A Deeper Look at DES Dwarf Galaxy Candidates: Grus i and Indus ii

With the advent of high-precision, large-area surveys such as the Sloan Digital Sky Survey (SDSS; York et al. 2000), the Dark Energy Survey (DES; The Dark Energy Survey Collaboration 2005), Pan-STARRS (Chambers et al. 2016), the Hyper Suprime-Cam Subaru Strategic Program (Homma et al. 2016), MagLites (Drlica-Wagner et al. 2016; Torrealba et al. 2018), and DECam Local Volume Exploration Survey (DELVE; Mau et al. 2020), the number of known faint satellite systems that orbit the Milky Way (MW) has dramatically increased (Drlica-Wagner et al. 2015; Bechtol et al. 2015; Koposov et al. 2015). The ambiguity of what constitutes a galaxy increases as more systems are discovered that lie between the traditional loci of globular clusters and galaxies. Additionally, these low-luminosity systems challenge spectroscopic studies due to their low number of bright member stars (Willman & Strader 2012).

Many of these satellites discovered in the past decade are categorized as ultra-faint dwarf (UFD) galaxies (Simon 2019). With M_V ≳ −8 mag (M_* ≲ 10⁵ M_⊙; Martin et al. 2008; Bullock & Boylan-Kolchin 2017), UFDs overlap with bright globular clusters (GCs) in the size–luminosity plane. Though they overlap in this parameter space, UFDs and GCs likely have different formation mechanisms (Forbes et al. 2018). From their internal stellar kinematics, GCs are consistent with having little or no dark matter, and may be remnants of nucleated dwarf galaxies or may follow a completely separate evolutionary path (Bassino et al. 1994).

In contrast, the stellar kinematics of UFDs exhibit high M/L_V ratios (i.e., M/L_V ∼ 10³; Simon & Geha 2007) and represent the faintest end of the galaxy luminosity function. Dynamical mass measurements are one of the primary distinguishing characteristics between UFDs and GCs. In comparison to low-luminosity GCs, UFDs have larger sizes (r_h ≳ 30 pc), larger velocity dispersions (σ ≳ 3 km s⁻¹), and significant metallicity spreads (σ_[Fe/H] ≳ 0.3 dex), as shown in Simon & Geha (2007) and Martin et al. (2007).

As the most dark matter dominated objects visible in the universe, UFDs provide crucial, empirical information about the nature of dark matter and hierarchical structure at the smallest-scales (Frenk & White 2012; Bullock & Boylan-Kolchin 2017). In Λ-cold dark matter cosmology, structure forms hierarchically, with the UFDs corresponding to the galaxies in the smallest of dark matter halos (Sawala et al. 2015; Wheeler et al. 2015; Wetzel et al. 2016). Discerning the exact nature of MW satellites is therefore our paramount observational method to better constrain and compare cosmological models to low-luminosity systems. Firmly establishing the newly-discovered satellites as UFDs, and measuring their mass-to-light ratios, requires spectroscopic studies of a significant sample of their stars (e.g., Li et al. 2018).

However, due to the faintness of these systems, spectroscopy is only possible for a small sample of their stars, making a robust determination of their mass-to-light ratios difficult to obtain. In addition to spectroscopic studies, information on the structural parameters and stellar populations of UFDs may be obtained through deep photometric studies. For faint overdensities of stars like UFDs, this requires targeted imaging and precise photometry, in order to distinguish members of the systems from background stars and galaxies (e.g., Martin et al. 2008; Muñoz et al. 2012; Brown et al. 2014; Conn et al. 2018a, 2018b; Jerjen et al. 2018; Mutlu-Pakdil et al. 2018).

In this work we seek to clarify the nature of two objects detected in the DES footprint, Grus i and Indus ii, with deep Magellan/Megacam imaging.

Grus i was discovered by Koposov et al. (2015); however, its status as a GC or UFD has not yet been totally disentangled due to its faintness (M_V = −3.4) and the lack of deep, wide-field photometry. Follow-up studies based on the deep but small Gemini/GMOS-S field of view (FOV) photometry (Jerjen et al. 2018) were not able to determine the properties of Grus i because of its extension (${r}_{h}=1\buildrel{\,\prime}\over{.} 77$, Koposov et al. 2015).

Martínez-Vázquez et al. (2019) obtained a precise distance to Grus i of D_⊙ = (127 ± 6) kpc (μ₀ = 20.51 ± 0.10 mag) from the detection of two RR Lyrae members. They find that this distance would imply a change of 5% in its previously calculated physical size, consistent with the estimate of Koposov et al. (2015). Given the large uncertainties in the previous determinations of physical size, deep and extended imaging in Grus i is needed to firmly confirm this.

