Expanding the Y Dwarf Census with Spitzer Follow-up of the Coldest CatWISE Solar Neighborhood Discoveries

How complete is our census of the Sun's closest neighbors? How far does the population of substellar objects born like stars extend into the planetary-mass regime? The Wide-field Infrared Survey Explorer (WISE; Wright et al. 2010), with its unique full-sky sensitivity at 4.6 μm, has unrivaled potential to answer these open questions by identifying the coldest brown dwarfs down to planetary masses. WISE-based discoveries already include the three nearest known brown dwarfs, among which is the coldest known brown dwarf (WISE 0855−0714), an ∼250 K planetary-mass object (Luhman 2014a, 2014b, 2013). Wright et al. (2014) estimated that between 3 and 34 WISE 0855−0714 analogs should be detectable with WISE, yet only perhaps one such candidate has thus far been found (Marocco et al. 2019).

The coolest brown dwarfs revealed by WISE will be key James Webb Space Telescope (JWST; Gardner et al. 2006) targets, overlapping in mass and temperature with extrasolar giant planets and providing simplified laboratories for modeling planetary atmospheres, free of the irradiation and contaminating glare from a primary star. Indeed, WISE has already established the existence of a new brown dwarf spectral class with T_eff ≲ 500 K (Y dwarfs; Cushing et al. 2011; Kirkpatrick et al. 2011). Although prior WISE brown dwarf searches have been highly successful (e.g., Cushing et al. 2011; Kirkpatrick et al. 2011, 2012, 2014, 2016; Griffith et al. 2012; Mace et al. 2013a; Luhman 2014a; Pinfield et al. 2014a; Schneider et al. 2016; Kuchner et al. 2017; Burningham 2018; Tinney et al. 2018), the Y dwarf census has stagnated in recent years at a sample size of just ∼25–30 objects.

The discoveries of nearly all known Y dwarfs can be traced back to WISE. The WISE 3.4 μm (W1) and 4.6 μm (W2) bands were designed for optimal sensitivity to the coldest brown dwarfs (Mainzer et al. 2011b), such that selecting for large W1–W2 color is a highly effective search strategy (e.g., Kirkpatrick et al. 2011, 2012; Griffith et al. 2012). Even more powerful is the combination of WISE-based color and motion criteria, eliminating stationary extragalactic contaminants (e.g., Luhman 2014a, 2014b). Pinfield et al. (2014b) illustrated that faint, unrecognized Y dwarfs remain to be found in the WISE imaging, identifiable with novel search criteria using WISE data alone. Follow-up techniques such as methane on/off imaging (Tinney et al. 2012, 2018) and adaptive optics imaging (Liu et al. 2012; Dupuy et al. 2015) have also proven effective at pinpointing Y dwarfs among the numerous WISE solar neighborhood discoveries.

Our CatWISE analysis (Eisenhardt et al. 2019) represents a major step toward realizing the entire WISE data set's full sensitivity for brown dwarf discovery. Mining the vast WISE imaging archive to its faintest depths is a formidable challenge, and prior WISE motion searches have generally been limited by restricting to bright single-exposure detections and/or the short half-year time baseline of 2010–2011 observations. CatWISE pushes several magnitudes deeper than the foregoing WISE-based motion surveys by jointly analyzing 4 yr of WISE and NEOWISE (Mainzer et al. 2011a, 2014) data spanning the 2010–2016 time period. CatWISE thereby provides long time baseline WISE proper motions for roughly a billion mid-infrared sources over the full sky.

Using the CatWISE Preliminary Catalog¹⁴ and drawing upon well-established faint moving-object selection/confirmation techniques (e.g., Lépine et al. 2002), we have performed an extensive motion-based search for previously undiscovered solar neighborhood constituents. As part of our ground- and space-based follow-up program, we have obtained Spitzer IRAC (Fazio et al. 2004; Werner et al. 2004) photometry of ∼170 newly discovered brown dwarf candidates suspected of being extremely cold and/or nearby. At CatWISE depths, the hallmark of such targets is the presence of a moving 4.6 μm (W2) source with no firmly detected 3.4 μm (W1) counterpart.

Detectable motion and a W2 magnitude alone are insufficient to estimate the basic parameters of most immediate interest for these brown dwarf candidates: spectral type, temperature, luminosity, distance, and near-infrared (NIR) flux. The mid-infrared color, whether W1–W2 from WISE or [3.6]–[4.5] from Spitzer, represents a critical diagnostic in obtaining estimates for all of these quantities, as both colors tend to increase monotonically toward later spectral types beyond mid-T (e.g., Patten et al. 2006; Kirkpatrick et al. 2011). For our targets with W1 nondetections, Spitzer provides the only opportunity to measure a mid-infrared color. With spectral type estimates based on IRAC colors in hand, luminosity, distance, and NIR flux estimates also follow.

Based on the Spitzer follow-up we present in this work, many of our discoveries have photometric spectral type estimates placing them within the 20 pc volume and/or very red [3.6]–[4.5] colors most consistent with spectral type Y. Among these, CWISEP J144606.62−231717.8 (hereafter CWISEP 1446−2317) stands out with an exceptionally large but highly uncertain [3.6]–[4.5] color of 3.71 ± 0.44 mag.

In Section 2 we briefly summarize the relevant characteristics of the WISE and NEOWISE missions. In Section 3 we provide a concise overview of CatWISE. In Section 4 we describe our selection of brown dwarf candidate targets for follow-up Spitzer photometry. In Section 5 we present the basic properties of our Spitzer photometry target sample. In Section 6 we explain our Spitzer observing strategy. In Section 7 we present our Spitzer color measurements. In Section 8 we combine WISE and Spitzer astrometry to confirm the motions of our brown dwarf candidates. In Section 9 we present complementary NIR photometry drawn from our ground-based follow-up observations and archival data sets. We discuss the combined implications of our photometric and astrometric analyses in Section 10. We conclude in Section 11.

Launched into low-Earth orbit in late 2009, WISE is a satellite-borne 40 cm aperture telescope. During early and mid-2010, WISE mapped the entire sky in four broad infrared bandpasses centered at 3.4 (W1), 4.6 (W2), 12 (W3), and 22 (W4) μm. Although observations in the two longest wavelength channels were discontinued by late 2010 due to cryogen depletion, WISE kept observing in W1 and W2 until 2011 February as part of the NEOWISE mission (Mainzer et al. 2011a). WISE was placed into hibernation from 2011 February until 2013 December, at which point it recommenced surveying in W1 and W2 thanks to the NEOWISE-Reactivation (NEOWISE-R; Mainzer et al. 2014) mission extension. WISE has continued observations since reactivation to this writing (2019 October). A typical sky location is observed during an ∼1 day time period once every 6 months, and WISE has now performed a total of more than 13 complete sky passes in W1 and W2. The high quality of W1 and W2 imaging has remained essentially unchanged throughout the entire WISE lifetime (Mainzer et al. 2014; Cutri et al. 2015).

Although NEOWISE-R has now supplied the vast majority of W1/W2 observations, the mission itself does not provide any coadded data products optimized for science beyond the inner solar system. As a result, AllWISE (Cutri et al. 2013) has remained the definitive coadded WISE catalog for many years, despite incorporating only the ∼13 months of prehibernation WISE imaging.

Our CatWISE archival data analysis program (Eisenhardt et al. 2019) has combined ∼4 yr of 2010–2016 WISE/NEOWISE data to build a deeper, longer time baseline successor to AllWISE at 3–5 μm. Whereas the AllWISE Source Catalog directly modeled WISE single-exposure images, CatWISE instead applies the AllWISE cataloging software to "unWISE" coadds (Lang 2014) as a computational convenience. By detecting W1/W2 sources in 4 yr depth unWISE stacks (Meisner et al. 2018a), CatWISE extracts 5σ sources to Vega magnitudes of W1 = 17.67 and W2 = 16.47, ∼0.6 mag deeper than AllWISE.¹⁵ CatWISE fits apparent linear motions for every source using a set of "time-resolved" unWISE coadds (Meisner et al. 2018b, 2018c). Each time-resolved coadd stacks the ∼1 day of WISE frames together at a given sky location during a single sky pass, sampling the motion at 6 month intervals. Such coaddition results in effectively no loss of motion information for objects in the solar neighborhood¹⁶ (μ ≲ 10'' yr⁻¹). By virtue of its >10× extended time baseline and 4× input imaging increase, CatWISE derives motions an order of magnitude more accurate than those of AllWISE for ∼900 million sources over the full sky.¹⁷ We should therefore expect CatWISE motion and/or color selections to reveal many nearby brown dwarfs not previously identified with AllWISE. This includes objects below the AllWISE detection limit and also those with AllWISE motion measurements too noisy to be statistically significant.

Artifact flagging is a key ingredient in WISE-based rare-object searches, where anomalies due to bright stars and blending dramatically outnumber the astrophysical sources of interest. Our brown dwarf searches take advantage of two complementary artifact-flagging capabilities provided by CatWISE: (1) CC flags inherited from AllWISE via cross-match and (2) unWISE ab_flags, similar to the CC flags but available even in cases when a CatWISE source has no AllWISE counterpart.¹⁸

Our Spitzer follow-up observations presented throughout this work were acquired as part of program 14034 (hereafter p14034; PI: Meisner). At the time of our Cycle 14 proposal submission (2018 March), Spitzer's final observations were expected to take place on 2019 November 30. Given Spitzer's impending retirement, we sought to fill our target list with the most exceptional WISE-based cold brown dwarf candidates selected by any/all means necessary, and as a result, virtually no emphasis was placed on sample uniformity/homogeneity.

Additionally, our search for Spitzer targets was performed as part of a larger effort to fully mine CatWISE (and unWISE/AllWISE) for moving-object discoveries, including earlier-type candidates accessible via our ground-based photometric/spectroscopic observing programs. Only a subset of our brown dwarf searches were specifically tailored to supply targets for our p14034 Spitzer campaign. Through the combination of multiple searches described in Sections 4.1–4.3, we discovered (and visually confirmed) a total of ∼2500 previously unpublished moving objects. Most of these ∼2500 discoveries are not appropriate Spitzer targets because they can be sensibly followed up from ground-based facilities. Discussion of additional CatWISE motion discoveries beyond those followed up via Spitzer p14034 is deferred to future papers.

