EDP Sciences
Free Access
Issue
A&A
Volume 592, August 2016
Article Number A120
Number of page(s) 17
Section Catalogs and data
DOI https://doi.org/10.1051/0004-6361/201628864
Published online 09 August 2016

© ESO, 2016

1. Introduction

Galactic globular clusters are interesting stellar systems in astronomy since they are fossils of early Galaxy formation and evolution. This makes them excellent laboratories in a wide range of topics from stellar evolution to cosmology, from observations to theory.

NGC 6715 (M 54) was discovered on July 24, 1778 by Charles Messier1. The cluster is in the Sagittarius dwarf spheroidal galaxy at a distance of 26.5 kpc from our Sun and 18.9 kpc from our Galactic centre. It has a metallicity [Fe/H] = 1.49 dex and a distance modulus (mM)V = 17.58 mag. The magnitude of its horizontal branch is VHB = 18.16 mag (2010 version Harris 1996). The stellar population and morphology of NGC 6715 have attracted the attention of several studies. Although NGC 6715’s position is aligned with the central region of the Sagittarius dwarf spheroidal galaxy, recent studies have shown that this cluster is not the nucleus of the mentioned galaxy (Layden & Sarajedini 2000; Majewski et al. 2003; Monaco et al. 2005; Bellazzini et al. 2008). Carretta et al. (2010) found that the metallicity spread in this cluster is intermediate between smaller-normal Galactic globular clusters and metallicity values associated with dwarf galaxies. It has a very peculiar colourmagnitude diagram with multiple main sequences, turnoff points, and an extended blue horizontal branch (HB) (Siegel et al. 2007; Milone 2015). Rosenberg et al. (2004) found that this cluster hosts a blue hook stellar population in its blue HB. It is thought that NGC 6715 might host an intermediate mass black hole (IMBH) in its centre (Ibata et al. 2009; Wrobel et al. 2011) as well.

Oosterhoff (1939) found that globular clusters can be classified into two groups based on the mean periods and the number ratios of their RR0 and RR1 stars. These are the Oosterhoff I (OoI) and Oosterhoff II (OoII) groups, and this is known as the Oosterhoff dichotomy. It was also found that the metallicity of the clusters plays an important role in this classification (Kinman 1959) and that it could be related with the formation history of the Galactic halo (see e.g. discussions in Lee & Carney 1999; Catelan 2004, 2009; Smith et al. 2011; Sollima et al. 2014).

The first Oosterhoff classification for NGC 6715 was done by Layden & Sarajedini (2000) by comparing the amplitudes and periods of 67 RR Lyrae in this cluster with the period-amplitude relation obtained for M3 (OoI) and M9 (OoII). NGC 6715 follows the relation defined by M3 and this implies that it is an OoI cluster. They also found that the mean periods of the RR Lyrae also agree with the OoI classification. Similarly, Sollima et al. (2010), in their study of variable stars in NGC 6715, discovered a new set of 80 RR Lyrae, and they used 95 RR0 and 33 RR1 (after excluding non-members or problematic stars) to find that this cluster shares some properties of both OoI and OoII clusters. For instance, the average period of its RR0 and RR1 stars was found to be intermediate between the values for the OoI and OoII classifications. We note that Catelan (2009) also classified this cluster as an intermediate Oosterhoff type based on its position in the PRR0-metallicity diagram.

Several time-series photometric studies focused on the variable star population (Rosino 1952; Rosino & Nobili 1958; Layden & Sarajedini 2000; Sollima et al. 2010; Li & Qian 2013), but owing to the high concentration of stars in its core, many variables have previously been missed because of blending. NGC 6715 is very massive and it is among the densest known globular clusters (Pryor & Meylan 1993). As a result, NGC 6715 also caught our attention as an excellent candidate for our study of globular clusters using electron-multiplying CCDs (EMCCDs) and the shift-and-add technique (Skottfelt et al. 2013, 2015a; Figuera Jaimes et al. 2016).

Section 2 summarises the observations, the reduction, and photometric techniques employed. Section 3 explains the calibration applied to the instrumental magnitudes. Section 4 contains the technique used in the detection and extraction of variable stars. Sections 5 and 6 show the methodology used to classify variables and the colourmagnitude diagrams employed, respectively. Section 7 presents our results and Sect. 8 discusses the Oosterhoff classification of this cluster. Our conclusions are presented in Sect. 9.

