The APOGEE-2 Survey of the Orion Star Forming Complex: I. Target Selection and Validation with early observations
J'Neil Cottle, Kevin R. Covey, Genaro Suárez, Carlos Román-Zúñiga, Edward Schlafly, Juan Jose Downes, Jason E. Ybarra, Jesus Hernandez, Keivan Stassun, Guy S. Stringfellow, Konstantin Getman, Eric Feigelson, Jura Borissova, J. Serena Kim, A. Roman-Lopes, Nicola Da Rio, Nathan De Lee, Peter M. Frinchaboy, Marina Kounkel, Steven R. Majewski, Ronald E. Mennickent, David L. Nidever, Christian Nitschelm, Kaike Pan, Matthew Shetrone, Gail Zasowski, Ken Chambers, Eugene Magnier, Jeff Valenti
DDraft version April 20, 2018
Typeset using L A TEX twocolumn style in AASTeX61
THE APOGEE-2 SURVEY OF THE ORION STAR FORMING COMPLEX:I. TARGET SELECTION AND VALIDATION WITH EARLY OBSERVATIONS
J’Neil Cottle, Kevin R. Covey, Genaro Su´arez, Carlos Rom´an-Z´u˜niga, Edward Schlafly, Juan Jose Downes,
2, 4
Jason E. Ybarra,
2, 5
Jesus Hernandez, Keivan Stassun, Guy S. Stringfellow, Konstantin Getman, Eric Feigelson, Jura Borissova,
9, 10
J. Serena Kim, A. Roman-Lopes, Nicola Da Rio, Nathan De Lee,
14, 6
Peter M. Frinchaboy, Marina Kounkel, Steven R. Majewski, Ronald E. Mennickent, David L. Nidever, Christian Nitschelm, Kaike Pan, Matthew Shetrone, Gail Zasowski, Ken Chambers, Eugene Magnier, and Jeff Valenti Department of Physics & Astronomy, Western Washington University, Bellingham WA 98225-9164 USA Instituto de Astronom´ıa, Universidad Nacional Aut´onoma de M´exico, Unidad Acad´emica en Ensenada, Ensenada BC, 22860 Mexico Max-Planck-Institut f¨ur Astronomie, K¨onigstuhl 17, D-69117 Heidelberg, Germany Centro de Investigaciones de Astronom´ıa, AP 264, M´erida 5101-A, Venezuela Department of Physics, Bridgewater College, 402 E College St, Bridgewater, VA 22812 Department of Physics & Astronomy, Vanderbilt University, Nashville, TN 37235, USA Center for Astrophysics and Space Astronomy, University of Colorado, UCB 389, Boulder, CO 80309, USA Department of Astronomy & Astrophysics, 525 Davey Laboratory, Pennsylvania State University, University Park, PA 16802, USA Instituto de F´ısica y Astronom´ıa, Facultad de Ciencias, Universidad de Valpara´ıso, Av. Gran Breta˜na 1111, Playa Ancha, Casilla 5030,Valapara´ıso, Chile Millenium Institute of Astrophysics, Av. Vicu˜na Mackenna 4860, 7820436, Macul, Santiago, Chile Steward Observatory, University of Arizona, Tucson, AZ 85721, USA Department of Physics, Universidad de La Serena, Cisternas, 1200 La Serena, Chile Department of Astronomy, University of Virginia, Charlottesville, VA 22904-4325, USA Department of Physics, Geology and Engineering Tech, Northern Kentucky University, Highland Heights, KY 41099, USA Department of Physics and Astronomy, Texas Christian University, Forth Worth, TX 76129, USA Universidad de Concepci´on, Departmento de Astronom´ıa, Casilla 160-C, Concepci´on, Chile Department of Physics, Montana State University, Bozeman, MT 59717, USA Unidad de Astronom´ıa, Universidad de Antofagasta, Av. Angamos 601, Antofagasta 1270300 Chile Apache Point Observatory and New Mexico University, P.O. Box 59, Sunspot, NM 88349, USA University of Texas at Austin, McDonald Observatory, Fort Davis, TX 79734, USA University of Utah, Dept. of Physics & Astronomy, 115 South 1400 East, JFB 201, Salt Lake City, UT 84112, USA Institute for Astronomy, University of Hawai’i, 2680 Woodlawn Drive, Honolulu, HI 96822, USA Space Telescope Science Institute, Baltimore, MD, 21218, USA
ABSTRACTThe Orion Star Forming Complex (OSFC) is a central target for the APOGEE-2 Young Cluster Survey. Existingmembership catalogs span limited portions of the OSFC, reflecting the difficulty of selecting targets homogeneouslyacross this extended, highly structured region. We have used data from wide field photometric surveys to produce aless biased parent sample of young stellar objects (YSOs) with infrared (IR) excesses indicative of warm circumstellarmaterial or photometric variability at optical wavelengths across the full 420 square degrees extent of the OSFC.When restricted to YSO candidates with
H < .
4, to ensure S/N ∼
100 for a six visit source, this uniformly selectedsample includes 1307 IR excess sources selected using criteria vetted by Koenig & Liesawitz and 990 optical variablesidentified in the Pan-STARRS1 3 π survey: 319 sources exhibit both optical variability and evidence of circumstellardisks through IR excess. Objects from this uniformly selected sample received the highest priority for targeting, butrequired fewer than half of the fibers on each APOGEE-2 plate. We fill the remaining fibers with previously confirmed [email protected] a r X i v : . [ a s t r o - ph . GA ] A p r and new color-magnitude selected candidate OSFC members. Radial velocity measurements from APOGEE-1 andnew APOGEE-2 observations taken in the survey’s first year indicate that ∼
90% of the uniformly selected targets haveradial velocities consistent with Orion membership.The APOGEE-2 Orion survey will include > Keywords: open clusters and associations: individual (Orion) INTRODUCTIONStar forming regions are invaluable astrophysical lab-oratories. The pre-main sequence binaries within theseregions enable stringent tests of stellar evolutionarymodels (e.g., Stassun et al. 2014), and the ages andkinematics of the full cluster population constrain thephysical processes during the formation and early evolu-tion of stars and clusters (e.g., Da Rio et al. 2014). Sur-veys of cluster populations with high-resolution, multi-object spectrographs provide the most efficient route tothe precise measurements of stellar and kinematic prop-erties that these tests require. Surveys conducted at op-tical wavelengths have provided important constraintson the membership, star formation histories, and dy-namical states of the optically accessible members ofrelatively compact clusters (Tobin et al. 2009; Jeffrieset al. 2014; Sacco et al. 2015), but investigations of em-bedded or extended complexes have thus far been lim-ited to smaller subsets of low-extinction (Rigliaco et al.2016) or centrally concentrated sources.The APOGEE (Apache Point Observatory GalacticEvolution Experiment) survey provides a unique oppor-tunity to study the most embedded and extended re-gions of active star formation. The SDSS-IV APOGEE-2 survey aims to obtain infrared spectra of hundredsof thousands of red giant stars, in all components ofthe Milky Way, in both hemispheres, to reconstruct theGalaxy’s star formation history (Zasowski et al. 2013;Majewski et al. 2015). The survey includes several ad-ditional ‘Goal Programs’ that are not included in theAPOGEE-2 survey’s formal science requirements butaddress other areas of scientific interest, including starformation. The APOGEE spectrographs’ infrared (1.51-1.7 µ m) sensitivity, multiplex capability, and wide fields-of-view enable efficient surveys of extended and embed-ded (A V ≥ σ T eff ∼
80 K; σ logg ∼ ∼ σ ∼ σ ∼ eff values are well correlated with several independent prox-ies for YSO evolutionary state, such as log g , total line-of-sight extinction, and mid-infrared SED slope. Stel-lar age estimates derived from APOGEE spectra couldbe quite useful for measuring the timescales of variousstar formation processes, such as changes in the struc-ture and composition of circumstellar disks (e.g., Haischet al. 2001; Andrews & Williams 2005; Hern´andez et al.2007).The APOGEE-2 Young Cluster Survey aims to pro-vide a homogeneous, high-quality catalog of RVs andstellar parameters for several nearby ( d < t <
