EDP Sciences
Free Access
Issue
A&A
Volume 576, April 2015
Article Number A14
Number of page(s) 19
Section Catalogs and data
DOI https://doi.org/10.1051/0004-6361/201425165
Published online 13 March 2015

© ESO, 2015

1. Introduction

Studying the properties of old star clusters in and around galaxies is one of the keystones for understanding the formation and evolution of galaxies (Ashman & Zepf 1992; Forbes et al. 1997; Côté et al. 1998; Brodie & Strader 2006; Tonini 2013). Because it is relatively easy to detect them out to large galactocentric distances, and because their host stellar populations are less complex than massive galaxies, star clusters are an accurate and relatively straightforward tool for unveiling the mechanisms that produced the current distribution and evolutionary properties of stars in the host galaxy.

The surroundings of massive galaxies are populated by a zoo of small stellar systems (SSSs): globular clusters (GCs), extended clusters (ECs), ultracompact dwarfs (UCDs), dwarf spheroidals (dSphs), dwarf ellipticals (dEs), compact ellipticals (cE), etc. (see, e.g., Forbes et al. 2013, and references therein). The characteristic magnitude, colors and half-light radii for some SSS classes are given in Table 1. We emphasize that the distinction between the different SSS types is sometimes not trivial, and somewhat arbitrary because of the lack of a sharp distinction between the classes of SSSs, as revealed, for example, by the scaling relations of mass, radius, luminosity, central surface brightness, or velocity dispersion (Drinkwater et al. 2004; Misgeld & Hilker 2011; Chiboucas et al. 2011; McConnachie 2012). A natural explanation of this lack of clear class-boundaries is that there is none. Indeed, the transformation processes occurring in dense environments may disrupt or transform massive SSSs, littering the galaxy field with the remains of disrupted system: low-mass SSSs, stellar streams, etc. (Bassino et al. 1994; West et al. 1995; D’Abrusco et al. 2013, 2014).

Characterizing the properties of the wealth of SSSs in the potential well of the host galaxy is fundamental for the understanding of their origin, and is an important tool for gauging the growth of the galaxy and, more in general, of cosmic structures.

In this context, our study is dedicated to the analysis of SSSs in NGC 3115, and is the first of a series aimed at analyzing SSSs in bright early-type galaxies, which are observed as part of the ongoing imaging VST survey VEGAS (“VST early-type galaxy survey”, distributed over many semesters; GTO-INAF program, P.I. Massimo Capaccioli).

An overview of VEGAS and its scientific aims is presented in Capaccioli et al. (in prep.). At completion, the survey will have collected detailed photometric information of ~100 bright early-type galaxies to study the galaxy light distribution out to ~15–20 effective radii. These galaxy regions are still almost unexplored in the CCD era, mainly because of the difficulties posed by the reduced detector field-of-view. The coupling of a dedicated survey telescope, the VST (Capaccioli & Schipani 2011), with a new-generation wide-field optical imager, the OmegaCAM (Kuijken 2011), offers a great opportunity to investigate this issue. Similar studies for the northern hemisphere are being carried out for the “Next generation Virgo cluster survey” (NGVS; Ferrarese et al. 2012), and the “Mass assembly of early-type galaxies with their fine structures” survey (MATLAS; Duc 2014).

The specific aim of the VEGAS-SSS series is to study and characterize the properties of the SSSs out to very large galactic radii, taking advantage of VEGAS imaging data. Small stellar systems, especially the GC systems, have been studied for decades, and progress has been limited not so much by telescope collecting area, but by the field of view and the image quality (both to reduce contamination and to reduce the exposure times). Thus, the use of 8 m and even 4 m telescopes is not compelling, at least for the photometry. In this paper we show the original achievements possible with wide-field imaging from a 2.6 m telescope.

thumbnail Fig. 1

VST/OmegaCAM g-band image of NGC 3115. North is up, east is left. The image size is 52.5′ × 52.5′. The black square and circle mark the position of NGC 3115B and KK 084.

Open with DEXTER

Table 1

Typical properties of various classes of SSS.

So far, except for the already mentioned ongoing studies from the NGVS and MATLAS surveys, the SSSs field population of only a few galaxies has been analyzed out to large galactocentric radii (Dirsch et al. 2003; Forbes et al. 2011; Usher et al. 2012; Blom et al. 2012), although typical studies did not go much beyond 30′ × 30′, which makes a robust estimate of the total background contamination difficult. Taking advantage of the large field of view of the VST, we here:

  • analyze the photometry in g and i bands for candidate GCs, UCDs, ECs, dSphs, etc. Furthermore, at completion, VEGAS will also include r data for most of the targets, and u for selected galaxies;

  • study the properties of various SSS populations as a function of galactocentric distance to limits not reached to date;

  • when possible, characterize the spatial extent of sources, with the specific purpose of increasing the efficiency in distinguishing between the various classes of SSSs;

  • provide catalogs of SSS candidates, which are essential for preparing spectroscopic follow-up campaigns based on samples suffering for low or, at least, controlled fore- or background contamination. To this aim, VEGAS-SSS data covering the u bands, possibly complemented with near-IR photometry, would be particularly efficient (Muñoz et al. 2014).

Here, we present the analysis of the g and i band of the field centered on NGC 3115, with the aim of describing the data reduction, the analysis tools, and performances of the telescope, and to anticipate the future exploitation of the survey. In particular, we mostly focus on the properties of the GC system in the galaxy. Throughout the paper we verify the reliability of the methodologies used by taking advantage of the large amount of literature data available for NGC 3115 (including results from HST observations and 8–10 m class telescopes) and present original results on SSSs topics made possible by the use of the large-format CCD mosaic. Indeed, NGC 3115, an isolated lenticular galaxy, is particularly interesting for testing the procedures used. Because of its proximity, the galaxy and its satellites were targeted by many photometric and spectroscopic studies (Hanes & Harris 1986; Capaccioli et al. 1987; Kundu & Whitmore 1998; Puzia et al. 2000, 2002; Norris et al. 2006; Arnold et al. 2011; Usher et al. 2012). Moreover, we recall that the GC system of the galaxy is the first system beyond the Local Group with a confirmed bimodal metallicity distribution, as shown by Brodie et al. (2012) from calcium triplet analysis and by Cantiello et al. (2014) using optical to near-IR photometry (see also Blakeslee et al. 2010; Yoon et al. 2011).

The paper is organized as follows: the next section briefly describes the observations and data reduction procedures. We introduce in Sect. 3 the data analysis and the full catalog, which provides the details of the photometric and morphological study of SSS candidates. Taking advantage of the large field-of-view of the images, we study in Sect. 4 the properties of the GC population versus galactocentric radius using a statistical background decontamination method. Section 5 is dedicated to the delicate matter of deriving SSS sizes. The final section provides a summary of our main conclusions and describes the perspectives for the forthcoming VEGAS-SSS studies.

2. Observations and data reduction

The VLT Survey Telescope VST is a wide-field optical-imaging telescope with a 2.6-m aperture, operating from the u to the z with a corrected field of view of 1 degree by 1 degree. Its single focal-plane instrument, OmegaCAM, is a large-format (16 k × 16 k pixels) CCD camera with a pixel scale of pixel-1.

The data reduction, including dither combination, vignetting, and exposure correction, astrometric solution, and photometric calibration was performed with the VST-Tube pipeline (Grado et al. 2012). Details about the overall data quality can be found in Capaccioli et al. (in prep.). In particular, the full width at half maximum (FWHM) of the point spread function (PSF) varies for < across the field of view, and the internal astrometric accuracy is ~ (the rms with respect to the USNO-B1 catalog is ~).

To improve the analysis of the spatial extent of the sources in the frame, we restricted our analysis to imaging data with an average PSF . With this choice the exposure time is reduced by ~30% in g and ~50% in i with respect to the total integration time available.

Basic properties for the target and optical observations are listed in Table 2. The g-band image of NGC 3115 is shown in Fig. 1.

Table 2

Main properties of NGC 3115.

To study SSSs, we need to minimize the contamination due to the light from NGC 3115. To model and subtract the galaxy, we used the ISOPHOTE/ELLIPSE task in IRAF/STSDAS (Jedrzejewski 1987)1. The modeling failed to match the central thick-disk region, which implied poor detection of the sources within the inner ~2′ area2. However, the central regions of the galaxy were accurately inspected using a ~10′ × 7′ mosaic obtained with the ACS camera onboard of the Hubble Space Telescope (HST; Jennings et al. 2014). The ACS study relies on data with similar wavelength coverage and g-band depth as the data used here. Given the higher resolution of HST data, we did not attempt to recover the sources in the poorly modeled central ~3.5′ × 1.5′ regions along the galaxy major axis.

3. Photometry and size estimates

To produce a complete catalog of all sources in the VST field of view, we independently ran SExtractor (Bertin & Arnouts 1996) on the galaxy-model-subtracted frame for each filter.

We obtained aperture magnitudes within a diameter aperture of six pixels (~ at OmegaCAM resolution), and applied aperture correction to infinite radius. The aperture correction, derived from the analysis of the curve of growth of bright isolated point-like sources (see Cantiello et al. 2005, 2011, for more details), is 0.52 ± 0.01 and 0.46 ± 0.01 mag in g and i. For extended sources, that is, for sources that are spatially more extended than the instrumental FWHM of the PSF (see below), we used the SExtractor Kron-like elliptical aperture magnitude. Finally, the photometric catalogs in the two bands were matched adopting matching radius. The final photometric catalog contained ~47 000 sources.

thumbnail Fig. 2

Extinction map of the area. The color bar shows the extinction-to-color mapping.

