Two Local Volume Dwarf Galaxies Discovered in 21 cm Emission: Pisces A and B
Erik J. Tollerud, Marla C. Geha, Jana Grcevich, Mary E. Putman, Daniel Stern
DDraft version October 30, 2018
Preprint typeset using L A TEX style emulateapj v. 5/2/11
TWO LOCAL VOLUME DWARF GALAXIES DISCOVERED IN 21 CM EMISSION: PISCES A AND B
Erik J. Tollerud , Marla C. Geha Jana Grcevich , Mary E. Putman , Daniel Stern Draft version October 30, 2018
ABSTRACTWe report the discovery of two dwarf galaxies, Pisces A and B, from a blind 21 cm H I search. Thesewere the only two galaxies found via optical imaging and spectroscopy of 22 H I clouds identified inthe GALFA-HI survey as dwarf galaxy candidates. They have properties consistent with being inthe Local Volume ( <
10 Mpc), and one has resolved stellar populations such that it may be onthe outer edge of the Local Group ( ∼ I -rich.The galaxies may thus be indicative of a large population of dwarfs at the limit of detectability thatare comparable to the faint satellites of the Local Group. Because they are outside the influence ofa large dark matter halo to alter their evolution, these galaxies can provide critical anchors to dwarfgalaxy formation models. Subject headings: galaxies: dwarf — galaxies: individual (Pisces A, B) — radio lines: galaxies —Local Group INTRODUCTION
The properties of faint dwarf galaxies at or beyondthe outer reaches of the Local Group (1 − L V (cid:46) L (cid:12) difficultto detect beyond the Local Group (Tollerud et al. 2008;Walsh et al. 2009; Hargis et al. 2014). Because of thesebiases, searching for nearby dwarf galaxies with method-ologies beyond the standard optical star count methodsare essential.This motivates searches for dwarf galaxies using the 21cm emission line of neutral hydrogen (H I ). While suchsearches cannot identify passive dwarf galaxies like mostLocal Group satellites, which lack H I (Grcevich & Put-man 2009; Spekkens et al. 2014), they have the potentialto find gas-rich, potentially starforming dwarf galaxies.This is exemplified by the case of the Leo P dwarf galaxy,found first in H I and later confirmed via optical imaging Astronomy Department, Yale University, P.O. Box208101, New Haven, CT 06510, USA; [email protected],[email protected] Department of Astrophysics, American Museum of NaturalHistory, Central Park West at 79th St., New York, NY 10024,USA; [email protected] Department of Astronomy, Columbia University, New York,NY 10027, USA; [email protected] Jet Propulsion Laboratory, California Institute of Technol-ogy, 4800 Oak Grove Drive, Mail Stop 169-221, Pasadena, CA91109, USA; [email protected] Hubble Fellow Visiting Astronomer, Kitt Peak National Observatory, Na-tional Optical Astronomy Observatory, which is operated by theAssociation of Universities for Research in Astronomy (AURA)under cooperative agreement with the National Science Founda-tion. (Giovanelli et al. 2013; Rhode et al. 2013).Here we describe two faint dwarf galaxies identifiedvia H I emission in the first data release of the Galac-tic Arecibo L-band Feed Array H I (GALFA-HI) survey(Peek et al. 2011). As described below, they are likelywithin the Local Volume ( <
10 Mpc) but just beyond theLocal Group ( (cid:38) §
2, wepresent the data used to identify these galaxies. In § § H = 69 . − Mpc − from WMAP9 (Hinshaw et al. 2013). DATA
The two galaxies we report on here were identified ini-tially as cold H I clouds with possibly galaxy-like proper-ties in DR1 of the GALFA-HI survey (Peek et al. 2011).Confirmation of these clouds as galaxies required addi-tional optical imaging and spectroscopy, which we de-scribe below. H I Detection
GALFA-HI was performed with the Arecibo Observa-tory 305-m telescope, using the ALFA feed array andthe GALSPECT spectrometer. GALFA-HI DR1 (Peeket al. 2011) includes velocities | V LSR | <
650 km s − , cov-ers 7520 square degrees of sky from δ = − ◦ to +38 ◦ ,has a channel spacing of 0 . − , and a spatial res-olution of 4 (cid:48) . The sensitivity of DR1 varies with posi-tion, but the majority of the objects cataloged wouldhave M HI < M (cid:12) if at 1 Mpc. The two candidatedwarfs were first found in a GALFA-HI DR1 catalogthat identified H I clouds with sizes < (cid:48) and velocityFWHMs <
