables became more complicated with the discovery of several parallel sequences of pulsators in the LMC (Ita et al. 2004) and multi-frequencies detected in their light curves (e.g., Soszyński & Wood 2013). In the M 33 galaxy, Soraisam et al. (2018) detected a well defined period–luminosity relation for RSGs pulsating in the fundamental mode, and a parallel sequence, likely of first-overtone pulsators 0.3 mag brighter.

In the Milky Way, Pierce, Jurcevic, and Crabtree (2000) report that 12 RSGs in the Perseus OB1 association follow the same period–luminosity relations determined for RSGs in the LMC and M 33 in the R, I, K bands. Kiss, Szabó, and Bedding (2006) analyze the light curves of about 40 Galactic M-type RSGs covering 61 yr and found that about 40% of them have two periods, a short period (<1000 d) and a long secondary period (LSP) greater than 1000 d. The same sample of Galactic stars and others from the LMC and M 33 (220 stars) were analyzed by Chatys et al. (2019), to find that, for variable RSGs, a period–luminosity relationship exists for the short periods and it appears to be universal and independent of metallicity. About 52% of the Galactic RSG sample has short periods and 47% long periods.

In the last decade, a large number of new Galactic RSGs have been reported in the literature (e.g., Messineo et al. 2017; Dorda et al. 2018). At the current time, in view of the forthcoming surveys, it is advisable to aim for a more precise and well established characterization of already detected objects, as those will constitute the reference frames for the new stars to be classified. Especially for those located towards the densest and obscured regions of the Milky Way, variability studies are of primary importance to confirm their nature, estimate their distance, and, therefore, their luminosity class.

2 The sample

The 57 late-type stars observed with ANDICAM are taken from the sample of 94 stars analyzed by Messineo et al. (2016, 2017).

The sample of Messineo et al. (2016) comprises stars from the GLIMPSE I North survey brighter than K_s = 7 mag and with A|$_{\it K_{\rm s}}$| > 0.4 mag, which satisfies the infrared color criteria advised by Messineo et al. (2012). Messineo et al. (2017) show that large equivalent widths of the CO at 2.29|$\, \mu$|m [EW(CO) > 45 Å] and lack of water vapor absorption featured in 62% of that sample. The sample is, therefore, mostly made up of red supergiants (RSGs) (see the discussion in Messineo et al. 2017).

The subsample observed with ANDICAM is listed in table 1, and comprises 55 stars with broad EW(CO) (>45 Å) and little water, which Messineo et al. (2017) label “EW>,” plus two other stars (MZM7 and MZM21), which are labeled “CO” by having EW(CO) larger than 37 Å.

Unfortunately, as described in the recent work of Messineo et al. (2021),¹ the EW(CO) alone is not a good luminosity indicator and distance estimates and variability information remain essential to determine the luminosity class.

In the following, the stars are called late-type stars, as their classification is based on the EW of the CO band-heads at 2.29|$\, \mu$|m from low-resolution spectra. A more solid classification can be foreseen with high-resolution spectra. However, the broadness of the EW(CO), together with the lack of water absorption, current distance uncertainty, and the small amplitudes presented here, suggest that they are all consistent with being RSGs.

2.1 Infrared photometry and distances

The collection of infrared photometric measurements and bolometric magnitudes are presented in the works of Messineo et al. (2017) and Messineo and Brown (2019) and listed in table 1.

In Messineo et al. (2016), distances are determined by comparing the target extinction with the extinction curves of nearby clump stars, which are primary indicators of distance. In Messineo and Brown (2019), Gaia DR2 parallaxes are matched to the infrared sources and used to infer their distances. A revised version of this Gaia catalog with EDR3 parallaxes was made available by M. Messineo and A. G. A. Brown (2021).² Kinematic distances using the Gaia DR2 velocities are only possible for seven sources. The absolute magnitudes in K_s, M_K, are calculated as K_s–A|$_{\it K_{\rm s}}$|–DM, where K_s are the 2MASS K_s magnitudes and A|$_{\it K_{\rm s}}$| is the interstellar extinction, which is derived from the observed H − K_s and J − K_s, by assuming the intrinsic colors of Koornneef (1983) and the extinction coefficients by Messineo et al. (2005), which assumes an infrared power law with an index of −1.9. DM are the distance moduli from the Gaia EDR3 parallaxes.