125 Myr) clusters and star forming regions. Mea-surements of the global velocity dispersions of these clus-ters, particularly at the largest cluster radii which areuniquely available from the APOGEE spectrograph’sfield of view, will enable new tests of the mechanismsby which young clusters form, thermalize, and often dis-perse (e.g., Stutz & Gould 2016). Consistent measure-ments of precise stellar properties, particularly (model-dependent) stellar ages, within and across cluster envi-ronments, will help distinguish between models that pre-dict fast (e.g. Elmegreen 2000) or slow (e.g. Krumholz& Tan 2007) timescales for the star formation process.The Orion Star Forming Complex (OSFC) is a partic-ularly important target for the survey: this region is acritical benchmark for studies of low- and high-mass starformation, but the region’s large angular extent ( > ) has typically limited observers to study isolatedsub-populations. APOGEE-2’s unique ability to effi-ciently observe thousands of YSOs across hundreds ofsquare degrees, and place heavily obscured YSOs on anequal footing with optically visible sources, provides theopportunity to assemble the first self-consistent diagno-sis of the global kinematics and star formation historiesof this rich star forming region. This effort builds on theIN-SYNC survey of the Orion A molecular cloud, led byDa Rio et al. (2016, 2017), by expanding the footprint ofthe survey to enable comparative studies of all the ma-jor sub-populations of this extended star forming region:Orion B, λ Ori, σ Ori, and Orion OB1. Selecting tar-gets across this extended region is no simple task, how-ever: existing membership catalogs typically span onlya portion of the region targeted by this program, andmerging members from multiple, disjoint literature cat-alogs would produce a complex selection function thatwould be difficult to account for in any analysis thatseeks to compare the populations of YSOs located indifferent sub-regions of the OSFC. Only by selecting tar-gets in a consistent and straightforward manner acrossthe full survey footprint can we be confident that anydifferences detected in stellar kinematics, ages, or otherstellar properties reflect intrinsic differences in the un-derlying population, rather than differences arising fromselection biases in the underlying catalogs.For example, consider the biases that may result fromrelying exclusively on a common youth indicator, suchas the presence of an IR excess due to circumstellar ma-terial. YSOs with IR excesses can be identified andclassified into broad evolutionary categories (i.e., Class0/I/II/III; see Adams et al. 1987; Whitney et al. 2003)with mid-IR photometry, which is now widely avail-able from wide-field surveys conducted by the
Spitzer and
WISE space telescopes (see, e.g., Allen et al. 2004;Gutermuth et al. 2008; Esplin et al. 2014; Koenig & Lei-sawitz 2014, KL14 hereafter). However, the evolution-ary stages associated with mid-IR excesses represent acritical phase, but not the entirety, of a star’s pre-mainsequence evolution. IR excesses typically disappear be-fore the star reaches the zero-age main sequence, as thecircumstellar material accumulates into larger bodies oris dispersed by radiative or dynamical effects. A typicaltime scale for disk dispersal is 10 years for stellar massesup to 3 M (cid:12) . For masses larger than 7 M (cid:12) , the time scaledrops by an order of magnitude (Gorti & Hollenbach2009). Censuses of YSOs that rely exclusively on theidentification of IR excess sources will, therefore, be sub-stantially incomplete for stellar populations as young as ∼ Gaia photometry and astrometry to analyzethe spatial distribution, kinematics, and star formationhistories of the YSOs and stellar populations that makeup the OSFC. Our first efforts, which are now under-way, use this homogeneous dataset to identify coher-ent spatial and kinematic substructures across the fullOSFC (Kounkel et al., in prep); subsequent analyses willprovide finer-grained diagnoses of the membership andproperties of each sub-population.This paper is structured as follows: In Section 2 wedescribe the observational data used to generate our cat-alog of uniformly selected candidate YSOs; we also sum-marize the APOGEE-2 observations that were obtainedin early 2016, which together with the legacy IN-SYNCOrion A observations provide a first indication of theyield of our targeting methods. In Section 3 we de-scribe the construction of the catalog of uniformly se-lected YSOs, presenting the selection of IR excess andoptically variable YSO candidates in Sec. 3.1 and 3.2 re-spectively. Section 4 summarizes the additional OSFCmembers/candidates that were targeted to increase thecompleteness of our final target lists, albeit with a sig-nificantly more complex selection function, when fibersremain to be filled in a given field. The process forproducing final target lists in each field is presented inSection 4.2. In Section 5 we quantify the yield of ouruniformly selected targets by analyzing radial velocitiesmeasured in the IN-SYNC survey of Orion A (Da Rioet al. 2016, 2017) and in the 2015-2016 APOGEE-2 ob-servations of Orion B, Orion OB1a/b and Lambda Ori.Finally, we summarize our conclusions and future anal-ysis plans in Section 6. OBSERVATIONS2.1.
Data from the Literature: WISE+2MASS pointsources
Near- and mid-infrared (NIR, MIR) photometry usedto select APOGEE-2 YSO targets was primarily drawnfrom the AllWISE Source Catalog (Cutri & et al. 2013).The AllWISE Source Catalog is a primary data prod-uct of the
Wide-field Infrared Survey Explorer (Wrightet al. 2010, WISE). The AllWISE data release combinesdata from the WISE cryogenic (Wright et al. 2010) andpost-cryo NEOWISE (Mainzer et al. 2011) missions; theresulting WISE detections were associated with near-infrared counterparts in the 2MASS Point and ExtendedSource Catalogs (Skrutskie et al. 2006) using a 3 arcsec-ond matching radius. Sources detected in both catalogsprovide reliable photometry in up to seven wavelengthbands spanning 1-22 µ m, as required for the algorithmdeveloped by KL14 to select candidate YSOs with in-frared excesses (Sec. 3.1).Additional YSO candidates were targeted for APOGEE-2 observations based on criteria that did not require aMIR detection (e.g., optical variability or location in anoptical/near-infrared color-magnitude diagram). NIRphotometry was still essential, however, to plan andprioritize the H-band APOGEE observations. For thispurpose, we used a 3 arcsecond matching radius to iden-tify counterparts to optical sources in the 2MASS pointsource catalog, which is effectively deeper for non-IR-excess-bearing pre-main sequence stars than the mid-IRflux-limited AllWISE catalog.2.2. Optical photometry: Pan-STARRS1 3 π Survey
We identified optically variable candidate YSOsthroughout the OSFC using multi-epoch optical pho-tometry from the Pan-STARRS1 (PS1) 3 π survey. ThePan-STARRS1 3 π survey was carried out using theGPC1 camera (Hodapp et al. 2004; Onaka et al. 2008;Tonry & Onaka 2009) mounted on a 1.8m telescope atHaleakala, Hawaii. The camera provides images of a 3 ◦ field of view in five broad filters: g P1, r P1, i P1, z P1and y P1, with effective wavelengths of 4800˚A 6200˚A , 7500˚A , 8700˚A and 9600˚A respectively. Source photom-etry is measured by the Pan-STARRS1 Image Process-ing Pipeline (Magnier 2006, 2007; Magnier et al. 2008,2013), which provides relative and absolute photometryaccurate to better than 1% (Tonry et al. 2012; Schlaflyet al. 2012). The photometry used in this analysis weredrawn from the third processing version release (PV3).2.3.
APOGEE Spectroscopy
Observations for the APOGEE-2 Young Cluster Sur-vey are collected with the APOGEE-2 northern spectro-graph (APOGEE-2N hereafter; Wilson et al. 2010) onthe 2.5m SDSS telescope at the Apache Point Observa-tory. The APOGEE-2N spectrograph enables the simul-taneous acquisition (via optical fibers plugged into pre-drilled aluminium plug-plates) of up to 300 moderate-to-high resolution (R ∼ µ m)spectra across a 1.5 deg. radius field-of-view. Spectraare processed by the APOGEE data reduction (Nideveret al. 2015) and APOGEE Stellar Parameter and Chem-ical Abundances (ASPCAP; Garc´ıa P´erez et al. 2016)pipelines, to produce basic APOGEE-2 data products,including calibrated spectra, stellar parameters (e.g., T eff , log g ), radial velocities, and bulk and individualelemental abundances (Holtzman et al. 2015). Analy-sis routines developed by Cottaar et al. (2014) can per-form an independent spectral analysis to extract jointconstraints for a suite of spectral parameters most im-portant for young stars: effective temperatures, surfacegravities, and radial velocities, as well as projected ro-tational velocities ( v sin i ), H -band veiling flux (r H ),and by combining the APOGEE-based stellar parame-ters with 2MASS photometric measurements, infraredcolor-excesses ( E ( J − H )). Table 1 . Fields in APOGEE-2 Orion Survey
Field R.A. Dec. Plate EpochName (deg.) (deg.) ID (MJD) λ Ori A 84.140 10.340 8879 2457406 · · · · · · · · ·
Table 1 continued
Table 1 (continued)
Field R.A. Dec. Plate EpochName (deg.) (deg.) ID (MJD)8884 2457677 λ Ori B 82.340 11.730 8885 24574098886 2457411 · · · λ Ori C 86.611 8.993 9482 2457685Orion B-A 86.654 0.134 8890 24574338891 24574438892 2457675 · · · · · · · · · · · · · · · · · ·
Orion A-B 83.550 -5.300 9533 2457737 · · · · · · · · · · · ·
Table 1 continued A B C D EF AB C DEF
ABCDE
A BCDE
A BC
A B
A B A BC
Lambda OriOrionOB1-ABOrion B Orion A
85° 80°10°5°0°-5°-10°
RA (J2000) D e c ( J ) Figure 1.