Complementary spectroscopic studies made of this system (Walker et al. 2016; Ji et al. 2019) were not able to decipher the nature of this object either, since the velocity dispersion could not be resolved because of the scarce sample of members detected.

Our second target, DES J2038–4609, was identified in Drlica-Wagner et al. (2015) as a low-confidence UFD candidate and will be referred to as Indus ii throughout the paper for convenience. The initial data for Indus ii were located in a survey region with atypical non-uniformity as they were taken part-way through the survey observations. The primary evidence for candidacy stems from a clump of apparent blue horizontal branch (BHB) stars at g ∼ 22. While Indus ii has been targeted in some dark matter indirect detection analyses (Albert et al. 2017), there are no other studies confirming the nature of the object. Given the uncertainty associated with this system, we chose to confirm whether this target was a gravitationally bound system due to Magellan/Megacam's FOV potentially covering 3 × r_h (Drlica-Wagner et al. 2015).

We follow similar methods to other studies that have confirmed the status of many MW satellites as dwarf galaxies (see, e.g., Sand et al. 2012; Crnojević et al. 2016; Kim et al. 2016; Luque et al. 2016; Carlin et al. 2017; Conn et al. 2018a, 2018b; Luque et al. 2018; Mutlu-Pakdil et al. 2018). Our data complements other studies by utilizing a larger FOV (necessary for the potentially larger extents), while still resolving magnitudes ∼3 mags deeper than the discovery papers.

This paper is structured as follows. In Section 2 we describe the Magellan/Megacam and DES observations, photometry, and catalog selection. We present the likelihood method used to infer structural parameters in Section 3. In Section 4 we report the results from the statistical analysis, and Section 5 compares our results to previous results and concludes this work.

2.1. Data

We observed Indus ii and Grus i over four nights in 2017 April with the Megacam instrument (McLeod et al. 2015) at the f/5 focus of the 6.5 m Magellan Clay telescope. Megacam is an imager composed of 36 CCDs of 2048 × 4608 pixels, creating a square array with a FOV of ∼24' × 24'(see Figure 1). The data were binned 2 × 2 resulting in a pixel scale of 016. Observations were dithered such that each image is offset by +5'' in R.A. and +13'' in decl. from the previous one. This reduces the impact of the small gaps between the CCDs.

Zoom In Zoom Out Reset image size

The data were reduced using the Megacam pipeline developed at the Harvard-Smithsonian Center for Astrophysics. ⁵² This pipeline includes tasks such as bias subtraction, flat-fielding, and cosmic ray correction. In addition, the pipeline derives astrometric solutions using the Two Micron All Sky Survey (2MASS) survey (Skrutskie et al. 2006). The images were then re-sampled, with a lanczos3 interpolation function, and combined with a weighted average using SWarp (Bertin et al. 2002). This process produced a final, stacked g- and r-band image for each object. An observing log can be found in Table 1.

Table 1. Observing Log of Magellan/Megacam Observations in the g- and r−bands for Grus i and Indus ii

Object	UT Date	Filter	N × texp	Seeing
			(s)	('')
Grus i	2017 Apr 23	g	7 × 300	0.7
	2017 Apr 24	r	8 × 300	0.9
Indus ii	2017 Apr 21	g	8 × 300	0.6
	2017 Apr 22	r	8 × 300	0. 5

Download table as:  ASCIITypeset image

2.2. Megacam Photometry

2.3. DES Photometry

2.4. Transformation from Megacam to DES

The matched objects found in the previous section were used to find a transformation between DES magnitudes and Megacam instrumental magnitudes. A color cut of ${({g}_{0}-{r}_{0})}_{\mathrm{DES}}\lt 1.2$ was applied to remove a clump of M0 and redder stars. We used only stars that had DES photometric errors less than 0.03 mag. These criteria ensure that a high-quality stellar sample is utilized in finding the magnitude system transformation.

To perform this transformation, we solve for the coefficients of the following equation using a generalized least squares regression:

$$\begin{eqnarray}&&{M}_{\mathrm{DES}}={m}_{\mathrm{instr}}+\beta +\alpha {({g}_{0}-{r}_{0})}_{\mathrm{DES}},\end{eqnarray} \tag{ 1 }$$

where β is the zero-point offset and α is the color coefficient. To find the true distribution of ${M}_{\mathrm{DES}}-{m}_{\mathrm{instr}}$, we run the catalog through a sigma-clipping algorithm based on the median absolute deviation. Stars that lie outside 3σ are clipped until the distribution converges. Equation (1) is then applied to all of the instrumental magnitudes found from ALLFRAME. A second-order fit was explored and deemed unnecessary. The coefficients of this fit are in the third and fourth columns in Table 2 and the difference between transformed Megacam magnitudes and DES magnitudes of stars used to find the transformation can be seen in Figure 2.

