LACERTA I AND CASSIOPEIA III. TWO LUMINOUS AND DISTANT ANDROMEDA SATELLITE DWARF GALAXIES FOUND IN THE 3π PAN-STARRS1 SURVEY

The salience of satellite dwarf galaxies for understanding galaxy formation in a cosmological context has been made all the more evident in the past decade with the discovery of numerous faint Local Group galaxies. These faint systems are not only important to understand the faint end of galaxy formation (e.g., Koposov et al. 2009; Revaz & Jablonka 2012; Vargas et al. 2013) but also their distribution around their host can test the hierarchical formation induced by the favored cosmological paradigm (e.g., Pawlowski et al. 2012; Ibata et al. 2013). The Andromeda satellite system is arguably an ideal system to study in order to tackle these questions as its proximity renders it easily observable, while our outsider's perspective facilitates its complete mapping, provided one has access to high-quality, multi-color, and wide-field photometry.

The number of known Andromeda satellite dwarf galaxies has soared over the last decade, thanks mainly to two large photometric surveys of this region of the sky. The Sloan Digital Sky Survey (SDSS) enabled the discovery of a handful of relatively bright (M_V ≲ −8.5) systems, initially from a dedicated stripe along the major axis of the Andromeda galaxy (And IX, And X; Zucker et al. 2004, 2007), which was later complemented by a more systematic coverage of parts of the region around M31 in Data Release 8 (And XXVIII, And XXIX; Slater et al. 2011; Bell et al. 2011). The Pan-Andromeda Archaeological Survey (PAndAS) is, on the other hand, a survey dedicated to the study of the stellar populations within the halo of the Andromeda galaxy, out to distances of ∼150 kpc. Alone, it has enabled the discovery of 16 unambiguous new Andromeda satellite dwarf galaxies (Martin et al. 2006, 2009; Ibata et al. 2007; McConnachie et al. 2008; Richardson et al. 2011) with total magnitudes ranging from M_V ≃ −6.0 to M_V ≃ −10.0.

The great success of these surveys in unveiling the M31 satellite system also highlights their coverage limitations. And VI, VII, and XXVIII are all examples of dwarf galaxies at projected distances of at least about 30° (corresponding to ∼400 kpc M31-centric distances), yet the SDSS has only surveyed regions that stay well away from the Milky Way disk, mostly south of M31, and the PAndAS is limited to a cone of ∼10° from M31. The Panoramic Survey Telescope and Rapid Response System 1 (Pan-STARRS1; Kaiser et al. 2010) is a photometric survey that does not have these limitations and, with final stacked data reaching slightly deeper than the SDSS (Metcalfe et al. 2013), it has the potential to provide a much more complete understanding of the M31 dwarf galaxy satellite system out to and beyond the virial radius of Andromeda (∼300 kpc).

In this paper, we present the discovery of the two dwarf galaxies found in stacked Pan-STARRS1 (PS1) imaging data. As they are located in the Lacerta and Cassiopeia constellations, but are also likely Andromeda satellites, we follow the dual-naming convention emphasized in Martin et al. (2009) to avoid ambiguity. We therefore dub the first dwarf galaxy both Lacerta I and Andromeda XXXI (Lac I/And XXXI) and the second one both Cassiopeia III and Andromeda XXXII (Cas III/And XXXII). The data set used for the discovery is summarized in Section 2, while the derivation of the dwarf galaxies' properties is described in Section 3, before we conclude in Section 4.

The PS1 3π survey (K. C. Chambers et al. 2013, in preparation) is a systematic imaging survey of the sky north of δ = −30° in five optical and near-infrared photometric bands (g_P1r_P1i_P1z_P1y_P1; Tonry et al. 2012). Conducted on a dedicated 1.8 m telescope located on Haleakala, in Hawaii, the system relies on a 1.4 Gpixel, 33 field-of-view camera for rapid mapping of the sky. Any location in the survey is observed repeatedly for a planned four times per year per filter, conditions permitting, with exposure times of 43/40/45/30/30 s in the g_P1/r_P1/i_P1/z_P1/y_P1 bands, respectively (Metcalfe et al. 2013). The median seeing values in these bands are 1.27/1.16/1.11/1.06/1.01 arcsec. Images are automatically processed through the image processing pipeline (Magnier 2006, 2007; Magnier et al. 2008) to produce the stacked data catalog.