We obtained p14034 observations for only the subset of our moving-object discoveries that would be extremely difficult to follow up from any platform other than Spitzer and/or for which Spitzer 3.6 μm (hereafter ch1) and 4.5 μm (hereafter ch2) photometry provides a critically informative diagnostic. We therefore selected only discoveries falling into one or more of the following categories when constructing our p14034 target list.

• 1.  
  Nonstationary objects detected exclusively in W2. Detectable motion (μ ≳ 200 mas yr⁻¹ at the typical W2 magnitude of our targets; Eisenhardt et al. 2019) suggests that such objects must be relatively nearby. For a moving source, nondetections in W1 and all other shorter wavelengths indicate an extremely cold temperature and low luminosity and hence a small distance. Moving objects detected solely in W2 are thus prime close-by Y dwarf candidates, and they were considered the highest-priority target class for our Spitzer p14034 campaign.
• 2.  
  Moving objects detected in W1 but still potentially within 20 pc. In some cases, the presence of a faint W1 counterpart indicated that a moving object was insufficiently red to plausibly be a Y dwarf. We nevertheless retained such objects for p14034 consideration if the photometric distance estimate implied by W2 and W1–W2 indicated that the object could potentially be a new member of the 20 pc sample. Completing the census of objects within this volume is crucial for space density and mass function analyses (Kirkpatrick et al. 2019a). Furthermore, Spitzer ch1/ch2 photometry can provide significantly refined photometric distance and spectral type estimates relative to those dependent upon low-significance W1 detections. For these reasons, we deemed d ≲ 20 pc candidates worthy p14034 Spitzer targets despite detection in W1.
• 3.  
  Candidate very late type (≳T0) common proper-motion (CPM) companion objects potentially in wide-separation visual binary systems. Possible CPM status was usually noted serendipitously via our standard visual inspection motion-vetting process (Section 4.4) and subsequently checked using the similarity between the CatWISE motion of the candidate secondary and that of the putative primary. Spitzer ch1/ch2 observations for CPM candidates can provide a relatively accurate photometric distance and astrometry for an improved motion estimate. These can then be compared against the distance/motion of the primary to better test the comoving pair hypothesis. Late-type wide CPM companions are rare, so it is also highly valuable to obtain Spitzer ch1/ch2 photometric data points for these critical benchmarks. Presentation of late-type CatWISE CPM discoveries followed up via Spitzer p14034 in Marocco et al. (2020).
• 4.  
  Objects with exceptionally large proper motion (μ ≳ 1'' yr⁻¹) and/or reduced proper motion¹⁹ (H_W2 ≳ 19.5). Very high (reduced) proper motion of a source first recognized in the mid-infrared may be indicative of several interesting phenomena, including a late-type subdwarf, a very cold nearby neighbor to the Sun, and/or large tangential space velocity.

A given object can sometimes fall under more than one of the above target categories. But no moving-object discovery was placed on the p14034 program without meeting at least one of the above four selection criteria. All of our targets were, at the time of Astronomical Observation Request (AOR) submission to the p14034 program, never-before-published discoveries; our Spitzer campaign was not intended to provide supplementary follow-up of known objects drawn from the literature. Additionally, we vetoed candidates based on checking the Spitzer Heritage Archive (SHA) for existing unpublished observations by other teams targeting recent brown dwarf discoveries (e.g., Backyard Worlds; Kuchner et al. 2017) so as not to wastefully duplicate Spitzer pointings.

In our target selection decision-making, we sought to pursue a "high-risk, high-reward" strategy. We declined to place many of our unmistakable late-type moving-object discoveries on p14034 because secure W1 detections made clear that they were simply T dwarfs at ≳30 pc. Instead of such "safe" objects, we opted to prioritize very faint W2-only candidates with Y dwarf potential, even those so marginal in WISE that they might ultimately turn out to be entirely spurious. The rationale is that mid-T dwarfs do not necessitate Spitzer's unique capabilities in the same way that Y dwarfs do, so we ought to observe as many Y dwarfs as possible before Spitzer's retirement, even at the expense of false positives. Similarly, to find the most superlative targets before Spitzer's retirement, we chose not to limit ourselves to searches based strictly on CatWISE but instead leveraged additional data products at hand (the AllWISE Catalog and unWISE coadds; see Sections 4.2 and 4.3, respectively).

The following subsections provide a detailed list of our many (14) distinct searches that contributed at least one target to the p14034 photometry presented in Table 1. For convenience, we refer to each search by a shorthand that consists of one capital letter followed by one digit. Table 1 lists which specific search(es) discovered each target in the "search method" column. Our searches generally rely on motion selection, color selection of objects that are very red in W1–W2, or some combination of the two. Each search's candidate identification yielded far more false positives than genuine newly discovered moving objects, so extensive visual inspection campaigns were performed to assess each candidate's possible motion by eye (Section 4.4). The dominant sources of false positives are bright star artifacts, blends, and statistical measurement fluctuations that lead to the impression of potentially large motion and/or red W1–W2 color.

4.1. CatWISE-based Selections

Most of our p14034 targets were discovered by directly mining the CatWISE Preliminary Catalog. We ran a total of 11 distinct moving-object searches on this catalog. Note that our searches were performed on early, unfinished versions of CatWISE, rather than the published CatWISE Preliminary Catalog described in Eisenhardt et al. (2019). In many cases, limited or no artifact flagging existed in these early CatWISE databases; this often shaped/constrained our tailoring of search criteria. Also, all CatWISE-based selections were run at a time when there was no distinction between sources now separated into the CatWISE "catalog" and "reject" tables.²⁰ Appendix A of Eisenhardt et al. (2019) provides column descriptions for the CatWISE quantities involved in our selection criteria.

Our 11 CatWISE-based searches fall under two broad categories: (1) traditional catalog queries, each implementing a set of hard cuts (Section 4.1.1), and (2) supervised machine-learning methods trained on human-verified late-type moving objects (Section 4.1.2).

4.1.1. CatWISE Catalog Cuts

Prior WISE motion surveys such as the AllWISE and AllWISE2 motion searches (Kirkpatrick et al. 2014, 2016) were performed via selections cutting on motion, color, and artifact-flagging catalog columns. Motivated by the success of these previous WISE-based motion surveys, we modeled eight of our search methods after this same general approach but now applied to the newly available CatWISE data set.

Selection method C1 combines CatWISE color and motion information to isolate objects that are both red in W1–W2 and have large W2 reduced proper motion (H_W2). The usage of H_W2 rather than simply proper motion itself is a way of prioritizing fast-moving objects of low luminosity, such as Y dwarfs and late-type subdwarfs. The C1 candidates are obtained from a full-sky CatWISE query requiring (w2snr > 20), (w2snr_pm > 20), (H_W2 > 15), (rchi2/rchi2_pm > 1.03), (w2rchi2_pm < 2), (W1–W2 ≥ 1.5), and (Q < 10⁻⁵). Here Q is a significance-of-motion metric defined in Section 3.4.1 of Kirkpatrick et al. (2014) as $Q={e}^{-{\chi }_{\mathrm{motion}}^{2}/2}$, where ${\chi }_{\mathrm{motion}}^{2}$ = (pmra/sigpmra)² + (pmdec/sigpmdec)². Note that typical WISE sources such as main-sequence stars have a color of W1–W2 ≈ 0 (Vega). The relatively high W2 signal-to-noise ratio (S/N) requirements stipulated as part of this query were necessary to keep the candidate sample size manageable, as this search was performed at a time when no CatWISE artifact flagging was available.

Search C2 is a variant of search C1 replacing the (W1–W2 ≥ 1.5) color cut with a proper-motion cut of μ > 05 yr⁻¹. Again, search C2 was executed without the benefit of any artifact-flagging information.

Search C3 implements a combination of the cuts from C1 and C2 and was performed after artifact-flagging columns had been added to our working CatWISE database. By leveraging the artifact flagging to remove many spurious candidates, C3 extended to a lower W2 S/N than C1/C2, specifically restricting to the range 10 < w2snr_pm ≤ 20. Also included in C3 was an additional requirement of (w2snr_pm > w1snr_pm). The significance of the motion criterion was made more stringent than in C1 and C2, requiring Q < 10⁻⁶. Both the proper-motion cut μ > 05 yr⁻¹ from C2 and the H_W2 > 15 reduced proper-motion cut from C1/C2 were applied. Lastly, search C3 sought to eliminate artifacts by requiring (ab_flags = "00").

Search C4 is a variant of C1 run after artifact flagging had been put in place. The CatWISE catalog cuts in C4 are the same as those in C1, aside from the following updates: C4 requires (ab_flags = "00"), flips the rchi2/rchi2_pm cut to (rchi2/rchi2_pm ≤ 1.03), and adds a new (w1snr_pm = null) W1 nondetection requirement.

Although based on CatWISE catalog cuts, searches C5–C8 are not immediate descendants of the C1–C4 approaches. Search C5 consists of an all-sky CatWISE query requiring Q < 10⁻⁵, (w2mpro > 13), (w2snr > 10), (k1 = 0), and (k2 = 3). Here k1 (k2) is a CatWISE column indicating, for each object, which scan direction(s) provided a successful photometric measurement in W1 (W2); k1 = 0 means that neither ascending nor descending WISE scans provided good W1 photometry, while k2 = 3 means that good W2 photometry was obtained for both WISE scan directions. This query is thus, in effect, implementing an alternative means of identifying W2-only moving objects within the CatWISE catalog. Search C5 was also conducted prior to the existence of any CatWISE artifact-flagging capabilities.