2. Data, reduction, and photometry

EMCCDs, also known as low light level charge-coupled devices (L3CCD; see e.g. Smith et al. 2008; Jerram et al. 2001), are conventional CCDs that have an extended readout register where the signal is amplified by impact ionisation before they are read out. Hence the readout noise is negligible when compared to the signal and very high frame-rates become feasible (10100 frames/s). This provides an opportunity to compensate for the blurring effect of the turbulence in the atmosphere. By shifting and adding the individual frames appropriately, it is possible to construct much higher resolution images than is possible using conventional CCD imaging from the ground. Furthermore, the dynamic range of the stacked images is greatly increased and the saturation of bright stars is therefore not an issue except for the very brightest stars in the sky. However, the main drawback with this technique is that the process, whereby the signal is amplified, also increases the photon noise component in the images by a factor of when compared to conventinal CCD images. We also need to be aware that EMCCD imaging data need to be calibrated in a different way to conventional CCD data. Several studies where EMCCDs have been used can be found in the literature, e.g. high-resolution imaging of exoplanet host stars in the search for unseen companions (Southworth et al. 2015; Ciceri et al. 2016; Bozza et al. 2016; Street et al. 2016; Evans et al. 2016) photometric and astrometric measurements of a pair of very close brown dwarfs (Mancini et al. 2015), and time-series photometry of crowded globular cluster cores aiming to complete the census of variable stars (Skottfelt et al. 2015a; Figuera Jaimes et al. 2016).

The data presented in this paper are the result of EMCCD observations performed over three consecutive years: 2013, 2014, and 2015 in April to September each year. The 1.54 m Danish telescope at the ESO Observatory in La Silla, Chile was used with an Andor Technology iXon+897 EMCCD camera, which has a 512 × 512 array of 16 μm pixels, giving a pixel scale of per pixel and a total field of view of ~45 × 45 arcsec2.

For this research, the EMCCD camera was set to work at a frame-rate of 10 Hz (ten frames per second) and an EM gain of 300e/photon. The camera was placed behind a dichroic mirror which works as a long-pass filter. Taking the mirror and the sensitivity of the camera into consideration, it is possible to cover the wavelength range from 650 nm to 1050 nm. This is roughly a combination of SDSS i′ + z filters (Bessell 2005). More details about the instrument can be found in Skottfelt et al. (2015b). The total exposure time employed for a single observation was ten minutes, which means that each observation is the result of shifting-and-adding 6000 exposures. The resulting point spread function (PSF) full width at half maximum (FWHM) in the reference image employed in the photometric reductions (see below) was 0.44′′, while the mean standard seeing calculated with the values reported at the observatory station during April to September 2013, 2014, and 2015 were 1.0′′, 1.2′′, and 0.9′′, respectively.

Table 1

Time-series I-band photometry for all known and new variables in the field of view covered in globular cluster NGC 6715.

In Fig. 1, histograms of the number of observations per night during each year are shown. Data in the left-hand panel correspond to 2013, data in the middle panel to 2014, and data in the right-hand panel to 2015. We aimed to always take two observations per night, although it was not always possible because of weather conditions or time slots needed for other projects, as is the case of the monitoring of microlensing events carried out by the MiNDSTEp consortium.

thumbnail Fig. 1

Histograms with the number of observations per night for globular cluster NGC 6715. Left: 2013. Middle: 2014. Right: 2015.

Open with DEXTER

Bias, flat-field, and tip-tilt corrections were performed using the procedures and algorithms described in Harpsøe et al. (2012). In particular, the tip-tilt correction allows high-resolution stacked images to be created, as described in detail in Skottfelt et al. (2015a), Figuera Jaimes et al. (2016). Briefly, the method uses the Fourier cross-correlation theorem where, for each bias- and flat-corrected exposure Ik(i,j) within an observation, we calculate the cross-correlation image between Ik(i,j) and an average of 100 randomly chosen exposures. The peak in the cross-correlation image gives the appropriate shift to correct the tip-tilt error. We then stack the shifted exposures according to batches grouped by image quality (a measure of image sharpness), to create ten-layer image cubes. The best quality layers from each cube (i.e. the sharpest layers) are extracted from all of the available observations to create a high-resolution reference image for subsequent difference image analysis (DIA). Finally, each ten-layer cube is stacked to create a single image that corresponds to each observation.

To extract the photometry in each of the stacked images, we used the DanDIA2 pipeline (Bramich 2008; Bramich et al. 2013), which is based on difference image analysis (DIA; Alard & Lupton 1998; Alard 2000). The pipeline works by aligning all images to the reference image, solving for a set of convolution kernels that were modelled as discrete pixel arrays, and subtracting the convolved reference image in each case to create a set of difference images. Stars are detected on the reference image and their reference fluxes fref (ADU/s) are measured using PSF photometry. Difference fluxes fdiff(t) (ADU/s) for each star detected in the reference image are measured in each of the difference images by optimally scaling the PSF model for the star to the difference image. The light curve for each star in instrumental magnitudes mins are built as shown in Eq. (1): (1)where ftot(t) is the total flux in ADU/s defined as (2)The quantity p(t) is the photometric scale factor used to scale the reference frame to each image, as part of the kernel model explained in Bramich (2008). An electronic table with photometric measurements and fluxes for all the variable stars presented in this work is available through the CDS3 database with the format illustrated in Table 1.