Zoom In Zoom Out Reset image size

Table 2. The Photometric Transformations Between Megacam Instrumental Magnitudes and the DES Photometric System

Object	Filter	β	α	std (Δmag) a	Sharp	Chi	# of Stars	Magnitude Range b
Grus i	g	7.554	−0.136	0.036	(−0.7, 1.2)	1.65	6743	(15.6, 26.7)
	r	7.651	−0.027	0.023	(−0.5, 0.7)	1.25	6743	(15.2, 26.3)
Indus ii	g	7.596	−0.167	0.028	(−0.5, 0.3)	2.05	5520	(15.2, 26.6)
	r	7.657	−0.029	0.021	(−0.7, 0.2)	4.91	5520	(14.8, 26.8)

Notes. The fifth column displays the difference between the transformed Megacam magnitudes and the DES magnitudes. The sixth (sharp range) and seventh (median chi-value) columns detail the morphological cuts made on the instrumental photometry. The last two columns detail the number of stars and magnitude range of the final Megacam+DES stellar catalogs after these cuts were applied.

^a The median absolute standard deviation of the difference between DES magnitudes and transformed Megacam magnitudes. ^b The faint magnitude limits correspond to S/N ∼ 5.

Download table as:  ASCIITypeset image

We created the final stellar catalog by applying morphological cuts using the statistics sharp and χ which were determined during the PSF fitting. Sharp can be approximated as sharp${}^{2}\sim {\sigma }_{\mathrm{obs}}^{2}-{\sigma }_{\mathrm{PSF}}^{2}$, where σ_obs is the observed photometric error and σ_PSF is the expected photometric error (Stetson 1987).

The second statistic, χ, is the ratio of observed pixel-to-pixel scatter over expected scatter, determined from the intrinsic scatter in the PSF models. Star galaxy separation begins to break down at fainter magnitudes, i.e., g₀ ∼ 25.5 and r₀ ∼ 24.75. The details of these cuts and the magnitude range of the final stellar catalogs can be found in the last four columns in Table 2. In addition, the final catalog's brighter magnitudes are supplemented by the DES stellar objects where Megacam saturates at g₀ ∼ 18 and r₀ ∼ 17.5. A portion of these catalogs can be seen in Tables 3 and 4.

Table 3. The Final Calibrated Stellar Catalog for Grus i—Sorted by Star ID

Star ID	R.A.	Decl.	g0,DES	σg	r0,DES	σr
	(deg)	(deg)	(mag)	(mag)	(mag)	(mag)
12543	344.134	−50.285	25.398	0.106	24.807	0.068
12845	344.139	−50.284	22.623	0.007	22.402	0.007
13343	344.138	−50.281	24.075	0.029	24.044	0.036
14597	344.103	−50.278	24.745	0.047	24.361	0.049
14730	344.168	−50.278	24.956	0.070	24.902	0.076
15406	344.187	−50.275	17.679	0.003	17.276	0.002

Note. This table is published in its entirety in machine-readable format. A portion is shown here for guidance regarding its form and content. All magnitudes are in the DES magnitude system.

Only a portion of this table is shown here to demonstrate its form and content. A machine-readable version of the full table is available.

Download table as:  DataTypeset image

Table 4. The Final Calibrated Stellar Catalog for Indus ii—Sorted by Star ID

Star ID	R.A.	Decl.	g0,DES	σg	r0,DES	σr
	(deg)	(deg)	(mag)	(mag)	(mag)	(mag)
17138	309.735	−46.284	25.845	0.125	25.532	0.090
17539	309.705	−46.283	25.911	0.117	25.658	0.100
17819	309.691	−46.282	21.532	0.007	20.891	0.006
18608	309.677	−46.278	21.129	0.005	20.183	0.010
18821	309.764	−46.278	25.005	0.064	24.897	0.052
19208	309.689	−46.276	21.143	0.006	20.198	0.010

Note. This table is published in its entirety in machine-readable format. A portion is shown here for guidance regarding its form and content.

Only a portion of this table is shown here to demonstrate its form and content. A machine-readable version of the full table is available.