Open with DEXTER

As a result of the large areal coverage, there is a non-negligible variation of Galactic extinction from one side to the other of the field (ΔAg ~ 0.07 mag). We obtained the local extinctions from the dust maps by Schlegel et al. (1998) and used the reddening factors from Schlafly & Finkbeiner (2011). The final extinction map is shown in Fig. 2. All colors and magnitudes are corrected for extinction unless otherwise stated. Other details on the photometric properties of the images analyzed (completeness and limiting magnitudes) are given in Appendix A.

The color–magnitude diagram of the full sample of g and i matched sources is shown in Fig. 3 (panel a).

thumbnail Fig. 3

Photometry and sizes of SSS candidates in NGC 3115. Panel a) color–magnitude diagram for the full sample of g and i matched sources (gray dots) and for the sources with size estimates (black points). Panel b) apparent size-versus-color diagram. Panel c) apparent size-versus-magnitude diagram.

Open with DEXTER

It is very important to emphasize that the selection of SSSs based on one single color, the (gi) , is inherently uncertain and results in a catalog with large fractions of contaminating sources (foreground stars and background galaxies, Durrell et al. 2014). The selection with an additional optical color would certainly reduce the fraction of contaminants, especially if u-band photometry is available. However, a contaminant-free catalog based on optical photometry is basically unattainable. It is useful to highlight, however, that the coupling of optical data with only one near-IR band is very effective in reducing the fraction of contamination to the GC and UCD catalogs to lower than ~5% (Muñoz et al. 2014).

To partly overcome the problem of selecting SSSs by relying on one optical color alone, one can use statistical decontamination techniques (see Sect. 4), and/or add another selection criterion: the physical extent of the source (Table 1). The methodology we adopted to derive object sizes is described below, while the effectiveness and the practical problems in using object sizes as a selection parameter are discussed in Sect. 5.

3.1. Size and shape measurements as compactness criterion

As shown in Table 1, when the half-light radius Rh of SSSs can be estimated, then the object shape can be used together with photometric properties to classify the system. However, size measurements can be very challenging, especially with ground-based imaging data. Furthermore, only angular sizes can be measured in general, which require the previous knowledge of the object distance to be transformed in linear scale. In spite of this, angular sizes and shapes have been estimated for a large sample of SSSs in different environments and with various ground- and space-based telescopes (e.g. Larsen 1999; Larsen & Brodie 2003; Jordán 2004; Cantiello et al. 2007; Caso et al. 2013; Puzia et al. 2014). Below we describe how object sizes were estimated for objects in the VEGAS-SSS fields.

Given the difficulty posed by the task, specific tools were designed and implemented to analyze the light profiles of sources with intrinsic sizes similar to or slightly smaller than the instrumental PSF, to estimate the intrinsic size of a source exceeding some instrumental-dependent size limit. For VEGAS-SSS we choose to adopt Ishape3 to obtain structural parameters (in particular Rh and the axis ratio b/a) of SSSs. Ishape is optimized for modeling the light distribution for marginally resolved sources down to 1/10 of the FWHM of the PSF (Larsen 1999; Larsen & Richtler 2000). In this context, NGC 3115 is one of the most attractive targets in the survey, being also one of the nearest. At the adopted distance of 9.4 Mpc (Tonry et al. 2001, using the updated calibration zeropoint from Cantiello et al. 2013), and given the FWHM of the images (Table 2), Ishape can be used to determine the physical extent of objects with Rh ≥ 3.5 pc. For the reasons explained in Sect. 5, we also took into account objects down to Rh ~ 2 pc.

Measuring source size below the FWHM is particularly demanding in terms of the signal-to-noise ratio (S/N). We checked that the currently available i-band data did not provide an adequate S/N, hence we used the Rh estimates derived from the g band throughout.

Ishape reaches a convergence for ~30 000 of the input ~47 000 sources in the photometric catalog. The size-color and size-magnitude plots for the sample of sources for which we have structural parameters are shown in Fig. 3 (panels b and c).

More details on Ishape runs are given in Appendix B.

Table 3

Photometry and size estimates for the matched gi catalogs.

3.2. Final catalog

The final catalog resulting from the color, size, and shape criteria is given in Table 3. The catalog contains the full list of ~47 000 sources matched in the g and i catalogs. For each source, the following parameters are reported: Col. 1 VEGAS-SSS ID; Cols. 2 and 3 right ascension and declination (J2000); Col. 4 galactocentric distance; Col. 5 g magnitude and error; Col. 6 SExtractor CLASS_STAR parameter in the g band CSg; Col. 7 i magnitude and error; Col. 8 SExtractor CLASS_STAR parameter in the i band CSi; Col. 9 (gi) color; Col. 10 local reddening; Col. 11 S/N from Ishape; Col. 12 FWHM of the source; Col. 13 effective radius; Col. 14 object minor to major axis ratio (b/a); Col. 15 notes. The absolute value of Rh in pc depends on the distance adopted, thus it is incorrect for all unknown contaminating fore- and background sources. In Sect. 5 we discuss the percentage of contamination expected on the basis of a comparison with data from the literature.

4. GC population properties as a function of galactocentric distance: statistical decontamination of the sample

In this section, we analyze the color and magnitude distribution of SSSs in the field of NGC 3115. Because they dominate the SSS population in the galaxy core, we focus in particular on GCs, making use of statistical decontamination of background sources. To have a better statistics for the background subtraction, we used the entire VEGAS-SSS catalog of g and i matched sources (~47 000 objects), and selected good GC candidate sources a) in the color range for 0.4 ≤ (gi) ≤ 1.25 mag (e.g., Faifer et al. 2011; Kartha et al. 2014; Vanderbeke et al. 2014); b) maximum photometric error Δ(gi) = 0.15 mag for color analysis (Δmg = 0.5 mag for magnitude analysis); c) SExtractor star-galaxy ⟨ CS ⟩ ≥ 0.2, to avoid contamination from sources that are trivial background galaxies; d) and mg ≥ 18 mag, that is, sources ~4 σTOM brighter than were not taken into account (see below). Thus, for the analysis presented in this section we did not apply any restriction on Rh.

4.1. Background determination

thumbnail Fig. 4

Surface density histograms versus color. Panel a) Surface density for sources within RgalRbg. Darker color refers to areas with smaller galactocentric radii. Panel b) same as left, but for background sources at Rgal>Rbg. For the sake of clarity, the dotted histogram shows the density histogram times a factor 10. Panel c) residual surface density, Σ(RgalRbg) − Σbg.

Open with DEXTER

Our approach relies on the assumption that all sources beyond some limiting galactocentric radius, Rbg, are foreground or background contaminants, that are uniformly spatially distributed across the field. In particular, we adopted Rbg = 23′ (see also Sect. 5), which corresponds to ~65 kpc at the distance of the galaxy. Taking as reference the GC systems in the Milky Way and M 31, we estimate that a fraction of ~2–3% GCs brighter than mg ~ 25 mag (the approximate 90% limiting magnitude in g, see Fig. 3 and Appendix A) might still be in the background sample because of their large galactocentric distances Rgal ≳ 65 kpc. The Galaxy has seven GCs, over 157, at Rgal ≳ 65 kpc (Harris 1996, 2010 release), only three of them are brighter than the detection limit of our photometric catalog. This implies that, if placed at the distance of NGC 3115, and for random viewing angles, 2% of the MW GCs would be included in the background sample. The GCs catalog of M 31 by Galleti et al. (2004; RBCv5), selected using optical to near-IR color cuts (Muñoz et al. 2014), contains 447 GC candidates, none of which at galactocentric distance larger than 35 kpc. On the other hand, Huxor et al. (2014) discovered 59 new GCs at large galactocentric distances in Andromeda using CFHT/MegaCam data of the PAndAS survey: 19 of them would be brighter than our magnitude cut and are located at a projected distance larger than 65 kpc from the galaxy center. This corresponds to a fraction of ~3% of the total, assuming a total population of at least 700 GCs (638 from RBCv5, 59 from PAndAS).

Adopting Rbg = 23′ means that ~40% of the image is used to analyze contamination. The possible future addition of more bands will allow increasing the inner radius for the selection of GCs (more in general, of SSS satellites), which will allow using a smaller fraction of the detector to characterize the contamination.

Under this assumption, the difference between the surface density at galactocentric distance RgalRbg and Rgal>Rbg gives the residual density of sources in NGC 3115, mainly GCs.

We proceeded as follows: we first estimated the surface density of background objects per square arcminute at a given color (or magnitude), Σbg, by adopting the selection criteria (a)–(d) given above, plus the galactocentric distance. Then, the total surface density of objects within elliptical concentric regions, Σ(RgalRbg), was estimated using the same criteria on color (or magnitude), adopting different inner radii, starting from Rgal = 2′ out to Rgal = Rbg, with 1′ steps. The geometry of the ellipses, with constant ellipticity ϵ = 0.5 and position angle PA = 45°, was assumed following the results of Capaccioli et al. (in prep.) (see also Arnold et al. 2011; Jennings et al. 2014). In the following, Rgal is the semimajor axis, if not stated otherwise.

The overdensity of sources at given colors (or magnitude) associated with NGC 3115 is finally estimated as the difference ΣHost ≡ Σ(RgalRbg) − Σbg.

4.2. Color distribution

Table 4

Results of GMM at various galactocentric radii.

The panels in Fig. 4 show the density histograms Σ(RgalRbg), Σbg, and ΣHost versus color (from left to right, respectively). In each panel, darker color refers to regions with smaller inner radii. The histograms after the first innermost radius were smoothed for the sake of clarity. In the first panel of the figure, the density distribution shows a dip at (gi)~ 0.9 mag and two well-defined peaks at (gi)~ 0.75 and 1.00 mag whose prominence decreases, but does not reach zero as larger radii are considered.