2.1. Astrometry and a finding chart

To create a reference image with the astrometric information for each star in the field covered in NGC 6715, we used the celestial coordinates available in the ACS Globular Cluster Survey4 (see Anderson et al. 2008), which were uploaded for the field of the cluster through GAIA (Graphical Astronomy and Image Analysis Tool; Draper 2000). An (x,y) shift to match their respective stars in our reference image was applied. Stars lying outside the field of view and those without a clear match were removed and the (x,y) shift was refined by minimising the squared coordinate residuals. A total of 305 stars over the entire field was used to guarantee that the astrometric solution applied to the reference image considered enough stars. The radial root mean square (rms) scatter obtained in the residuals was ~0.̋028 (~0.3 pixels). This astrometrically calibrated reference image was used to produce a finding chart for NGC 6715 on which we marked the positions and identifications of all variable stars studied in this work (Fig. 5). Finally, a table with the equatorial J2000 celestial coordinates of all variables is given in Table A.1.

3. Photometric calibration

The photometric transformation of instrumental magnitudes to the standard system was accomplished using information available in the ACS Globular Cluster Survey, which provides calibrated magnitudes for selected stars in the fields of 50 globular clusters extracted from images taken with the Hubble Space Telescope (HST) instruments ACS and WFPC.

By matching the positions of the stars in the field of the HST images with those in our reference image, we obtained the photometric transformation shown in Fig. 2. The I magnitude obtained from the ACS (see Siegel et al. 2007) is plotted versus the instrumental i′ + z magnitude obtained in this study. The red line is a linear fit with slope unity yielding the zero point labelled in the title, where N is the number of stars used in the fit and R is the correlation coefficient obtained. Owing to the substantial differences between the i′ + z and standard I wavebands, there are non-linear colour terms in the transformation that we have not accounted for. However, we have opted for an approximate absolute photometric calibration since variable star discovery and classification do not require a precise calibration. Furthermore, our non-standard i′ + z waveband precludes the possibility of using our RR Lyrae light curves for physical parameter estimation.

thumbnail Fig. 2

Standard I magnitude taken from the HST observations as a function of the instrumental i′ + z magnitude. The red line is the fit that best matches the data and it is described by the equation in the title. The correlation coefficient is 0.999.

Open with DEXTER

4. Variable star searches

Light curves for a total of 1405 stars were obtained with the DanDIA pipeline in the field covered by the reference image. To detect and extract the variable stars from all the non-variables, three automatic (or semi-automatic) techniques were employed. They are described in Sects. 4.14.3.

4.1. Root mean square

A diagram of root mean square (rms) magnitude deviation against mean I magnitude (see top Fig. 3) was constructed for the cluster. In this diagram, we measure not only the photometric scatter for each star, but also the intrinsic variation of the variable stars over time, which gives them a higher rms than the non-variables. The classification is indicated by the colour as detailed in Table 2. To select candidate variable stars, we fit a polynomial to the rms values as a function of magnitude and flag all stars with an rms greater than two times the model value.

4.2. SB statistic

A detailed discussion can be found about the benefits of using the SB statistic to detect variable stars (Figuera Jaimes et al. 2013) and RR Lyrae with Blazhko effect (Arellano Ferro et al. 2012). The SB statistic is defined as (3)where N is the number of data points for a given light curve and M is the number of groups formed of time-consecutive residuals of the same sign from a constant-brightness light curve model (e. g. “mean or median”). The residuals ri,1 to ri,ki correspond to the ith group of ki time-consecutive residuals of the same sign with corresponding uncertainties σi,1 to σi,ki. The SB statistic is larger in value for light curves with long runs of consecutive data points above or below the mean, which is the case for variable stars with periods longer than the typical photometric cadence.

A plot of SB versus mean I magnitude is given for NGC 6715 (see bottom Fig. 3), variable stars are plotted in colour. To select candidate variable stars, the same technique employed in Sect. 4.1 was used, with the threshold also set at two times the model SB values.

4.3. Stacked difference image

Based on the results obtained using the DanDIA pipeline, a stacked difference image was built for NGC 6715 with the aim of detecting the difference fluxes that correspond to variable stars in the field of the reference image. The stacked image is the result of summing the absolute values of the difference images divided by the respective pixel uncertainty (4)where Sij is the stacked image, Dkij is the kth difference image, σkij is the pixel uncertainty associated with each image k, and the indexes i and j correspond to pixel positions.

All of the variable star candidates obtained by using the rms and SB diagrams explained in Sects. 4.1 and 4.2 were inspected visually in the stacked image and by blinking the difference images to confirm or refute their variability.

thumbnail Fig. 3

Root mean square (rms) magnitude deviation (top) and SB statistic (bottom) versus the mean I magnitude for the 1402 stars detected in the field of view of the reference image for NGC 6715. Coloured points follow the convention adopted in Table 2 to identify the types of variables found in the field of this globular cluster.