At the time of this paper, the region around Andromeda had been observed for three consecutive seasons from 2010 May, with up to 12 exposures per filter. Chip gaps, poor observing conditions, and technical problems, however, mean that a varying number of exposures are available for stacking at any given location. For point sources, this leads to 10σ-completeness limits⁸ of 22.1 in the r_P1 band and 22.2 in the i_P1 band for the surroundings of Lac I. Around Cas III, these limits are 22.0 and 21.9, respectively. Checking the spatial completeness from the presence of Two Micron All Sky Survey stars, bright enough to be observed with PS1, yields an area coverage factor >95% in either band.

Flags were used to remove spurious detections that are above the saturation limit, blended with other sources, thought to be defective, saturated pixels, or are considered significant cosmic rays or extended sources. Sources for which either the point-spread function fit is bad, the moments, the local sky, or the local sky variance could not be measured, were also removed from the data set. Finally, detections where the peak lands on a diffraction spike, a ghost, a glint, or near a chip edge were discarded as well, in addition to sources with signal to noise below five.

The photometric calibration of the stacked PS1 data is still under development with spurious offsets in the photometry for subsets of detections. In the region of Lac I, a comparison of the stacked photometry with that of the single exposure PS1 observations, calibrated at the 0.01 mag level by Schlafly et al. (2012), yields no significant offset in the i_P1 band, whereas the r_P1 band has ∼20% of sources offset by ∼0.2 mag. In the Cas III surroundings, a small fraction of measurements are affected at the 0.05 level in the r_P1 band and at the 0.1 level in the i_P1 band. However, since the majority of point sources have a reliable photometry, we have decided to keep these discrepant sources in the catalog.

Throughout, the distance to M31 is assumed to be $779^{+19}_{-18}{\rm \,kpc}$ and the uncertainties on this value are modeled by drawing directly from the posterior probability distribution function provided by Conn et al. (2012). All magnitudes are dereddened using the Schlegel et al. (1998) maps with the extinction coefficients for the PS1 filters provided by Schlafly & Finkbeiner (2011): $A_{g_{P1}}/E(B-V) = 3.172$, $A_{r_{P1}}/E(B-V) = 2.271$, and $A_{i_{P1}}/E(B-V) = 1.682$.

In a first pass at searching for unknown M31 satellites within 30° from their host, a map was produced from the catalog sources selected in the color–magnitude diagram (CMD) to loosely correspond to potential red giant branch (RGB) stars at the distance of Andromeda (0.0 ≲ (r_P1 − i_P1)₀ ≲ 0.8, i_{P1, 0} < 20.8). This map showed a handful of high-significance peaks, including the well-known M31 dwarf galaxies And I, II, III, V, VI, and LGS3. The CMD of significant peaks, which did not overlap with known satellites, was inspected by eye, leading to the clear discovery of two compact stellar systems located at ICRS $(\alpha,\delta)=(22^\mathrm{h}58^\mathrm{m}16\buildrel{\mathrm{s}}\over{.}3,+41^\circ 17^{\prime }28^{\prime \prime })$ and $(\alpha,\delta)=(0^\mathrm{h}35^\mathrm{m}59\buildrel{\mathrm{s}}\over{.}4,+51^\circ 33^{\prime }35^{\prime \prime })$ that have no counterparts in the NASA/IPAC Extragalactic Database or Simbad. We designate them as Lac I/And XXXI and Cas III/And XXXII and show the location of the two dwarf galaxies comparatively to the M31 satellite dwarf galaxy system in Figure 1.