Download table as:  DataTypeset image

Figure 3 shows the color–magnitude diagrams (CMDs) created for both Grus i and Indus ii using the final calibrated stellar catalog as it was described in this section. The uncertainties show that the photometric signal-to-noise ∼10 to a depth ∼3 magnitudes below that of the discovery papers.

Zoom In Zoom Out Reset image size

We utilize the Ultra-faint Galaxy Likelihood toolkit to determine structural parameters and the best-fitting isochrones for Grus i and Indus ii. Here we review the aspects of UGaLi that are important for our analysis, and refer to Bechtol et al. (2015) and the appendix of Drlica-Wagner et al. (2020) for a more detailed description.

Our data sample consists of the magnitude and the error on the magnitude in two filters, ${{ \mathcal D }}_{c,i}=\{{g}_{i},{\sigma }_{{g}_{i}},{r}_{i},{\sigma }_{{r}_{i}\}}$, and the spatial positions of the stars ${{ \mathcal D }}_{s,i}=\{{\alpha }_{i},{\delta }_{i}\}$. We define the probability distribution for the structural parameters as u_s , and the probability distribution for the parameters of the isochrone as u_c . The total probability distribution function (PDF) for the data ${{ \mathcal D }}_{i}=\{{{ \mathcal D }}_{s,i},{{ \mathcal D }}_{c,i}\}$ given the model parameters θ is then

$$\begin{eqnarray}&&u({{ \mathcal D }}_{i}| {\boldsymbol{\theta }})={u}_{s}({{ \mathcal D }}_{s,i}| {{\boldsymbol{\theta }}}_{s})\times {u}_{c}({{ \mathcal D }}_{c,i}| {{\boldsymbol{\theta }}}_{c}).\end{eqnarray} \tag{ 2 }$$

This probability distribution is defined such that the integral of it over the entire spatial and magnitude domain is unity.

For the structural properties, u_s , we assume an elliptical Plummer model, with a projected density distribution (Plummer 1911; Martin et al. 2008),

$$\begin{eqnarray}&&{\rm{\Sigma }}(R)\propto {\left[1+{\left(\displaystyle \frac{{R}_{i}}{{R}_{p}}\right)}^{2}\right]}^{-2}.\end{eqnarray} \tag{ 3 }$$

Here R_i is the elliptical radius coordinate from the center of the galaxy, and R_p is the Plummer-scale radius (equivalent to the 2D azimuthally-averaged halflight radius, ${r}_{h}={a}_{h}\sqrt{1-\epsilon }$). There are five model parameters that describe the Plummer profile: the centroid coordinates (α₀, δ₀,), the semimajor halflight radius (a_h ), the ellipticity (), and the position angle (ϕ). The density distribution is further related to spatial position by

$$\begin{eqnarray}\begin{array}{rcl}{R}_{i} & = & \left\{{\left[\displaystyle \frac{1}{1-\epsilon }\left({X}_{i}\cos \phi -{Y}_{i}\sin \phi \right)\right]}^{2}\right.\\ & & {\left.-{({X}_{i}\sin \phi +{Y}_{i}\cos \phi )}^{2}\right\}}^{\displaystyle \frac{1}{2}}\end{array}\end{eqnarray} \tag{ 4 }$$

and spatial position is related to the object centroid by

$$\begin{eqnarray}&&{X}_{i}-{X}_{0}=({\alpha }_{i}-{\alpha }_{0})\cos ({\delta }_{0})\end{eqnarray} \tag{ 5 }$$

and

$$\begin{eqnarray}&&{Y}_{i}-{Y}_{0}={\delta }_{i}-{\delta }_{0}.\end{eqnarray} \tag{ 6 }$$

For the isochrone properties, u_c , we calculate the PDF by binning the color–magnitude information over a grid of isochrones that are weighted by a Chabrier IMF (Chabrier 2003) and have a fixed solar alpha abundance. These isochrones are described in terms of all metallicities are reported as [Fe/H] $={\mathrm{log}}_{10}\left(\tfrac{Z}{{Z}_{\odot }}\right)$, with Z_⊙ = 0.0152.

The grid of PARSEC isochrones are representative of old metal-poor stellar populations, i.e., 0.0001 < Z < 0.001, 1 Gyr < τ < 13.5 Gyr, and 16.0 < m −M < 25.0 to fit the CMD properties of each object (Bressan et al. 2012). We check that our results do not depend on the isochrone model by comparing to Dotter (2016) and find that they are insensitive to this specific assumption.