Unfortunately, fractional parallactic errors are large for these types of cool sources and distances, with resulting errors in the distance moduli mostly between 0.8 and 1.0 mag. However, a comparison with the values inferred using the extinction is interesting because it further confirms the source location in the inner Galaxy, as shown in figure 1. The final Gaia release will narrow down these errors.

3 Observations

To monitor the fluxes of the 57 late-type stars, the infrared ANDICAM camera mounted on the 1.3 m telescope of Cerro Tololo in Chile was used. The telescope is operated by the SMARTS consortium, and a total of 28.7 nights were allocated to this program (2016B-0106) by the Telescope Allocation Committees (TAC) for the National Optical Astronomy Observatory. A CCD camera was attached for simultaneous optical observations.

4 J-band observations

The infrared array of ANDICAM consists of 1024 × 1024 pixels (512 × 512 pixels with a pixel scale of 0276 pix⁻¹ after a default 2 × 2 binning) and has a field of view of |${2{^{\prime }_{.}}4} \times {2{^{\prime }_{.}}4}$|⁠. We used the J-filter. The observing sequence is made with the classical seven dithered positions (the dither scale parameter was set to 40; i.e., each dithered position is within 20″ from the center, or, equivalently, the center moves within a box of 40″ × 40″). Eight exposures were taken for each star. Each exposure consists of two coadds and their integration times ranged from 4–30 s. This allowed us to obtain an excellent sky subtraction (using a robust mean with a 2σ clipping). Each frame was sky-subtracted and flat-fielded.

The world coordinate system of each J-band exposure was created using, as a reference system, the coordinates of bright 2MASS J point sources (J > 14 mag) detected by ANDICAM.

The peaks of the stellar counts vary from a few hundred to 3000, as recommended in the ANDICAM manual. A few frames were discarded because of saturation.

Typically, the full width half-maximum (FWHM) of the PSF ranged from 1″ to |${1{^{\prime \prime}_{.}}4}$|⁠. Aperture photometry was performed using the DAOPHOT (Stetson 1987) version available in the NASA IDL Astronomy User’s Library (Landsman 1993). Aperture photometry was performed on each frame (at each dithered position) as the targeted stars are among the brightest in the field and the derived magnitudes were averaged.

4.1 J-band flux calibration

Photometric calibration was performed in a relative manner. The measured magnitudes were registered on those measured in the first (reference) epoch. The absolute calibration was done using field stars with known 2MASS J-band magnitudes brighter than 13 mag. The calibrators were visually inspected and a few stars with a larger dispersion discarded (see figure 2). In a given field, the average standard deviation of the calibrator ANDICAM J magnitudes ranges from 0.001–0.044 mag, as listed in table 2.

The average J-band magnitudes of the targets range from 6.9–11.6 mag, as listed in table 2.

5 I-band observations

The ANDICAM CCD detector was also used for simultaneous I-band observations, taken in staring mode. The detector with 1024 × 1024 pixels covers a field of view of |${6{^{\prime }_{.}}33} \times {6{^{\prime }_{.}}33}$|⁠. For each observation, six to seven frames were acquired. The integration time was set to 8 s, and we reached a peak of 50 counts on a 16 mag star with a seeing of |${1{^{\prime \prime}_{.}}1}$|⁠.

The CCD frames are distributed by the observatory after corrections for bias and flat-field. The individual frames were combined with a 10σ clipping.

For the astrometric calibration of each observation, DENIS data points brighter than I = 15 mag (or 13 mag in the denser field) were overlaid on the CCD image. The absolute astrometric solution is ≈|${0{^{\prime \prime}_{.}}15}$| accurate. For the 12 missing fields, 2MASS J-band data points were used.

Aperture photometry was performed using the DAOPHOT (Stetson 1987) version available in the NASA IDL Astronomy User’s Library (astron). For each field, the FWHM was measured and the aperture radius set to FWHM*0.5+1.5 pix, and the sky annulus taken from FWHM*0.5+2 to FWHM*0.5+4 pix.

5.1 I-band flux calibration

The target magnitude calibration was done in a relative manner by using field stars with known flux; the absolute calibration only affects the global zero point, but not the magnitude variations.