For background sources (Fig. 4, Σbg middle panel) the density distribution does not show relevant features and appears to be nearly flat in the color interval shown. As expected, the color distribution of the difference diagram (Fig. 4, ΣHost right panel) shows two distinct color peaks at all radii.

To investigate the properties of the color distributions in panel (c) at each given radius, we used the Gaussian mixture modelling code (GMM; Muratov & Gnedin 2010)4. More specifically, we randomly populated the difference distribution, ΣHost, with a fixed number of sources (Nsim ~ 1500), and then ran the GMM code on the repopulated sample.

The results of the GMM run are given in Table 4, where for each Rgal we report the position of the peak and width of the blue and red distributions and the fraction of GCs associated with each populations (in parentheses). In some cases the total fraction does not equal one because of a minor very red peak. Figure 5 shows the positions of the blue and red peaks, the standard deviation of each distribution, and the fraction of objects associated with each peak (given by symbol size).

In the table we also report the number of GC candidates selected according to the (a)–(d) selection criteria given above (, column). From the average surface density of contaminants Σbg = 0.056 ± 0.014 [N/ arcmin2] and the listed, the expected number of GCs corrected for contamination at each elliptical radius can be easily derived. As an example, the area with Rgal ~ 6–8′ roughly corresponds to the ACS area inspected by Jennings et al. (2014), and is expected to contain ~310–390 GCs, to be compared with the 360 candidates found with ACS.

thumbnail Fig. 5

Position and width of the blue and red GCs (blue squares and red empty circles) at different Rgal as obtained from GMM. Symbol size is proportional to the fraction of objects associated with each peak. A fit to the data is shown with dot-dashed lines for the two subpopulations. Gray dots show spectroscopic confirmed GCs from Arnold et al. (2011). Full green points mark the running mean (median from equal number of data) of gray points. Gray long-dashed lines mark the rolling fits of the blue and red GC peaks as derived by Arnold et al. (2011) obtained from a combination of spectroscopically and photometrically selected GCs.

Open with DEXTER

We note that the position of the two peaks and their width are consistent at all radii inspected, and agree very well with the recent similar analyses on the same host galaxy (Faifer et al. 2011; Usher et al. 2012). A closer inspection of the data in Fig. 5 and Table 4 reveals important features. First, a color-Rgal correlation is observed for the blue GC component (Pearson correlation coefficients rxy ~ − 0.8), with a ~0.06 mag color difference between the inner and outer region. There is no or only a very weak color-radius correlation for the red GCs (rxy ~ − 0.25). Furthermore, the fraction of sources in the red subpopulation shows a slight but significant decrease with respect to the blue one at large radii. The width of the two sequences is relatively stable with radius, with the blue distribution being broader at all radii.

These properties support a scenario where blue GCs are associated with the galaxy halo, while red ones are more centrally concentrated and associated with the bulge stellar component in the galaxy (Kissler-Patig 1997; Côté et al. 1998; Forte et al. 2005; Liu et al. 2011).

To study the population of GCs associated with NGC 3115 excluding the GC contaminants from the neighboring fainter galaxies, we also carried out several tests by rejecting all GC candidates within 2–3 from KK 084 and NGC 3115B (see Fig. 1). The first galaxy, KK 084, is a dSph centered at Rgal ~ 5.5′ from NGC 3115, and a non-negligible population of GC candidates with specific frequency SNNGC 100.4(MV + 15) = 10 (Harris & van den Bergh 1981; Puzia & Sharina 2008). In spite of the relatively high SN, the net effect on the properties of the GC system in NGC 3115 is negligible. None of the sources in NGC 3115B fall in the elliptically shaped area of NGC 3115 inspected here.

In Fig. 5 we added data from Arnold et al. (2011), which are part of the SLUGGS survey (Brodie et al. 2014). The mean VEGAS-SSS color obtained with the statistical decontamination approach matches the color of the spectroscopically confirmed GCs very well.

Arnold et al. (2011) also derived the radial profiles out to Rgal ~ 10′ by combining the spectroscopic sample with a photometric sample corrected for contamination using “an iterative Monte Carlo scheme” (gray lines in Fig. 5). The VEGAS-SSS and SLUGGS color profiles for blue GCs match well at all common radii.

We note that at Rgal ≥ 6′ the color profiles from Arnold et al. mostly depend on the properties of the photometric sample, thus the transition appearing in both the blue and red GC profiles at 6 ≥ Rgal ≥ 8′ is strongly weighted toward the photometric sample.

For the blue GCs, the difference between the mean from VST and the color reported by Arnold et al. for spectroscopically confirmed GCs at Rgal ≤ 6′ is Δ(gi)blue< 0.01 mag. For the red GC component the difference is Δ(gi)red ~ 0.03 mag.

By coupling the spectroscopic and photometric samples (gray dashed lines), Arnold et al. found that the red GCs are on average bluer at larger galactocentric distances than blue GCs. A radial trend in the red GCs from SLUGGS data mostly appears beyond Rgal ~ 6′, where the photometric sample dominates the spectroscopic one. Furthermore, the red GC profile is nearly flat for Rgal ≤ 5.5′ and Rgal ≥ 7.5′, with a ~0.07 mag color transition in between.

Overall, the radial color profiles of GCs from the VEGAS-SSS and SLUGGS are consistent if one takes into account the error envelopes, the intrinsic width of the distribution at fixed Rgal, and the different analysis approaches adopted.

The good agreement appears even more striking if one takes into account that the data from Arnold et al. (2011) were obtained by coupling gri-band imaging data from Suprime-Cam at the 8.2 m Subaru telescope and spectroscopy from the 10 m Keck-II telescope with DEIMOS.

In conclusion, the comparison shown in Fig. 5 provides strong evidence in support of the efficiency of the approach adopted here to analyze the properties of the GC system out to more than ~20 galactic effective radii. It also shows that original results are obtained, even with the photometry in only two passbands, when using the wide-field imaging data from the 2.6 m VST telescope.

4.3. Surface density profiles

The radial profiles of the projected surface density for GC candidates are shown in Fig. 6. The surface density at each radius is obtained as the difference between the total density of sources with RgalRbg and the background density.

Taking advantage of the results obtained with GMM on the blue and red GCs, we also analyzed the radial density profiles of the blue and red subpopulations. When the GCs are divided into subpopulations by adopting a sharp blue-to-red separation at gi = 0.9, the radial profile for the red GCs appears to be steeper than that for the blue GCs. Moreover, both density profiles very closely follow a r1/4 de Vaucouleurs profile (dotted lines), and both are shallower than the galaxy light profile, showing a behavior similar to that of other galaxies (e.g., NGC 4636 and NGC 3923; Dirsch et al. 2005; Norris et al. 2012).

thumbnail Fig. 6

Surface density profiles of blue, red, and total GC population (blue, red and black lines, respectively). The galaxy surface brightness profile in g band from Capaccioli et al. (in prep.) is also reported with green long-dashed line. The linear fit to the surface density is shown with dotted lines (blue-dotted, red-dashed for the blue and red GCs). The scale of the galaxy profile is arbitrary.

Open with DEXTER

thumbnail Fig. 7

Surface density histograms versus magnitude. Panels a) to c): as in Fig. 4, but magnitude is used instead of color. The vertical dotted line shows the position of the TOM. The ~0.2 mag tolerance for the peak position (gray shaded area) is also shown as well as the best-fit Gaussian to the GCLF (green line; mag, σTOM = 1.14). Panel d): edge and second-run edge (Edge2) functions.

Open with DEXTER

The steeper starlight gradient than seen for the GC density (blue or total GC density) suggests that the GC system of NGC 3115 extends farther than the surface brightness profile of the galaxy halo. This result is consistent with the general picture of the GC system being spatially more extended than the host galaxy (Harris 1991; Harris et al. 2000; Forbes et al. 2006; Alamo-Martínez et al. 2012; Kartha et al. 2014).

Another feature in Fig. 6 is the matching density profile for red GCs with the galaxy light profile at Rgal ≥ 7.5′ (), while the surface density of GCs at smaller radii is slightly lower. This depletion has previously been observed in galaxies brighter than NGC 3115 (e.g., Dirsch et al. 2005; Goudfrooij et al. 2007) and is associated with the higher efficiency of GC-disruption mechanisms in the inner galaxy regions (dynamical friction, two-body relaxation, and GC tidal shocking; Vesperini 2001; Goudfrooij et al. 2007). It suggests that the galaxy has undergone a relatively quiescent evolution, without major star-forming events, which would have increased the inner density of red GCs.

4.4. Luminosity function: GCLF

We adopted the same approach as we used for colors in Sect. 4.2 to analyze the luminosity function of sources in the field, with the specific purpose of inspecting the GC luminosity function (GCLF) to independently estimate the galaxy distance modulus (Harris 2001) and furthermore derive the position of the turnover magnitude TOM, , as a function of galactocentric distance.

Figure 7 shows the surface density distribution obtained as described in the previous section, with the difference that in this case we used the total g magnitude instead of the (gi) color. The GCLFs derived were corrected for radial-dependent completeness as described in Appendix A. Panel a in the figure shows various local maxima in the density distribution, whereas the distribution of background sources in panel b has a power-law increase with a drop between mg ~ 24 and 25.5 mag, due to the completeness limit given by the adopted selection criteria. The density distribution of sources in the host galaxy, shown in panel c, reveals a major peak at mg ~ 22.75 mag.