Search C6 is a full-sky CatWISE query requiring (w2snr > 10), w1flux < (w2flux$\cdot {10}^{-1.5/2.5}-2\cdot$w1sigflux), (rchi2/rchi2_pm > 1.03), (w2rchi2_pm < 2), Q < 10⁻⁶, and AllWISE CC flags = "0000" when available. The AllWISE CC flags were gathered via a CatWISE–AllWISE positional cross-match. One notable aspect of this search is that a color cut of effectively (W1–W2 > 1.5) is implemented in terms of fluxes rather than magnitudes to avoid the complications associated with, e.g., quoting magnitudes in cases of zero or negative W1 flux (as can happen for a very red W1 nondetection). In the CatWISE catalog, both w1flux and w2flux have units of Vega nanomaggies.²¹

Search C7 is a variant of search C6 that lowers the W2 S/N threshold to (w2snr > 5) in an attempt to push fainter. To balance out the large influx of sources at relatively low S/N, the color criterion was made more stringent at effectively (W1–W2 > 2.5), with the actual flux-based cut being w1flux <(0.1 · w2flux−2 · w1sigflux). The significance-of-motion criterion was loosened to (Q < 10⁻⁵) for search C7. We additionally required that AllWISE CC flags not contain a capital letter (when an AllWISE cross-match was available) and (ab_flags = "00").

Search C8 is a pure color selection. We required that candidates not be significantly detected at W1 (w1snr < 3) and be well detected in W2 (w2snr > 10) in CatWISE. A negative cross-match (25 radius) against AllWISE sources with (w3mpro < 13) was used to remove extragalactic contaminants that tend to be red in W2–W3 color. Search C8 made use of CatWISE ab_flags to require that candidates not be flagged as W2 ghosts, W2 latents, or W2 diffraction spikes.

4.1.2. CatWISE Machine Learning

Search method M1 is described fully in Section 3 of Marocco et al. (2019). In brief, M1 uses the XGBoost software package (Chen & Guestrin 2016) to perform supervised machine learning on the CatWISE catalog. A training set was constructed from the CatWISE sources corresponding to known high proper motion late-T and Y dwarfs, with the goal of finding other CatWISE entries displaying similar properties. Search M1 is restricted to CatWISE objects that are faint and red by only classifying the subset of CatWISE rows with (w2mpro > 14) and also satisfying

$$\begin{eqnarray}&&\begin{array}{l}(w1{mpro}-(3\times w1{sigmpro}))\\ \quad -\,(w2{mpro}+(3\times w2{sigmpro}))\geqslant 1.0.\end{array}\end{eqnarray} \tag{ 1 }$$

This enforces a requirement that each retained source would have a color of (W1–W2 ≥ 1) even if the CatWISE-reported magnitudes turned out to be 3σ bright in W2 and 3σ faint in W1.

Search M2 is a variant of search M1, with the classifier trained on a sample including hitherto identified p14034 targets rather than a sample consisting exclusively of previously published late-type brown dwarfs.

Search M3 is also a modified version of search M1 but removing the W1–W2 color criterion in Equation (1). The motivation for this variant is the possibility of recovering overlooked late-type moving objects with WISE color measurements corrupted due to blending with background sources, plus the potential to find additional fast-moving sources irrespective of W1–W2 color.

4.2. AllWISE-based Selections

4.3. unWISE-based Selection

4.4. Visualization Tools

Extensive visual inspection of moving-object candidates delivered by the searches described in Sections 4.1–4.3 played a vital role in providing a high-purity sample of brown dwarf targets for our Spitzer p14034 campaign. In total, we visually inspected ∼130,000 candidates in the course of the searches described in Sections 4.1–4.3. A major factor enabling our discoveries based on CatWISE, AllWISE, and unWISE was our usage of visualization tools/aids that leveraged the full W1/W2 time baseline afforded by the combination of prehibernation and postreactivation WISE/NEOWISE imaging.

4.4.1. Finder Charts

We created a new, customized version of the multipanel, multiwavelength finder chart program used by prior WISE motion searches such as Kirkpatrick et al. (2016) and Kirkpatrick et al. (2016); for an example, see Figure 1 of Schneider et al. (2016). We added two sets of W1/W2 time-resolved unWISE coadd cutouts for each candidate, with one set at the beginning of the prehibernation WISE mission and one at the end of the third year of the postreactivation NEOWISE mission. Taking advantage of the ∼6.5 yr WISE–NEOWISE time baseline in this way allowed us to perceive the source motion (or lack thereof) using only W1/W2 data at widely spaced epochs, rather than needing to obtain an appreciable time baseline by comparison of WISE images to shorter-wavelength data sets.

4.4.2. WiseView Interactive Blinker

Our visual inspection workflow relied heavily on a new visualization tool called WiseView (Caselden et al. 2018) not previously available for WISE moving-object searches such as the AllWISE and AllWISE2 motion surveys. In contrast to static multiwavelength finder charts, WiseView is an interactive browser-based interface for creating customized animated blinks of time-resolved unWISE coadds. Numerous blink parameters are tunable in real time, including the band(s) shown (W1, W2, or both), central sky position, frame rate, stretch, and field-of-view size. Figure 1 illustrates an example of the WiseView interface as employed when vetting one of our W2-only brown dwarf discoveries.

Zoom In Zoom Out Reset image size

4.4.3. DESI Imaging Viewer

To select moving objects detected in W2 but not at any shorter wavelengths, we made extensive use of the DESI pre-imaging "Legacy Surveys" sky viewer²² to inspect red-optical survey images. This viewer allows for interactive exploration of wide-area survey data sets with deep z- and Y-band imaging, in particular DECaLS/MzLS (z ≈ 23.0 AB at 5σ over ∼1/3 of the sky; Dey et al. 2019) and Dark Energy Survey DR1 (z ≈ 23.4 AB and Y ≈ 22.2 AB at 5σ over ∼1/8 of the sky; Abbott et al. 2018). Visible z and/or Y counterparts were generally treated as evidence that a moving-object candidate was either insufficiently red to be a Y dwarf or else extragalactic if the red-optical counterpart appeared extended.

4.4.4. IRSA Finder Chart

In some cases where conclusively confirming/denying motion by eye proved difficult, we consulted AllWISE W3 and W4 images via the IRSA Finder Chart application.²³ Because our motion candidates are so faint (median W2 ∼ 15.9; see Section 5), a strong counterpart at W3 and/or W4 would only be expected in the case of a stationary extragalactic source, not for a late-type brown dwarf. We therefore avoided selecting sources seen to have coincident W3/W4 emission as p14034 targets.

4.4.5. PanSTARRS-1 Cutouts

Before placing candidates on the p14034 target list, we inspected PanSTARRS-1 image cutouts²⁴ for objects north of δ ≈ −30° (Chambers et al. 2016). Visible detections in PanSTARRS-1 were in general used to veto potential Y dwarf candidates; given the faintness of our sample in W2, detection in any PanSTARRS-1 filter would be inconsistent with a Y dwarf spectral type.

We filled our allocated 40.5 hr of Cycle 14 Spitzer time with 174 unique brown dwarf candidate targets, each of which received a single corresponding AOR as detailed in Section 6. As discussed in Section 4, these 174 targets represent only a small subset of the visually vetted moving-object discoveries yielded by our searches. Figure 2 shows that our targets are much fainter than those of previous WISE motion surveys, including the AllWISE and AllWISE2 searches (Kirkpatrick et al. 2014, 2016). The median W2 magnitude of our targets is 15.93, with a dispersion of 0.47 mag. For comparison, the W2 single-exposure depth is W2 ≈ 14.5, which has represented the faint limit of prior WISE-based motion searches (e.g., Luhman 2014a; Schneider et al. 2016). Figure 3 shows the W2 S/N distribution of our Spitzer photometry sample. The median (mean) W2 S/N is just 16.2 (18.4), with a dispersion of 8.6. Note that these values indicate the total W2 S/N when combining 4 yr of WISE/NEOWISE imaging; detections in any time slice of the available W2 imaging will generally be of even lower significance.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figure 4 shows the spatial distribution of our p14034 targets, which, as expected, are scattered fairly uniformly across the entire sky while preferentially avoiding the confused Galactic plane.

Zoom In Zoom Out Reset image size

5.1. Candidates with Archival ch1/ch2 Data

5.2. Targets as Yet Unobserved by Spitzer

We based our observing strategy on those of prior Spitzer campaigns designed to measure the colors of WISE-selected late-T and Y dwarf candidates (e.g., programs 70062 and 80109; PI: Kirkpatrick). These foregoing Spitzer programs generally observed substantially brighter objects than those comprising our p14034 target list and showed that coaddition of five 30 s ch1 dithers typically achieves an S/N of 10 (5) at a Vega magnitude of ch1 = 18.0 (18.75).

The boundary between the ch1−ch2 colors of late-T and Y dwarfs occurs at ch1−ch2 ≈ 2.4 (e.g., Kirkpatrick et al. 2019a), and our primary goal was to obtain ch1 imaging deep enough to distinguish between late-T and Y dwarfs. We therefore attempted to engineer our number of ch1 dithers such that a 5σ ch1 detection would always establish a color of at least ch1−ch2 = 2.75 mag.

For each target, we chose either 5, 7, 9, 11, or 13 ch1 dithers (at 30 s dither^–1) so that we would achieve a ch1 S/N of at least 5 for ch1−ch2 = 2.75, under the assumption that the ch2 mag would be equal to the W2 mag. According to this strategy, W2 ≤ 16 candidates receive five ch1 dithers, and the break point between five and seven dithers is W2 = 16.0. Then the break point between seven and nine dithers occurs at W2 = 16.18, and so forth. We exercised some case-by-case discretion in bumping up the exact number of dithers chosen but always used a ch1 S/N of 5 at ch1−ch2 = 2.75 to enforce a minimum number of ch1 dithers.

For each target, both ch1 and ch2 were obtained as part of the same AOR. This minimizes slew overheads and ensures that our color measurements cannot be corrupted by any long timescale variability aliasing into images of the same target taken at large time separation. We always acquired the same number of 30 s dithers in both ch1 and ch2 for each target. This effectively maintains a fixed ch2 S/N of ∼75 for the high-significance ch2 detection across all members of our sample, which ensures the high value of each target's ch2 detection for astrometry (Section 8.4). We dithered with a "cycling" pattern of medium scale.²⁵

The Spitzer p14034 imaging analyzed in this work was acquired between 2018 October 21 and 2019 August 16.