Field plan for APOGEE-2 Young Cluster Sur-vey in Orion. OSFC main regions are indicated with boldlabels. In each subregion, individual fields are identified byletters. Plates indicated with a dashed line are only partiallyobserved at the time of publication and are expected to becompleted in Winter 2017-18. Fields in Orion A (shown asgreen circles) were first observed in APOGEE-1, as reportedby Da Rio et al. (2016), with additional observations plannedfor APOGEE-2. The background image is a mosaic from theWISE 12 µm WSSA survey (Meisner & Finkbeiner 2014)
Table 1 (continued)
Field R.A. Dec. Plate EpochName (deg.) (deg.) ID (MJD) · · · · · ·
The OSFC is the largest region targeted in theAPOGEE-2 Young Cluster Survey. The complete 16field plan for the APOGEE-2 Orion Survey is outlinedin Table 1 and shown schematically in Figure 1. Thefirst APOGEE-2 observations in Orion were conductedin January and February of 2016, and comprise fifteen1-hour ‘visits’ for twelve distinct plate designs in fiveof our planned fields, sampling three sub-populationswithin Orion: λ Ori, Orion B, and the Orion OB1a/bregion (covering σ -Ori and 25 Ori). Additional obser-vations, comprising the bulk of the planned surveys forthe λ Ori, Orion B, and Orion OB1ab fields, were com-pleted during the 2016-2017 observing season. We listthe MJDs of all observations acquired for APOGEE-2Orion fields through the summer of 2017 in Table 1.Other star forming regions and young clusters targetedby this APOGEE-2 Goal Science program are Taurus,NGC 2264, Alpha Per, and the Pleiades; additionalregions may be able to be added to the program in2018-2019, depending on survey progress.APOGEE observations in crowded environments facean important trade-off between sample completenessand typical S/N. The width of APOGEE-2N fibersand their plugging ferrules prevent observations of twosources separated by less than 72 (cid:48)(cid:48) on a single APOGEE-2N plate. A more complete sample of crowded clusterstars can be achieved by designing distinct plates foreach visit, so that objects whose fibers would collide ona single plate design can be observed on separate plates.Allocating a fiber on only one plate design, however,limits the S/N that can be achieved for a source of agiven magnitude. In particular, a typical 1-hour platevisit returns S/N ≥
100 for sources with
H <
11 mag,whereas a S/N ∼
100 spectrum for an H ∼
12 magsource must be built up by co-adding three separate 1-hour (and S/N ∼
45) spectra. We choose to limit oursample to sources with
H < . ∼
100 (see Fig. 22 by Nide-ver et al. 2015, for APOGEE’s S/N performance as afunction of magnitude for 6 visit sources). For more in-formation on the standard APOGEE plate design andobserving strategies, see Zasowski et al. (2013).To reduce the limitation imposed by fiber crowdingand maximize the number of bona-fide OFSC membersobserved in this program, we designed several distinctplates for each of the fields in the Orion Complex. Toachieve a more uniform S/N in the final combined spec- A ‘visit’ is a contiguous sequence of short exposures on thesame plate, and represents the fundamental unit of the APOGEEobserving schedule and survey plan; see Zasowski et al. (2013) formore details. tra of sources of different brightnesses, we identified theminimum number of visits we sought to achieve for eachsource as a function of its H-band magnitude: one visitsuffices for
H <
11 mag sources, whereas three visits isthe goal for 11 ≤ H < H ≥ ≤ ≥
40 fibers available fortelluric and sky subtraction purposes.To quantify spatial biases in our survey due toAPOGEE-2N’s 72 (cid:48)(cid:48) fiber collision limit, we performeda simple nearest neighbor analysis on our final targetedsample. We computed the nearest neighbor distances(
Dnn ) for the complete sample of candidates in theobserved fields (i.e., including all candidates prior tothe the final fiber assignment stage) and for the finalsample of targets which were targeted for observationwith at least one visit. We then computed the ratioof the two distributions as a function of nearest neigh-bor distance to quantify the completeness of the targetsample. The results, which are shown in Figure 2, indi-cate that the sample is indeed less complete for smallernearest neighbor separations. Our strategy to use mul-tiple independent observations of a field has mitigatedthe effect, however, such that the completeness doesnot drop until nearest neighbor distances significantlybelow the 72 (cid:48)(cid:48) fiber collision limit: the effect is mostclearly seen at a typical separation of ∼ (cid:48)(cid:48) , where thecompleteness declines from ∼
60% for
Dnn > (cid:48)(cid:48) to ∼
40% for
Dnn < (cid:48)(cid:48) . Our completeness is even higherfor uniformly selected YSOs, which received the highestpriority for fiber assignment; even in the dense OrionA + B fields, we achieved completeness of (cid:38)
85% for
Dnn > (cid:48)(cid:48) and (cid:38)
60% for
Dnn < (cid:48)(cid:48) .As the number of visits required to meet our signal-to-noise goal is a direct function of source magnitude, wealso performed this nearest neighbor analysis for subsetsof bright ( H <
11) and faint (
H >
11) targets. The com-pleteness levels achieved for these subsets of our sampleare also shown in Figure 2, and indicate that the sam-
Figure 2.
Ratio between the nearest neighbor distance dis-tributions for all candidates in the observed fields and thosetargets with at least one visit. Symbols show the complete-ness for different subsets of the sample, including the com-plete sample (red dots), bright stars (
H <
11; blue asterisks),and faint stars (
H >
11; green crosses). The mean complete-ness of each sample falls at
Dnn < (cid:48)(cid:48) . ple of bright stars is ∼ ∼
20% for sources with nearest neighbor separationsless than 35 (cid:48)(cid:48) and 25 (cid:48)(cid:48) , respectively. We emphasize thatour scientific goals do not require a complete survey,but rather an homogeneously selected sample of youngstars. Nevertheless, our sample should include 40-60%of spatially resolved wide binaries with separations assmall as 2000 AU (5 (cid:48)(cid:48) resolution). Even in the densestregions such as Orion A and B, our targets include morethan half of the stars and the sample is representativeenough to describe well the properties of the full stellarpopulation. METHODS FOR UNIFORMLY SELECTINGYOUNG STELLAR OBJECTSTo select targets in each APOGEE-2 OSFC field, weadopted a multi-tier prioritization scheme. The highestpriority targets were YSO candidates selected in a con-sistent manner across the entire survey footprint usingdata from wide-field surveys. In this section we describe the construction of the catalog of ‘uniformly selected’candidate YSOs that provide these highest priority tar-gets; in future sections we describe how these uniformlyselected targets are supplemented with previously con-firmed members, and new candidates selected via (loweryield) optical+near-infrared color-magnitude criteria.3.1.
IR Excess: The Koenig & Leisawitz2MASS+WISE Algorithm
KL14 developed a method to select YSOs with evi-dence of circumstellar disks revealed in all-sky 2MASSand WISE photometry. Koenig et al. (2015) thenspectroscopically validated the fidelity of this selectionmethod, finding that ∼
80% of the candidate YSOs in σ and λ Ori selected by this method are indeed prob-able or likely members of the OSFC. In this section,we briefly summarize the selection method developed byKL14, which we outline schematically in Figure 3. Wethen validate the algorithm’s yield with training sets ofknown YSOs in Taurus and Orion, before performing ablind search for sources across the full Orion complex.The first several steps in the KL14 classificationscheme serve to eliminate low quality photometric de-tections, or sources that appear consistent with a non-YSO nature. The KL14 algorithm is designed to assesssources that are detected in both WISE and 2MASS;in practice, the depth of the 3-band (W1, W2, W3)WISE detections is the primary factor limiting the sizeof the sample that can be classified using the KL14 tech-nique. For sources with WISE counterparts, the KL14algorithm first applies a series of photometric qualitycuts to reject sources with low quality detections inW1 or W2 (i.e., null measurement uncertainty in W1[w1sigmpro = null] or W2 [w2sigmpro = null]) or a highreduced χ given the source’s signal-to-noise ratio inthat band. Figure 4 provides an example of applyingthis latter cut to W1 detections, where sources are re-jected if w1 χ > W snr − . Similar cuts are applied toW2, W3, and W4 photometry; see Sec. 3.1.1 in KL14for the relevant equations. After rejecting potentiallyspurious WISE detections, the KL14 algorithm rejectssources identified by color-magnitude cuts in the (W1-W2) vs. (W2 - W3) and (W1-W3) vs W1 space as likelyStar Forming Galaxies (SFGs) and Active Galactic Nu-clei (AGN), respectively.After eliminating low-quality or likely contaminantsources, the next steps in the KL14 classification schemeuse a series of NIR/MIR color-color cuts to identifylikely YSOs in a range of evolutionary states. Sourceswhose WISE or 2MASS colors are consistent with aClass I and Class II classification (based on cuts in the H − K S vs (W1 - W2) and the (W1 - W2) vs (W2 - Start 3478 YSO Candidates from Megeath+ (2012) Have WISE counterpart
Yes: 2470 No:1008
Lost due to Crowding Quality Cuts on W1/W2
Yes: 1947 No:523
Fail Quality Cuts Fail SFG test?
Yes: 1870 No:77
Lost as SFGs Fail AGN test?
Yes: 1465 No: 405 Yes: 5 No: 400
WISE Class I?
No: 1315 Yes: 150
WISE Class II?
Yes: 1013 No: 302
Yes: 87 No: 215
Class I or II? TransiVon Disk?
Yes: 14
WISE Class I
No: 201
TransiVon Disks
Lost potenVally Class 3
Lost as AGNs WISE Class II 2MASSClass I 2MASSClass II Fail AGB test? Found YSO candidates Lost as AGBs W4 Class I?
Yes: 1250 No: 19
W4 Class I
Figure 3.
Summary flowchart of the KL14 algorithm (see description in Sec. 3.1) as applied to the catalog of IR-excess sourcescompiled by Megeath et al. (2012) from Spitzer photometry of Orion A & B. Sources in the Megeath et al. (2012) catalog thatare not recovered by the KL14 classification are primarily lost either because they lack a WISE counterpart, have low qualitymeasurements in W1 and W2, or are classified by the KL14 system as likely AGN. The number of sources from the Megeathet al. (2012) catalog that survive or are discarded at each step of the KL14 classification is shown in the flowchart with the‘Yes:’ and ‘No:’ labels, respectively. Of the 3478 YSOs in the Megeath et al. (2012) catalog, 1250 are classified as candidateYSOs by the KL14 algorithm and 2228 either lacked the required data or received a non-YSO classification.
W3) color-color spaces; see Figure 5 for an example of the latter cut) are provided a provisional YSO classifi-0
W1 S/N ratio -1 W χ Fail KL14 Quality CutPass KL14 Quality Cut
Figure 4.
W1 S/N ratio (W1 snr ) vs. W1 reduced χ value(W1 χ ) for 2470 WISE counterparts, with the KL14 qualitycut ( w χ = ( W snr − /
7) overlaid as a red line. More than21% (523/2470) of the WISE sources fail the quality cut andare eliminated from further classification. This quality cutreduces contamination at a cost of reduced completeness inactive star forming regions. cation. The remaining sources, whose W1-W3 SpectralEnergy Distribution (SED) did not indicate the presenceof an IR excess, are then flagged as provisional transi-tion disk candidates and added to the provisional YSOlist if they satisfy the requirements as given in KL14Sec. 4.2.2. A final set of possible Class I sources are ex-tracted from the likely AGN sample with color cuts on(W1 - W2), (W2 - W4), and (W3 - W4) colors. Thesesources are added to a provisional YSO list, which isthen subjected to a final screening that uses the W1vs (W1 - W2) color-magnitude and (W1 - W2) vs (W3- W4) color-color diagrams to identify and remove anyremaining Asymptotic Giant Branch (AGB) candidates.Sources that pass all the above tests are ultimately iden-tified as likely YSOs, and given a likely Class designationbased on their mid-IR SED slopes.3.1.1.