To inspect the discontinuity in the luminosity function due to the TOM, we adopted the quantitative method introduced by Lee (1993) to identify the position of the RGB tip in Galactic resolved GCs. The results of this edge-detection method (based on the Sobel filter, see Appendix A) are shown in panel d of Fig. 7. Although the uncertainties in the surface density and their propagation in the definition of the edge and second-edge functions are certainly large, the diagrams highlight an inflection point (edge) and a maximum (edge2) around mg ~ 22.75 mag, as expected at the TOM (see Fig. A.25).

In Fig. 7, we also show the Gaussian GCLF with arbitrary peak normalization, assuming a turnover magnitude , with σTOM = 1.14 derived from Jordán et al. (2009, Eq. (18)). A ~0.2 mag tolerance area around is also shown. We adopted the absolute value for the turnover magnitude from the ACS Virgo Cluster Survey for galaxies with MB< − 18, mag and estimate a distance modulus μ0 = 29.95 ± 0.3, which agrees with the literature distance of the galaxy (Table 2).

Furthermore, thanks again to the large area inspected, we also probed the variation of to large projected galactocentric radii. The data in Fig. 7 (panel c) reveal a TOM essentially constant over the spatial scales inspected, as also found in other galaxies (Jordán et al. 2007).

On the basis of the results shown in Figs. 7 and 4, we deduce a low contamination rate in the regions within Rgal ≤ 8′ (first six darker curves in the figures), as the luminosity and color surface density of contaminants Σbg shown in the central panels can be one order of magnitude lower than the density in the inner regions. This implies that the contamination rate of the VEGAS-SSS catalog for NGC 3115 is quite low for the innermost ~8′. As an example, the background density at mg ~ 23 mag is Σbg ~ 0.3 [N/ arcmin2], while the density of sources within Rgal ≤ 8′ is ~3.5 times higher and is even about seven times higher at Rgal ≤ 4′.

thumbnail Fig. 8

GCLF for the blue and red GC components analyzed separately. The upper panels, from left to right, show the total density distribution Σ(RgalRbg), the density after subtracting for background contamination ΣHost, and the edge and edge2 functions. The vertical dashed line marks the approximate position of the TOM, as obtained from the edge functions. The green line shows a Gaussian with peak at mag. Lower panels: as upper ones, but for the red GC components, with mag.

Open with DEXTER

thumbnail Fig. 9

Magnitude comparison between VEGAS-SSS and ACS. Left panels: a) and b) comparison for the full list of ACS to VEGAS-SSS matching sources (gray dots). The running mean and the corresponding rms are shown with black circles and error bars. The median difference, the rms and the number of objects matched are also labeled. Panels c) and d) are the same as the upper panels, but for sources with magnitude and photometric error cuts as labeled. Right: as left panels, but only GC candidates in the ACS catalog are considered.

Open with DEXTER

We also inspected how the TOM differs between red and blue GCs. This test is not often possible and feasible here thanks to the large area inspected. After dividing the blue and red GCs by adopting a sharp color separation at (gi) = 0.9, we carried out the analysis described above on the luminosity functions of the blue and red GCs. The results are shown in Fig. 8, where the total luminosity function Σ(RgalRbg), the background-corrected luminosity function ΣHost, and the edge and edge2 diagrams are shown for the blue and red GCs (upper and lower panels). Even though the samples adopted are numerically smaller than before, the corrected GCLF still shows the peak around the same of the total GC population. From estimating the position of the TOM with the edge functions (right panels in the figure), the interesting point here is that there appears to be a ~0.2 mag offset between the TOM of red and blue GCs, with the red system being fainter. From the point of view of stellar population models, if the GCs mass function is universal across metallicity, the Gaussian mean of the blue GCs is expected to be brighter than that of the red one (Ashman et al. 1995; Di Criscienzo et al. 2006; Jordán et al. 2007). Observationally, our result confirms previous evidence obtained from data with much smaller surface coverage (Whitmore et al. 1995; Puzia et al. 1999; Peng et al. 2009). Additional improvements on this will be enabled by analyzing new galaxies in the VEGAS-SSS sample, with the possible inclusion of u- and r-band data in the SSSs selection process.

As a final comment, we highlight that the depth, in terms of absolute magnitude, and the image quality for the other galaxies in the VEGAS sample will be similar to the one inspected here, thus we expect that using the tools presented here6, we will reliably analyze the color distributions and study GCs luminosity function for all other targeted galaxies out to unreached galaxy effective radii. Moreover, for objects at larger distances the background-decontamination methods described in this section will probably be more effective because of the larger galactocentric distances inspected.

5. Object sizes: comparison with literature and analysis

In this section we present a detailed analysis of the properties of the SSS in the field of NGC 3115, and concentrate on the sample of objects with Rh estimates from Ishape.

5.1. Comparing photometry and sizes derived from VEGAS-SSS and HST/ACS

As already mentioned, using the photometry and size measurement tools described in Sect. 3.1, we finally worked with a catalog containing ~30 000 sources. We compared our measurements with the estimates by Jennings et al. (2014), based on HST/ACS gF475W and zF850LP observations. At the distance of the galaxy, all sources in the field of NGC 3115 with Rh ≥ 1 pc appear to be resolved at the pixel resolution of the ACS. Thus we assume that the Rh measured from ACS data represents the true distribution of Rh for the GCs in the galaxy within the limited common area.

The matched GC list from Jennings et al. (2014) with the VST catalog (using radius) contains ~270 of the 360 candidates7. Nearly ~70% of the unmatched sources are GCs located in the central galaxy regions, where we did not model and subtract the galaxy light profile. The remaining ~30% of missing objects are faint sources, undetected in the shallower i image, or objects blended with bright neighbors. The number of missed sources beyond the central regions drops to ~10 if only the g-band catalog is considered (again faint or blended sources). The g-band ACS and VEGAS-SSS photometry agrees very well, as shown in Fig. 9 for magnitude and color. In the figure, the full ACS catalog (Jennings, priv. comm.) and the GC-only sample are considered separately. For sake of homogeneity, the comparison with ACS was made using constant extinction. The large scatter for the full sample (left panels) is due to the extended background sources, whose aperture magnitude does not represent a good estimate for the total magnitude, for either ACS or VEGAS-SSS. Comparison of photometry for the GC-only (right panels in the figure) indicates a negligible mean residual difference in magnitude and color.

thumbnail Fig. 10

Size comparison between VEGAS-SSS and ACS. Left panels: effective radii for objects common to the ACS and VEGAS-SSS catalogs (full sample with gray dots, reference sample with black circles). For each sample, the median ratio between the ACS and VEGAS-SSS Rh is reported, the rmsMAD and the number of sources used are given in parentheses. The histograms shown in the upper and right insets use the same color coding as the central panel and are normalized to 1 at peak value for the full sample. The effective radii from ACS data are used for the selection (unshaded area). The diagonal line represents the 1:1 relation. Right panels: as left panels, but the selection is based on Rh from VEGAS-SSS.

Open with DEXTER

We also compared the Rh estimates from VEGAS-SSS with those from the ACS. In the comparison one must note that the list of SSS candidates is not contaminant-free in either catalogs, as it includes background sources whose (linear) Rh estimates are incorrect because they were derived according to the distance of the galaxy. In spite of this, the ACS VEGAS-SSS comparison is still valid since the same distance modulus is assumed in both analysis. Moreover, the size estimates are not equally good for the full sample of objects measured. Thus, we defined a reference sample of VEGAS-SSS candidates with reliable structural and photometric parameters by adopting the following criteria derived on the basis of the comparison with ACS photometry and object shape:

  • Ishape S/N ≥ 30 (Larsen 1999);

  • total relative error on Rh ≤ 30%;

  • for each source where the iteration to derive Rh was successful, Ishape provides a cutout of the image with the object analyzed, the model brightness profile, the residuals between them, and the weighting map (Larsen 1999). To reject sources with large residuals (see Fig. C.4), but otherwise good S/N and FWHM error, we chose a limit of median/rms ≤0.3 for good candidates after various experiments where we inspected the statistical properties of the residual cutouts. This criterion applied for objects with contaminating neighbors or structures not accounted for by the previous criteria;

  • as in Sect. 4 we adopted a color range 0.4 ≤ (gi) ≤ 1.25 mag both for GCs and UCDs;

  • maximum photometric uncertainty Δmg = 0.15 mag;

  • axial ratio, b/a ≥ 0.3. (van den Bergh & Morbey 1984; Blakeslee & Barber DeGraaff 2008; Cantiello et al. 2009).

Figure 10 shows the size comparison of VEGAS-SSS to ACS for the full and reference samples. A first evidence is the “coma”-shaped distribution of data. This behavior highlights the expected lack of accuracy of Ishape for objects with effective radii below 1/10 the FWHM, that is, Rh,VST ≲ 3.5 pc at the distance of the galaxy.

When only sources in the reference sample and with Rh,ACS ≥ 3.5 pc are used (black filled circles in Fig. 10, left panels), the median ratio of ACS and VEGAS-SSS Rhs is 1.02, while the rmsMAD (rms derived from the median absolute deviation) of the ratio is ~0.59. Thus, for the 29 matched objects in the reference sample, the median and standard deviation of the mean are 1.02 ± 0.11, which provides a satisfactory agreement for the ACS and VEGAS-SSS samples when limited to the reference sample. Nevertheless, we must highlight that the Rh estimates for single objects can differ by up to a factor ~5 even for Rh,ACS ≥ 3.5 pc. More details on the differences between the size estimates for the ACS and VEGAS-SSS for extended objects are given in Appendix C.