All of our Spitzer photometry/astrometry analyses presented in this work are based on custom mosaics built from the single-dither basic calibrated data (BCD) images with MOPEX (Makovoz & Khan 2005; Makovoz & Marleau 2005). We created one such custom mosaic per band (ch1, ch2) per AOR. Relative to using the default "PBCD" mosaics supplied for each AOR via SHA, creating our own custom mosaics provided us extra freedom to, for example, reject occasional problematic single-frame IRAC images with, e.g., a cosmic ray contaminating the targeted brown dwarf candidate. The algorithmic rejection of single-frame outliers such as cosmic rays also appears to be much better overall in our custom mosaics than in the PBCD stacks. In a handful of cases, we excised one or two BCD frames from our custom mosaics due to the presence of a cosmic ray contaminating the faint ch1 counterpart (CWISEP 0035−1532, CWISEP 0403−4916, CWISEP 1434−1344, CWISEP 2251−0740, and CWISEP 2355+3804).

Extraction and photometry of sources within our custom Spitzer mosaics proceeded as described in Section 4 of Marocco et al. (2019). In brief, we use the MOPEX/APEX software to detect sources in our custom mosaics and perform both point response function (PRF)-fit and aperture photometry (including the application of an aperture correction) in both ch1 and ch2. Note that our photometry was always run independently in ch1 and ch2; we did not employ a forced photometry approach. By default, we used an S/N = 5 source detection threshold. In the case of two very red objects (CWISEP 1434−1344 and CWISEP 1446−2317), obtaining a ch1 counterpart extraction required lowering the ch1 detection threshold to S/N = 2.

Table 1 lists our ch1 and ch2 photometry results. The quoted magnitude values and their uncertainties are derived by averaging the aperture-based and PRF-fit quantities for each target in each Spitzer band. Table 1 also reports the W1 and W2 photometry for each target. With only a small number of exceptions, such as our two discoveries not detected by CatWISE, this WISE magnitude information is drawn from the CatWISE columns w1mpro_pm, w1sigmpro_pm, w2mpro_pm, and w2sigmpro_pm.

7.1. Spurious Candidates

7.2. Candidates with Two Spitzer Counterparts

We seek to use motion as a proxy for confirming that an object is a nearby brown dwarf. High-significance motion establishes solar neighborhood membership, whereas objects consistent with remaining stationary may be, e.g., of extragalactic origin. Because we lack spectroscopic confirmations and our most interesting p14034 targets are detected only in WISE and Spitzer, detailed astrometric analysis is needed to best determine whether each source is indeed moving. In this section, we explain how we have combined astrometry from both WISE and Spitzer to best identify the subset of our brown dwarf candidates that have statistically significant proper motions. The inclusion of Spitzer astrometry is a critical component of this analysis, since our Spitzer data point provides a completely independent cross-check on the perceived WISE-based motion used to select our candidates; if the Spitzer detection "lines up" along the WISE astrometric trajectory, then this gives us strong reassurance that the candidate was selected due to true motion rather than a rare fluke in the WISE data.

Using the methodology described here, we limit our astrometric analysis to fits of apparent linear motion; we do not seek to obtain/constrain the parallaxes of our targets. In general, fitting parallaxes for brown dwarfs as faint as our targets will require multiple epochs of Spitzer (or similarly precise) observations sampling both sides of the parallactic ellipse (e.g., Kirkpatrick et al. 2019a), whereas our p14034 imaging provides just a single Spitzer astrometric epoch. Additionally, as discussed in Section 8.3, we typically must coadd data acquired from both sides of WISE's orbit in order to obtain W2 detections of our exceedingly faint targets, meaning that our WISE astrometry is generally unsuitable for parallax fitting.

Our astrometric analysis incorporates ch2 and W2 but never ch1 or W1. This is because our targets are very red in both ch1−ch2 and W1–W2. For our p14034 Spitzer imaging, where both bands received the same total exposure time, the S/N of each target's ch2 detection will be much higher than that of its ch1 detection. Further, because the ch1 and ch2 images of a given target are nearly contemporaneous, folding in ch1 does not offer the possibility of an appreciably extended time baseline, and in combination with ch2 astrometry, it would merely lead to a negligible improvement in the p14034 Spitzer positional precision. The same considerations apply with regard to W1: data acquisition in W1 and W2 is simultaneous, so W1 astrometry would provide only a set of much less precise positions at the same epochs as our W2 astrometric data points.

All Gaia-recalibrated W2 and ch2 (R.A., decl.) coordinates quoted throughout this paper are in the International Celestial Reference System (ICRS). The Spitzer and WISE positions reported in Tables 3 and 4 are relative rather than absolute; we did not attempt to correct for the typically very small parallaxes of our astrometric calibration sources.

8.1. Gauging Significance of Motion

There are various ways one could imagine quantifying significance of motion. In this work, we opt for a simple, intuitive metric that has been applied in the course of past WISE-based moving-object analyses:

$$\begin{eqnarray}&&{\chi }_{\mathrm{motion}}^{2}={({\mu }_{\alpha }/{\sigma }_{{\mu }_{\alpha }})}^{2}+{({\mu }_{\delta }/{\sigma }_{{\mu }_{\delta }})}^{2}.\end{eqnarray} \tag{ 2 }$$

This ${\chi }_{\mathrm{motion}}^{2}$ statistic has previously been used during, e.g., the AllWISE and AllWISE2 motion surveys (Kirkpatrick et al. 2014, 2016). It tends to increase with larger (absolute) linear motion components, as well as with decreasing uncertainties on the linear motion measurements. It can also be thought of as corresponding to a false-alarm rate, $Q={e}^{-{\chi }_{\mathrm{motion}}^{2}/2}$, where Q is the probability of a statistical fluke causing ${\chi }_{\mathrm{motion}}^{2}$ to be exceeded.

In this work, we set the threshold for WISE+Spitzer "motion confirmation" at Q = 10⁻⁵, which corresponds to ${\chi }_{\mathrm{motion}}^{2}\,=23.03$. Ignoring the relatively high-precision Spitzer astrometric data points available from p14034 follow-up, a substantial fraction of our sample's targets (46% = 79/173) have CatWISE ${\chi }_{\mathrm{motion}}^{2}$ less than this threshold, illustrating the critical need to combine ch2 and W2 astrometry toward better confirming/refuting source motions.

8.2. Strategy for Combining WISE and Spitzer Astrometry

8.3. WISE Astrometry

It is challenging to obtain a time series of WISE astrometric detections for moving sources as faint as our targets. By selection, our brown dwarf candidates tend to be completely undetected in W1. In W2, they typically have an S/N of just ∼15 even when combining 4 yr of WISE/NEOWISE data (see Figure 3). As a result, it is almost never possible to extract single-exposure W2 astrometry for any target in our sample, and we do not attempt to do so. Furthermore, in most cases, it is not possible to obtain W2 detections of our targets even in time-resolved unWISE coadds that stack together the ≳12 W2 exposures at each sky location during each single 6 month WISE sky pass. Therefore, we often must perform source extraction on W2 stacks that combine multiple WISE sky passes.

The latest full-sky unWISE data release (Meisner et al. 2019) provides time-resolved coadds that bin W2 exposures into a series of single WISE sky passes, incorporating both the prehibernation time period (2010–2011) and the first 4 yr of NEOWISE-R observations (2013–2017). These time-resolved unWISE coadds form the starting point for our W2 astrometry analysis. Ideally, we would also have access to such unWISE coadds for the 2018 time period, but these have not yet been generated. However, since the 2018 NEOWISE-R data are only slightly earlier in time than our Spitzer p14034 imaging, there would be relatively little marginal benefit attained by including 2018 W2 data; this additional W2 imaging would not increase our overall WISE+Spitzer time baseline and would only contribute a relatively weak astrometric constraint adjacent in time to our much higher precision p14034 Spitzer data point. With the Meisner et al. (2019) unWISE data set, we typically have 5 yr of W2 imaging available at each sky location, corresponding to 10 time-resolved W2 coadds, which are labeled e000, e001, ..., e009. For concreteness, one representative cadence of such time-resolved unWISE coadds at fixed sky locations is e000 ∼ 2010.4, e001 ∼ 2010.9, e002 ∼ 2014.4, ..., e009 ∼ 2017.9.

In gathering astrometric detections for our brown dwarf candidates, we always attempt extractions from all available e??? W2 unWISE coadds. Because this still leaves many of our faint targets undetected, we also generate and perform source extraction on a set of W2 "meta-coadd" time slices. This set of meta-coadd slices is listed in Table 5. The "pre" slice stacks all prehibernation e??? W2 unWISE coadds together to form a deeper 2010–2011 meta-coadd. Analogously, the "post" slice stacks all posthibernation e??? W2 unWISE coadds together, resulting in a deep 2013–2017 coadd. The "post?_1yr" meta-coadd slices stack the posthibernation e??? W2 unWISE coadds within a series of four nonoverlapping 1 yr time intervals. Lastly, the "post?_2yr" meta-coadd slices stack the postreactivation e??? W2 unWISE coadds within a series of two nonoverlapping 2 yr time intervals.

Table 5.  unWISE Meta-coadd Time Slices

Time Slice Name	Approx. Time Period
	(Calendar Years)
pre	2010.0–2011.1
post	2014, 2015, 2016, and 2017
post0_1yr	2014
post1_1yr	2015
post2_1yr	2016
post3_1yr	2017
post0_2yr	2014 and 2015
post1_2yr	2016 and 2017

Download table as:  ASCIITypeset image

We generate our W2 meta-coadds for the full 156 × 156 unWISE coadd tile footprint containing each target. This has multiple advantages relative to considering only small postage stamps about our targets. First, it allows us to obtain a large number of Gaia DR2 (Gaia Collaboration et al. 2018) calibrator sources with which to produce refined world coordinate system (WCS) solutions for each W2 time slice (Section 8.3.1). Second, it provides sufficient numbers of bright point sources to accurately model the W2 point-spread function (PSF) within each time slice.