Zoom In Zoom Out Reset image size

The top left panels of Figure 2 present the CMD of stars within two half-light radii of Lac I (see below for the estimation of its structural parameters); for comparison, the panels also include the CMD of a field region with the same coverage but offset by a degree toward the southeast. Lac I produces a clear overdensity of stars within the RGB color selection, typical of stars close to the tip of the red giant branch (TRGB) at the distance of M31. Isolating these stars yields the map shown to the right of the CMDs and reveals that they are also clumped in a clear spatial overdensity, with a size of a few arcminutes, typical of M31 satellite dwarf galaxies. The same is true for Cas III (top right panels), despite a more severe contamination from the prominent Milky Way disk.

Zoom In Zoom Out Reset image size

For comparison, the bottom panels of Figure 2 show the same set of plots for the well-known Andromeda I dwarf galaxy, also present in the stacked PS1 data, albeit in a region of deeper data. The CMDs of both And I and its background region are less populated owing to the lower Galactic foreground contamination (b = −167 for Lac I and b = −112 for Cas III versus b = −248 for And I), but otherwise display a similar overdensity of stars when compared to the two discoveries.

3.1. Structure

The structural parameters of Lac I and Cas III are determined following the maximum likelihood technique presented by Martin et al. (2008) and updated in N. F. Martin et al. (2013, in preparation) to allow for a full Markov Chain Monte Carlo (MCMC) estimate of the parameters, allowing for missing data. For systems that contain relatively few stars, a likelihood-based estimate of the structural parameters is far less biased and much more robust than estimates that rely on binning the data (Martin et al. 2008). In this instance, the algorithm explores the likelihood of a family of exponential models with a constant background level to reproduce the distribution of PS1 stars around Lac I or Cas III. The parameters of the dwarf galaxy models are its centroid, its exponential half-light radius (r_h), the position angle of its major axis (θ), its ellipticity ( = 1 − b/a, with a and b being, respectively, the major and minor axis scale lengths of the system), and the number of member stars within the chosen CMD selection box (N*), which is tied to the background level through the number of stars in the data set. Uncertainties on these parameters are determined from the one-dimensional, marginalized likelihood distribution functions and the location at which the likelihood reaches e^−0.5 ≃ 0.61 of the peak value. For a Gaussian distribution, the corresponding parameter values bound the 1σ confidence interval.

The likelihood distribution functions for these parameters, constrained on the PS1 data of the region within 25' of the centroid of Lac I or Cas III, are presented in Figure 3 for the most relevant parameters. These are well constrained and, in particular, the high significance of N* in both cases leaves no ambiguity to the existence of the new dwarf galaxies. Their radial density profiles are also displayed in the same figure and, for both dwarf galaxies, we compare the data, binned following the most likely centroid, ellipticity, and position angle, with the exponential model of most likely half-light radius and number of stars. In both cases, the very good agreement between the two is testimony to the quality of the fit.

Zoom In Zoom Out Reset image size

The measured structures of Lac I and Cas III are rather typical of bright Andromeda satellite galaxies with moderate ellipticities (0.42  ±  0.07 and 0.50  ±  0.09, respectively). The size of the two systems is on the high side of the size distribution of M31 satellite dwarf galaxies: $r_h = 4.2^{+0.4}_{-0.5}$ arcmin or $912^{+124}_{-93}{\rm \,pc}$ for Lac I and $r_h = 6.5^{+1.2}_{-1.0}$ arcmin or 1456 ± 267 pc for Cas III (see below for their heliocentric distance estimates).

The properties of Lac I and Cas III are summarized in Table 1 and the MCMC chains are available on request to the corresponding author.