With the above model, we can define the Poisson log-likelihood

$$\begin{eqnarray}&&\mathrm{log}{ \mathcal L }=-\lambda {N}_{s}-\sum _{i}^{\mathrm{stars}}\mathrm{log}(1-{p}_{i}),\end{eqnarray} \tag{ 7 }$$

where λ, the stellar richness, is a normalization parameter representative of the total number of member stars with M_* > 0.1 M_⊙ in the satellite, N_s is the fraction of observable satellite member stars, and _pi is the probability that a star is a member of the satellite.

Because we choose to normalize the signal PDF to unity, we can interpret λ as the total number of stars in the satellite (observed + unobserved). The membership probability is given by

$$\begin{eqnarray}&&{p}_{i}=\displaystyle \frac{\lambda {u}_{i}}{\lambda {u}_{i}+{b}_{i}},\end{eqnarray} \tag{ 8 }$$

where b_i is the background density function (for more details see Appendix C in Drlica-Wagner et al. 2020). We take the background density function to be independent of spatial position in our region of interest (ROI).

The empirical background density function, b_i , is determined from an annulus ($7\buildrel{\,\prime}\over{.} 2\lt r\lt 12^{\prime}$) surrounding our target ROI (r < 72). We require the ROI to be ≳2 × r_h . This is the maximum ROI that still allows for the background annulus to contain ∼3 × r_h (Martin et al. 2008), where r_h is from Drlica-Wagner et al. (2015); Koposov et al. (2015). Figure 1 depicts these regions as blue circles. Any non-stellar objects that still contaminate the data at greater magnitudes are expected to do so equally over the entire FOV and therefore averaged within b_i .

With λ allowed to vary and b_i held fixed, we simultaneously explore the whole parameter space with flat priors for all parameters except r_h (an inverse prior). With UGaLi, we run an MCMC chain with 100 walkers, 12,000 steps, and 1000 burn-in (Foreman-Mackey et al. 2013). Absolute V-band magnitude is determined following the prescription of Martin et al. (2008)

4.1. Grus i

Column 1 of Table 5 lists the parameters obtained from the median peak likelihood of the posterior distributions (see Figure 4). With our improved parameters for Grus i, we find it to be consistent with an extended ultra-faint dwarf galaxy that resides at the faint edge (μ ∼ 30 mag) of the galaxy locus in the size–luminosity plane. Figure 5 shows the relationship between GCs and UFDs in this parameter space and also shows the updated location of Grus i. Table 5 lists the parameters that were derived in previous works (both Grus i and Indus ii are represented).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Table 5. Photometric and Spectroscopic Parameters of Grus i and Indus ii Found in the Literature Prior to This Work

	Grus i							Indus ii
	This Work
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
α2000 (deg)	${344.166}_{-0.006}^{+0.007}$	344.1765	⋯	344.1700	344.1797	⋯	⋯	309.76
δ2000 (deg)	$-{50.168}_{-0.005}^{+0.006}$	−50.1633	⋯	−50.1641	−50.1800	⋯	⋯	−46.16
t-value (σ)	21.3	10.1	⋯	⋯	⋯	⋯	⋯	∼5.7
MV (mag)	−4.1 ± 0.3	−3.4 ± 0.3	⋯	⋯	⋯	⋯	⋯	−4.3 ± 0.19
D⊙ (kpc)	${125}_{-12}^{+6}$	120	⋯	115 ± 6	⋯	⋯	⋯	214 ± 16
rh (arcmin)	${4.16}_{-0.74}^{+0.54}$	${1.77}_{-0.39}^{+0.85}$ b	⋯	⋯	0.81 ± 0.66 b	⋯	⋯	${2.9}_{1.0}^{1.1}$ b
rh (pc)	${151}_{-31}^{+21}$	${62}_{-13.6}^{+29.8}$	⋯	⋯	28.3 ± 23.0	⋯	⋯	181 ± 67
	${0.44}_{-0.10}^{+0.08}$	${0.41}_{-0.28}^{+0.20}$	⋯	⋯	0.45 ± 0.30	⋯	⋯	<0.4
ϕ (deg)	${153}_{-7.0}^{+8.0}$	4 ± 60	⋯	⋯	23 ± 18	⋯	⋯	⋯
m − M (mag)	${20.48}_{-0.22}^{+0.11}$ a	20.4	⋯	20.30 ± 0.11	⋯	⋯	20.51 ± 0.10 c	⋯
τ (Gyr)	${13.26}_{-0.25}^{+0.18}$	⋯	⋯	${14.0}_{-1.0}^{+1.0}$	⋯	⋯	⋯	⋯
[Fe/H] (dex)	$-{1.88}_{-0.03}^{+0.09}$	⋯	$-{1.42}_{-0.42}^{+0.55}$	$-{2.5}_{-0.3}^{+0.3}$	−2.5 ± 0.3	−2.57, −2.50	⋯	⋯
σ[Fe/H] (dex)	⋯	⋯	<0.9	⋯	⋯	⋯	⋯	⋯
$\left\langle {v}_{\mathrm{LOS}}\right\rangle$ (km s−1)	⋯	⋯	$-{140.5}_{-1.6}^{+2.4}$	⋯	⋯	⋯	⋯	⋯
${\sigma }_{{v}_{\mathrm{LOS}}}$ (km s−1)	⋯	⋯	<9.8	⋯	⋯	⋯	⋯	⋯
M⊙/LV,⊙	⋯	⋯	<2645	⋯	⋯	⋯	⋯	⋯