Open with DEXTER

Table 2

Convention used in the variable star classification of this work based on the definitions of the General Catalogue of Variable Stars (Samus et al. 2009).

5. Variable star classification

To define the type of variation in each of the variable stars found, several steps were done. First, we used their position in the colourmagnitude diagram (CMD; Fig. 4) as a reference for their evolutionary stage as most types of variable stars are placed in very well defined zones in the CMD. Second, we implemented a period search for each of the light curves by using the string method (Lafler & Kinman 1965) and by minimising the χ2 in a Fourier analysis fit. Periods found and light curve shapes were also taken into account. Finally, to classify the variable stars, we used the conventions defined in the General Catalogue of Variable Stars (Samus et al. 2009).

In Table 2, the classification, corresponding symbols, and colours used in the plots throughout the paper are shown.

6. Colour magnitude diagram

Since our sample only has data available for one filter, we decided to build the colourmagnitude diagram (CMD; see Fig. 4) by using the information available from the HST images at the ACS Globular Cluster Survey. The data used correspond to the V and I photometry obtained in Sirianni et al. (2005). The CMD was useful in classifying the variable stars, especially those with poorly defined light curves such as long period variables and semi-regular variables, as well as corroborating cluster membership.

7. NGC 6715/C1851-305/Messier 54

The details of all variable stars in our FoV that are discussed in this section are listed in Table A.1, and all light curves are plotted in Fig. A.1.

thumbnail Fig. 4

Colour magnitude diagram of the globular cluster NGC 6715 built with V and I magnitudes available in the ACS globular cluster survey extracted from HST images.

Open with DEXTER

7.1. Known variables

This globular cluster has of the order of 200 known variable stars listed in the Catalogue of Variable Stars in Galactic Globular Clusters (CVSGGC; Clement et al. 2001). Most of them are of the RR Lyrae type but a few are irregular (long-period type), semi-regular, W Virginis, eclipsing binaries, and SX Phoenicis. To date, four studies report variable star discoveries in this globular cluster: V1-V28 from Rosino (1952), V29-V82 from Rosino & Nobili (1958), V83-V117 from Layden & Sarajedini (2000), and V118-V211 from Sollima et al. (2010).

thumbnail Fig. 5

Finding chart for the globular cluster NGC 6715. The image used corresponds to the reference image constructed during the reduction. All known variables and new discoveries are labelled with their V numbers. Known variables, or new variables discovered in this work, are plotted with black symbols. Variables discovered by the OGLE survey are plotted with blue symbols. Those variables that were candidate variables from Montiel & Mighell (2010) are plotted using squares. Otherwise symbols are circles. Green squares and red circles, both without labels, are candidate variables from Montiel & Mighell (2010) and McDonald et al. (2014), respectively, for which we do not detect variability in our survey. Image size is ~40 × 40 arcsec2.

Open with DEXTER

In the field of view covered by our reference image there are only five known variable stars (V112, V160, V173, V181, V192). All of them lie towards the edges of the image, as can be seen in Fig. 5. The star V112 was previously classified as irregular. However, we were able to find a period of ~100 days in the variability of this star and, because of this, we reclassified it as a semi-regular variable. For V160, we were not able to produce a good phased light curve using the published period of 0.6194848 d. The discovery observations by Sollima et al. (2010) cover a time baseline of only six days. With our time baseline of more than two years, our derived periods are much more precise and the period found is in agreement with that found by the Optical Gravitational Lensing Experiment (OGLE, Udalski et al. 1992; Soszyński et al. 2014, see below). For V160, we list the OGLE period of 0.62813716 d in Table A.1. For V173, we improved the period estimate over that from Sollima et al. (2010). For V181, we find a very different period with respect to the one estimated by Sollima et al. (2010). The new period of 0.877072 d makes this RR Lyrae the one with the longest period in the cluster. The phased light curve in Fig. A.1 is somewhat noisy because this variable is highly blended with a brighter star. The case of V192 is particularly interesting because the star was classified as RR1 with a period of 0.3986799 d. However, at the astrometric position reported for this star we found a RR Lyrae type RR0 with a period of 0.600373 d. This is also in agreement with the period and classification found by OGLE (see below), which we list in Table A.1. Again, the phased light curve of this variable is noisy because of blending with a brighter star. It is clear then that the Sollima et al. (2010) periods and RR Lyrae classifications are not robust based on relatively few observations. We discuss the consequences of this later on.