Table 1. Derived Properties of Lac I/And XXXI and Cas III/And XXXII

	Lac I	Cas III
α (ICRS)	$22^\mathrm{h}58^\mathrm{m}16\buildrel{\mathrm{s}}\over{.}3$	$0^\mathrm{h}35^\mathrm{m}59\buildrel{\mathrm{s}}\over{.}4$
δ (ICRS)	+41°17'28''	+51°33'35''
ℓ (°)	101.1	120.5
b (°)	−16.7	−11.2
E(B − V)a	0.133	0.193
(m − M)0	24.40 ± 0.12	24.45 ± 0.14
Heliocentric distance (kpc)	$756^{+44}_{-28}$	$772^{+61}_{-56}$
M31-centric distance (kpc)	275 ± 7	$144^{+6}_{-4}$
MV	−11.7 ± 0.7	−12.3 ± 0.7
μ0 (mag arcsec−2)	25.8 ± 0.8	26.4 ± 0.8
Ellipticity	0.43 ± 0.07	0.50 ± 0.09
Position angle (N to E; °)	−59 ± 6	−90 ± 7
rh (arcmin)	$4.2^{+0.4}_{-0.5}$	$6.5^{+1.2}_{-1.0}$
rh (pc)	$912^{+124}_{-93}$	1456 ± 267

Note. ^aFrom Schlegel et al. (1998).

Download table as:  ASCIITypeset image

3.2. Distance

The magnitude of the TRGB is a very good distance indicator in the case of Andromeda dwarf galaxies (e.g., Conn et al. 2011), especially in the absence of deep photometric data. The tip's location also has the benefit of being relatively insensitive to the metallicity and age of stellar populations, as long as these are reasonably old and metal-poor, a condition that is fulfilled for all known M31 dwarf spheroidal galaxies. We apply the likelihood technique implemented by Slater et al. (2011) to determine the distance to And XXVIII, later confirmed to be accurate from deeper observations of this system (C. T. Slater et al. 2013, in preparation). The TRGB of Lac I is estimated at $i_{\mathrm{P1},0}^\mathrm{TRGB} = 20.95\,{\pm}\, 0.07$ and that of Cas III at $i_{\mathrm{P1},0}^\mathrm{TRGB} = 21.00\,{\pm}\, 0.10$. In both cases, the measurement is in good agreement with the dwarf galaxy's luminosity function, as shown in Figure 4.

Zoom In Zoom Out Reset image size

The TRGB magnitude has been estimated at $M_{i,\mathrm{SDSS}}^\mathrm{TRGB}=-3.44\pm 0.10$ by Bellazzini (2008). Its conversion onto the PS1 photometric system using the color equations of Tonry et al. (2012) yields $M_{i,\mathrm{P1}}^\mathrm{TRGB}=-3.45\,{\pm}\, 0.10$ and allows us to calculate the distance modulus of Lac I: (m − M)₀ = 24.40  ±  0.12. This distance modulus corresponds to a heliocentric distance of $756^{+44}_{-28}{\rm \,kpc}$ and, when also accounting for the M31 distance uncertainty, an M31-centric distance of 275  ±  7 kpc.⁹ Given the dearth of isolated Local Group dwarf galaxies, it still appears likely that Lac I is an Andromeda satellite despite its large M31-centric distance; its properties are similar to other remote satellites such as And VI, VII, XXVIII, or XXIX, which have now all been confirmed as Andromeda satellites from radial velocity measurements (McConnachie 2012; Tollerud et al. 2013).

In the case of Cas III, the distance modulus is (m − M)₀ = 24.45 ± 0.14, leading to a heliocentric distance of $772^{+61}_{-56}{\rm \,kpc}$ and an M31-centric distance of $144^{+6}_{-4}{\rm \,kpc}$. At this distance, Cas III is even more likely to be a member of the Andromeda satellite system. Of course, radial velocities are needed to add evidence that both Lac I and Cas III are truly bound to M31.

3.3. Total Magnitude

We report the discovery of two new dwarf galaxies, Lac I/And XXXI and Cas III/And XXXII, from stacked PS1 imaging data. At a heliocentric distance of $756^{+44}_{-28}{\rm \,kpc}$, Lac I is located in the environs of the Andromeda galaxy, at an M31-centric distance of 275 ± 7 kpc, making it a likely satellite of Andromeda. Cas III is located closer to M31 in projection and, combined with our measurement of its heliocentric distance ($772^{+61}_{-56}{\rm \,kpc}$), is well within the M31 virial radius with an M31-centric distance of only $144^{+6}_{-4}{\rm \,kpc}$. The membership of both dwarf galaxies to the M31 satellite system shall be confirmed kinematically through our ongoing spectroscopic program of both Lac I and Cas III. The discovery of Cas III is of particular interest as it is located on the rotating disk of dwarf galaxies that includes about half of Andromeda's dwarf galaxies (Conn et al. 2013; Ibata et al. 2013).