Notes. The columns for Grus i, in order, are from this work, Koposov et al. (2015); Walker et al. (2016); Jerjen et al. (2018); Muñoz et al. (2018); Ji et al. (2019); Martínez-Vázquez et al. (2019). Column 8 describes Indus ii as found in the discovery paper, Drlica-Wagner et al. (2015).

^a A systematic uncertainty of 0.1 mag was added to account for the difference between best-fits for Bressan et al. (2012) and Dotter (2016) isochrones. ^b Semimajor halflight radii converted from azimuthally-averaged radii with $\sqrt{1-\epsilon }$ factor. ^c This distance measurement is based on two RR Lyrae stars found in Grus i.

Download table as:  ASCIITypeset image

Previous spectroscopic studies by Walker et al. (2016) and Ji et al. (2019) find the brightest potential member stars are very metal-poor (e.g., [Fe/H] ∼ −2.3), which is in agreement with the photometric metallicity found in Jerjen et al. (2018). In contrast, this and other photometric studies, such as Koposov et al. (2015), have found it to be less metal-poor, i.e., [Fe/H] ≲ −2. In agreement with these studies, several fainter potential member stars were found to be less metal-poor (Walker et al. 2016).

With a a_h ∼ 202 pc and M_V ∼ −4.1, we find Grus i is both larger and brighter than estimates from previous works (Koposov et al. 2015; Jerjen et al. 2018; Muñoz et al. 2018). While photometric metallicities are not as reliable as spectroscopic, according to our results, Grus i is among the more metal-rich ([Fe/H] ∼ −1.9) UFDs found to date (Simon 2019). As previous analyses have found mixed results, a larger spectroscopic sample is required to confirm the metallicity of this object.

This discrepancy between spectroscopic and photometric metallicities has been seen in previous studies (see Section 4.3 in Caldwell et al. 2017). In that case, it was considered more likely that the spectroscopic results were systematically metal-poor. There was very good agreement with the isochrone calculated with the photometric metallicity and probable member stars.

Figure 6 shows the CMD of the stars within 2 × r_h and the CMD of the background (see the first and second panel, respectively). The last three panels of Figure 6 are Hess diagrams of the background stellar density, the stellar density within 2 × r_h , and the difference of the two (see panels 3, 4, and 5, respectively). Overlaid in the first, fourth, and fifth panels is a PARSEC isochrone representative of an old, metal-poor population with τ = 13.3 Gyr and [Fe/H] = −1.9. This isochrone agrees with the best-fit properties of Grus i inferred from the maximum-likelihood distribution.

Zoom In Zoom Out Reset image size

For Grus i the background-subtracted Hess diagram shown in the fifth panel of Figure 6 clearly illuminates MSTO and MS that are well-represented by the inferred properties. Less obvious, but still well-populated, the isochrone clearly delineates an HB and RGB population. It should be noted that some potential members can still be seen in the second top-row panel (background CMD) of Figure 6 due to Grus i's large extent.

In the left two panels of Figure 7, we show the distribution of the UGaLi membership probabilities in sky coordinates and color–magnitude space. These membership probabilities were determined as described in Section 3. The three panels in this figure show that our inferred parameters describing the stellar population and morphology of Grus i are consistent with a theoretical Plummer profile.

Zoom In Zoom Out Reset image size

These probabilities were further used to create a binned and weighted density profile as seen in the far right panel of Figure 7. There are an equal number of stars in each bin. It can be seen that the binned data fits well over the Plummer model profile shown as the dashed line.