Recently, Montiel & Mighell (2010) announced 50 RR Lyrae candidates based on observations taken with the HST. However, the data obtained consist of 12 epochs covering only 8 hours, which made the study unsuitable for a period search and certainly some RR Lyrae stars will have been missed with such a short time baseline. No light curves were presented in their paper. Of these 50 candidate variable stars, 17 lie outside of our field of view (VC2, VC11-VC18, VC22, VC24, VC38, VC44, VC45, VC47-VC49). As pointed out in the Catalogue of Variable Stars in Galactic Globular Clusters (Clement et al. 2001), 11 of these candidates are previously known variables, thus VC2 = V127, VC11 = V162, VC12 = V163, VC13 = V95, VC14 = V164, VC15 = V142, VC17 = V129, VC18 = V179, VC44 = V46, VC45 = V148, and VC47 = V76 (see also Appendix B). For eight of their candidates in our field of view, we could not detect variations in our difference images at their coordinates (VC3, VC6, VC19, VC21, VC29, VC42, VC43, VC50), and we therefore cannot confirm their variability. We plot their positions in Fig. 5 with a green square5. Three of their candidates in our field of view, VC28, VC34, and VC46, are the known variables V181, V160, and V192, respectively. We confirm the variable nature of the remaining 22 candidates in our field of view, and we have assigned them “V” numbers as part of our study (see Sect. 7.2). Table A.1 lists their VC identification in Col. 2. We classify 18 of them as RR Lyrae stars, one as an eclipsing binary, and are unable to classify VC27, VC32, and VC33. We plot their positions in Fig. 5 with a square symbol.

The study by McDonald et al. (2014) using the VISTA survey also covered the cluster and presents candidate variables based on typically ~1213 epochs spread over ~100200 days, which was insufficient to derive periods for many of them. Of the short period candidate variables listed in their Table 1, 18 are inside the field of view covered by our reference image. None of the bright irregular variables listed in their Table 2 and faint candidates listed in their Table 3 are inside the field covered in our study. Positions of these stars inside our field of view are plotted in Fig. 5 with a red circle. It is worth noting that all of these candidates are located more toward the edges of the reference image. We detect variability in only two of the McDonald et al. (2014) candidates within our field of view (SPVSgr18550405-3028580 and SPVSgr18550386-3028593, which we assign V identifications as V229 and V263, respectively, see Sect. 7.2).

Similarly, the field of this cluster was also covered by OGLE, particularly with their OGLE-IV survey (Soszyński et al. 2014; Udalski et al. 2015). In this survey, we find 15 of the variable stars studied in our work, of which two are previously known variables (V160 and V192), and 13 are new discoveries by OGLE (we assigned the following V identifications: V227-V229, V233, V236, V237, V240, V244, V246, V250-V253). The OGLE light curves for these stars typically have ~150 epochs covering a baseline of ~2.5 yrs and OGLE derived precise periods for them. With our data, we were also able to find the same periods and type of classification assigned by OGLE. The positions of these stars are plotted in Fig. 5 with a blue colour. Again, it is worth noting that all of these variables are located more toward the edges of the reference image. In the particular case of V229, V244, and V246, the pipeline was not able to detect these stars in the reference image but their variation is clear in the difference images. Their differential fluxes against phase are plotted in Fig. A.1. Finally, epochs, periods, mean magnitudes, amplitudes, and classifications for these three stars were taken from the OGLE database, although the number of data points listed in Table A.1 correspond to our light curves.

In Appendix B (Table B.1), we provide the cross-identifications for the previously known RR Lyrae stars in NGC 6715 between the CVSGGC (Clement et al. 2001), the variable star candidates from Montiel & Mighell (2010), and the OGLE RR Lyrae stars (Soszyński et al. 2014).

7.2. New variables

After employing the methods described in Sect. 4, we were able to extract 67 new variable stars in the core of NGC 6715 of which 30 are RR Lyrae, one is a W Virginis star (CWA), 21 are irregular (long-period type), three are semi regular, one is an eclipsing binary, and 11 remain without classification.

7.2.1. RR Lyrae

V213-V226, V230-V232, V234-V235, V238-V239, V241-V243, V245, V247-V249, V254-V255: these 30 newly discovered variable stars are clear RR Lyrae variables. Their positions in the CMD, light curve shapes, periods, and amplitudes corroborate their variability type. We found that 17 are pulsating in the fundamental mode (RR0); eight are pulsating in the first overtone (RR1); one is a double-mode pulsator (RR01) and four RR Lyrae remain with an uncertain subtype (3 RR0? and 1 RR1?). As shown in Table A.1, their periods range from ~0.28 d to ~0.76 d, with amplitudes between 0.06 and 1.69 mag.

The periodogram analysis for V221 showed two predominant frequencies typical of double mode RR Lyrae stars, one equivalent to the fundamental period P0 = 0.459608 d, and one equivalent to the first overtone period P1 = 0.343828 d giving a period ratio P1/P0 = 0.748, which falls into the expected ratio range of ~0.725 to ~0.748 for this type of pulsating RR Lyrae stars (Netzel et al. 2015; Moskalik 2013; Cox et al. 1983). Further data for a more detailed analysis of this star will be useful to corroborate its pulsational properties.