Validation in Taurus
As a first test of the completeness and fidelity of theKL14 selection method, we examined the algorithm’sability to recover YSOs with evidence of circumstellardisks in the Taurus star forming region. 2MASS andWISE counterparts were identified for 414 known Tau-rus members compiled by Esplin et al. (2014) using a1 (cid:48)(cid:48) matching radius; sources lacking unique detections inboth 2MASS and all four WISE bands were eliminated
W2-W3 W - W SFGClass IClass II
SFG cutsClass I CutsClass II CutsMain Sequence
Figure 5. (W1-W2) vs. (W2-W3) color-color diagram, pop-ulated by candidate YSOs identified by Megeath et al. (2012)that meet the KL14 WISE photometry quality cuts shownin Figure 4. Colored lines indicate regions used by KL14 toclassify Class I (red lines) and Class II (green lines) YSOs, aswell as Star Forming Galaxies (SFGs; blue lines). Of these1140 sources, 77 objects with W1 >
13 are classified as SFGsand removed by the KL14 algorithm prior to classification.Of the remaining sources, 150 are classified as Class I and1013 as Class II. The KL14 selection selects Class I and ClassII objects with high fidelity. WISE colors of typical main se-quence sources, as tabulated by Davenport et al. (2014), areshown as a yellow locus. from further consideration. To evaluate the number offalse positives that the KL14 algorithm produces in re-gions free of active star formation, we also analyzed acatalog of 2MASS-WISE detections in a 2 square de-gree off-cloud field (control region; Dec: 24 - 26 deg;RA: 75 - 77.5 deg). The WISE color cuts used to assignYSO classifications to sources in the Esplin et al. (2014)catalog and to identify potential contaminants in the off-cloud region are shown in Figure 6. Of the 156 bona fideTaurus members in the Esplin catalog that satisfy theW1-W2 vs. W3-W4 criteria for Class II YSOs shown inFigure 6, 135 satisfy the criteria in the KL14 algorithmfor identification as a YSO candidate; by contrast, only7 (0.037%) of the color-selected YSO candidates in theoff-cloud region are retained as likely YSOs accordingto the KL14 criteria. As Figure 7 shows, simple color-cut selection techniques flag a smaller, but non-trivial,number of candidate Class II YSOs in the off-cloud re-gion than within Taurus itself; utilizing the more com-plex, multi-dimensional KL14 selection algorithm, how-1
W3-W4 W - W Control Region with SNR > 2Class1 IClass IIClass IIITransDisk
Figure 6. (W1-W2) vs. (W3-W4) color-color diagram for all sources in the control region with SNR > Figure 7.
Histograms of H magnitudes for known Class IIYSOs in Taurus from Esplin et al. (2014)(blue histograms)and candidate Class II sources identified in the off-cloud con-trol region (red histograms). Open histograms show sourcesidentified with the simple (W1-W2) vs. (W3-W4) color cutsshown in Figure 6; filled histograms show sources of eachtype identified as candidate YSOs by the KL14 selection al-gorithm. ever, preserves the vast majority of known Class II YSOsin Taurus while excluding nearly all candidate Class IIsources in the off-cloud region. This test provides a firstdemonstration that the KL14 algorithm accurately re-tains a high fraction of known YSOs with infrared ex-cesses, without spuriously flagging a large number ofnon-YSO interlopers or contaminants. 3.1.2. Validation in Orion
Megeath et al. (2012) analyzed Spitzer/IRAC obser-vations of the Orion A & B clouds to compile a censusof dusty YSOs throughout the high-extinction regions ofthe OSFC. The areas targeted by our APOGEE surveyof the OSFC include the full extent of the Megeath et al.(2012) catalog, making it a valuable resource for testingthe completeness of the KL14 selection method. We findour implementation of the KL14 selection, (a sample ofwhich is shown in Table 5) process identifies 44% and50% of the Class I and II objects, respectively, cata-loged by Megeath et al. (2012) and Fang et al. (2013) inOrion’s high extinction regions.The most significant factors limiting the KL14 algo-rithm’s recovery of YSOs identified by Megeath et al.(2012) are the combined effects of crowding and WISE’slow angular resolution ( ∼ . (cid:48)(cid:48) in W3; ∼ (cid:48)(cid:48) in W4). Asseen in Table 2, 1008 sources (28% of the Megeath et al.(2012) catalog) lack a unique WISE counterpart, whichprevents their recovery with the KL14 algorithm. AsFigure 8 demonstrates, the frequency of WISE counter-parts to Megeath et al. (2012) sources drops for smallernearest neighbor distances. Searching the 2MASS cat-alog, we find that 5-15% of 2MASS detections in themagnitude range of interest for our catalog (7 < H <
13) are located within 6 (cid:48)(cid:48) of another 2MASS source, andmay thus may not be resolvable into separate sourcesin WISE imaging. Neighboring sources are typically22-4 magnitudes fainter in H than the candidate tar-get, however, suggesting that the merged WISE coun-terpart will likely be dominated by the emission fromthe 7 < H <
13 2MASS source, providing a reason-ably accurate description of the source’s spectral energydistribution and avoiding a false/spurious identificationas a potential IR excess source. Figure 9 compares thespatial distributions of Spitzer identified YSOs in Orionthat do and do not possess WISE counterparts, show-ing that sources lacking WISE counterparts are prefer-entially found in sub-regions of Orion with high stellarnumber densities, as well as high extinctions and ele-vated IR backgrounds from associated nebulosity. Thesefactors, and visual inspection of a representative sampleof the sources that lack WISE counterparts, point tothe combined influence of crowding and elevated back-grounds in suppressing the recovery of WISE counter-parts for the YSOs cataloged by Megeath et al. (2012),and by extension, the remainder of our catalog.Aside from a complete lack of a WISE catalog coun-terpart, the presence of low-quality WISE photometryis the second most common reason that bona fide YSOsin the Megeath et al. (2012) catalog fail to be recov-ered by the KL14 selection algorithm. As described inSec. 3.1, the KL14 selection method applies quality cutsto WISE photometry such as those shown in Figure 4,which accounts for 523 of the YSOs in the Megeath et al.(2012) catalog which are not recovered in our parsing ofthe WISE catalog. A final 697 YSOs from the Megeathet al. (2012) catalog are excluded from a YSO classi-fication by the KL14 algorithm based on conservativecuts in W1 vs W1 - W2 color-magnitude and (W1 -W2) vs (W3 - W4) color-color space to eliminate extra-galactic SFGs and AGN, or as potential Galactic AGBstars. This count also includes sources lost due to fail-ing the tests for Transition Disk classification as definedin KL14. As this test shows, these cuts are conserva-tive, and exclude some bona fide YSOs. We retain thecuts in generating our sample of WISE-identified can-didate YSOs, however, choosing to sacrifice some likelyYSOs to maintain a lower level of contamination acrossour full sample. Both Figure 3 and Table 2 summarizethe results of applying the multidimensional selection tothe Megeath et al. (2012) catalog and demonstrate thenumber of YSOs recovered with our selection.3.2.
Pan-STARRS-Variability Selected Young Stars
To extend the uniform sample to include diskless ClassIII pre-main sequence stars, we utilize multi-epoch Pan-STARRS1 3 π (PS1 3 π ; for more on the 3 π survey design,see sections 3.2 and 6 in Chambers et al. 2016) pho- Spitzer [4.5] N e a r e s t N e i g h b o r D i s t a n c e ( a r c s e c ) o f o b j e c t s Megeath+ (2012) YSOsMegeath+ YSOsw/ WISE counterparts
Figure 8.
Nearest neighbor distance as a function of Spitzer[4.5] magnitude for sources identified as YSOs by Megeathet al. (2012), distinguishing between all YSOs in the catalog(blue) and the subset with unique WISE counterparts (greenpoints). Sources without a Spitzer [4.5] detection are shownat [4.5] = 0. The histograms in the top and right sub-panelsshow the number of YSOs in the Megeath et al. (2012) cata-log with and without WISE counterparts as functions of eachaxis. The ratio of WISE counterparts to Megeath sourcesremains relatively constant for Spitzer [4.5] magnitude (topsub-panel) which suggests that incompleteness near or belowWISE’s flux limit is not responsible for the missing WISEcounterparts for some Megeath sources. The ratio steadilydecreases, however, for sources with smaller nearest neighbordistances (right sub-panel), reaching a minimum of about 0.5at nearest neighbor distances close to WISE’s angular reso-lution. tometry to identify optically variable pre-main sequencestars.To select optically variable pre-main sequence mem-bers of the OSFC from the PS1 3 π catalog, we firstimposed color-magnitude cuts informed by optical pho-tometry of known YSOs in Orion compiled by Megeathet al. (2012) and Fang et al. (2013). Based on the lo-cations of YSOs in those catalogs, we extracted PS1 3 π sources that satisfied the following cuts:1 . g − H ) + 4 . < g < . g − H ) + 10 . . r − H ) + 7 . < r < . r − H ) + 12 . Table 2.