The posterior distributions and maximum-likelihood peak values are shown in Figure 4. While some of the properties shown in Figure 4 agree with previous works (see Table 5) within the uncertainties (e.g., centroid coordinates, ellipticity, distance modulus), others have shifted slightly in this work (i.e., r_h ), changing some of the derived properties.

Jerjen et al. (2018) find two small overdensities at [(α − α₀), (δ − δ₀)] ≈ [+0.2, −0.5] (arcmin) and [(α − α₀), (δ − δ₀)] ≈ [−0.6, +0.8] (arcmin)–with extents of 22 × 25 pc and 13 × 28 pc, respectively. It is interesting to note that we do not find obvious evidence of the two slight overdensities or diffuse centroid found in Jerjen et al. (2018). Our centroid shift does not seem to be significant or dependent on any lack of dense central overdensity as can be seen in Figure 8. The dashed yellow line in this figure indicates the halflight radius created with our inferred parameters.

Zoom In Zoom Out Reset image size

The r_h (416) found in this work is larger than previous works by more than a factor of 2. Our larger FOV (see Figure 1) allows us to more accurately constrain the local background contamination and is likely the reason for the change in extent. Additionally, we find Grus i to be about one magnitude brighter (M_V ∼ −4.1 mag) than previously thought (Koposov et al. 2015), while the distance to the object is in agreement with the recently updated distance determination based on RR Lyrae stars (Martínez-Vázquez et al. 2019).

It should be noted that Jhelum, a nearby stellar stream (D_⊙ ∼ 13 kpc, m − M ∼ 15.6; Shipp et al. 2018), potentially contaminates the FOV. In order to test this, we cut potential stream member stars from our catalog and performed the same analysis on the new catalog. These potential members were chosen based on Jhelum's spatial footprint and location in color–magnitude space. The width of the area in color–magnitude space was chosen to account for our photometric uncertainties. The distance of the Jhelum stream (D_⊙ ∼ 13 kpc) compared to how far Grus i is precludes any physical association between the two. The results from this analysis were similar within uncertainties. Therefore, we determined that the presence of the stream does not significantly affect our analysis.

4.2. Indus ii

We confirm the status of Grus i as a likely dwarf galaxy with the results of an MCMC algorithm and fit to a Plummer density model with deep Magellan/Megacam follow-up photometry of the objects Grus i and DES J2038–4609. This photometry reaches ∼2–3 magnitudes deeper than the discovery data, allowing us to derive (i) improved distance, which is in agreement with the distance obtained using RR Lyrae distance indicators (Martínez-Vázquez et al. 2019), (ii) luminosity, 1 mag brighter than Koposov et al. (2015), and (iii) structural parameters, particularly finding that the r_h is two times larger. We find that DES J2038–4609 is a false positive that was flagged due to a chance projection of an overdensity of stars.

Grus i is an extended (${r}_{h}\approx 4\buildrel{\,\prime}\over{.} 16$), elliptical ( = 0.44) dwarf galaxy with a distance of 125 kpc. Like other dwarfs, Grus i has an old single, stellar population (13.3 Gyr) with low metallicity ([Fe/H] = −1.9). Its luminosity (M_V ≈ −4.1) and azimuthal halflight radius (151 pc) place it at the lower edge of the dwarf locus in the size–luminosity plane (see Figure 5).

Our analysis complements previous studies with a larger FOV and deeper photometry, allowing us to confirm the suggestion of Walker et al. (2016) that Grus i was likely larger in extent. ⁵³ We also find that Grus i is slightly less metal-poor than most UFDs, with [Fe/H] ≲ −1.9 although not as metal-rich as suggested in Walker et al. (2016).

In this work, we reach the necessary FOV (24' × 24') which allows us to improve upon and find new structural parameters of Grus i. Martin et al. (2008) determined that a FOV three times the halflight radius is necessary to accurately constrain the structural properties of UFDs. Therefore, with r_h ∼ 416', our 24' × 24' FOV is just large enough to derive accurate structural parameters.

Previous spectroscopic studies find mixed results with respect to the average metallicity of this object. There are two well-measured, brighter member stars that are consistent with old, metal-poor UFDs ([Fe/H] ∼ −2.5), but Walker et al. (2016) found five faint stars that suggested a metallicity of [Fe/H] ∼ −1.4. It is possible that these fainter stars have a systematic uncertainty or bias that cause Grus i to appear more metal-rich than it is.