We create W2 meta-coadds by performing an inverse variance–weighted sum of the contributing e??? unWISE coadds, making use of the unWISE -invvar-m inverse variance maps. Using these same inverse variance weights, we also create a corresponding map of the mean MJD for each meta-coadd, enabling us to quote MJD values corresponding to our W2 astrometric detections.

Our modeling of the time-resolved unWISE coadds and meta-coadds, including source detection and centroiding, was performed using the crowdsource crowded field photometry pipeline (Schlafly et al. 2019). The crowdsource pipeline has proven adept at modeling unWISE W1 and W2 images during creation of the full-sky unWISE Catalog (Schlafly et al. 2019). It derives a PSF model for each unWISE image it processes and reports profile-fit astrometry that is equivalent to flux-weighted centroiding because the nominal PSF center is defined to coincide with the PSF model's flux-weighted centroid.

For the unWISE tile footprint containing each target, we ran crowdsource on all W2 time-resolved coadds (e??? time slices) and all meta-coadds (Table 5 time slices). Next, we proceeded to select the subset of these crowdsource detections that are counterparts to each brown dwarf candidate and will ultimately be combined with Spitzer astrometry during our final linear motion fits. We began by identifying a visually vetted set of prehibernation W2 crowdsource counterparts, one for each target. In combination with our Spitzer positions (Section 8.4), this allowed us to bracket each target's ∼2010–2019 trajectory and derive a crude linear motion estimate. Using this preliminary motion estimate, we then identified all crowdsource detections near the moving object's trajectory in all time slices. We visually inspected all such potential counterparts, removing severe blends and static contaminants.

The last step in selecting crowdsource detections for our final WISE+Spitzer motion fits is to pare down the full set of available detections for each target into a list that incorporates information from each WISE sky pass exactly once.²⁶ In this context, our faintest targets are the simplest. These objects will only have W2 detections in the deepest meta-coadds on each side of the WISE hibernation boundary: the "pre" and "post" time slices. Indeed, the simple combination of "pre" and "post" crowdsource astrometry was adopted for 104 of our 167 targets,²⁷ as can be seen in Table 3. For brighter targets, we have the freedom to choose the specific set of time slices adopted. For instance, bright targets will have e000, e001, and "pre" time slice detections available. We cannot use all of these in our joint WISE+Spitzer fits, since this would effectively double count W2 imaging during the 2010–2011 time period. In these situations, we carefully constructed lists of crowdsource detections for each target that omitted as few W2 sky passes as possible while never double counting. In doing so, we enforced a preference for shorter time slices, with the rationale being that longer time slices incur more smearing of the moving object, which is nonoptimal for centroid measurements. Table 3 lists the W2 time slices employed for each target in our production WISE+Spitzer linear motion fits. On average, ∼3 W2 detections per target are used.

8.3.1. W2 Astrometric Recalibration

The time-resolved unWISE coadds inherit low-level (up to a few hundred mas) astrometric systematics by virtue of propagating the single-exposure WISE astrometry without modification (Meisner et al. 2018b, 2019). We therefore sought to improve the accuracy of our crowdsource W2 astrometry by recalibrating it to Gaia DR2.

For the crowdsource catalog corresponding to each time slice of each unWISE tile footprint, we seek to compute a scalar astrometric offset along each sky direction so as to bring our W2 centroids into best agreement with Gaia DR2. In practice, this is accomplished by adding a small offset to each of the two CRPIX components in the native W2 unWISE coadd WCS.

Each coadd from which we draw a brown dwarf candidate W2 detection covers an ∼2.4 deg² sky area, resulting in an abundance of available Gaia DR2 calibrators. To assemble a set of Gaia–crowdsource calibration sources, we cross-match the full list of crowdsource detections against a subset of Gaia DR2 rows using a 2'' radius. We require that our Gaia DR2 astrometric calibrators have Gaia proper motions available, and we use these to propagate each calibrator's position to the mean epoch of the W2 coadd under consideration. We do not attempt to correct for the Gaia calibrator parallaxes, which have a mean amplitude of only ∼1–2 mas. The median number of Gaia DR2 calibrators employed per W2 coadd is 6880.

With Gaia positions at the relevant epoch in hand, we calculate the two-element shift that needs to be applied to the native CRPIX to zero out the median offsets between the Gaia calibrators and their crowdsource matches along the coadd x and y pixel directions. We then apply this offset to create a slightly modified CRPIX value that, in combination with the other native W2 WCS parameters, provides a recalibrated astrometric solution most consistent with Gaia at zeroth order.

The mean amplitude of the per-coordinate offsets applied to the native CRPIX values is ∼50 mas. Using the recalibrated WCS for each coadd, the typical bright end scatter (assessed with 10 < W2 < 11.6 unsaturated sources) relative to the Gaia "truth" is just 44 mas (41 mas) in R.A. (decl.). This is a very small fraction of the ∼65 W2 PSF FWHM (∼1/150 FWHM), providing confidence in the astrometric fidelity of our W2 (meta-)coadds.

The systematics floor of our W2 astrometry as characterized by the bright end scatter is very small compared to the typical per-coordinate statistical uncertainties on our W2 crowdsource centroids, which have a median value of 515 mas. The W2 centroid statistical uncertainties are large because of the broad W2 PSF and low S/N of our crowdsource W2 detections (median S/N = 9.8). Figure 5 provides a visual illustration of the large statistical noise inevitably present in our W2 centroids.

Zoom In Zoom Out Reset image size

Table 3 provides the positions and associated metadata of all W2 detections that feed into our final WISE+Spitzer linear motion fits. The σ_R.A. and σ_decl. values quoted each result from summing the statistical centroid uncertainty and bright end scatter systematics floor in quadrature (even though the former strongly dominates over the latter). Here σ_R.A. is in angular rather than coordinate units.

8.4. Spitzer Astrometry

Whereas our typical per-coordinate W2 centroid precision is larger than 500 mas, we should expect the ch2 per-coordinate centroid precision to be characteristically an order of magnitude smaller. This is simply due to the much higher S/N of our target detections in ch2 (median S/N ≈ 77) than in W2 (median S/N ≈ 10) and ∼3× narrower ch2 PSF. Because our centroid uncertainties are so much smaller in ch2 than in W2, the uncertainties on our eventual WISE+Spitzer linear motion measurements will be almost entirely dictated by the W2 positional uncertainties. Thus, we need not become fixated on obtaining Spitzer astrometry that achieves the ch2 imaging's theoretically optimal precision floor.

Kirkpatrick et al. (2019a) described a procedure for measuring Spitzer ch2 astrometry while achieving a systematics floor of just 10 mas coordinate^–1. This entailed a rather involved analysis that proceeded directly from the set of individual Spitzer BCD frames contributing to each AOR. On the other hand, Martin et al. (2018) demonstrated that a systematics floor of ∼20 mas coordinate^–1 could be achieved via a substantially more convenient procedure based on astrometric measurements performed on MOPEX mosaics. Because reducing our ch2 systematics floor from ∼20 to ∼10 mas would negligibly decrease the uncertainties on our WISE+Spitzer linear motions, and we already have MOPEX mosaics/extractions in hand for each target's field (Section 7), we opt to perform a mosaic-based Spitzer astrometric analysis.

Our approach for deriving recalibrated Spitzer ch2 astrometric measurements from the p14034 imaging is largely modeled after the procedure laid out in Martin et al. (2018). Our starting point is the list of ch2 MOPEX/APEX pixel coordinate centroids and (R.A., decl.) world coordinate values for all sources in each of our AORs. These are the same MOPEX/APEX catalogs used previously to obtain our ch2 magnitudes for each target in Section 7. The world coordinates natively provided by MOPEX/APEX rely on a WCS solution calibrated to the Two Micron All Sky Survey (2MASS) and typically show offsets of several tenths of a mosaic pixel (06 pixel^–1) relative to Gaia. The systematics inherent in the initial 2MASS WCS solutions are thus much larger than the systematics floor attainable through astrometric recalibration to Gaia.

8.4.1. ch2 Astrometric Recalibration

The primary challenge in recalibrating each ch2 mosaic's astrometry is identifying a sufficient number of Gaia DR2 calibration sources (N_calib) with high-S/N ch2 counterparts. In selecting astrometric calibrators, we always restrict to the "nonflanking" region of each p14034 ch2 mosaic (the portion of the mosaic built from single-frame BCD images that contain the brown dwarf candidate's location). Given our dither strategy, this results in a full frame coverage sky area from which to select Gaia calibrators of only 13–15 arcmin², depending on the number of dithers.

Ideally, we would be able to obtain at least 10 Gaia DR2 calibrators in each ch2 mosaic's nonflanking, full-coverage sky region with very high S/N ch2 counterparts (S/N ≥100) and Gaia proper motions available. However, this was possible for just 67 of our 164 ch2 mosaics (see the first row of Table 6). In cases where this ideal set of calibrator selection criteria (which we refer to as method 1) yielded fewer than 10 calibrators, we tried a sequence of somewhat loosened cuts in order to always obtain at least five Gaia calibrators.²⁸ We sequentially tried each set of selection criteria listed in Table 6, in order of ascending "method number" (leftmost column in Table 6), until a set of selection cuts yielded ${N}_{\mathrm{calib}}\geqslant {N}_{\mathrm{calib},\min }$. Method 2 is the same as our ideal set of cuts but reduces the ch2 counterpart S/N threshold to 50. Method 3 further reduces the ch2 S/N threshold to 30 and additionally reduces the minimum required number of Gaia calibrators from 10 to five (as specified in the ${N}_{\mathrm{calib},\min }$ column of Table 6).