Megeath YSOs Successfully Selected by KL14 Algorithm
Total Sources Sources Sources SourcesTest Tested Still Viable Eliminated ClassifiedWISE counterpart 3478 2470 1008 –Quality Cut 2470 1947 523 –SFG Test 1947 1870 77 –AGN Test 1870 1465 405* –WISE Class I 1465 1315 – 150WISE Class II 1315 302 – 10132MASS Test 302 215 – –2MASS Class I 87 67 – 202MASS Class II 67 – – 67Transition Disk 215 – 201 14w4 Class 1 405 – 400 5AGB Test 1269 1250 19 –Total Candidates 3478 – 2228 1250
Note —*These sources are not eliminated until they are passed through the W4Class I test. RA D e c . Megeath+ (2012) YSOsMegeath+ YSOs w/oWISE counterparts
Figure 9.
Locations of YSOs in Orion identified by Megeathet al. (2012), split into sets of sources that do (blue points)and do not (red points) possess WISE counterparts. Mostsources lacking counterparts are in regions with high numberdensities of YSOs leading to substantial crowding.
PS1 3 π sources are not shown, as they densely fill theallowed area.The PS1 3 π survey provides multi-band, multi-epoch grizY photometry for millions of stars, QSOs and galax-ies, but the time domain is not densely sampled: in r'-H r ' r > 1.2 (r - H) +7.5r < 1.375 (r - H) + 12.625APOGEE Faint Limit PanSTARRS Saturation Limit Figure 10. r - H vs. r color magnitude diagram of membersidentified by Fang et al. (2013), along with color-magnitudecuts shown in green and purple that bracket these knownmembers, which can be applied to Pan-STARRS1 photom-etry (not shown) to identify and select YSO candidates. Inblue and red are the limits for Pan-STARRS saturation andAPOGEE faintness, respectively. Fang et al. (2013) sourcesmainly reside between the purple and green cuts. The queryof the PS1 3 π dataset identified sources (omitted here forclarity) whose mean colors and magnitudes placed themwithin the green and purple lines. the PS1 data analyzed here, a typical source was ob-served ∼ ζ N u m b e r o f S o u r c e s PanSTARRSKnown YSO catalogs:Megeath+ (2012),Fang+ (2013),Pillitteri+ (2013)
Figure 11.
Histograms of ζ = log( σ F err F √ N ) for all PS1 3 π sources in the OSFC footprint that meet the color-magnitudecriteria in Fig. 10 (blue), and the subset of catalogs previ-ously identified as a YSOs as identified by Fang et al. (2013),Megeath et al. (2012), and Pillitteri et al. (2013) (yellow).The red line represents the 3-sigma variability cut; for val-ues of ζ larger than this limit, YSOs represent a substantialfraction of the sample. Welch-Stetson statistic (Welch & Stetson 1993; Stetson1996). We instead identified bona-fide variables usinga relatively simple, logarithmic variability metric com-puted by comparing a source’s observed variability toits median photometric precision as reported by the PS1pipeline: ζ = log (cid:18) σ F err F √ N (cid:19) where σ F is the observed standard deviation in mea-surements over all epochs, err F is the expected error perepoch and N is the number of epochs in all PanSTARRSfilters . Computing this metric for each of the PS1bands ( grizY ), we compared the ζ values for the fullsample of PS1 sources to those measured for previouslyidentified Orion members; Figure 11 shows this compar-ison for ζ values for PS1 i -band detections. As Figure This metric was originally intended to utilize a filter-specificvalue of N, but due to a coding error, the total number of observa-tions across all PanSTARRS filters was used in the calculation onwhich target selection was based. This implementation nonethe-less preserves ζ ’s utility as a relative variability indicator, however,particularly as calibrated against values of ζ measured for knownYSOs. We therefore report the results from the ζ values calculatedin this way, and as used in the target selection process. Number of Filters with Detected Variability N u m b e r o f s o u r c e s r e t u r n e d Megeath+ (2012)Pillitteri+ (2013)Fang+ (2013)IR ExcessAll PanStarrsSources
Figure 12.
Numbers of sources in the Orion A region thatare flagged as photometrically variable from PanSTARRSdata as a function of the number of filters the variabilitymust be detected in. Catalogs of known YSOs begin to leveloff near three filters, while the number of sources selectedfrom the PS1 3 π survey continues to increase as the numberof filters required for a variability classification decreases.
11 indicates, no value of ζ can be used to select a sam-ple composed only of known Orion members. Above athreshold of ζ = 0 .
6, however, which represents a ∼ σ detection of variability, known YSOs do represent a sig-nificant fraction of the total sample. Conversely, belowthe ζ = 0 . ζ .The contamination of non-OSFC members in a vari-ability selected sample can be further reduced, however,by requiring sources to meet the ζ = 0 . bona fide intrinsic variability will meet the cut inall, or nearly all, filters with reliable measurements. Todetermine the number of filters in which variability mustbe detected in order to classify a source as a likely YSO,we examined how the identification of known YSOs de-pends on the number of filters considered. Figure 12compares the number of sources that meet the ζ thresh-old in n or more filters, when selecting from the full(color-magnitude restricted) PS1 3 π sample or from sub-sets of previously cataloged Orion members. The num-ber of candidate variables extracted from the full PS13 π catalog drops by approximately an order of magni-tude when the number of filters a source must exhibitvariability in is raised from one to three; by contrast, thenumber of previously known members that are flagged5as candidate variables only drops by a factor of 2 whenraising the threshold from one filter to three. Furtherincreasing the number of filters in which a source mustexhibit variability to be considered a candidate YSOproduces similar reductions in the samples of variablesextracted from the full PS1 3 π catalog versus the sub-sets of known YSO members, and thus no gain in theratio of bona fide YSOs to non-YSO contaminants. Wetherefore conservatively adopted a 3-filter requirementfor a source to be classified as a bona fide variable and acandidate pre-main sequence star. We also investigatedpotential biases due to color dependences in the pho-tometric variability of known YSOs, and see no clearcolor dependencies in our sample. Aside from a mod-est ( ∼ r filter, potentially dueto variations in H α line emission from either chromo-spheric activity or mass accretion, all other filters flag aconsistent 52-57% of YSOs as photometric variables.Requiring variability detections in three filters is anattempt to balance the competing demands of identify-ing as many candidate members as possible, while alsoreducing contamination of the sample by photometri-cally variable non-members. Based on the tests shownin Figure 12, we expect that nearly 50% of the pre-mainsequence stars in this magnitude range are properly clas-sified as variable by our selection technique. We can-not eliminate contamination completely, however, andas such we expect that our final sample of optically vari-able candidate pre-main sequence stars still contains asubstantial numbers of other types of photometric vari-ables, such as background AGB stars. SELECTED TARGETS AND FINAL PLATEDESIGNS IN ORIONThe presence or absence of strong selection effects inthe final sample observed in this survey will stronglyinfluence the types of questions that the survey can ad-dress, and the complexity of the analysis that must beconducted to correct for those selection effects. As aresult, in assembling our final target lists and plate de-signs, we assigned the highest priority to objects in the‘uniform sample’ constructed from the union of the IR &variability selected catalogs whose construction was de-scribed in Section 3. In Sec. 4.1 we describe the processby which these IR & variability selected sources wereincluded in the APOGEE-2 Orion plate designs; in Sec.4.2, we describe how additional sources were added tothese designs to fill the remaining optical fibers, provid-ing a larger sample of bona fide members, albeit with amore complex and heterogeneous selection function. 4.1.
A Uniform Sample of IR/Variability SelectedYSOs Spanning All APOGEE-2 Orion Fields
Applying the KL14 algorithm as described in Section3.1 to sources in the AllWISE catalog within a ∼ < R.A. <
90 degand − . < Dec. <
15 deg, we identified an initialsample of 2699 candidate YSOs throughout Orion. Theresulting catalog of YSOs with evidence of circumstellardisks is given as Table 5 of the appendix and shown ingreen in Figure 13, with the locations of each APOGEEfield indicated for reference. For targeting purposes, werestrict this catalog to sources with
H < ∼
100 and enables the measurement of robust stellar pa-rameters such as T eff and log g (see Fig. 22 by Nide-ver et al. 2015, for APOGEE’s S/N performance as afunction of magnitude for 6 visit sources). Restrict-ing the catalog to sources with H < σ Ori cluster;more diffuse populations such as the λ Ori and OrionOB1a/b associations are also visible, albeit at somewhatlower contrast.Selecting sources in the color-magnitude filtered PS13 π catalog that exhibited photometric variability (i.e. ζ ≥ .
6) in three or more filters identifies 3697 likelyvariables within the same ∼
400 square degree footprintused for the identification of IR excess sources. All 3697likely variables are listed in Table 6 of the appendix, butas with the IR selected candidates we target only thosewith
H <
Merging Targets to Produce Final Plate Designs
In most APOGEE-2 Orion Survey fields, the catalogof H ≤ (cid:48)(cid:48) fiber collision radius.As a result, after assigning fibers to H ≤ Right Ascension (degrees) D e c li n a t i o n ( d e g r ee s ) Lambda OriOrion BOB1abOrion A
Variable SelectionIR Excess SelectionSelected by both methods
Figure 13.