Both dwarf galaxies are quite large ($r_h=912^{+124}_{-93}{\rm \,pc}$ and r_h = 1456 ± 267 pc), yet this is not unexpected for dwarf galaxies with this brightness. Brasseur et al. (2011) estimated that, for a total magnitude of M_V = −12.0, Andromeda dwarf galaxies have an average size of 775 pc, with a lognormal dispersion of 0.23, implying that Lac I is barely above the average size and Cas III is just slightly more than one dispersion away. One also has to keep in mind that, of course, it is the largest systems that are the most likely to have so far avoided detection, biasing new discoveries toward larger systems than were discovered from photographic surveys.

To a certain degree, the only unexpected properties of Lac I and Cas III are their total magnitudes, M_V = −11.7 ± 0.7 and −12.3  ±  0.7, respectively, which are an order of magnitude more luminous than any member of the cohort of Local Group dwarf galaxies discovered in the last decade. Although surprising at first, one must remember that, prior to PS1, the region of the two discoveries had only been systematically observed in the optical by photographic surveys. Regardless, one might wonder why such bright dwarf galaxies were not discovered earlier and, in particular, why the searches based on photographic plates that led to the discovery of, for example, the fainter And VI (Armandroff et al. 1999; Karachentsev & Karachentseva 1999), did not also pick up Lac I. The explanation lies in the large half-light radii of the new systems, which are two to three times larger than that of And VI, leading to their much fainter surface brightness (μ₀ ∼ 26 mag arcsec⁻²). In fact, if one is to look meticulously at the location of Lac I in the POSS plates, one can see a very faint fuzz of unresolved light that betrays the presence of the dwarf galaxy (Figure 5). The signal of this detection was likely too easily confused with the wispy Milky Way dust filaments, ever more present at this latitude, and disregarded at the time. Cas III, with its even larger size, does not appear as a significant detection in the POSS plates.

Zoom In Zoom Out Reset image size

Providing a detailed assessment of the stellar populations present in Lac I and Cas III is beyond the PS1 data and the scope of this discovery paper, even though we can note that the dwarf galaxies do not appear to harbor a population of very young and blue stars (≲ 200 Myr).

The strong signal of Lac I and Cas III in the PS1 data, along with the detectability of fainter systems such as And XXIX, highlights the ability of the PS1 3π stacked survey to expand our knowledge of the Andromeda satellite system and, beyond, that of the 1 Mpc heliocentric sphere. The new observations should prove particularly interesting in the region around M31 not observed by the deeper PAndAS, nor by the SDSS that shied away from the close-by Milky Way plane.

N.F.M. gratefully acknowledges the CNRS for support through PICS project PICS06183. N.F.M. and H.-W.R. acknowledge support by the DFG through the SFB 881. C.T.S. and E.F.B. acknowledge support from NSF grant AST 1008342.

The Pan-STARRS1 Surveys (PS1) have been made possible through contributions of the Institute for Astronomy, the University of Hawaii, the Pan-STARRS Project Office, the Max Planck Society and its participating institutes, the Max Planck Institute for Astronomy, Heidelberg, and the Max Planck Institute for Extraterrestrial Physics, Garching, The Johns Hopkins University, Durham University, the University of Edinburgh, Queen's University Belfast, the Harvard-Smithsonian Center for Astrophysics, the Las Cumbres Observatory Global Telescope Network Incorporated, the National Central University of Taiwan, the Space Telescope Science Institute, the National Aeronautics and Space Administration under grant No. NNX08AR22G issued through the Planetary Science Division of the NASA Science Mission Directorate, the National Science Foundation under grant No. AST-1238877, and the University of Maryland.