For every epoch, the extracted catalog of point source was cross-correlated with that of the reference epoch (usually the first epoch) and flux calibrated. Most of the observed fields (57 minus 12) were covered by the DENIS survey (Epchtein et al. 1994). The DENIS observations in the Gunn-I filter saturate at around 10 mag and have a 3σ detection limit at 19 mag, e.g., as described in Messineo et al. (2004), as shown in figure 3. Therefore, DENIS fully covers the range of interest, as the I magnitudes of the detected targets range from 10.8–17.5 mag. DENIS point sources with 10 < I magnitude < 13 mag were used to determine the night zero point; the median of the differences between the instrumental magnitudes and the DENIS I magnitude was adopted.

ANDICAM mounts a KPNO-I filter, while the DENIS survey made use of a Gunn-I filter, as shown in figure 4. The average difference between the I-band magnitudes of the standard stars by Landolt (2009) in the Johnson–Kron–Cousins system and the DENIS I-band magnitudes is 0.015 mag with σ = 0.045 mag. This offset was not applied.

The average difference between the I-band magnitudes by Landolt (2009) and the SDSS I-band magnitudes is 0.50 mag with σ = 0.10 mag when 16 < I band < 13 mag. However, unfortunately, only three target fields were covered by the SDSS survey DR12 (Alam et al. 2015). The average difference between the I-band magnitudes by Landolt (2009) and the Pan-STARRS I-band magnitudes (mean PSF AB) from 16–13 mag is 0.46 with σ = 0.09 mag. The standard stars by Landolt (2009) are not covered by the Galactic plane VPHAS+ survey (Drew et al. 2014). However, for most of the observed fields, I magnitudes are available from VPHAS+, and average shifts from 0–0.3 mag are measured between the DENIS and VPHAS+ I (Vega system) magnitudes. The bright tails of stars detected in ANDICAM were saturated in the Pan-STARRS and VPHAS+ I-band catalogs.

The absolute photometric calibration was refined by analyzing the time behaviors of field stars with DENIS 13 < I < 10 mag, and retaining as calibrators those stars with smaller ANDICAM time variations [std_cal(I)_j], i.e., with std_cal(I)_j values within the field mean 〈std_cal(I)〉 plus 1.5 times their dispersion; std_cal(I)_j varies from 0.013–0.048 mag.

For 12 fields calibrated with Pan-STARRS (because of not being covered by DENIS), stars from 15.5 < I < 13 mag were used. An example of the adopted calibrators is illustrated in figure 5. The computed average I-band magnitudes of the targets are listed in table 2, along with some parameters (e.g., internal standard deviations of calibrator magnitudes and external standard deviations of field stars detected in the I band by ANDICAM as well as by the DENIS, VPHAS+, and Pan-STARRS surveys) to illustrate the uncertainties on the absolute calibration.

Only fields where the targeted stars were detected in at least two epochs were analyzed further (15 stars were below the detection threshold).

6 J- and I-band variations

In order to assess the existence of significant variations in the brightness of the targeted stars, the σ of the target J magnitudes, σ_*(J), are compared with the σ of the calibrator stars; indeed, the targets are among the brightest stars detected in the J band. In the I band, the targets are faint; therefore, σI_circa is calculated with field stars at the target I magnitude, and compared with σ_*(I).

In the J-band analysis, three targets (MZM42, MZM56, and MZM85) have σ_*(J) values larger than 3.5 times the 〈std_cal(J)〉 of their calibrators, and larger than their quoted error bars; this calculation permits the detections of variations larger than 0.07 mag, because the mean of the variations of the calibrators is 0.025 mag. In the I-band data, six stars appear significantly variable (>3.5 〈σI_circa〉) (MZM40, MZM42, MZM50, MZM75, MZM84, and MZM85). The mean σI_circa is 0.13 mag.

Despite the small numbers of detected variables, correlated trends appear in several I-band and J-band light curves, as shown in figures 6 and 7. 16 targets (out of the 42 detected in both bands) have a Pearson correlation coefficient between I-band and J-band measurements above 50%, and the σ_*(I) or σ_*(J) of their I-band or J-band measurements is larger than twice the σ of corresponding field stars with similar magnitudes (〈std_cal(J)〉 or 〈σI_circa〉). In this latter calculation, 38% of the sample shows variations, MZM06, MZM09, MZM10, MZM16, MZM20, MZM22, MZM33, MZM34, MZM40, MZM42, MZM50, MZM59, MZM75, MZM83, MZM84, and MZM85. The use of multi-wavelength data improves the detection of variables and the correlated patterns make it more solid and reliable. When using combined IJ bands, the variable detection threshold in a single band is lowered (from 3.5σ to 2σ) to detect more variables.