35 km s − (Saul et al. 2012). From the Saulet al. (2012) sample of 1964 clouds, Grcevich (2013) iden-tified 51 candidate galaxies with fluxes and sizes similarto the known gas-rich Local Group dwarf galaxies (par-ticularly Leo T). The two candidates presented here were a r X i v : . [ a s t r o - ph . GA ] N ov also identified by Saul et al. (2012) as being likely galax-ies because they cannot be easily associated with knownhigh velocity cloud (HVC) complexes or Galactic gas inposition-velocity space. Optical Imaging
We performed follow-up optical imaging of 22 of the H I clouds from Grcevich (2013). These observations wereperformed with the pODI instrument on the WIYN Tele-scope in the g and r -band filters, with integration timesof 600-1200 sec per filter per target. Standard imagingreductions were performed by the ODI Portal, Pipeline,and Archive facility. These include bias subtraction, flat-fielding, and alignment of individual Orthogonal Trans-fer Array (OTA) cells into chips. The SWarp program(Bertin et al. 2002) was used to combine the individualexposures, and DAOPHOT (Stetson 1987) was used toperform PSF-fitting photometry on stars in the field.Most of the H I clouds did not have optical coun-terparts with morphologies like nearby galaxies withinthe ∼ (cid:48) GALFA-HI beam. Those in the Sloan Digi-tal Sky Survey (SDSS, Ahn et al. 2014) footprint showneither diffuse features like the galaxies described below,nor point source overdensities to the limit of the DR 10catalog. Similarly, our deeper pODI imaging showed nei-ther overdensities nor Red Giant Branch (RGB) featuresin the color-magnitude diagrams (CMD) down to r (cid:46) > (cid:46) § −
40” from the H I emission, this is wellwithin the 4 (cid:48) uncertainty from the GALFA-HI beam. Allother optical counterparts within the beam are less likelyto be associated with the H I ; they either appear stellaror are consistent with being distant background galax-ies (and hence at too high a redshift to match the H I ).Furthermore, the H α emission discussed in the next sec-tion is clearly associated with these optical conterparts,and its velocity is consistent with the H I , confirmingthe association between the optical objects and the H I cloud. Optical Spectroscopy
To confirm that the galaxies visible in Figure 1 cor-respond to the H I clouds discussed in § Fig. 1.—
Upper panels: gr color composite images of dwarfsPisces A (left) and Pisces B (right) from pODI on WIYN. Theimages are 1 (cid:48) tall, N is up, and E is left. The slit for the Palo-mar optical spectroscopy is shown as the gray lines. Lower panels:GALEX AIS images at the same location and orientation as theupper panels. For Pisces A (left), only NUV imaging is available,while for Pisces B (right), the image is an NUV/FUV color com-posite. α and the upper panels displayingthe H I emission.The optical spectra reveal H α emission. Other emis-sion lines were not detected, but flux estimates of otherlines assuming typical starforming dwarf galaxies suggestthey should have S / N (cid:46)
1, due to poor observing con-ditions. We fit the H α emission with Gaussian profiles,yielding a central velocity offset from the H I by only3 ±
34 and 10 ±
35 km s − (see Table 1). This is wellwithin the H α σ uncertainties , implying that the op-tical galaxies are indeed associated with the H I clouds.Because flux calibration was not possible for these ob-servations due to non-photometric conditions, we cannotquantify the magnitude of star formation implied by theemission. However, the presence of any detectable H α emission implies star formation is ongoing (or only ceasedwithin the last ∼
10 Myr) in at least some part of these Uncertainties on the optical velocities are much higher thanfor the H I due to the lower resolution of the optical spectrum ocal Dwarfs from HI 3galaxies.