For the typical VEGAS target the selection will only rely on Rh measurements from VST images. The right panels of Fig. 10 show the same data as the left panels, using the Rh,VST values for the selection instead of Rh,ACS. The comparison of ACS to VEGAS-SSS worsens, as the Rh,ACS to Rh,VST mean ratio and standard deviation of the mean are 0.57 ± 0.06. It is interesting to note that, taking Rh,VST = 2 pc as lower limit, we obtain (rms = 0.55, 46 objects). This suggests that even though the nominal limit for Ishape is 1/10 of the FWHM, or ~3.5 pc at the distance of NGC 3115, the tool allows separating stars from extended sources down to 2 pc.

thumbnail Fig. 11

Rh distribution for ACS and VEGAS-SSS samples. From left to right: Rh distribution for the full list of matched ACS and VEGAS-SSS sources, for the sample selected using Rh,ACS, and for the sample selected using Rh,VST (see text). Shaded histograms refer to Rh,ACS distributions, solid thick lines to Rh,VST. All histograms are normalized to one at peak value.

Open with DEXTER

thumbnail Fig. 12

Magnitude-size diagram. Left panel: magnitude versus effective radius for the VEGAS-SSS field centered on NGC 3115. Color-coding for black and gray dots is as in previous figures, with the addition of the best sample selection (red empty circles). Yellow regions show the mean loci of the labeled SSS classes. The number of UCD and GC candidates, NUCD and NGC, for the reference and best sample (given in parenthesis) are also reported. Right panel: as left panel, but only sources with RgalRbg are plotted. The blue arrow shows the direction in which the points are shifted if the object lies at larger distance. We included the ACS sample using green symbols: GCs shown with circles, UCDs with squares. For the ACS sample, spectroscopically confirmed GCs from Arnold et al. (2011) are plotted as solid symbols.

Open with DEXTER

To inspect this matter more closely, Fig. 11 shows the Rh distributions for a) the full list of matched sources from ACS and VEGAS-SSS (left panel, ACS data in gray, VST data as a thick black line); b) objects in the reference sample selected using Rh,ACS ≥ 3.5 pc (middle panel); and c) objects in the reference sample selected using Rh,VST ≥ 3.5 pc (right panel). The ACS and VEGAS-SSS distributions appear to be quite similar in the middle panel (case b). The Rh distributions based on VEGAS-SSS half-light radii (right panel, case c) shows a shift, with VEGAS-SSS radii being on average larger, and lacking the peak at Rh ~ 2 pc seen in the ACS data. This behavior is due to the sources more compact than 3.5 pc, which are scattered over the entire 3.5–20 pc interval when Rh estimates from VST are used. In particular, the list of common SSS candidates in the reference sample reaches from 28, with the selection based on the ACS radii, to 39 with the Rh selection from VST data. On this basis, we estimate that for values of Rh ≥ 3.5 pc the reference sample contains ~30% objects with unreliable effective radii, spread across the entire Rh distribution. Of course the sample of matched objects with high-quality VEGAS-SSS sizes is relatively small (28 or 39, depending on the selection), which makes hard to generalize the results of the comparison for the entire VST field of view.

Finally, with the aim of deriving a catalog of GC candidates from the only VST data, and estimating the contamination taking as reference the ACS GCs list, we carried out the following blind test: we adopted the selection criteria given at the beginning of this section, with the additional requirement that GC candidates must have 2 ≤ Rh ≤ 8 pc (we adopted the same Rh as used for GC by Jennings et al. 2014, for ACS data), and matched this VEGAS-SSS list with the sample of GC candidates from ACS. The results is that ~20% of the candidates (~10 over ~50) are absent from the GC list from ACS8. In contrast, adopting as lower limit Rh = 3.5 pc, the number of matching sources is ~30 and the contamination is nearly doubled. In other words, the test points out that the results from Ishape allow distinguishing between compact and extended sources down to Rh = 2 pc, although the exact value of the effective radius is reliable only above ~3.5 pc.

In conclusion, using the reference sample obtained from the coupling of photometric and spatial extent properties, the present VEGAS-SSS catalog can be used to:

  • i)

    obtain a list of GC candidates, selected in therange 2 ≤ Rh(pc) ≤ 8, with an expected contamination of ~20%, poorly populated because of the narrow selection adopted. The number of GC candidates over the entire area inspected, ~52.5′×52.5′ (or ~145 kpc × 145 kpc), selected on the given photometric and size criteria is NGC ~ 220. However, for radii below 3.5 pc, the Rh are only used as an effective binary selection criterion (i.e., Rh ≥ 2 (<2), meaning extended (point-like) source), because this limit is smaller than the nominal limit of the tool;

  • ii)

    a catalog of extended objects with Rh ≥ 3.5 pc, which has a contamination of ~30% objects with unreliable Rh estimates.

5.2. GC and UCD population properties based on color- and size-selection criteria

Figure 12 shows the size-versus-magnitude diagram obtained using the reference sample, that is, with the selection criteria described in the previous section, and adopting for all objects in the field the same distance modulus. In the right panel, we plot only SSSs at galactocentric distance RgalRbg = 23′, corresponding to ~65 kpc at the distance of NGC 3115. The approximate regions for GCs, UCDs, ECs, dSphs, dEs, and cEs are shown and labeled (yellow regions; mean loci are taken from Brodie et al. 2011 and Brüns & Kroupa 2012).

Figure 12 (left panel) shows that a large fraction of selected objects in the reference sample falls within the avoidance region at Mg ~ − 7.5 mag and Rh ≳ 50 pc (Forbes et al. 2013). The situation does not seem to improve much even if only sources in the reference sample and with S/N ≥ 60 are considered (best sample, hereafter). On the other hand, there are fewer sources in the avoidance area if only sources within RgalRbg are taken (Fig. 12, right panel), but these are still significantly many. Moreover, we must highlight the many UCD candidates even for the best sample and for RgalRbg sources (NUCD = 137).

Forbes et al. (2013) found that the avoidance zone is the result of a selection bias and confirmed the presence of various SSSs within the region, but very many sources in the avoidance area are probably background galaxies (some of them are recognizable by eye). As shown by the arrow in Fig. 12 a background source should move toward larger absolute radii and be brighter when larger distance moduli are considered.

thumbnail Fig. 13

Radial surface density distribution of sources in the reference (black histograms) and best (red histogram) samples. Panel a) all Rh values are taken. The r1/4 fit to data is shown with gray lines. The dotted lines show the histograms without correction for areal coverage. The vertical long-dashed line is the position of limiting galactocentric radius Rbg. Panel b) as upper, but only GC candidates are plotted. The green-dashed line shows the galaxy surface brightness (from Capaccioli et al., in prep., g band, arbitrary scale). Panel c) as panel a), but for UCD candidates.

Open with DEXTER

In right panel of Fig. 12, we also plot the GC (green circles) and UCD (squares) data from Jennings et al. (2014). The spectroscopically confirmed GCs and UCDs from Arnold et al. (2011) are plotted as solid symbols. Two interesting elements here are i) the nice overlap of the overdensity region for spectroscopically confirmed GCs and VEGAS-SSS selections at −7.5 ≤ Mg ≤ − 9 and Rh ~ 2–2.5 pc; and ii) the UCDs from the ACS sample outside the region where they typically occur. Three UCD candidates from the ACS sample lie at Mg ~ − 9 mag and Rh ~ 80 pc, that is, within the zone of avoidance (if any). Two other UCDs have Mg ~ − 6.5 mag and Rh ~ 10 pc, which typically is associated with the EC region (see also the discussion in Appendix C). This clearly shows that the distinction between the different SSS types is not trivial and sometimes contains elements of arbitrariness.

To inspect this problem in more detail, we analyzed the surface density distribution of sources versus galactocentric radius and versus Rh. Figure 13 shows the radial surface density distribution for the reference and best samples, for GC candidates (2 ≤ Rh (pc) ≤ 8, panel b), and for UCD candidates (8 <Rh (pc) ≤ 100, panel c)9. In each panel we also report the total number of objects selected for the reference and best sample (the latter in parentheses). The surface density for the sample with no selection on sizes (panel a) shows an obvious correlation with Rgal, and a flattening at RgalRbg, suggesting that sources beyond this radius are most likely background galaxies or foreground stars. The radial density profile for GC candidates follows a de Vaucouleurs density profile out to Rbg, as for the galaxy light, providing additional proof of the actual membership of the objects selected to comprise the GC population, and supporting the role of the object-size analysis carried out here. However, one must not neglect the fraction of background sources. The reference sample does not show a tendency for a radial trend in the distribution of objects with UCD-like radii. The result is expected, given the few expected UCD-candidates and the many contaminants. It is noteworthy, however, that the UCDs in the best sample show hints of a radial trend.

thumbnail Fig. 14

Rh surface density distribution for the reference sample (black lines) and for the best sample (in red, Σ in units of number of objects per square arcminute). The panels, from upper to lower, show the surface density over the full inspected frame for sources at RgalRbg for the background area at Rgal>Rbg, and for the difference between the inner and outer 23. The vertical dashed line marks the 3.5 pc limit.

Open with DEXTER

By integrating the fitted de Vaucouleurs r1/4 density profiles from zero to Rbg for both GC and UCD density profiles after subtracting the total number of background sources we find for the reference (best) GC sample NGC ~ 113 (~42) and for the best UCD sample NUCD ~ 30. The comparison of these numbers, in particular for the GCs, with the numbers in Fig. 12 (right panel), confirms our previous results that the contamination for the reference GC sample is ~30%, and also indicates that the best sample suffers from very little contamination. The numbers are quite different for UCDs, given the higher confusion with extended background sources. In this case, the coupling with data in other passbands will greatly reduce the contamination.