The Gaia variables listed by Mowlavi et al. (2021) were detected with a detection threshold of 0.06 mag in the G band. It appears that variables with amplitudes in the G band larger than 0.08 mag are all retrieved. There are eight ANDICAM variables with estimated G variations below the 0.06 mag threshold for variability. While the Gaia variables retrieved by ANDICAM have average amplitudes in the I band of 0.27 mag (0.10 mag in the J band) and are mostly >0.21 mag, variables found by ANDICAM, but not in Mowlavi et al. (2021), have average I-band amplitudes of 0.16 mag (0.09 in the J band) and are mostly <0.21 mag. We conclude that the detection of variables is complete for I-band amplitudes larger than 0.21 mag.

While G-band amplitudes are determined for stars with a wide range in the G band (from 10–20 mag), the variables reported in AAVSO are brighter than a G magnitude of 14.5.

24 (out of 42) stars, 57% of the sample, have J and I variations below the adopted detection threshold (2σ).

The standard deviations of the targets and calibrators are listed in table 2.

6.1 Periodograms

Periods are obtained as the highest power in the Lomb–Scargle periodogram, which is an elaborated version of the classical periodogram for unevenly sampled data (Scargle 1982; Horne & Baliunas 1986). The scargle.pro routine in the NASA IDL Astronomy User’s Library was used (Landsman 1993). The analysis of the I-band Lomb–Scargle periodogram yields periods for eight stars (out of the 16 variables in section 6), which are listed in table 3. Only periods with power levels corresponding to false alarm probabilities (the probability of the periods being false) below 50% were considered. The power of MZM83 corresponds to a false alarm probability (FAP) of 43.6%, while the probability remains below 25% for the other seven stars. Their periods range from 167.6–433.3 d.

A higher FAP brings a better census of variables; however, it may bring false cases. A FAP value of 10% is commonly used in time-series photometry, as mentioned in the work of VanderPlas (2018). FAP values of 50% are also found in the literature, for example, in the spectroscopic time-series analysis of Cincunegui, Díaz, and Mauas (2007). Besides the Scargle method, a fitting of the light curve was performed and the sinusoidal curves are evident.

The eight periodograms and phased light curves are shown in the Appendix. In conclusion, 38% of the 42 targets with ANDICAM IJ detections are found to be variable, and 19% show periodicity.

6.2 Amplitudes of the variations

16 late-type stars (out of 42 detected in both IJ bands) appear to have significant light variations when compared with surrounding field stars, as described in section 6. The measured variations range from 0.038–1.076 mag in the I band, and from 0.031–0.517 mag in the J band, and are listed in table 2. As mentioned in section 6, the variations in the I and J bands appear correlated. For pairs of simultaneously taken I and J magnitudes of the 16 variables, the I variations, ΔI, are plotted against the J variations, ΔJ, in figure 6. The ΔI values are 2.2 ± 0.1 times larger than the ΔJ values, and 1.292 ± 0.004 larger than those in the I − J colors, Δ(I − J).

For the eight stars found to be periodic variables, the least-squares fitting of data by sinusoidal curves of the type |${I(t)} = \frac{\Delta I}{2} \sin [2 \pi (\frac{t}{\rm Per} + \omega _{\rm o} )] + { \langle {I}\rangle }$| was performed. For each curve, three parameters were estimated; |$\frac{\Delta I}{2}$| is the semi-amplitude of the pulsation (i.e., the difference in absolute value between the mean value and the minimum or maximum deviation), ω_o is the value at zero phase (at the maximum), and 〈I〉 is the mean magnitude of the pulsator. The |$\frac{\Delta I}{2}$| values are listed in table 3, and shown in the Appendix.

7 Previously known variables

For nine targets, amplitudes are reported in the AAVSO International Variable Star database (VSX)(Watson et al. 2006), four of which were not detected as significantly varying in this work. They range from 0.10–0.64 mag.