160 180 200 220 240 260 280 300 320 v helio [km / s] F [ m J y ] Pisces A 21 cm
236 km s −
520 540 560 580 600 620 640 660 680 v helio [km / s] F [ m J y ] Pisces B 21 cm
615 km s − λ [ ] F l u x [ a r b i t r a r y un i t s ] Pisces A H α
236 km s − λ [ ] F l u x [ a r b i t r a r y un i t s ] Pisces B H α
615 km s − Fig. 2.—
Spectra of the dwarf galaxies. The upper panelsshow the GALFA-HI spectra for the clouds near Pisces A (upper-left) and Pisces B (upper-right). The lower panels show the sky-subtracted longslit optical spectra near H α for Pisces A (lower-left)and Pisces B (lower-right). In all panels, the dashed black verticallines are the emission features redshfited to 236 and 615 km s − forPisces A and Pisces B, respectively. Both optical spectra show H α emission at velocities consistent with the H I peak, confirming thatthe optical galaxies correspond to the H I clouds. The negativeflux spikes at 6530 and 6610˚A in the Pisces A optical spectrum arecaused by difficult-to-subtract OH sky lines. DISTANCES
The basic details of the two objects described in § I . Pisces A in particular has a very simi-lar H I line width as Leo P ( W ∼
25 km s − ), althoughboth it and Pisces B are optically somewhat fainter thanLeo P, while having similar 21 cm flux. Further compar-isons require the answer to a crucial question: what isthe distance to Pisces A and Pisces B? We consider thisquestion for each galaxy in turn below. Pisces A
We consider two scenarios for the distance to PiscesA. The first is based on the assumption that Pisces Ais in the Hubble flow. That is, its distance is simply D = v/H = 3 . .
72 Mpc, McQuinn et al. 2013). Pisces Ais relatively close in projection to M31, so in this sce-nario, d M31 = 1 . ∼ .
06 Mpc from Mc-Connachie 2012), placing Pisces A just beyond the edgeof the Local Group.
Pisces B
Pisces B has a higher radial velocity than Pisces A(615 km s − from the H I ), yielding a larger distance inthe Hubble flow scenario (8 . I line width ( W ∼
45 km s − ). This scenariois considered in the third column of Table 1.In the pODI imaging (upper-right panel of 1), Pisces Bappears to contain several potential point sources. How-ever, Pisces B has more diffuse light than Pisces A, mak-ing it difficult to obtain accurate photometry and esti-mate a distance based on stellar CMDs. Additionally,the much stronger H α emission apparent in Figure 2(lower-right panel), as well as its detection in the UVwith GALEX (FUV = 18 .
93, NUV = 18 .
87, DonovanMeyer et al., submitted), means that at least some ofthe brighter point sources may be unresolved H II re-gions rather than distinct stars. That said, if these pointsources are resolved stars, it is possible that Pisces B issomewhat closer than the Hubble flow distance implies,and instead has a substantially positive peculiar velocity.In the absence of deeper and higher-resolution imagingto resolve this question, in the fourth column of Table 1,we simply consider the limiting case that Pisces B is asclose as the Hubble Flow scenario for Pisces A. DISCUSSION AND CONCLUSIONS
In Figure 3, we show Pisces A and B in the context ofnearby dwarf galaxies. These include the Local Groupdwarfs (as compiled in McConnachie 2012), the SHIELDsample of nearby 10 − M (cid:12) H I -rich galaxies (Can-non et al. 2011), and Leo P. The upper panel demon-strates that, depending on the distance, the galaxiesdescribed here may be among the faintest known star-forming galaxies. They also overlap in their basic struc-tural properties with Local Group dwarfs, although theyare somewhat more compact. Unlike the Local Groupgalaxies, however, they are well beyond the virial radiiof any large dark matter halo like that of the MW orM31. Hence, they are crucial data points both for un-derstanding how star formation functions at the lowestluminosities and as possible progenitors of the faint (pre-dominantly passive) dwarf satellites of the Local Group.Additionally, the lower panel of Figure 3 shows theyhave a slightly higher H I mass relative to typical dwarfsof the same luminosity. The resulting gas fractions (lastrow of Table 1), while high, are not necessarily surprising, TABLE 1Key observed properties of the dwarf galaxies.
Pisces A Pisces BDistance-independent Properties(1) R.A. (J2000) 00 h m . s h m . s ◦ (cid:48) . (cid:48)(cid:48)
01 +11 ◦ (cid:48) . (cid:48)(cid:48) ◦ ) 108.52 133.83(4) b ( ◦ ) -51.03 -51.16(5) m r (mag) 17 . ± .
05 17 . ± . g − r ) (mag) 0 . ± .
08 0 . ± . R eff ,r ( (cid:48) ) 7 . ± . . ± . ◦ ) 111 ± ± F HI (Jy km s − ) 1 . ± .
07 1 . ± . M HI (10 M (cid:12) /D ) 2 . ± . . ± . v helio , HI (km s − ) 236 ± . ± W HI (km s − ) 22 . ± . ± v helio , opt (km s − ) 240 ±
34 607 ± f gas . ± .