7.2.2. W Virginis

V256: particularly interesting is the case of this star, since its variation (and position in the CMD) do not follow the pattern found for the other variable stars studied and classified in this work. We found a very well phased light curve with a period of ~14.771 d and an amplitude of 0.71 mag. This is the only bright variable star on the blue side of the colourmagnitude diagram far away from the red giant branch.

The properties found in the variation of this star and its position in the CMD match very well with the W Virginis-type of variable star described in Samus et al. (2009), particularly with the subtype CWA, which has periods longer than eight days (see also Wallerstein 2002). Although these types of stars have not been commonly found in globular clusters in contrast to RR Lyrae stars, they are not entirely uncommon. In the statistics of variable stars in Galactic globular clusters reported by Clement et al. (2001), it is possible to notice that 60 variable stars are Cepheids, which include Population II Cepheids, anomalous Cepheids, and RV Tauri stars. V256 is the first CWA star discovered in this cluster.

7.2.3. Irregular

V260-V280: these 21 stars are located at the top of the red giant branch, as shown with blue squares in Fig. 4. Their amplitudes range from 0.05 to 0.46 mag. We found no clear periods for these stars. Owing to this and also their position in the colourmagnitude diagram, we classified them as irregular. Light curves for all these variables are found in Fig. A.1.

7.2.4. Semi regular

V257-V259: based on the position of these three stars in the colourmagnitude diagram (see Fig. 4), their periods, and the shape of their light curves, we have classified them as semi-regular. Light curves for these variables may be found in Fig. A.1. Their amplitudes range from ~0.04 to ~0.45 mag and their periods span between ~20 and ~150 d.

7.2.5. Eclipsing binary

V212: the light curve variations for this star are very similar to those presented in eclipsing binary systems. We found that the amplitude of the deeper eclipse is of the order of ~0.8 mag and the amplitude of the secondary eclipse is ~0.5 mag. The phased light curve shown in Fig. A.1 represented a period of P = 0.202144 d.

7.2.6. Other variable stars

V281-V285: these five stars are clear variable stars. They show clear variability by blinking the difference images. They have amplitudes between 0.45 and 0.94 mag. Several attempts to determine periods for these stars were done without success. As a result, their light curves in Fig. A.1 are plotted against HJD. We note that the variable source V281 is only ~0.24 arcsec from the photometric centre of the cluster, as measured by Goldsbury et al. (2010).

V286: towards the beginning of the 2013 data, we found that the flux of this star was increasing to a maximum at ~2 456 464.9081 d of about ~39 400 ADU/s on the flux scale of the reference image, corresponding to a peak magnitude of ~15.04 mag. After that, its flux strongly decreased during the rest of the observational campaign. During 2014 and 2015, we found that the object seems to be at baseline and it is not detected in the original images.

The centre of NGC 6715 is ~4.45 arcseconds from this source. Also, at a distance of 3.73 arcseconds, there is an X-ray source studied by Wrobel et al. (2011) using data from Chandra and Hubble Space Telescope. Given the relatively small astrometric uncertainties in these positions, V286 is not associated with either. The nature and classification of this variable will need further studies, and it will remain without classification in this work.

V287-V291: these five stars were not detected by the pipeline in the reference image. Unfortunately, there are no light curves available in the OGLE database for these stars, but their variations are clear in the difference images. In Fig. A.1 their differential fluxes against HJD are plotted. As we were not able to produce phased light curves for them, they will remain without classification until future studies are done.

8. Oosterhoff dichotomy

In this work, we discovered 30 new RR Lyrae, and OGLE also discovered 17 more (of which 13 are inside our FoV). After removing the stars without a secure classification, there are 33 new RR0 and nine new RR1 stars, which represent a significant increment in the known RR Lyrae population. The vast majority of these new RR Lyrae stars are cluster members since we have studied the core of NGC 6715. Hence it is pertinent to recalculate the mean periods and number ratios of the RR Lyrae stars to see if they modify the current conclusions about the Oosterhoff type of NGC 6715.

However, we note that NGC 6715 lies projected against the Sagittarius dwarf spheroidal galaxy and behind the Galactic bulge. Therefore, the sample of RR Lyrae stars in the field of the cluster is a mixture of cluster members, bulge stars, and Sagittarius dwarf spheroidal (Sgr dSph) stars. Towards the centre of NGC 6715, the cluster member RR Lyraes dominate, but limiting our sample to the cluster core also reduces the number of RR Lyraes that can be used to calculate the mean periods and number ratios. Hence, in Fig. 6, we have plotted the total number of RR Lyrae stars, the mean period of the RR0 stars, the mean period of the RR1 stars, and the number ratio of the RR1 to all RR Lyrae stars (nRR1/(nRR0 + nRR1)), all as a function of distance from the cluster centre (RA(J2000) = 18:55:03.33; Dec(J2000) = 30:28:47.5; Goldsbury et al. 2010). To convert angles on the sky into distances in parsecs, we used the distance from the Sun to NGC 6715 of 26500 pc (Harris 1996, 2010 version). The inner and outer red vertical lines are the tidal radii estimated by McLaughlin & van der Marel (2005) based on King (1966) and Wilson (1975) models, respectively. The horizontal black lines correspond to the OoI (solid) and OoII (dashed) types given by (Smith 1995) in his Table 3.2. We only used RR Lyrae stars with certain classifications (i.e. published phased light curves with reliable period estimates).