Figure 14 shows the Rh distributions for the reference and best samples, normalized to the area inspected: full detector area, objects within the RgalRbg area, objects outside Rgal>Rbg, and the difference between the latter (from panel a to panel d).

The reference sample over the entire VST area (upper panel) shows a slightly increasing surface density for increasing Rh up to ~50 pc, while the Rh distribution for the best sample is rather flat. The differences of the surface density for objects within Rbg or in the outside area, shown in panels b and c of Fig. 14, are quite obvious, especially for the overdensity of objects with GC-like radii, 3.5 ≤ Rh (pc) ≤ 8. Because of the contamination, the density of sources with GC-like radii is non-zero in the outer radius (panel c). This is in part due to the expected fraction of galaxy GCs that might lie at large galactocentric distances (see Sect. 4), while most of the contribution comes from background contamination. Indeed, the mean density of objects with 3.5 ≤ Rh (pc) ≤ 8 at radii Rgal>Rbg (panel c) is ~30% of the density at RgalRbg (panel b). This result confirms our previous estimate of the fraction of contamination for the color- and size-selected reference sample. Furthermore, we note that the density in the background region is 15% of the density in the inner regions for the best sample.

The differences between inner and outer density are better visible in panel d of Fig. 14, where we show ΔΣ ≡ Σ(RgalRbg) − Σ(Rgal>Rbg) versus Rh. Here, the surface density of sources with Rh ≳ 70 pc is consistent with zero. More specifically, the ΔΣ distribution for the reference sample (black histogram) is generally consistent with zero density from Rh ≳ 50 pc (with some possible candidates at Rh ~ 55–60 pc), and for Rh ~ 8 pc, while for the best sample various regions are compatible with zero density (e.g Rh ~ 7–10 pc, ~20–30 pc, and 40 pc). These results imply that the surface density of objects with such Rh values is constant across the inspected area, as expected for a uniform background contamination. In other words, panel d suggests that most sources falling in the zone of avoidance (Fig. 12) are background galaxies. Second, the over-abundance of sources with Rh that have GC-like radii appears clearly both for the reference and best samples. Furthermore, for the reference sample, we find a positive density of sources around the characteristic Rh values of UCDs (between 10 and 40 pc), which confirms the membership to this class for some of the selected objects. This overdensity, however, is weaker for the best sample, and possibly consistent with zero in some cases.

6. Summary

We presented the first results of the VEGAS survey for the specific science case of small stellar systems (SSSs). We described the methodology for the photometry and the size analysis of SSS candidates in the field of NGC 3115, a well-studied lenticular galaxy, and showed the potential of the survey in providing original results on SSS-related science.

The VEGAS survey will collect the deep g and i imaging of bright early-type galaxies with cz ≤ 4000 km s-1, possibly complemented with r for most of the targets, and also with u-band observations for selected galaxies. One of the great advantages of VEGAS-SSS is the use of wide-field imaging, ~1 square degree, which allows studying the properties of SSSs out to very large galactocentric distances, with an accurate characterization of the background-contaminating objects. For the specific case of NGC 3115 we inspected the properties of SSSs out to ~23′, which is more than twenty times the effective radius of the galaxy.

We first analyzed the properties of the GCs system. Because GCs are the population of SSSs numerically most abundant in the galaxy, their properties can be derived using solely photometric information, color, and magnitudes by comparing the surface density of sources in the inner galaxy regions with the density in the outer regions. Our results can be roughly divided into two groups: i) results that repeat previous analyses, giving us the opportunity to confirm the reliability of this study; ii) new results allowed by the use of the wide-field imaging. In the first group we include:

  • the GCs have a bimodal (gi) distribution with peaks at~0.75 and ~1.0 mag;

  • red GCs are more centrally concentrated than blue GCs;

  • as for the galaxy light, the radial density of GCs follows a de Vaucouleurs r1/4 profile, but with a shallower slope;

  • the turnover magnitude of the g-band GCLF, , coupled with the calibration from the ACSVCS survey, implies a distance modulus μ0 = 29.95 ± 0.3 that agrees well with the literature.

These achievements support the results of previous studies, some of which were carried out with 8–10 m class telescopes, and are further complemented by the following compelling results:

  • the color bimodality extends to more than ~20 galaxy effective radii;

  • the blue GCs show a tendency toward bluer color at larger galactocentric radii Rgal, while red GCs seem to have a nearly constant color with Rgal;

  • the galaxy light has a steeper density profile than the GCs, whether the blue or total fractions of GCs is taken into account;

  • the slope of the surface density profile for red GCs at Rgal ≥ 7.5′ matches that for the galaxy light, while a red GC overdensity appears in the inner galaxy regions;

  • the ratio of blue to red clusters shows a trend with Rgal, with the fraction of blue GCs being slightly larger at larger radii;

  • by separately analyzing the blue and red GCs, we find a mag, with the blue TOM being brighter;

  • we do not find an obvious dependence of with Rgal.

Both the color and luminosity properties obtained are consistent with similar existing studies of the GC system in other early-type galaxies.

The bimodal GC system, with blue GCs more extended than the galaxy stellar light, and a deficiency of red GCs in the inner regions, has previously been observed in other early-type galaxies brighter than NGC 3115 and supports a scenario where blue GCs are associated with the galaxy halo, while red ones are more centrally concentrated and associated with the bulge stellar component in the galaxy. The overall observed properties might suggest that the galaxy has undergone a relatively quiescent evolution, without major star-forming events.

Adding the spatial extent of the sources to the color information gives an additional criterion for selecting SSSs, in particular GCs and UCDs. We used Ishape to determine the effective radius Rh of slightly extended objects in the field. By comparing our estimates for the objects in the reference sample with those in the literature obtained from ACS data, we found on average a satisfactory agreement. However, the Rh estimates for single objects can differ by up to a factor ~5 between ACS and VEGAS-SSS. Furthermore, the result is sensitive to the Rh estimate taken as reference (ACS or VEGAS-SSS) because of the contamination of the VEGAS-SSS sample. The various comparisons with the literature and with inner and outer galaxy regions suggest that the level of fore- and background contamination of our reference sample is ~30%, possibly reduced to one half for the (poorer) best sample. Future studies with new VEGAS-SSS u- and r-band data will be used to constrain the properties of other, less populated classes of SSSs, such as cEs, in the field of NGC 3115 even more tightly.

In spite of the large uncertainties posed by the estimate of Rh, the results obtained are encouraging, suggesting that similar analyses might be successfully carried out for the other targets in the survey, although at larger distances, studying the sizes will be limited to the most extended SSSs (UCDs, cEs), excluding the GC component for most of the targets beyond ~10 Mpc distance.

The results of this work on one hand confirm existing studies and support the validity of the analyses scheme developed here using data from the 2.6 m VST telescope. On the other hand, they provide new and independent results – especially for the GC properties out to the previously unreached galactocentric distance of ~65 kpc – showing the great potential for future applications to other VEGAS targets, in particular for the part of the sky not accessible to similar facilities.

As a final remark we highlight that at survey completion, for most of the VEGAS targets observations in at least one more passband other than g and i will be available. The selection of SSSs with another optical color would certainly reduce the percentage of contaminants, especially if u-band photometry is included. However, a contamination-free catalog based on purely optical photometry is basically unattainable. Since the coupling of optical data with just one near-IR band is very effective in reducing the fraction of contaminants to the GC and UCD populations to 5%, the VEGAS-SSS catalogs will be perfectly suited to be complemented with single-band near-IR imaging (e.g., with a large-format near-IR imager like VISTA), to define the most complete and clean SSS catalogs possible, essential, for example, for future spectroscopic follow-up.


1

IRAF is distributed by the National Optical Astronomy Observatory, which is operated by the Association of Universities for Research in Astronomy (AURA) under cooperative agreement with the National Science Foundation.

2

A test with GALFIT (Peng et al. 2002), another program designed for modeling two-dimensional brightness profiles, also failed to model the central galaxy regions. We also obtained a galaxy-subtracted frame as described in Jordán et al. (2007) and Cantiello et al. (2014, which modeled NGC 3115 from near-IR data). This method uses the SExtractor spline background derived from the image logarithm, and provided very flat residuals. However, the latter procedure affects the shape of slightly extended objects, which makes it unsuitable for the purposes of the present study.

3

The software can be downloaded at http://baolab.astroduo.org/. We used release 0.94.1d.

4

GMM uses the likelihood-ratio test to compare the goodness of fit for double-Gaussians versus a single-Gaussian. For the best-fit double model, it estimates the means and widths of the two components, their separation DD in terms of combined widths, and the kurtosis of the overall distribution. It also provides uncertainties based on bootstrap resampling. In addition, the GMM analysis provides the positions, relative widths, and fraction of objects associated with each peak.

5

The edge functions show two other possible TOM-point candidates located at mg ~ 22.2 and 23.2 mag (see Appendix A). However, both magnitude values are ruled out as TOM peaks by the shape of the GCLF.

6

Except for inspecting the size and shape which will hardly be possible for GCs in distant galaxies, but which can be done for the UCDs.

7

We found a systematic shift in RA, . In our analysis we applied the correction to ACS data.

8

The result is not very sensitive to the particular choices of Ishape input parameters (see Fig. B.2).

9

To estimate the density of sources, the effective area coverage in each annulus is corrected for the annular area outside the image and for the central uninspected regions (dashed histograms show the uncorrected distributions). Poisson statistics is adopted to estimate the errors.

10

This equation, suggested in the Ishape handbook, provides results nearly identical to the one used by other authors (e.g., Blakeslee & Barber DeGraaff 2008).