In the work of Mowlavi et al. (2021), there are estimates of the G-band amplitudes for 15 of the ANDICAM targets (10 of which were classified as variable in this work); they are based on the G-band photometric errors and range from 0.06–0.39 mag. These stars are not listed in the Gaia DR2 table of long-period variables (LPV). This brings to 59% the fraction of variable targets and known amplitudes.

For seven targets, periods from 109–457 d are listed in the International Variable Star Index VSX (Watson et al. 2006), four of which are not detected as significantly varying in this work. MZM83 has a period of 148.5 d in the AAVSO catalog; in the ANDICAM data a low power peak appears at 279 d with a high probability (43%) of being false. The other six AAVSO stars have no periods detected in ANDICAM.

This increases from 19% to 33% the number of targets with known periods.

8 M_K vs. period of known Galactic RSGs

To verify the newly obtained periods, the M_K vs. period diagram of the targets is compared with that of well studied variable RSGs in figure 8. For known variables, the periods are taken from Chatys et al. (2019) and the M_K values are those calculated by M. Messineo and A. G. A. Brown (2021)² with EDR3 Gaia distances and are obtained as described by Messineo and Brown (2019). The distances adopted are based on EDR3 Gaia parallaxes and are the geometric distances by Bailer-Jones et al. (2021). The short periods (<2000 d) appear to describe a clear sequence in this plane. The sequence appears much improved with the use of EDR3 Gaia distances, and has σ = 0.32 mag. The sequence is consistent with the fit made for RSGs in Perseus OB1 by Pierce, Jurcevic, and Crabtree (2000), as well as with the fit obtained for RSGs in M 31 by Soraisam et al. (2018).

At this stage, in the M_K–period plane, the distribution of the 14 ANDICAM targets with determined periods appears consistent with that of known RSGs, within errors.

9 Summary and remarks

ANDICAM observations of a sample of late-type stars were obtained over a 1054 d time period (2.9 yr). 57 bright late-type targets from the sample of Messineo et al. (2017) were observed in the J bands and 42 were detected in the simultaneously taken I-band images. It appears that at least 38% of the ANDICAM sample is made of variable stars, 47% when including additional information in AAVSO, or 59% when also considering the Gaia amplitude estimates reported by Mowlavi et al. (2021). Furthermore, 19% have detected periodic behaviors in the ANDICAM data and 33% when including the AAVSO periods.

The targeted late-type stars have average ANDICAM 〈J〉 from 6.85–11.65 mag and ANDICAM 〈I〉 from 10.82–17.49 mag. However, despite their faintness in the I band, the I band is more suitable than the J band for detecting variables. Indeed, the magnitude variations measured in the I band are correlated with those seen in the J band and are a factor of 2.2 larger. In the I band, the differences between the minimum and maximum magnitudes of each star range from 0.04–1.08 mag, and from 0.03–0.52 mag in the J band.

The ANDICAM data presented here indicate that the I magnitudes and I − J colors of the targets vary in a correlated manner and that ΔI ∝ 1.29Δ(I − J). In pulsating large-amplitudes stars, the amplitudes are known to decrease with increasing wavelength, as the envelope expands and cools down every radial pulse (Reid & Goldston 2002). For example, in Mira stars the V-band amplitudes can even be 8 mag (Reid & Goldston 2002), and the J-band amplitudes are about 1.6 times larger than those measured in the K_s band (Messineo et al. 2004). This effect is caused by a changing opacity in the stellar atmosphere (Reid & Goldston 2002), due to molecular bands. When a late-type star pulses, it expands and cools down; it reaches its minimum light when the atmospheric opacity is at the maximum value (Reid & Goldston 2002). When a long-period variable pulses, it cyclically changes its spectral type. As mentioned by Pierce, Jurcevic, and Crabtree (2000), for RSGs the measured light variations are found to be a function of the filter used. The I-band spectrum is dominated by TiO molecular bands, which are extremely sensitive to temperature variations, while the J-band spectrum is not affected by TiO bands. The observed correlation between the ANDICAM color variations, Δ(I − J), and the magnitude variations are, therefore, consistent with the expected behavior for radial pulsation. For normal (static) giants and RSGs, a change of spectral type from M2–M5 (M3.5 ± 1.5) corresponds to a color change Δ(I − J) ≈ 0.26 mag (Johnson 1966). Variable RSGs could have larger color changes, due to their larger radii and the large convective cells present in their turbulent atmospheres.