Figure 6 clearly shows that within the range of the two estimates of the tidal radii, where the RR Lyrae stars from the cluster still dominate, the values of PRR0, PRR1 and nRR1/(nRR0 + nRR1) are intermediate between those expected for OoI and OoII clusters. We obtain PRR0⟩ = 0.613529 d, PRR1⟩ = 0.334107 d, and nRR1/(nRR0 + nRR1) = 0.24050633 when calculated for the 79 RR Lyraes within the outer estimate of the tidal radius (i.e. 380 pc). The contaminating RR Lyrae populations have (PRR0, PRR1) = (0.556 d, 0.310 d) and (PRR0,PRR1) = (0.574 d, 0.322 d) for the bulge and Sgr dSph, respectively (Soszyński et al. 2011; Cseresnjes 2001), placing them in the OoI region in the PRR0-metallicity diagram of Catelan (2009). This explains why the values of PRR0 and nRR1/(nRR0 + nRR1) tend towards the OoI values beyond the outer estimate of the tidal radius. The value of PRR1 is hardly influenced though, because there are only six RR1 stars beyond the outer estimate of the tidal radius. Hence the new RR Lyrae discoveries in this paper have served to confirm that NGC 6715 is of intermediate Oosterhoff type.

thumbnail Fig. 6

The number of RR Lyraes, mean periods, and number ratios as a function of the distance from the cluster centre. The red lines correspond to the tidal radii calculated by McLaughlin & van der Marel (2005).

Open with DEXTER

9. Conclusions

The globular cluster NGC 6715 turns out to be a very interesting stellar system, for which images with the highest angular resolution have ever been obtained so far with ground-based telescopes. The use of the EMCCD and shift-and-add technique was demonstrated to be an excellent procedure to minimise the effect of the atmospheric turbulence, which is one of the main constraints when doing ground-based observations. Thanks to this and the use of difference image analysis, it was possible to obtain high-precision time series photometry in the core of this cluster down to I ~ 18.3 mag.

A total of 1 405 stars in the field covered by the reference image were statistically studied for variable star detection. We presented light curves for 17 previously known variables that were found toward the edges of the reference image (16 RR Lyrae and 1 SR). We also discovered 67 new variable stars, which consist of 30 RR Lyrae, 21 irregular (long-period type), three semi-regular, 1 W Virginis, one eclipsing binary, and 11 unclassified stars. We estimated periods and ephemerides for all variable stars in the field of our reference image. Our new RR Lyrae star discoveries help confirm that NGC 6715 is of intermediate Oosterhoff type. Finally, our photometric measurements for all variable stars studied in this work are available in electronic form through the Strasbourg Astronomical Data Centre.


2

DanDIA is built from the DanIDL library of IDL routines available at http://www.danidl.co.uk

5

The Montiel & Mighell (2010) coordinates differ by RA ~ 0.3′′ and Dec ~0.8′′ from our coordinates and we have corrected for this in Fig. 5.

Acknowledgments

Our thanks go to Christine Clement for clarifying the known variable star content in NGC 6715 and the numbering systems of the variable stars while we were working on these clusters. This support to the astronomical community is very much appreciated. The Danish 1.54 m telescope is operated based on a grant from the Danish Natural Science Foundation (FNU). This publication was made possible by NPRP grant # X-019-1-006 from the Qatar National Research Fund (a member of Qatar Foundation). The statements made herein are solely the responsibility of the authors. K.H. acknowledges support from STFC grant ST/M001296/1. G.D. acknowledges Regione Campania for support from POR-FSE Campania 2014-2020. D.F.E. is funded by the UK Science and Technology Facilities Council. T.H. is supported by a Sapere Aude Starting Grant from the Danish Council for Independent Research. Research at Centre for Star and Planet Formation is funded by the Danish National Research Foundation. T.C.H. acknowledges support from the Korea Research Council of Fundamental Science & Technology (KRCF) via the KRCF Young Scientist Research Fellowship. Programme and for financial support from KASI travel grant number 2013-9-400-00, 2014-1-400-06 & 2015-1-850-04. N.P. acknowledges funding by the Gemini-Conicyt Fund, allocated to project No. 32120036 and by the Portuguese FCT – Foundation for Science and Technology and the European Social Fund (ref: SFRH/BGCT/113686/2015). CITEUC is funded by National Funds through FCT – Foundation for Science and Technology (project: UID/Multi/00611/2013) and FEDER – European Regional Development Fund through COMPETE 2020 – Operational Programme Competitiveness and Internationalisation (project: POCI-01-0145-FEDER-006922). OW and J. Surdej acknowledge support from the Communauté française de Belgique – Actions de recherche concertées – Académie Wallonie-Europe. This work has made extensive use of the ADS and SIMBAD services, for which we are thankful.