Table 6.  Cuts Defining Spitzer–Gaia Astrometric Calibrator Samples

Method Number	Min. ch2 S/N	Min. BCD Cov. Frac.	Gaia PM Required	${N}_{\mathrm{calib},\min }$	Number of AORs
1	100	1.0	Yes	10	67
2	50	1.0	Yes	10	28
3	30	1.0	Yes	5	57
4	100	1.0	No	10	1
5	50	1.0	No	10	0
6	30	1.0	No	5	6
7	100	0.5	Yes	10	0
8	50	0.5	Yes	10	0
9	30	0.5	Yes	5	5

Download table as:  ASCIITypeset image

Methods 4, 5, and 6 are the same as methods 1, 2, and 3, respectively, but with the requirement of Gaia DR2 proper-motion availability dropped (see the Boolean "Gaia PM required" column of Table 6). Our inability to correct for calibrator motion between the Gaia 2015.5 epoch and our Spitzer ch2 epoch a few years later is regrettable but a necessary compromise when resorting to methods 4–6. The typical motions of our Gaia DR2 calibrators are very small, so this should have a negligible impact on our final WISE+Spitzer linear motion results, and only seven of 164 AORs end up using Gaia calibrators lacking proper motions.

Methods 7, 8, and 9 are the same as methods 1, 2, and 3, respectively, but reduce the minimum mosaic BCD frame coverage requirement from full coverage to at least 50% coverage. While this is not ideal, only five of 164 AORs needed to employ this lowered frame coverage requirement.

The "method number" column of Table 4 specifies which set of Gaia DR2 calibrator selection criteria was used in determining each row's recalibrated Spitzer ch2 position. The N_calib column of Table 4 lists the number of Gaia DR2 calibrators employed for each recalibrated ch2 position measurement. The minimum (maximum) number of astrometric calibrators per AOR is five (92). The median (mean) number of astrometric calibrators per AOR is 12 (16). In all cases where our selection criteria demand that Gaia DR2 calibrators have proper motions available, we use these proper motions to propagate the Gaia calibrator positions to the Spitzer ch2 epoch.

Having selected a set of Gaia DR2 calibrators for each AOR, we proceed to refit six parameters of each ch2 mosaic's WCS: all four elements of the CD matrix and the two CRPIX components. The bright end systematics floor achieved via our recalibrated ch2 mosaic WCS solutions is very similar to that of Martin et al. (2018). The median per-mosaic bright end scatter is 25 (23) mas in R.A. (decl.). The 16th–84th percentile ranges are 15–44 mas in R.A. and 14–39 mas in decl. For comparison, Martin et al. (2018) cited a typical systematics floor of ∼15–40 mas for their ch2 mosaic WCS recalibration.

Table 4 lists the recalibrated ch2 (R.A., decl.) position obtained for each target, including metadata such as the AOR used and the MJD. The uncertainties σ_R.A. and σ_decl. are computed by summing the per-AOR bright end scatter in quadrature with the statistical uncertainty on each target's ch2 centroid measurement.

8.5. WISE+Spitzer Linear Motion Fits

For each brown dwarf candidate, we gather its combined list of WISE and Spitzer positions, corresponding MJD values, and positional uncertainties from Tables 3 and 4. The typical combined WISE+Spitzer time baseline for targets with p14034 imaging available is ∼8.6 yr. Along each coordinate direction (R.A. and decl.), we fit a linear model to the combined list of WISE and Spitzer positions as a function of MJD in order to measure μ_α and μ_δ. Throughout this paper, the quoted μ_α values are in angular rather than coordinate units; i.e., they already have the cos(δ) factor multiplied into them. Through these same per-coordinate linear fits, we also obtain parameters α₀ and δ₀, the object's (R.A., decl.) coordinates at a fiducial time MJD₀. This is the inverse variance–weighted mean MJD of the contributing astrometric data points, and it typically ends up being similar to the MJD of the Spitzer observation, since the ch2 positional uncertainties are much smaller than those in W2. Our linear fits are performed using weighted linear least-squares and so naturally produce uncertainties on the fiducial location and best-fit linear motion components via simple matrix algebra. The measurement uncertainties fed to the weighted linear least-squares routine are the σ_R.A. and σ_decl. values provided in Tables 3 and 4. No rescaling of the σ_R.A., σ_decl. positional uncertainties is performed.

We do not allow for any outlier rejection in our per-coordinate linear motion fits. All detections used in our WISE+Spitzer motion fits were visually vetted, so there should be no need for outlier rejection.

For our targets that turned out to have two Spitzer counterparts (Section 7.2), we performed separate linear motion fits for each Spitzer counterpart, where the two motion fits both use the same set of WISE detections (since, in such cases, the brown dwarf candidate appears as just a single object in WISE).

Table 7 lists our linear motion fit results for all targets. The μ_α, ${\sigma }_{{\alpha }_{0}}$, and ${\sigma }_{{\mu }_{\alpha }}$ values are all in angular rather than coordinate units. The total motion μ_tot values are calculated by summing the R.A. and decl. linear motion components in quadrature, and the quoted μ_tot uncertainties are based on first-order propagation of the ${\sigma }_{{\mu }_{\alpha }}$, ${\sigma }_{{\mu }_{\delta }}$ errors.