As shown by Kiss, Szabó, and Bedding (2006) and Chatys et al. (2019), variable RSGs are of late-type (M-type). A broad correlation is found between the stellar luminosity and the R-band amplitudes of 120 RSGs (Soraisam et al. 2018). In turn, as average stellar temperatures decrease with increasing luminosity, a positive correlation also exists between the stellar temperatures and amplitudes (Messineo & Brown 2019, 2020). The sample analyzed here is small and no clear trend is observed between luminosity and amplitude. However, the small amplitudes are in agreement with the supergiant classification. Indeed, the most luminous AGB stars (e.g., super-AGBs) are expected to be large-amplitude pulsators (O’Grady et al. 2021).

New periods are determined for eight late-type stars and range from 167–433 d, and seven periods are available from AAVSO. The time baseline of the ANDICAM data does not allow us to check for long secondary periods (LSP), which are typically longer than 2000 d. In the Galaxy, LSP periods are seen in 50% of RSGs (Kiss et al. 2006; Chatys et al. 2019).

The sample does not contain large-amplitude variables, which are bright luminous AGBs. For the targets found to be periodic variables, their distribution in the period–luminosity diagram suggests that they are RSGs, consistent with the work of Messineo et al. (2017). However, distance errors are still large and it is better to re-check with the final Gaia parallaxes. Gaia will also release spectra and G-band light curves.

Due to the significant uncertainties in their distances, time-series measurements appear to offer a promising means of assessing the stellar luminosity class of obscured inner Galactic objects.

Funding

This work was partially supported by the National Natural Science Foundation of China (NSFC- 11773025), and USTC start-up grant KY2030000054.

Acknowledgements

An incredible amount of work was done by the CTIO/SMARTS team. This work is dedicated to a peaceful world. This work has made use of data from the European Space Agency (ESA) mission Gaia (http://www.cosmos.esa.int/gaia), processed by the Gaia Data Processing and Analysis Consortium (DPAC, http://www.cosmos.esa.int/web/gaia/dpac/consortium). Funding for the DPAC has been provided by national institutions, in particular those institutions participating in the Gaia Multilateral Agreement. This publication makes use of data products from the Two Micron All Sky Survey, which is a joint project of the University of Massachusetts and the Infrared Processing and Analysis Center/California Institute of Technology, funded by the National Aeronautics and Space Administration and the National Science Foundation. DENIS is a joint effort of several institutes predominantly located in Europe. It has been supported mainly by the French Institut National des Sciences de l’Univers, CNRS, and French Education Ministry, the European Southern Observatory, the State of Baden-Wuerttemberg, and the European Commission under networks of the SCIENCE and Human Capital and Mobility programs, the Landessternwarte, Heidelberg and Institut d’Astrophysique de Paris. The Pan-STARRS1 Surveys (PS1) and the PS1 public science archive have been made possible through contributions by the Institute for Astronomy, the University of Hawaii, the Pan-STARRS Project Office, the Max-Planck Society and its participating institutes, the Max Planck Institute for Astronomy, Heidelberg and the Max Planck Institute for Extraterrestrial Physics, Garching, the Johns Hopkins University, Durham University, the University of Edinburgh, the Queen’s University Belfast, the Harvard-Smithsonian Center for Astrophysics, the Las Cumbres Observatory Global Telescope Network Incorporated, the National Central University of Taiwan, the Space Telescope Science Institute, the National Aeronautics and Space Administration under Grant No. NNX08AR22G issued through the Planetary Science Division of the NASA Science Mission Directorate, the National Science Foundation Grant No. AST-1238877, the University of Maryland, Eotvos Lorand University (ELTE), the Los Alamos National Laboratory, and the Gordon and Betty Moore Foundation. VPHAS+ is based on observations made with ESO Telescopes at the La Silla or Paranal Observatories under programme ID(s) 177.D-3023(B), 177.D-3023(C), 177.D-3023(D), 177.D-3023(E). This research has made use of the VizieR catalog access tool, CDS, Strasbourg, France, and SIMBAD database. This research has made use of NASA’s Astrophysics Data System Bibliographic Services (ADS).

Phased light curves

In figure 9, the phased light curves are shown. Periods are estimated for eight variables.

Footnotes

References