Acknowledgments

The optical imaging is collected at the VLT Survey Telescope using the Italian INAF Guaranteed Time Observations. The data reduction for this work was carried out with the computational infrastructures of the VST Center at Naples (VSTceN). We gratefully acknowledge INAF for financial support to the VSTceN. Part of this work was supported by PRIN-INAF 2011 (P.I.: G. Marconi), FIRB-MIUR 2008 (P.I.: G. Imbriani), PRIN-INAF 2011 (P.I.: A. Grado). M.P. acknowledges finical support from project FARO 2011 from the University of Naples Federico II. D.A.F. thanks the ARC for financial support via DP130100388. We are grateful to John P. Blakeslee and Zach Jennings for useful discussions related to this work. This research has made use of the NASA/IPAC Extragalactic Data-base (NED) which is operated by the Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration. This research has also made use of the SIMBAD database, operated at CDS, Strasbourg, France, and of the HyperLeda database (http://leda.univ-lyon1.fr).

References

Appendix A: Completeness and the edge detection functions

Appendix A.1: Completeness correction

The completeness function of the g-band images was determined by adding artificial stars to the original images and then reprocessing them as described in Sect. 3. The ratio between the number of artificial stars added and the number of stars recovered provides the completeness estimate. We added stars using a grid pattern with ~20″ increments in x and y. Since the field is dominated by the light from NGC 3115, the correction for magnitude completeness depends on the angular distance from the galaxy center (e.g., Cantiello et al. 2007). The radial-dependent completeness function is shown in Fig. A.1. To correct the luminosity functions, the number of objects at a given magnitude is multiplied by 1 /f using the proper function at each galactocentric distance.

thumbnail Fig. A.1

Completeness function estimated at different galactocentric radii. The label refers to the mean radius of the annulus in arcminutes. The dotted lines show the 90% completeness limit.

Open with DEXTER

Appendix A.2: Edge-detection filter

thumbnail Fig. A.2

Upper panel: analytic GCLF (green line), background galaxies (blue), and completeness function (red) are combined in the total expected total luminosity function (black solid line). The GCLF peak (at x = 22.8) and 50% completeness limit (at x = 25.5) are shown with dashed vertical lines. Middle panel: edge detection function applied to the analytical formula of the total luminosity function. Lower panel: second-run edge detection.

Open with DEXTER

In Fig. A.2 we analyze the behavior of the edge-detection function on a composite function similar to the one expected for the sources in the field of NGC 3115. The function inspected is the sum of a Gaussian GCLF (Harris 2001) and a power law for background galaxies (Tyson 1988; Bernstein et al. 2002) times a completeness-like smoothed step function (green, blue, and red line, respectively). The edge-detection function, in a first approximation, is a derivative function and shows an inflection point at the GCLF turnover magnitude (Fig. A.2, middle panel). A second run of the edge function – edge2, roughly a second derivative – reaches a local extremum at the TOM. Thus, in first approximation, the turnover of the GCLF can be found in correspondence of an inflection and a local extremum in the edge and edge2 functions.

Appendix B: Some details on Ishape

Ishape uses a PSF subsampled by a factor 10 relative to the resolution of the science image. To model the PSF we used the DAOPHOT package within IRAF and, to reduce the chance of contaminating the PSF modeling with GCs in the galaxy, we included in the list of PSF candidates unsaturated sources with gi ≤ 0.3, gi ≥ 1.7, and mg ≥ 18 mag. To account for PSF variations across the image, we set DAOPHOT VARORDER = 2, which means that the PSF is quadratically variable over the image. Then, the frame was divided into a grid of 5 × 5 equal subframes and the model PSF for Ishape was evaluated in the center of each subframe.

Within Ishape, we adopted the “KING30” profile, that is, the King (1962) model with concentration parameter c = 30, which is typical for marginally resolved GCs and UCDs (Larsen & Richtler 2000; Blakeslee & Barber DeGraaff 2008).

Table B.1

Main Ishape parameters used for the analysis.

To determine the best parameters for Ishape, we performed a reference run and various tests changing the input parameters. Table B.1 gives the main parameters for the reference run (g1 label). The other tests were obtained as follows: for test#1 we adopted the DAOPHOT Penny1 PSF instead of the Moffat25 (label g2); for test#2 and #3 (labels g3 and g4) we used Ishape fitting radius 9 pixels and 15 pixels; test#4: the maximum FWHM is set to 40 pixels (label g5); test#5: does not fit an elliptical model, circular symmetry is used instead (label g6). In all cases, except for test #6, the FWHM is transformed to circularized effective radius Rh = 1.48·FWHMKING30·0.5·(1 + wy/wx), where FWHMKING30 and wy/wx are the full width, and the axis ratio fitted by Ishape10. For test #6 we used Rh = 1.48·FWHMKING30. Figure B.1 shows the results of the Ishape tests. The data in the figure show that, in general, there can be even a factor 10 difference between Rh estimates with different Ishape input parameters. Nevertheless, for the reference sample (see Sect. 5), the effect of changing fitting parameters implies a median change on Rh of 10%. We also inspected the correlation between the radius and magnitude of the sources, and did not find any correlation.

Figure B.2 shows the comparison of VEGAS-SSS to ACS for the various tests. The data in the figure show that the results with reference g1 run are broadly consistent with the other tests. We note that, in choosing the best parameters for Ishape, we also took into account the number of sources successfully analyzed by the tool. For the g1 test the input catalog contained ~47 000 sources, and the spatial parameters were obtained for ~30 000. This number can decrease significantly for other choices of the input parameters – most notably in test g4.

thumbnail Fig. B.1

Comparison of effective radii with different Ishape input parameters. Gray circles refer to the full list of matched sources, black dots to the reference sample. The numbers in each panel show the median Rh,test/Rh,ref ratio, with the rmsMAD and the number of objects used in parenthesis, labels are color coded.

Open with DEXTER

thumbnail Fig. B.2

As in the left panel of Fig. 10, but for the different Ishape tests (as labeled).

Open with DEXTER

Appendix C: Some UCDs in Jennings et al. (2014)

thumbnail Fig. C.1

Upper panel: comparison of ACS and VEGAS-SSS aperture-corrected magnitudes for the GCs and UCDs. Dots and shaded histogram refer to GCs, UCDs are shown with empty circles (with symbol size scaled to Rh) and thick solid line histogram. The number of matched sources is reported together with the median and rms of the VEGAS-SSS to ACS mg difference. Lower panel: as upper, but the SExtractor AUTO magnitude is used for both GCs and UCDs.

Open with DEXTER

thumbnail Fig. C.2

Comparison of VEGAS-SSS g-band photometry (gVST label) with our photometry of ACS data (gACS tw) and with measurements of Jennings et al. (2014) (gACS J14). UCDs are shown with empty circles, GCs with dots. Left panel shows the mismatch for UCDs between VEGAS-SSS and ACS measurements from Jennings et al. (2014). Right panel: as left, but our measures for the ACS pointings are used.

Open with DEXTER

thumbnail Fig. C.3

Upper panels: comparisons of effective radii for UCD candidates obtained by Jennings et al. (2014) and VEGAS-SSS results. Upper left panel: comparison between VEGAS-SSS and ACS sizes from Jennings et al. (2014). Upper right panel: comparison between the Rh from VEGAS-SSS, and measurements with 2Dphot. Lower panel: photometric comparison for the same UCDs in upper panels. VEGAS-SSS photometry is taken as reference. The difference with respect to ACS data from Jennings et al. (2014), magnitudes derived with Ishape, and magnitudes from 2Dphot are shown with black circles, green triangles, and blue empty squares. In all panels symbols size are scaled to the Rh from Jennings et al. (2014).

Open with DEXTER

As discussed in Sect. 5 and shown in Fig. 9, we found a good match with ACS photometry. Figure C.1 shows that the photometric match is not as good for some of the UCD candidates in the catalog of Jennings et al. (empty circles). The mismatch cannot be simply explained by the different aperture correction, since, as described in Jennings et al. (2014, Sect. 2.4.1), the largest aperture correction is 0.94 mag, and we find differences up to ~2 mag.

To understand the problem, we downloaded one of the ACS pointings of NGC 3115 (choosing the one with the most UCDs over the frame), and independently derived the photometry of SSSs candidates using the same methods and tools as described in Sect. 3. Figure C.2 shows the comparison VEGAS-SSS magnitudes (gVST) with our photometry from ACS images (gACS tw) and with Jennings et al. (2014) (gACS J14). From this test, we find that, while the agreement for GC photometry is still acceptable, the large difference between ACS and VEGAS-SSS UCD data disappears (right panel in the figure). The large scatter for the GCs is mainly due to the use of only one of the ACS pointings available.

thumbnail Fig. C.4

Two examples of UCD in common with Jennings et al. (2014) showing large photometric scatter with respect to VEGAS-SSS. Left and middle panels: VST and ACS g image. Right panels: Ishape residual cutout. The UCD candidate is highlighted with a green circle in the VEGAS-SSS panels. Upper and lower panels refer to UCD10 and UCD20 in the catalog (15 and 10 zoom box).

Open with DEXTER

As an additional check, we also compared magnitudes and effective radii for VEGAS-SSS for the UCDs in common with the ACS pointing analyzed, using a different photometric tool, 2Dphot (La Barbera et al. 2008). We found a good match between the results of our standard procedures and those from 2Dphot (Fig. C.3, upper right and lower panels).

A visual inspection of some UCD candidates from Jennings et al. (2014) reveals possible problems with the identification of sources. The cases shown reveal that UCD10 (the object with the largest difference in the Figs. C.1, C.2) and UCD20 from the list of Jennings et al. are actually a spiral galaxy and an object immersed in system with clear merging features (tidal streams?).