Note added in proof. During the final stages of the refereeing process for this paper, Hamanowicz et al. (2016) posted a draft paper on the arXiv preprint website on the subject of variable stars in Sgr dSph and M 54. They used OGLE imaging data with conventional CCDs, covering a much larger field of view than our work, although our images are of better resolution having been obtained with EMCCDs. While both sets of results are complementary, there are however a substantial number of variables that have been independently discovered by ourselves and Hamanowicz et al. (2016). These stars are indicated in Col. 2 of Table A.1 with the term O16. Both teams agreed to work together to form a consistent naming convention for the new variable stars with the aim of avoiding confusion for future researchers. We believe that the positive scientific spirit shown by both teams to solve this issue to the benefit of the astronomical community is commendable.

References

Appendix A: Additional tables and figures

Table A.1

Ephemerides and main characteristics of the variable stars in the field of globular cluster NGC 6715.

thumbnail Fig. A.1

Light curves of the known and new variables discovered in globular cluster NGC 6715. Red, blue, and yellow triangles correspond to the data obtained during the years 2013, 2014, and 2015, respectively. For V229, V244, V246, V286-V291, we plot the quantity fdiff(t) /p(t) since a reference flux is not available.

Open with DEXTER

Appendix B: Cross-identification between variables

Table B.1

Cross-identification between the RR Lyrae variables listed by the CVSGGC (Clement et al. 2001), the variable star candidates from Montiel & Mighell (2010), and the OGLE RR Lyrae stars (Soszyński et al. 2014).

All Tables

Table 1

Time-series I-band photometry for all known and new variables in the field of view covered in globular cluster NGC 6715.

Table 2

Convention used in the variable star classification of this work based on the definitions of the General Catalogue of Variable Stars (Samus et al. 2009).

Table A.1

Ephemerides and main characteristics of the variable stars in the field of globular cluster NGC 6715.

Table B.1

Cross-identification between the RR Lyrae variables listed by the CVSGGC (Clement et al. 2001), the variable star candidates from Montiel & Mighell (2010), and the OGLE RR Lyrae stars (Soszyński et al. 2014).

All Figures

thumbnail Fig. 1

Histograms with the number of observations per night for globular cluster NGC 6715. Left: 2013. Middle: 2014. Right: 2015.

Open with DEXTER
In the text
thumbnail Fig. 2

Standard I magnitude taken from the HST observations as a function of the instrumental i′ + z magnitude. The red line is the fit that best matches the data and it is described by the equation in the title. The correlation coefficient is 0.999.

Open with DEXTER
In the text
thumbnail Fig. 3

Root mean square (rms) magnitude deviation (top) and SB statistic (bottom) versus the mean I magnitude for the 1402 stars detected in the field of view of the reference image for NGC 6715. Coloured points follow the convention adopted in Table 2 to identify the types of variables found in the field of this globular cluster.

Open with DEXTER
In the text
thumbnail Fig. 4

Colour magnitude diagram of the globular cluster NGC 6715 built with V and I magnitudes available in the ACS globular cluster survey extracted from HST images.

Open with DEXTER
In the text
thumbnail Fig. 5

Finding chart for the globular cluster NGC 6715. The image used corresponds to the reference image constructed during the reduction. All known variables and new discoveries are labelled with their V numbers. Known variables, or new variables discovered in this work, are plotted with black symbols. Variables discovered by the OGLE survey are plotted with blue symbols. Those variables that were candidate variables from Montiel & Mighell (2010) are plotted using squares. Otherwise symbols are circles. Green squares and red circles, both without labels, are candidate variables from Montiel & Mighell (2010) and McDonald et al. (2014), respectively, for which we do not detect variability in our survey. Image size is ~40 × 40 arcsec2.

Open with DEXTER
In the text
thumbnail Fig. 6

The number of RR Lyraes, mean periods, and number ratios as a function of the distance from the cluster centre. The red lines correspond to the tidal radii calculated by McLaughlin & van der Marel (2005).

Open with DEXTER
In the text
thumbnail Fig. A.1

Light curves of the known and new variables discovered in globular cluster NGC 6715. Red, blue, and yellow triangles correspond to the data obtained during the years 2013, 2014, and 2015, respectively. For V229, V244, V246, V286-V291, we plot the quantity fdiff(t) /p(t) since a reference flux is not available.

Open with DEXTER
In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.