Table 7.  WISE+Spitzer Linear Motion-fitting Results

Name	α0	δ0	${\sigma }_{{\alpha }_{0}}$	${\sigma }_{{\delta }_{0}}$	MJD0	${\mu }_{\alpha }$	μδ	μtot	${\chi }_{\mathrm{motion}}^{2}$	χ2	No. dof	${\chi }_{\nu }^{2}$
	(deg, ICRS)	(deg, ICRS)	(mas)	(mas)		(mas yr−1)	(mas yr−1)	(mas yr−1)
CWISEP J000006.01−704851.2	0.0238910	−70.8144313	42	36	58,656.07	−277 ± 40	−153 ± 38	317 ± 40	63.3	0.08	2	0.04
CWISEP J000110.81−093215.5	0.2954347	−9.5379786	26	27	58,552.36	345 ± 72	−194 ± 73	396 ± 72	30.2	3.34	2	1.67
CWISEP J001146.07−471306.8	2.9424406	−47.2186142	47	58	58,486.49	325 ± 63	−117 ± 63	346 ± 63	30.2	0.29	2	0.15
CWISEP J003507.77−153233.8	8.7825466	−15.5432421	25	32	58,394.44	20 ± 41	−553 ± 41	553 ± 41	181.8	14.45	8	1.81
CWISEP J003915.43+360939.0	9.8140272	36.1605309	27	23	58,577.21	−127 ± 78	−227 ± 79	260 ± 79	10.9	1.75	2	0.87
CWISEP J004158.35+381811.9	10.4937097	38.3033679	35	42	58,437.39	297 ± 51	−8 ± 52	297 ± 51	34.1	0.54	2	0.27
CWISEP J005802.63+723330.3	14.5100692	72.5585540	64	54	58,534.48	−474 ± 61	196 ± 59	513 ± 60	72.1	6.72	2	3.36
CWISEP J010247.48−654226.4	15.6969790	−65.7075044	62	41	58,453.38	−186 ± 85	−35 ± 82	189 ± 85	5.0	1.88	6	0.31
CWISEP J010527.69−783419.3	16.3685566	−78.5724495	48	23	58,280.05	333 ± 21	−180 ± 18	378 ± 21	340.3	22.84	2	11.42
CWISEP J010650.61+225159.1	16.7106177	22.8660649	22	35	58,385.82	−167 ± 34	−278 ± 35	324 ± 35	87.6	13.37	8	1.67
CWISEP J012735.44−564110.5	21.8976927	−56.6862172	27	57	58,455.16	4 ± 40	8 ± 40	9 ± 40	0.1	3.59	2	1.80
CWISEP J012748.35−631056.1	21.9518596	−63.1818606	47	42	58,445.67	660 ± 37	307 ± 35	728 ± 37	392.2	27.96	8	3.49
CWISEP J014607.55−375705.6	26.5324150	−37.9509194	21	21	58,541.43	510 ± 53	384 ± 54	638 ± 53	143.4	1.29	2	0.65
CWISEP J015613.24+325526.6	29.0568626	32.9235107	46	46	58,419.89	1138 ± 84	−373 ± 84	1197 ± 84	204.0	8.55	8	1.07
CWISEP J020103.10+293801.8	30.2636947	29.6333653	34	31	58,456.75	508 ± 67	−235 ± 69	560 ± 67	69.4	1.65	2	0.82
CWISEP J020938.72+180427.7	32.4119873	18.0741597	32	45	58,451.63	487 ± 61	−161 ± 63	513 ± 61	70.0	5.14	8	0.64
CWISEP J021243.55+053147.2	33.1813030	5.5299336	29	30	58,588.49	−102 ± 63	56 ± 65	116 ± 63	3.4	4.05	6	0.67
CWISEP J021921.66−265451.8	34.8407666	−26.9139810	74	73	58,517.37	353 ± 74	484 ± 76	599 ± 76	62.8	5.39	2	2.70
CWISEP J022122.41−564125.0	35.3440503	−56.6902923	24	21	58,454.17	243 ± 43	33 ± 42	245 ± 43	32.3	3.58	2	1.79
CWISEP J022513.27+154854.8	36.3056408	15.8152988	132	81	58,384.82	229 ± 54	85 ± 52	245 ± 54	20.7	4.24	8	0.53
CWISEP J022631.82−203439.4	36.6321750	−20.5783558	70	73	58,362.17	−325 ± 65	−538 ± 65	628 ± 65	92.7	1.99	6	0.33
CWISEP J022935.43+724616.4 N	37.3982422	72.7714650	78	72	55,436.54	−144 ± 28	−145 ± 27	204 ± 28	54.0	43.37	4	10.84
CWISEP J022935.43+724616.4 S	37.3968899	72.7711547	74	67	55,368.95	139 ± 28	75 ± 27	158 ± 28	32.7	13.11	4	3.28
CWISEP J023842.60−133210.7	39.6774081	−13.5375104	61	54	58,569.96	−158 ± 74	−742 ± 76	758 ± 76	100.6	3.11	6	0.52
CWISEP J024204.91−225604.6	40.5224841	−22.9344630	29	31	58,586.09	1049 ± 77	169 ± 79	1062 ± 77	192.6	0.10	2	0.05
CWISEP J024710.25−145809.9	41.7926505	−14.9694556	67	43	58,571.14	−62 ± 88	89 ± 90	108 ± 89	1.5	5.86	2	2.93
CWISEP J024810.75−694127.9	42.0450051	−69.6911310	50	34	58,522.07	7 ± 59	−31 ± 55	32 ± 56	0.3	2.03	2	1.01
CWISEP J025747.92−205602.7	44.4496962	−20.9337098	23	35	58,452.11	55 ± 49	229 ± 50	235 ± 50	21.9	4.74	2	2.37
CWISEP J031130.28+035931.8	47.8768496	3.9924660	61	64	58,450.57	595 ± 105	138 ± 111	611 ± 106	33.4	6.58	2	3.29
CWISEP J031557.05+203552.4	48.9878240	20.5979615	42	39	58,577.83	62 ± 46	82 ± 48	102 ± 48	4.6	4.88	2	2.44
CWISEP J031908.60+081120.4	49.7855761	8.1888656	75	73	58,434.99	−243 ± 102	205 ± 106	318 ± 104	9.4	11.45	2	5.73
CWISEP J031935.50−041231.7	49.8985275	−4.2088561	37	41	58,600.25	519 ± 92	−99 ± 94	528 ± 92	33.2	0.95	2	0.48
CWISEP J032109.59+693204.5	50.2936284	69.5342819	37	40	58,443.81	972 ± 61	−297 ± 63	1016 ± 61	275.2	11.62	8	1.45
CWISEP J034336.27+184025.8	55.9007363	18.6736946	48	49	58,599.52	−61 ± 71	25 ± 74	66 ± 71	0.9	21.63	2	10.82
CWISEP J034514.82+173528.1	56.3120041	17.5909526	28	28	58,606.63	152 ± 52	−148 ± 54	212 ± 53	16.1	0.00	2	0.00
CWISEP J034755.11+123051.9	56.9797369	12.5149653	31	26	58,435.00	20 ± 37	380 ± 38	380 ± 38	98.3	5.27	8	0.66
CWISEP J034904.05−462827.9	57.2673653	−46.4736606	32	35	58,452.20	236 ± 54	494 ± 54	548 ± 54	103.3	2.51	2	1.26
CWISEP J040106.67+085748.5	60.2782164	8.9629526	29	38	58,452.79	357 ± 61	−359 ± 64	507 ± 62	65.9	2.71	8	0.34
CWISEP J040235.55−265145.4	60.6494799	−26.8635386	60	53	58,440.51	755 ± 32	−571 ± 33	947 ± 33	832.5	16.93	6	2.82
CWISEP J040324.67+185729.6	60.8526375	18.9581692	56	41	58,452.93	−98 ± 81	−95 ± 84	137 ± 83	2.7	3.57	2	1.79
CWISEP J040351.00−491605.6	60.9630254	−49.2679072	38	41	58,426.72	256 ± 44	157 ± 45	300 ± 44	46.2	3.65	2	1.82
CWISEP J041025.10+033807.2	62.6046231	3.6352786	33	31	58,627.10	−97 ± 83	−72 ± 87	121 ± 84	2.1	2.39	2	1.19
CWISEP J042455.68+000221.4	66.2322875	0.0392211	43	31	58,589.11	153 ± 39	−70 ± 40	169 ± 39	18.8	2.01	8	0.25
CWISEP J042404.54+665011.2	66.0202303	66.8363662	34	48	58,422.26	395 ± 39	7 ± 41	395 ± 39	104.2	2.19	2	1.10
CWISEP J043034.27+255653.7	67.6433968	25.9481134	31	32	58,559.18	401 ± 27	−197 ± 29	447 ± 27	267.5	0.87	2	0.44
CWISEP J043309.31+100902.9	68.2890654	10.1502782	27	36	58,424.36	174 ± 34	−384 ± 36	422 ± 35	141.9	6.99	8	0.87
CWISEP J044330.73+693828.3	70.8783551	69.6410275	50	46	58,623.07	111 ± 76	−156 ± 79	191 ± 78	6.0	11.28	2	5.64
CWISEP J044719.61+202158.1	71.8317332	20.3651770	32	29	58,618.86	41 ± 49	−620 ± 52	621 ± 52	144.6	5.47	8	0.68
CWISEP J050521.29−591311.7	76.3403687	−59.2215888	33	33	58,448.04	461 ± 36	−1003 ± 35	1104 ± 36	964.6	7.88	2	3.94
CWISEP J052346.34−545314.7	80.9439898	−54.8868202	42	37	58,455.92	419 ± 49	518 ± 51	667 ± 50	176.7	1.98	2	0.99
CWISEP J053644.82−305539.3	84.1868335	−30.9275925	32	24	58,457.06	46 ± 45	5 ± 46	47 ± 45	1.1	1.93	2	0.97
CWISEP J054233.06+793459.1	85.6379950	79.5831471	36	29	58,540.27	6 ± 58	75 ± 61	75 ± 61	1.5	1.18	2	0.59
CWISEP J055816.68−450233.6	89.5694062	−45.0425330	81	44	58,352.04	−81 ± 39	69 ± 39	106 ± 39	7.4	2.02	8	0.25
CWISEP J060132.96−592227.3	90.3874560	−59.3743566	50	52	56,767.53	−209 ± 20	414 ± 21	464 ± 21	497.4	8.30	6	1.38
CWISEPR J062436.84−071147.2	96.1534148	−7.1962981	39	27	58,446.95	−90 ± 28	106 ± 29	139 ± 28	24.2	6.76	8	0.84
CWISEP J062742.27−215908.1	96.9262310	−21.9860615	30	32	58,501.69	25 ± 58	−376 ± 62	377 ± 62	36.8	0.99	2	0.49
CWISEP J063257.49+274629.4	98.2393490	27.7750194	41	40	58,480.14	14 ± 67	92 ± 71	93 ± 71	1.7	5.16	2	2.58
CWISEP J063428.10+504925.9	98.6173188	50.8225904	38	33	58,487.08	190 ± 61	−1057 ± 64	1074 ± 64	280.6	6.35	8	0.79
CWISEP J063845.48−615937.2	99.6888430	−61.9933801	70	46	58,446.38	−311 ± 67	286 ± 69	423 ± 68	38.8	0.71	2	0.35
CWISEPR J065144.62−115106.1	102.9358948	−11.8519416	27	27	58,511.18	−74 ± 41	−196 ± 43	210 ± 43	23.8	4.86	8	0.61
CWISEP J070055.19+783834.0	105.2368732	78.6417616	46	46	58,482.39	814 ± 96	−966 ± 103	1263 ± 100	160.8	0.81	2	0.41
CWISEP J070214.84−544041.7	105.5618914	−54.6790057	45	47	58,491.25	−83 ± 67	−558 ± 71	564 ± 71	62.8	0.21	2	0.10
CWISEP J071626.02−371951.1	109.1083915	−37.3307807	39	34	58,512.21	37 ± 54	78 ± 58	86 ± 57	2.3	0.20	2	0.10
CWISEP J071813.30−061421.1	109.5559281	−6.2391499	69	68	58,485.30	466 ± 90	141 ± 96	487 ± 91	28.7	0.02	2	0.01
CWISEP J082400.43+075019.9	126.0016447	7.8389405	75	72	58,512.95	−43 ± 123	105 ± 132	113 ± 131	0.8	3.66	4	0.92
CWISEP J084726.55+233558.1	131.8607674	23.5992073	37	36	58,516.78	43 ± 66	−218 ± 71	222 ± 71	9.7	0.02	2	0.01