We analyze DES J2038–4609 with background-subtracted Hess diagrams and UGaLi and find that the distribution of stars does not correlate with any isochrone or Plummer model. A chance alignment of possible BHB stars contributed to the original detection of DES J2038–4609 as a candidate satellite. Constraining power for this overdensity comes from a set of BHB stars. This feature can be seen in our data set as well, but does not match a corresponding MSTO that is consistent with a UFD. We conclude that DES J2038–4609 is not consistent with either a dwarf galaxy or a globular cluster.

Ongoing follow-up studies of UFDs will continue to have important implications for our understanding of near-field cosmology. As the most dark matter dominated objects and the only resolved examples of these old, relatively pristine stellar populations, it is important that their nature is well understood. Understanding their nature contributes to the characterization of the satellite population of the MW, leading to more accurate inferences about galaxy formation physics and the nature of dark matter. Studies with deep and wide follow-up photometry, such as this one, are useful to help characterize these faint objects.

This paper has gone through internal review by the DES collaboration. The authors thank Taylor A. Hutchison, Jonathan H. Cohn, and Peter S. Ferguson for their insightful conversation and support throughout this analysis. S.A.C. acknowledges support from the Texas A&M University and the George P. and Cynthia Woods Institute for Fundamental Physics and Astronomy. A.B.P. acknowledges support from NSF grant AST-1813881. J.D.S. acknowledges support from NSF grant AST-1412792. This research made use of Astropy, ⁵⁴ a community-developed core Python package for Astronomy (Astropy Collaboration et al. 2013; The Astropy Collaboration et al. 2018). This research made extensive use of arXiv.org and NASA's Astrophysics Data System for bibliographic information.

Funding for the DES projects has been provided by the U.S. Department of Energy, the U.S. National Science Foundation, the Ministry of Science and Education of Spain, the Science and Technology Facilities Council of the United Kingdom, the Higher Education Funding Council for England, the National Center for Supercomputing Applications at the University of Illinois at Urbana-Champaign, the Kavli Institute of Cosmological Physics at the University of Chicago, the Center for Cosmology and Astro-Particle Physics at the Ohio State University, the Mitchell Institute for Fundamental Physics and Astronomy at Texas A&M University, Financiadora de Estudos e Projetos, Fundação Carlos Chagas Filho de Amparo à Pesquisa do Estado do Rio de Janeiro, Conselho Nacional de Desenvolvimento Científico e Tecnológico and the Ministério da Ciência, Tecnologia e Inovação, the Deutsche Forschungsgemeinschaft and the Collaborating Institutions in the Dark Energy Survey.

The Collaborating Institutions are Argonne National Laboratory, the University of California at Santa Cruz, the University of Cambridge, Centro de Investigaciones Energéticas, Medioambientales y Tecnológicas-Madrid, the University of Chicago, University College London, the DES-Brazil Consortium, the University of Edinburgh, the Eidgenössische Technische Hochschule (ETH) Zürich, Fermi National Accelerator Laboratory, the University of Illinois at Urbana-Champaign, the Institut de Ciències de l'Espai (IEEC/CSIC), the Institut de Física d'Altes Energies, Lawrence Berkeley National Laboratory, the Ludwig-Maximilians Universität München and the associated Excellence Cluster Universe, the University of Michigan, the National Optical Astronomy Observatory, the University of Nottingham, The Ohio State University, the University of Pennsylvania, the University of Portsmouth, SLAC National Accelerator Laboratory, Stanford University, the University of Sussex, Texas A&M University, and the OzDES Membership Consortium.

Based in part on observations at Cerro Tololo Inter-American Observatory, National Optical Astronomy Observatory, which is operated by the Association of Universities for Research in Astronomy (AURA) under a cooperative agreement with the National Science Foundation.

The DES data management system is supported by the National Science Foundation under grant Nos. AST-1138766 and AST-1536171. The DES participants from Spanish institutions are partially supported by MINECO under grants AYA2015-71825, ESP2015-66861, FPA2015-68048, SEV-2016-0588, SEV-2016-0597, and MDM-2015-0509, some of which include ERDF funds from the European Union. IFAE is partially funded by the CERCA program of the Generalitat de Catalunya. Research leading to these results has received funding from the European Research Council under the European Union's Seventh Framework Program (FP7/2007–2013) including ERC grant agreements 240672, 291329, and 306478. We acknowledge support from the Brazilian Instituto Nacional de Ciência e Tecnologia (INCT) e-Universe (CNPq grant 465376/2014-2).

This manuscript has been authored by Fermi Research Alliance, LLC under Contract No. DE-AC02-07CH11359 with the U.S. Department of Energy, Office of Science, Office of High Energy Physics.