Spatial distribution of candidate YSOs selected via 2MASS+WISE photometry and PS1 3 π multi-epoch photometry,relative to the plate boundaries of the APOGEE-2 YSO Orion program. tailed discussion of the membership catalogs and se- lection processes most relevant to each of Orion’s sub-regions.74.2.1. Prioritization Scheme
We make use of a bitmask consisting of binary flags tostore and sort the relative priorities assigned to each ofthe potential targets in this program. Targets within agiven membership catalog, or satisfying the criteria as-sociated with a given selection algorithm, are assigneda bit corresponding to a binary value of 2 n ; the value ofthe priority flag used for each catalog is outlined in Ta-ble 3. We then determine the prioritization of all targetswithin a given field according to the sum total of all thebinary priority flags that have been set for each source.In this system, sources present in catalogs correspond-ing to higher values of n will be assigned to fibers first;in this way, the value of the bit assigned to a catalogserves mainly to order the input catalogs according toour qualitative assessment of their potential contribu-tion based on several factors, including the total size ofthe catalog, as well as the completeness and contamina-tion of its described sample. Among sources with thesame bit set, any bits set for lower priority catalogs willprovide a slight increase to the value of the total prior-ity value, and cause that source to be targeted beforesources that are absent from the lower priority catalogs.The prioritization method we have adopted ensuresthat our uniformly selected sources do not compete forfibers with other, less homogeneously selected members,and thus protects the simplicity of the uniform sample’sselection function. As an example, a source in the OrionA cloud that was included in the Megeath et al. (2012)and Pillitteri et al. (2013) catalogs, and also selected as alikely YSO by our implementation of the KL14 method,would be assigned bitwise priority flags of 2 , 2 , and2 , and would have a final priority value of 2 + 2 + 2 = 352. A source that was only selected by ourKL14 implementation would have a final priority valueof 2 , whereas a source that was not selected in our KL14implementation but was included in the catalogs com-piled by Megeath et al. (2012) and Pillitteri et al. (2013)would have a final priority of 2 + 2 = 96. Applying ourprioritization scheme to these three sources, the sourcedetected in all three catalogs would be targeted first, fol-lowed by the KL14-only source, and finally by the sourcedetected by Megeath et al. (2012) and Pillitteri et al.(2013). To ensure our most consistent selection meth-ods are applied across Orion’s full footprint, we assignthe highest priority values to the candidate YSOs weidentify on the basis of their 2MASS+WISE and PS13 π photometry. Due to a miscommunication amongthe team, different maximum bit values were used forthe Orion OB1a/b and λ -Ori/Orion AB regions: 2 forthe λ -Ori and Orion AB regions, and 2 for the OrionOB1a/b regions. Nonetheless, the 2MASS+WISE and PS1 selected targets were assigned the maximum andpenultimate bit values in each region’s ranking scale, sothe relative prioritization of all target classes remainsconsistent across the full OSFR. Table 3 documents thevarious catalogs and selection criteria from which wedraw targets in our Orion survey.4.2.2. λ Ori
As one of the older and sparser regions within Orion, λ Ori yielded only 147 H < π selectionmethods across three distinct APOGEE-2 fields. Tofill the remaining fibers in these fields, we first tar-geted nearly 400 confirmed cluster members identifiedby Dolan & Mathieu (2001) and Hern´andez et al. (2010)on the basis of optical spectroscopy and mid-infraredexcesses, respectively. The catalogs produced by Bar-rado y Navascu´es et al. (2007) from Spitzer IRAC pho-tometry and Barrado et al. (2011) yielded no new H < λ Ori members selected as follows: we definedan empirical locus in the optical-infrared V vs. V − K color-magnitude diagram traced by members from the(Hern´andez et al. 2010) catalog as well as high likelihoodcandidates in the 3XMM-DR4 source catalog (Rosenet al. 2015) and sources with WISE+2MASS colors in-dicative of a mid-IR excess (( K − W > Orion OB1a/b
The Orion OB1 association is one of the largest andnearest sites of recent star formation, and includes the ∼ −
10 Myr old OB1a and ∼ − < . π selection (178 and115, respectively). We also included 153 X-ray sourcesfrom the 3XMM-DR5 Source catalog Rosen et al. (2016)with 2MASS counterparts that lie within the OB1abfields. Most of these X-ray sources are located withinthe Ori OB1a/b-A, Ori OB1a/b-E and Ori OB1a/b-Fplates (see Fig. 15). To sample the brighter end of the8 DM02 ¡ K [mag]8101214161820 V D M [ m a g ] D&M02D&M02 selectionXMMXMM selectionIR excess
Figure 14. V vs. V − K color-magnitude diagram of sourcesin the λ Ori field from the catalog of Dolan and MathieuDolan & Mathieu (2002). Known members of the associa-tion (XMM X-ray and infrared excess sources) trace the em-pirical locus defined by the two dashed lines, from which weselected additional sources to fill vacant fibers in APOGEEplate designs. cluster sequence, we selected 108 highly probable stel-lar members of Orion OB1 from the Kharchenko et al.(2005) catalog. These sources populate all plates exceptOri OB1a/b-A (see Fig. 15). To fill the plates with bonafide members, we considered several studies to include381 spectroscopically confirmed stellar members, whichare mainly focused in the σ Ori cluster and the 25 Ori-onis stellar group: 130 and 56 members from Brice˜noet al. (2005, 2007) respectively, 178 from Hern´andezet al. (2014), 11 from Downes et al. (2014)
Table 3 . Priority Flag
Selection Priority Sources Sources PrimarySource Bit Flagged a Assigned b Assignment c Full Footprint d WISE+2MASS IR excess (KL14 algorithm) 2
990 531 212 λ Ori
WISE+2MASS IR excess (KL14 algorithm) 2
143 123 123Pan-STARRS variability 2
73 62 24Hern´andez et al. (2010) 2
291 285 276Dolan & Mathieu (2001) 2
177 164 117Barrado y Navascu´es et al. (2007) 2
48 46 0Franciosini & Sacco (2011) 2
50 48 8Barrado et al. (2011) 2
45 45 3Rosen et al. (2015) XMM 2
191 177 43Koenig et al. (2015) 2
156 136 5Dolan & Mathieu (2002) + WISE 2
61 53 2Dolan & Mathieu (2002) + CMD 2 Table 3 continued9
61 53 2Dolan & Mathieu (2002) + CMD 2 Table 3 continued9 Table 3 (continued)
Selection Priority Sources Sources PrimarySource Bit Flagged a Assigned b Assignment c Orion OB1a/b
WISE+2MASS IR excess (KL14 algorithm) 2
240 177 177Pan-STARRS variability 2
165 137 57Megeath et al. (2012) Spitzer IR excess 2
105 35 5Kharchenko et al. (2005) 2
109 109 108Brice˜no et al. (2005) 2
144 137 130Brice˜no et al. (2007) 2
56 56 56Hern´andez et al. (2014) 2
308 200 178Hern´andez et al. (2007) 2
91 2 2Downes et al. (2014) 2
15 11 11Bouy et al. (2014) 2
67 41 3Su´arez et al. (2017) 2
18 6 6XMM 2
314 237 172USNO+2MASS selection 2
101 100 62Caballero (2008) 2
338 179 161
Orion A&B
WISE+2MASS IR excess (KL14 algorithm) 2
857 738 738Pan-STARRS variability 2
406 355 132Megeath et al. (2012) Spitzer IR excess 2
492 446 287Broos et al. (2013) Chandra 2
209 144 43Caballero (2008) 2
194 77 59Hern´andez et al. (2007) 2
222 78 23Hern´andez et al. (2014) 2
400 231 139Brice˜no et al. (2005) 2
16 11 11Bouy et al. (2014) 2
115 25 5USNO+2MASS Selection 2
400 271 216 a ‘Sources flagged’ indicates how many sources within each region were identified by a particular selectionmethod/membership catalog. b ‘Sources assigned’ indicates how many sources within each region were identified by a particular se-lection method/membership catalog and assigned fibers after resolving potential fiber collisions. c ‘Primary assignment’ indicates the number of sources within each region that were identified by aparticular selection method/membership catalog and were assigned a fiber, but were not identified byany higher priority catalogs, such that the given catalog is ‘responsible’ for their targeting. d Full Footprint values do not correspond to the sum of the values reported for fields spanning individualsub-regions. Full Footprint ‘Sources Flagged’ are drawn from a much larger effective area, while targetsin the overlap region between fields assigned to different sub-regions [i.e., OrionB-B and OrionOB1ab-A] are included in the ‘Sources Assigned’ and ‘Primary Assignment’ fields for multiple sub-regions. and 6 from Su´arez et al. (2017). Additionally, we in-cluded 62 sources from the MYStIX catalog (Broos et al.2013), 161 members from the study of Caballero (2008),2 members from the list of Hern´andez et al. (2007),and 3 highly probable young stellar photometric can- didates from Bouy et al. (2014). The remaining fiberswere assigned to 2269 photometric candidates selectedaccording to their position in the I vs I − J diagramfor the Ori OB1a/b-E and Ori OB1a/b-F plates wherethere is minimum extinction, and in the I vs I − K Figure 15.
Spatial distribution of targets in the Ori OB1a/bplates. The color and size of the symbols indicate the sourcecatalog and the priority of the targets, respectively. From topto bottom the abbreviations in the plot legend correspond to:2MASS+WISE (IR excess), PS1 3 π (variability), 3XMM-DR4 catalog, Kharchenko et al. (2005), Brice˜no et al. (2005,2007), Hern´andez et al. (2014), Downes et al. (2014), Su´arezet al. (2017), Bouy et al. (2014) and our USNO-2MASS selec-tion. The gray points show the USNO+2MASS photometryinside the OB1a/b-A plate area. Orion A & B
As the youngest and densest sub-regions within Orion,the Orion A & B clouds have been extensively ob-served at IR and X-ray wavelengths, providing an ex-tensive suite of existing membership catalogs and multi-wavelength observations to support target selection.