02 0 . ± . D (Mpc) 3.5 1.7 8.9 3.5(16) M r (mag) -10.6 -9.0 -12.7 -10.7(17) R eff ,r (pc) 123 60 241 95(18) M ∗ (10 M (cid:12) ) 12.2 2.87 80.8 12.5(19) M HI (10 M (cid:12) ) 34.3 8.09 301 46.6(20) M tot (10 M (cid:12) ) 60.2 14.2 502 77.7Rows (1) through (14) are properties that do not require an assumed distance. (1) through (4) provide the locations of the object in J2000Equatorial and Galactic coordinates. Rows (5), (6), (7), and (8) are the r -band apparent magnitude, g − r color (extinction corrected usingthe Schlafly & Finkbeiner 2011 correction to Schlegel et al. 1998), half-light radius (on-sky), and position angle (East of North). (9) is thetotal H I line flux, and (10) is the corresponding H I mass for a galaxy at 1 Mpc. Rows (11) and (12) are the central velocity and FWHMof the H I line from GALFA-HI, and (13) gives the central velocity of the H α emission. (14) is the gas fraction f gas = 1 . M HI /M tot (seeE.g. McGaugh 2012 for a discussion of the 1.4 factor). The remaining rows are properties that require an assumed distance (see § r -band absolute magnitude andphysical half-light radius. Row (18) gives the stellar mass, assuming M ∗ /L V M (cid:12) /L (cid:12) ∼ r - to V -band conversion. Row(19) is the H I mass, and (20) is the total mass ( M ∗ + 1 . M HI ). as both the mean and scatter in gas fractions increase forfainter dwarfs (Geha et al. 2006, Bradford et al. in prep).Furthermore, these galaxies were discovered initially viaH I , so detection biases favor higher H I masses. Never-theless, the existence of a significant nearby populationof faint galaxies with high H I content may offer signifi-cant constraints on dwarf galaxy formation models.Motivated by the existence of these galaxies, we areidentifying further candidate nearby, faint, starformingdwarfs in the SDSS. As discussed in § ∼
100 show morphologies potentially consistentwith nearby dwarf galaxies. An ongoing spectroscopicfollow-up campaign on these objects shows that some doindeed have H α emission consistent with local galaxies,and we will present these results in a future paper. Comparison to Simulation
Detailed use of these galaxies as data points for galaxyformation will require more firm distances, only possiblewith deeper and/or higher resolution photometry (whichwill be obtained by our approved Cycle 22
Hubble SpaceTelescope program). However, the detection of PiscesA may have interesting interpretations in a cosmological context. To demonstrate this, we consider simulationsfrom the ELVIS suite, designed to resemble the MilkyWay/M31 pairing of the real Local Group (Garrison-Kimmel et al. 2014). We mock observe dark matter ha-los in a frame with v tan = 220 km s − and d = 8 kpc(relative to the host halo). GALFA-HI can distinguishhigh velocity cloud complexes from galaxy candidates at v helio (cid:38)
90 km s − (Saul et al. 2012), so we only con-sider halos with corresponding mock v helio . We also ig-nore halos inside the host’s virial radius, and with stellarmasses < M ∗ /M (cid:12) < . This yields 13 ± − I -selected faint dwarfs is roughly the same as thatof corresponding ELVIS halos, even these low-mass halostypically have H I if they are beyond the Local Group. Stellar masses in ELVIS were assigned via the abundancematching prescription of Garrison-Kimmel et al. (2014). ocal Dwarfs from HI 5 M V l og ( R e ff / p c ) log( L V /L fl ) log( M ∗ /M fl ) l og ( M H I / M fl ) Pisces APisces BLeo PStarforming LGPassive LGSHIELD
Fig. 3.—
Upper panel: Comparison of Pisces A and Pisces B tonearby dwarf galaxies in size vs. luminosity. Lower panel: Compar-ison of Pisces A and Pisces B to nearby galaxies in stellar mass vs.H I mass. Local Group galaxy properties are from McConnachie(2012). Leo P properties are from McQuinn et al. (2013) (the sizedescribed there is for an outer detectable extent, and thus is anupper limit on R eff ). The SHIELD sample of low-mass H I -richnearby dwarfs is from Cannon et al. (2011). The shaded bars showthe properties of Pisces A (cyan) and Pisces B (blue) between thetwo distance scenarios discussed in the text. For the upper panel,we use the Jester et al. (2005) transformations to convert from r -band magnitudes for Pisces A and Pisces B. For the lower panel,we make the estimate M ∗ /L V M (cid:12) /L (cid:12) ∼ Conclusions