All Tables

Table 1

Typical properties of various classes of SSS.

Table 2

Main properties of NGC 3115.

Table 3

Photometry and size estimates for the matched gi catalogs.

Table 4

Results of GMM at various galactocentric radii.

Table B.1

Main Ishape parameters used for the analysis.

All Figures

thumbnail Fig. 1

VST/OmegaCAM g-band image of NGC 3115. North is up, east is left. The image size is 52.5′ × 52.5′. The black square and circle mark the position of NGC 3115B and KK 084.

Open with DEXTER
In the text
thumbnail Fig. 2

Extinction map of the area. The color bar shows the extinction-to-color mapping.

Open with DEXTER
In the text
thumbnail Fig. 3

Photometry and sizes of SSS candidates in NGC 3115. Panel a) color–magnitude diagram for the full sample of g and i matched sources (gray dots) and for the sources with size estimates (black points). Panel b) apparent size-versus-color diagram. Panel c) apparent size-versus-magnitude diagram.

Open with DEXTER
In the text
thumbnail Fig. 4

Surface density histograms versus color. Panel a) Surface density for sources within RgalRbg. Darker color refers to areas with smaller galactocentric radii. Panel b) same as left, but for background sources at Rgal>Rbg. For the sake of clarity, the dotted histogram shows the density histogram times a factor 10. Panel c) residual surface density, Σ(RgalRbg) − Σbg.

Open with DEXTER
In the text
thumbnail Fig. 5

Position and width of the blue and red GCs (blue squares and red empty circles) at different Rgal as obtained from GMM. Symbol size is proportional to the fraction of objects associated with each peak. A fit to the data is shown with dot-dashed lines for the two subpopulations. Gray dots show spectroscopic confirmed GCs from Arnold et al. (2011). Full green points mark the running mean (median from equal number of data) of gray points. Gray long-dashed lines mark the rolling fits of the blue and red GC peaks as derived by Arnold et al. (2011) obtained from a combination of spectroscopically and photometrically selected GCs.

Open with DEXTER
In the text
thumbnail Fig. 6

Surface density profiles of blue, red, and total GC population (blue, red and black lines, respectively). The galaxy surface brightness profile in g band from Capaccioli et al. (in prep.) is also reported with green long-dashed line. The linear fit to the surface density is shown with dotted lines (blue-dotted, red-dashed for the blue and red GCs). The scale of the galaxy profile is arbitrary.

Open with DEXTER
In the text
thumbnail Fig. 7

Surface density histograms versus magnitude. Panels a) to c): as in Fig. 4, but magnitude is used instead of color. The vertical dotted line shows the position of the TOM. The ~0.2 mag tolerance for the peak position (gray shaded area) is also shown as well as the best-fit Gaussian to the GCLF (green line; mag, σTOM = 1.14). Panel d): edge and second-run edge (Edge2) functions.

Open with DEXTER
In the text
thumbnail Fig. 8

GCLF for the blue and red GC components analyzed separately. The upper panels, from left to right, show the total density distribution Σ(RgalRbg), the density after subtracting for background contamination ΣHost, and the edge and edge2 functions. The vertical dashed line marks the approximate position of the TOM, as obtained from the edge functions. The green line shows a Gaussian with peak at mag. Lower panels: as upper ones, but for the red GC components, with mag.

Open with DEXTER
In the text
thumbnail Fig. 9

Magnitude comparison between VEGAS-SSS and ACS. Left panels: a) and b) comparison for the full list of ACS to VEGAS-SSS matching sources (gray dots). The running mean and the corresponding rms are shown with black circles and error bars. The median difference, the rms and the number of objects matched are also labeled. Panels c) and d) are the same as the upper panels, but for sources with magnitude and photometric error cuts as labeled. Right: as left panels, but only GC candidates in the ACS catalog are considered.

Open with DEXTER
In the text
thumbnail Fig. 10

Size comparison between VEGAS-SSS and ACS. Left panels: effective radii for objects common to the ACS and VEGAS-SSS catalogs (full sample with gray dots, reference sample with black circles). For each sample, the median ratio between the ACS and VEGAS-SSS Rh is reported, the rmsMAD and the number of sources used are given in parentheses. The histograms shown in the upper and right insets use the same color coding as the central panel and are normalized to 1 at peak value for the full sample. The effective radii from ACS data are used for the selection (unshaded area). The diagonal line represents the 1:1 relation. Right panels: as left panels, but the selection is based on Rh from VEGAS-SSS.

Open with DEXTER
In the text
thumbnail Fig. 11

Rh distribution for ACS and VEGAS-SSS samples. From left to right: Rh distribution for the full list of matched ACS and VEGAS-SSS sources, for the sample selected using Rh,ACS, and for the sample selected using Rh,VST (see text). Shaded histograms refer to Rh,ACS distributions, solid thick lines to Rh,VST. All histograms are normalized to one at peak value.

Open with DEXTER
In the text
thumbnail Fig. 12

Magnitude-size diagram. Left panel: magnitude versus effective radius for the VEGAS-SSS field centered on NGC 3115. Color-coding for black and gray dots is as in previous figures, with the addition of the best sample selection (red empty circles). Yellow regions show the mean loci of the labeled SSS classes. The number of UCD and GC candidates, NUCD and NGC, for the reference and best sample (given in parenthesis) are also reported. Right panel: as left panel, but only sources with RgalRbg are plotted. The blue arrow shows the direction in which the points are shifted if the object lies at larger distance. We included the ACS sample using green symbols: GCs shown with circles, UCDs with squares. For the ACS sample, spectroscopically confirmed GCs from Arnold et al. (2011) are plotted as solid symbols.

Open with DEXTER
In the text
thumbnail Fig. 13

Radial surface density distribution of sources in the reference (black histograms) and best (red histogram) samples. Panel a) all Rh values are taken. The r1/4 fit to data is shown with gray lines. The dotted lines show the histograms without correction for areal coverage. The vertical long-dashed line is the position of limiting galactocentric radius Rbg. Panel b) as upper, but only GC candidates are plotted. The green-dashed line shows the galaxy surface brightness (from Capaccioli et al., in prep., g band, arbitrary scale). Panel c) as panel a), but for UCD candidates.

Open with DEXTER
In the text
thumbnail Fig. 14

Rh surface density distribution for the reference sample (black lines) and for the best sample (in red, Σ in units of number of objects per square arcminute). The panels, from upper to lower, show the surface density over the full inspected frame for sources at RgalRbg for the background area at Rgal>Rbg, and for the difference between the inner and outer 23. The vertical dashed line marks the 3.5 pc limit.

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

Completeness function estimated at different galactocentric radii. The label refers to the mean radius of the annulus in arcminutes. The dotted lines show the 90% completeness limit.

Open with DEXTER
In the text
thumbnail Fig. A.2

Upper panel: analytic GCLF (green line), background galaxies (blue), and completeness function (red) are combined in the total expected total luminosity function (black solid line). The GCLF peak (at x = 22.8) and 50% completeness limit (at x = 25.5) are shown with dashed vertical lines. Middle panel: edge detection function applied to the analytical formula of the total luminosity function. Lower panel: second-run edge detection.

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

Comparison of effective radii with different Ishape input parameters. Gray circles refer to the full list of matched sources, black dots to the reference sample. The numbers in each panel show the median Rh,test/Rh,ref ratio, with the rmsMAD and the number of objects used in parenthesis, labels are color coded.

Open with DEXTER
In the text
thumbnail Fig. B.2

As in the left panel of Fig. 10, but for the different Ishape tests (as labeled).

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

Upper panel: comparison of ACS and VEGAS-SSS aperture-corrected magnitudes for the GCs and UCDs. Dots and shaded histogram refer to GCs, UCDs are shown with empty circles (with symbol size scaled to Rh) and thick solid line histogram. The number of matched sources is reported together with the median and rms of the VEGAS-SSS to ACS mg difference. Lower panel: as upper, but the SExtractor AUTO magnitude is used for both GCs and UCDs.

Open with DEXTER
In the text
thumbnail Fig. C.2

Comparison of VEGAS-SSS g-band photometry (gVST label) with our photometry of ACS data (gACS tw) and with measurements of Jennings et al. (2014) (gACS J14). UCDs are shown with empty circles, GCs with dots. Left panel shows the mismatch for UCDs between VEGAS-SSS and ACS measurements from Jennings et al. (2014). Right panel: as left, but our measures for the ACS pointings are used.

Open with DEXTER
In the text
thumbnail Fig. C.3

Upper panels: comparisons of effective radii for UCD candidates obtained by Jennings et al. (2014) and VEGAS-SSS results. Upper left panel: comparison between VEGAS-SSS and ACS sizes from Jennings et al. (2014). Upper right panel: comparison between the Rh from VEGAS-SSS, and measurements with 2Dphot. Lower panel: photometric comparison for the same UCDs in upper panels. VEGAS-SSS photometry is taken as reference. The difference with respect to ACS data from Jennings et al. (2014), magnitudes derived with Ishape, and magnitudes from 2Dphot are shown with black circles, green triangles, and blue empty squares. In all panels symbols size are scaled to the Rh from Jennings et al. (2014).

Open with DEXTER
In the text
thumbnail Fig. C.4

Two examples of UCD in common with Jennings et al. (2014) showing large photometric scatter with respect to VEGAS-SSS. Left and middle panels: VST and ACS g image. Right panels: Ishape residual cutout. The UCD candidate is highlighted with a green circle in the VEGAS-SSS panels. Upper and lower panels refer to UCD10 and UCD20 in the catalog (15 and 10 zoom box).

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.