Figure 16. I vs I − K color-magnitude diagram of the se-lected targets for the APOGEE-2 Ori OB1a/b-A plate. Theblack solid curve corresponds to the empirical isochrone de-fined following the confirmed members and the high proba-bility young stellar candidates, and the brown curves are the1, 3, 5 and 10 Myr isochrones from Baraffe et al. (2015). Thered dashed lines indicate the APOGEE-2 limits. The labelsand references are the same as in Figure 15. Figure 17. I vs I − J color-magnitude diagram of the se-lected targets for the APOGEE-2 Ori OB1a/b-E plate. Allsymbols and references are consistent with Figure 15 andFigure 16. Nonetheless, nearly a thousand sources meet the crite-ria for our 2MASS+WISE and PanSTARRS selection inthis region; combined with 450 and 288 sources drawnfrom the Spitzer and XMM based catalogs compiled byMegeath et al. (2012) and Pillitteri et al. (2013), respec-tively, these catalogs provide more than 80% of the tar-gets we identify in these clouds. To these 1711 sources,we add 49 likely YSOs from
Chandra exposures of ofthe Orion Nebula and Flame Nebula (Broos et al. 2013;Getman et al. 2014, Getman 2015, private communi-1cation), and 128, 62 and 144 confirmed or candidate σ Ori members compiled by Caballero (2008), Hern´andezet al. (2007) and Hern´andez et al. (2014), respectively. RESULTS: EVALUATING YSO YIELD WITHAPOGEE RADIAL VELOCITIES5.1.
Velocity distributions of cluster members andGalactic thin and thick disk stars in the λ Ori Afield
To test the efficiency of our target selection, we com-pared the heliocentric radial velocities (RV) measuredfrom targets in the early 2016 observations of the λ OriA field to the RV distribution expected for the field pop-ulation along the same line of sight. The RV distribu-tions of the field populations from the Galactic thin andthick disks, as well as the Galactic halo, were simulatedusing the Besan¸con Galaxy model (Robin et al. 2003).This model has been extensively tested within the sen-sitivity limits of 2MASS and appropriately reproducesthe luminosity functions from the different componentsof the Galaxy (Robin et al. 2003). We assume the Be-san¸con model’s standard Galactic parameters; that is,we compute RV distributions based on models that as-sume the structural and kinematic properties listed byRobin et al. (2003) in their Tables 1-4 for the young andold disk populations.The simulation was performed for the full area cov-ered by the λ Ori plate ( ∼ < H < .
8) comparable tothat of our selected targets. RV errors were assignedto each simulated source by drawing from a distribu-tion matched to the RV precision of the APOGEE-2 ob-servations by fitting an exponential function to the ob-served RVs of previously known members (which sharea narrow range of intrinsic RVs) as a function of H-band magnitude. Figure 18 shows the RV distributionmeasured for APOGEE-2 targets in the λ Ori-A field(plates 8879 and 8880), together with the simulated RVdistributions for Galactic thin and thick disk stars alongthe same line-of-sight. The simulation predicts only twohalo stars, which are not shown.The RV distribution of the 363 targets observed withAPOGEE-2 in the λ Ori-A field during early 2016 showsa strong peak, indicating the presence of a distinct kine-matic population whose central velocity agrees with thatpreviously measured for λ Ori members (24.5 km/sec;Dolan & Mathieu 2001). The Besan¸con model predictsthat the RV distributions of thin and thick disk starshave central velocities near that of the λ Ori population,but much larger velocity dispersions, such that ≤ Rv [Km/s] [ % ] −80 −60 −40 −20 0 20 40 60 80 100 120 Thick disc (313)Thin disc (6439)Apogee−2 (363) Uniformly selected (34) Bin: 5 Km/s
Figure 18.
Distributions of heliocentric RVs along the lineof sight to the λ Ori-A field. The sample sizes in these dis-tributions range from as small as 34 objects, up to nearly6500: to place these on the same scale, the y-axis indicatesthe fraction of each distribution that falls within each bin.The black distribution shows all candidate YSOs observedwith APOGEE-2 in the Lambda Ori A field; the red distri-bution shows the subset of those YSOs that were uniformlyselected. Blue and orange distributions show respectively,the results from the simulations with the Besan¸con Galaxymodel for the thin and thick Galactic discs. Dashed linesindicate the range of the RVs measured for bona fide mem-bers of the major sub-populations of Orion: λ Ori (Dolan& Mathieu 2001), σ Ori (Jeffries et al. 2006) and Orion A(Tobin et al. 2009; Da Rio et al. 2016). The RV distribu-tions of the thin and thick disk populations are significantlybroader than those measured in Orion, such that ≤
20% ofthe Galactic contaminants accessible to our program wouldbe expected to exhibit RVs within the 20-35 km/sec range weuse to kinematically select candidate Orion members. Labelsindicate the total number of stars in each sample and the binsize of the distributions. The distributions were normalizedin order to emphasize their differences. by previously confirmed Orion members. Nonetheless,a substantial number of targets observed by APOGEEexhibit RVs more consistent with the field populationthan with cluster membership; this is not unexpected,particularly in the λ Ori field, where many low-yieldphotometric candidates were added to complete platesafter fibers had been assigned to the smaller samples ofuniformly selected and previously confirmed members(see Table 3).25.2.
Velocity distribution of uniformly selected targetsin all observed fields
To investigate the yield of our uniformly selectedcandidate YSOs, we examined RVs measured by theAPOGEE pipeline for all uniformly selected candidateYSOs with APOGEE-1 or APOGEE-2 spectra observedbefore April 2016. Of the 815 uniformly selected candi-date YSOs with APOGEE spectra, 210 exhibited an IRexcess and optical variability; 493 exhibited only an IRexcess, and 112 exhibited only optical variability. Themajority (70%) of these spectra were originally observedas part of the IN-SYNC survey of Orion A (Da Rio et al.2016, 2017), but the sources they represent meet the cri-teria for inclusion in the uniformly selected sample pre-sented in this work. The remaining 30% of this first setof spectra of uniformly selected targets consists of YSOsin the Lambda Ori, Orion OB1ab, and Orion B regionsthat were observed in Winter 2015-2016 as part of theAPOGEE-2 YSO Goal Science Program. Only 20% ofthe 2700 YSOs targeted by Da Rio et al. (2016) in OrionA were selected for inclusion in the uniformly selectedsample, however, demonstrating that the uniformly se-lected sample’s strength will be in providing a homoge-neous and representative, but not complete, sample ofthe stellar populations that constitute the OSFC. Manyof the 80% of the Da Rio et al. (2016) targets that arenot included in the uniformly selected sample are stillincluded in the APOGEE-2 target list, however, as can-didates from the literature were simply included at alower priority than the uniformly selected YSO candi-dates.Comparing the heliocentric RVs measured by theAPOGEE pipeline for the uniformly selected sources tothe mean velocities previously measured for key sub-populations in Orion provides support for the fidelityof our target selection methods. As shown in Figure19, the distribution of RVs measured for APOGEE ob-served sources is centered at ∼
25 km/s, with sharpboundaries at ∼
20 and ∼
32 km/s. This distributionagrees well with the central velocities previously mea-sured for sub-regions in Orion, which range from 24.5km/s in λ Ori (Dolan & Mathieu 2001) to 31 km/s in σ Ori (Jeffries et al. 2006). Notably, there are are also veryfew sources at RVs <
20 and >
32 km/sec, where Figure18 indicates we should expect to see a fair fraction ofany Galactic thin and thick disk contaminants in theuniformly selected sample. To more quantitatively eval-uate the yield of our uniformly targeted sample of can-didate YSOs, we consider sources with heliocentric ve-locities between 20 km/s and 35 km/s as ‘RV-confirmedYSO candidates’. We note that final confirmation as agenuine YSO requires a more detailed consideration of
Table 4.
RV confirmation foruniformly selected candidatesby field
Field Number NumberName Confirmed Rejected λ Ori A 69 14Orion B 120 26OB1-ab 7 6Orion A 531 26 the source’s spectroscopically determined stellar param-eters, which we defer to future papers analyzing thissample. Of the current sample of APOGEE-observeduniformly selected sources, 89% meet this definition ofan RV-confirmed YSO. The numbers of RV-confirmedcandidates by field are shown in Table 4. Only 11%of the uniformly selected sample exhibit RVs more con-sistent with membership in the Galactic thin and thickdisk than the OSFC; further examination of the sampleto remove the lowest confidence RV measurements mayfurther reduce the fraction of genuine contaminants. CONCLUSIONSWe have designed, implemented, and validated thetargeting algorithms for the APOGEE-2 survey of theOrion Star Forming Complex (OSFC). The overarchingstrategy of the targeting effort is to maximize the yieldof bona fide cluster members, while preserving a sub-sample whose simple selection function is optimized forcomparative analyses across the entire complex. Theuniformly selected stars provide this latter subset, andwere prioritized in targeting to ensure they did not needto ‘compete’ with other target classes with more com-plex selection biases. The remainder of the targetedsample is subject to a highly heterogeneous set of se-lection effects, but the expanded sample will be usefulfor analyses that are less sensitive to those effects – e.g.,measurements of the chemical composition within a spa-tially compact region, where the selection function iseffectively simpler.In detail, we have:1. applied the criteria developed by KL14 to identifyYSOs with warm circumstellar material through-out the OSFC. Using catalogs of previously iden-tified YSOs in Taurus and Orion, we validated thecriteria’s ability to identify YSOs with 2-24 µ m IRexcesses. Applying these cuts to the full catalogof 2MASS+WISE sources across the 420 square3 N u m b e r o f S o u r c e s Total Confirmed Fraction: 0.89IR Confirmed Fraction: 0.91Var. Confirmed Fraction: 0.86
IR Excess: 703 sourcesVariable: 322 sources
Figure 19.
Radial velocities of uniformly selected sources with APOGEE spectra obtained during or before March 2016.Sources include YSOs in all Orion’s sub-regions: Orion A (as previously reported by Da Rio et al. 2016, 2017), Orion B,OrionOB1a/b and Lambda Ori. Sources included in the uniformly selected sample are shown in two histograms, with 1 km/sbins: the RV distribution of sources selected due to an evidence of circumstellar disk through IR excess are shown in cyan, whilesources flagged as optically variable are shown in purple. The RV distribution of variability selected sources is less centrallypeaked, indicating a lower return of high-likelihood kinematic members for this selection method. degree OSFC, we identify 2699 likely YSOs withIR-excesses. Of these, 1307 are brighter than the H ≤ > σ optical variability in asingle Pan-STARRS filter, and requiring ≥ σ vari-ability in multiple Pan-STARRS bands further bi-ases the resultant sample towards bona fide YSOs.We identify 3697 optical variables throughout theOSFC with ≥ σ variability in at least three fil-ters, and mean positions in the g − H vs. g and r − H vs. r color-magnitude diagrams consistentwith membership in the OSFC. Of these, 990 arebrighter than H ≤ λ T a b l e . K o e n i g S e l e c t e d R . A . D ec . W σ W W σ W W σ W W σ W J σ J H σ H K s σ K s ( d e g . )( d e g . )( m ag )( m ag )( m ag )( m ag )( m ag )( m ag )( m ag )( m ag )( m ag )( m ag )( m ag )( m ag )( m ag )( m ag ) C l a ss . - . . . . . . . . . . . . . . . II . - . . . . . . . . . . . . . . . II . - . . . . . . . . . . . . . . . II . - . . . . . . . . . . . . . . . I . - . . . . . . . . . . . . . . . II . - . . . . . . . . . . . . . . . II . - . . . . . . . . . . . . . . . II . - . . . . . . . . . . . . . . . I . - . . . . . . . . . . . . . . . II . - . . . . . . . . . . . . . . . II N o t e — T h e f u ll t a b l e i s pub li s h e d e l ec tr o n i c a ll y ; w e s h o w t h e fi r s t li n e s h e r e a s a n e x a m p l e o f t h e t a b l e ’ s c o n t e n t a nd f o r m a t . T a b l e . O p t i c a l V a r i a b l e s i n t h e O S F C R . A . D ec . σ g e rr g σ r e rr r σ i e rr i σ z e rr z σ y e rr y ( d e g . )( d e g . ) N v a r ζ g ( J y )( J y ) ζ r ( J y )( J y ) ζ i ( J y )( J y ) ζ z ( J y )( J y ) ζ y ( J y )( J y ) N o b s . - . . . e - . e - . . e - . e - . . e - . e - . . e - . e - . . e - . e - . - . . . e - . e - . . e - . e - . . e - . e - . . e - . e - . . e - . e - . - . . . e - . e - . . e - . e - . . e - . e - . . e - . e - . . e - . e - . - . . . e - . e - . . e - . e - . . e - . e - . . e - . e - . . e - . e - . - . - . . e - . e - . . e - . e - - . . e - . e - . . e - . e - . . e - . e - . - . . . e - . e - . . e - . e - . . e - . e - . . e - . e - . . e - . e - . - . . . e - . e - . . e - . e - . . e - . e - . . e - . e - . . e - . e - . - . . . e - . e - . . e - . e - . . e - . e - . . e - . e - . . e - . e - . - . - . . e - . e - - . . e - . e - . . e - . e - . . e - . e - . . e - . e - . - . . . e - . e - . . e - . e - . . e - . e - . . e - . e - . . e - . e - N o t e — T a r g e t ss e l ec t e d f r o m o p t i c a l v a r i a b ili t y . T h e f u ll t a b l e i s pub li s h e d e l ec tr o n i c a ll y , w e s h o w t h e f i r s t li n e s h e r e a s a n e x a m p l e o f t h e t a b l e ’ s c o n t e n t a nd f o r m a t . A ll f l u x m e a s u r e m e n t s a r e r e p o rt e d i nun i t s o f J a n s ky s . Adams, F. C., Lada, C. J., & Shu, F. H. 1987, ApJ, 312, 788Allen, L. E., Calvet, N., D’Alessio, P., et al. 2004, ApJS,154, 363Alves de Oliveira, C., & Casali, M. 2008, A&A, 485, 155Andrews, S. M., & Williams, J. P. 2005, ApJ, 631, 1134Baraffe, I., Homeier, D., Allard, F., & Chabrier, G. 2015,A&A, 577, A42Barrado, D., Stelzer, B., Morales-Calder´on, M., et al. 2011,A&A, 526, A21Barrado y Navascu´es, D., Stauffer, J. R., Morales-Calder´on,M., et al. 2007, ApJ, 664, 481Bouy, H., Alves, J., Bertin, E., Sarro, L. M., & Barrado, D.2014, A&A, 564, A29Brice˜no, C., Calvet, N., Hern´andez, J., et al. 2005, AJ, 129,907Brice˜no, C., Hartmann, L., Hern´andez, J., et al. 2007, ApJ,661, 1119Broos, P. S., Getman, K. V., Povich, M. S., et al. 2013,ApJS, 209, 32Caballero, J. A. 2008, A&A, 478, 667Chambers, K. C., Magnier, E. A., Metcalfe, N., et al. 2016,ArXiv e-prints, arXiv:1612.05560Cottaar, M., Covey, K. R., Meyer, M. R., et al. 2014, ApJ,794, 125Cottaar, M., Covey, K. R., Foster, J. B., et al. 2015, ApJ,807, 27Covey, K. R., Ag¨ueros, M. A., Law, N. M., et al. 2016,ArXiv e-prints, arXiv:1601.07237Cutri, R. M., & et al. 2013, VizieR Online Data Catalog,2328Da Rio, N., Tan, J. C., & Jaehnig, K. 2014, ApJ, 795, 55Da Rio, N., Tan, J. C., Covey, K. R., et al. 2016, ApJ, 818,59—. 2017, ArXiv e-prints, arXiv:1702.04113Davenport, J. R. A., Ivezi´c, ˇZ., Becker, A. C., et al. 2014,MNRAS, 440, 3430Dolan, C. J., & Mathieu, R. D. 2001, AJ, 121, 2124—. 2002, AJ, 123, 387Downes, J. J., Brice˜no, C., Mateu, C., et al. 2014, MNRAS,444, 1793Elmegreen, B. G. 2000, ApJ, 530, 277Esplin, T. L., Luhman, K. L., & Mamajek, E. E. 2014,ApJ, 784, 126Fang, M., Kim, J. S., van Boekel, R., et al. 2013, ApJS,207, 5Fernandez, M. A., Covey, K. R., De Lee, N., et al. 2017,PASP, 129, 084201Foster, J. B., Cottaar, M., Covey, K. R., et al. 2015, ApJ,799, 136 Franciosini, E., & Sacco, G. G. 2011, A&A, 530, A150Garc´ıa P´erez, A. E., Allende Prieto, C., Holtzman, J. A.,et al. 2016, AJ, 151, 144Getman, K. V., Feigelson, E. D., & Kuhn, M. A. 2014,ApJ, 787, 109Gorti, U., & Hollenbach, D. 2009, ApJ, 690, 1539Grankin, K. N., Bouvier, J., Herbst, W., & Melnikov, S. Y.2008, A&A, 479, 827Grankin, K. N., Melnikov, S. Y., Bouvier, J., Herbst, W., &Shevchenko, V. S. 2007, A&A, 461, 183Gutermuth, R. A., Myers, P. C., Megeath, S. T., et al.2008, ApJ, 674, 336Haisch, Jr., K. E., Lada, E. A., & Lada, C. J. 2001, ApJL,553, L153Hern´andez, J., Morales-Calderon, M., Calvet, N., et al.2010, ApJ, 722, 1226Hern´andez, J., Hartmann, L., Megeath, T., et al. 2007,ApJ, 662, 1067Hern´andez, J., Calvet, N., Perez, A., et al. 2014, ApJ, 794,36Hodapp, K. W., Kaiser, N., Aussel, H., et al. 2004,Astronomische Nachrichten, 325, 636Holtzman, J. A., Shetrone, M., Johnson, J. A., et al. 2015,AJ, 150, 148Jeffries, R. D., Maxted, P. F. L., Oliveira, J. M., & Naylor,T. 2006, MNRAS, 371, L6Jeffries, R. D., Jackson, R. J., Cottaar, M., et al. 2014,A&A, 563, A94Kharchenko, N. V., Piskunov, A. E., R¨oser, S., Schilbach,E., & Scholz, R.-D. 2005, A&A, 440, 403Koenig, X., Hillenbrand, L. A., Padgett, D. L., &DeFelippis, D. 2015, AJ, 150, 100Koenig, X. P., & Leisawitz, D. T. 2014, ApJ, 791, 131Krumholz, M. R., & Tan, J. C. 2007, ApJ, 654, 304Magnier, E. 2006, in The Advanced Maui Optical andSpace Surveillance Technologies Conference, E50Magnier, E. 2007, in Astronomical Society of the PacificConference Series, Vol. 364, The Future of Photometric,Spectrophotometric and Polarimetric Standardization,ed. C. Sterken, 153Magnier, E. A., Liu, M., Monet, D. G., & Chambers, K. C.2008, in IAU Symposium, Vol. 248, A Giant Step: fromMilli- to Micro-arcsecond Astrometry, ed. W. J. Jin,I. Platais, & M. A. C. Perryman, 553–559Magnier, E. A., Schlafly, E., Finkbeiner, D., et al. 2013,ApJS, 205, 20Mainzer, A., Bauer, J., Grav, T., et al. 2011, ApJ, 731, 53Majewski, S. R., Schiavon, R. P., Frinchaboy, P. M., et al.2015, ArXiv e-prints, arXiv:1509.054209