CWISEP J085348.15+112921.5	133.4506930	11.4891927	33	46	58,500.74	42 ± 47	−131 ± 51	138 ± 51	7.4	4.84	2	2.42
CWISEP J085820.46+500834.4	134.5850031	50.1428221	54	33	58,501.60	−141 ± 63	−27 ± 67	143 ± 63	5.1	0.12	2	0.06
CWISEP J085908.26+152527.1	134.7844045	15.4238600	29	47	58,492.65	−81 ± 40	−263 ± 43	275 ± 43	41.6	2.52	2	1.26
CWISEP J085938.95+534908.7	134.9117592	53.8186842	35	44	58,511.53	−230 ± 53	−298 ± 56	377 ± 55	46.7	5.67	8	0.71
CWISEP J090547.50+700239.8	136.4486544	70.0445205	27	21	58,547.11	196 ± 50	−10 ± 53	197 ± 50	15.3	7.26	2	3.63
CWISEP J090536.35+740009.1	136.4033148	74.0008111	31	48	58,494.02	465 ± 67	−1490 ± 71	1561 ± 70	493.5	0.44	2	0.22
CWISEP J091558.51+254713.2	138.9938542	25.7865366	39	44	58,517.10	120 ± 56	−206 ± 59	238 ± 59	16.6	4.79	6	0.80
CWISEP J093111.03+232502.1	142.7956140	23.4165857	32	33	58,528.96	−295 ± 69	−474 ± 73	558 ± 72	60.0	0.24	2	0.12
CWISEP J093236.66−180029.3	143.1521859	−18.0083844	35	31	58,547.74	−431 ± 56	−182 ± 60	468 ± 57	67.3	0.17	2	0.08
CWISEP J093852.89+063440.6	144.7211265	6.5770243	46	50	58,511.93	453 ± 54	−731 ± 57	860 ± 56	236.1	1.81	8	0.23
CWISEP J094005.50+523359.2	145.0222817	52.5660222	46	24	58,522.22	−343 ± 50	−299 ± 53	455 ± 52	78.1	2.95	10	0.29
CWISEP J094615.56+351434.3	146.5645538	35.2420221	58	51	58,502.54	−229 ± 61	−634 ± 65	674 ± 65	108.9	1.06	2	0.53
CWISEP J094742.83+384619.3	146.9283136	38.7717593	90	50	58,501.73	−135 ± 72	−212 ± 75	251 ± 74	11.6	0.26	2	0.13
CWISEP J094930.41+663937.2	147.3766137	66.6600392	54	45	58,515.07	−5 ± 59	−298 ± 62	298 ± 62	23.5	1.98	2	0.99
CWISEP J094957.15−422017.1	147.4885119	−42.3380471	105	104	58,439.05	261 ± 71	229 ± 74	347 ± 73	22.9	1.16	2	0.58
CWISEP J095930.71−401046.8	149.8781840	−40.1803387	111	110	58,512.73	−131 ± 101	−681 ± 106	693 ± 106	43.2	2.06	4	0.51
CWISEP J100629.01+105408.5	151.6205056	10.9023902	35	30	58,547.59	−259 ± 47	12 ± 49	259 ± 47	30.8	4.05	8	0.51
CWISEP J100854.84+203136.6	152.2283025	20.5265504	37	58	58,515.53	−136 ± 51	−208 ± 55	248 ± 54	21.1	0.37	2	0.18
CWISEP J101841.86+513108.8	154.6745261	51.5193433	41	44	58,531.16	50 ± 68	265 ± 72	270 ± 72	14.1	6.07	2	3.03
CWISEP J102201.27+145520.2	155.5045106	14.9220742	33	41	58,538.06	−537 ± 61	−189 ± 65	569 ± 61	86.8	1.16	8	0.14
CWISEP J103453.14+161228.0	158.7211188	16.2076055	34	62	58,435.26	−235 ± 29	−143 ± 32	275 ± 30	87.2	0.25	2	0.12
CWISEP J103607.94−304253.1	159.0328176	−30.7148463	54	54	58,562.14	−196 ± 78	−116 ± 82	228 ± 79	8.3	5.52	4	1.38
CWISEP J104104.20+221613.6	160.2671291	22.2697391	47	42	58,585.15	−325 ± 111	−467 ± 118	569 ± 115	24.4	1.53	2	0.76
CWISEP J104446.56+001754.9	161.1928198	0.2984562	33	34	58,572.66	−852 ± 85	−160 ± 90	866 ± 85	103.1	3.08	4	0.77
CWISEP J104756.81+545741.6	161.9854882	54.9613395	38	38	58,525.41	−462 ± 72	−132 ± 76	480 ± 73	43.8	8.08	8	1.01
CWISEP J110021.08+094652.9	165.0868452	9.7811543	40	39	58,531.96	−626 ± 47	−40 ± 49	627 ± 47	179.4	1.16	2	0.58
CWISEP J111055.12−174738.2	167.7296340	−17.7944962	64	40	58,410.16	24 ± 28	−395 ± 27	396 ± 27	212.6	9.24	2	4.62
CWISEP J113010.21+313947.3	172.5411784	31.6615769	42	62	58,486.91	−1185 ± 35	−1476 ± 38	1893 ± 37	2684.6	11.72	18	0.65
CWISEP J120444.33−235926.8	181.1843852	−23.9914417	31	31	58,557.19	−263 ± 37	−473 ± 38	542 ± 38	208.2	0.18	2	0.09
CWISEP J121358.13+294237.0	183.4922016	29.7101348	40	39	58,551.73	−189 ± 64	9 ± 68	189 ± 64	8.6	14.08	8	1.76
CWISEP J122010.03+281431.3	185.0412133	28.2417212	32	32	58,576.01	−345 ± 50	−237 ± 52	419 ± 50	69.4	0.34	2	0.17
CWISEP J124138.41−820051.9	190.4111402	−82.0142553	50	45	58,106.00	230 ± 23	65 ± 23	239 ± 23	105.1	9.52	8	1.19
CWISEP J130255.54+191145.9	195.7315475	19.1957537	53	80	58,491.72	95 ± 44	−232 ± 47	251 ± 47	29.0	2.10	2	1.05
CWISEP J131252.97+341746.5	198.2210708	34.2959882	55	49	58,562.62	289 ± 74	−228 ± 78	368 ± 75	24.0	0.20	2	0.10
CWISEP J131208.16−105231.8	198.0338602	−10.8757384	41	52	58,593.28	−49 ± 80	−166 ± 84	173 ± 83	4.3	5.83	2	2.92
CWISEP J131221.97−310845.7	198.0907598	−31.1465777	31	31	58,602.89	−421 ± 48	−312 ± 51	524 ± 49	113.6	1.81	8	0.23
CWISEP J131350.91−440352.2	198.4612845	−44.0647872	38	32	58,592.27	−471 ± 47	−202 ± 50	513 ± 48	116.5	0.45	2	0.23
CWISEP J134143.75+574112.9	205.4323635	57.6866713	30	74	58,492.38	−149 ± 34	−22 ± 38	151 ± 34	19.6	1.75	2	0.87
CWISEP J135336.29−003756.6	208.4012529	−0.6322388	223	238	56,701.39	−653 ± 84	−448 ± 90	792 ± 86	84.6	3.53	6	0.59
CWISEP J135937.65−435226.9	209.9063082	−43.8744784	36	42	58,615.69	−296 ± 72	−205 ± 77	360 ± 74	23.9	1.48	4	0.37
CWISEP J140118.30+432554.2	210.3260378	43.4309089	42	30	58,353.31	−172 ± 33	−659 ± 34	682 ± 34	400.7	3.83	8	0.48
CWISEP J140247.83+102132.6	210.6991320	10.3591566	44	39	58,216.97	−140 ± 17	57 ± 17	151 ± 17	83.5	8.51	8	1.06
CWISEP J141206.85+234412.4	213.0275676	23.7369981	48	34	58,590.72	−625 ± 58	204 ± 61	657 ± 58	126.7	3.67	2	1.84
CWISEP J141400.68+163153.9	213.5031181	16.5321538	60	68	58,524.98	143 ± 51	181 ± 54	231 ± 53	19.2	29.84	2	14.92
CWISEP J142552.36+485151.3	216.4679680	48.8647060	58	60	58,545.20	−141 ± 69	272 ± 72	306 ± 71	18.5	9.42	2	4.71
CWISEP J143439.23−134421.4	218.6627039	−13.7397557	53	46	58,611.75	−473 ± 74	−152 ± 78	497 ± 75	44.4	0.72	2	0.36
CWISEP J144606.62−231717.8	221.5262966	−23.2894624	34	37	58,616.31	−860 ± 69	−954 ± 69	1285 ± 69	342.6	21.41	20	1.07
CWISEP J145837.91+173450.1	224.6573867	17.5806926	32	27	58,554.34	−469 ± 25	215 ± 26	516 ± 25	436.3	4.76	2	2.38
CWISEP J150252.82−304232.8	225.7194826	−30.7092702	25	33	58,380.33	−458 ± 32	−132 ± 34	477 ± 32	222.3	2.05	2	1.02
CWISEP J151140.51−835918.0	227.9188793	−83.9884181	43	40	58,469.21	−35 ± 69	−125 ± 73	130 ± 73	3.2	4.08	6	0.68
CWISEP J151521.22−215736.9	228.8385873	−21.9601747	165	175	56,794.73	255 ± 66	−429 ± 70	499 ± 69	52.3	6.33	6	1.05
WISEA J153429.75−104303.3	233.6209262	−10.7235637	27	29	58,634.28	−1254 ± 65	−2387 ± 69	2697 ± 68	1565.8	5.08	8	0.63
CWISEP J153859.39+482659.1	234.7478251	48.4490749	44	52	58,296.88	98 ± 32	−450 ± 34	461 ± 34	184.2	0.52	2	0.26
CWISEP J154151.59+523025.0 N	235.4656234	52.5068123	55	80	58,291.34	409 ± 28	−48 ± 31	412 ± 28	216.9	17.12	8	2.14
CWISEP J154151.59+523025.0 S	235.4645297	52.5062999	57	82	58,266.45	−14 ± 28	−374 ± 31	374 ± 31	148.2	54.35	8	6.79
CWISEP J160311.60−104620.4	240.7991406	−10.7735237	30	26	58,454.23	547 ± 52	−740 ± 55	920 ± 54	288.2	1.27	2	0.63
CWISEP J160835.01−244244.7	242.1462779	−24.7125120	33	26	58,406.40	290 ± 37	−71 ± 39	299 ± 37	64.6	1.52	8	0.19
CWISEP J161822.86−062310.2	244.5952072	−6.3863681	26	31	58,453.28	303 ± 51	−201 ± 54	364 ± 52	48.5	8.53	6	1.42
CWISEP J162225.92+370118.8	245.6073116	37.0213231	38	46	58,425.10	−361 ± 46	−473 ± 49	595 ± 48	154.2	5.38	8	0.67
CWISEP J163200.11+002108.6	248.0005765	0.3525295	47	46	58,454.71	−31 ± 90	252 ± 95	254 ± 95	7.1	4.97	2	2.49
CWISEP J165215.62+022918.5	253.0656696	2.4885946	58	57	58,442.79	406 ± 82	145 ± 87	432 ± 83	27.2	0.24	2	0.12
CWISEP J165359.67+214457.2	253.4987525	21.7488879	40	33	58,640.88	60 ± 69	−221 ± 72	229 ± 72	10.1	1.35	2	0.68
CWISEP J170918.83+000950.5	257.3283154	0.1643273	27	27	58,656.23	−102 ± 55	175 ± 58	202 ± 57	12.5	1.37	2	0.68
CWISEP J172104.42+595047.7	260.2683795	59.8466722	58	40	53,261.59	−10 ± 16	−32 ± 17	33 ± 17	4.0	1.43	2	0.72
CWISEP J175746.31+195112.6	269.4428220	19.8535267	33	32	58,459.19	39 ± 72	27 ± 76	47 ± 73	0.4	3.20	2	1.60
CWISEP J182358.73−740246.0	275.9965126	−74.0463976	55	52	58,383.86	371 ± 58	−309 ± 60	483 ± 59	67.3	0.69	2	0.34
CWISEP J185658.80+601351.4	284.2443326	60.2296897	30	35	58,474.24	−228 ± 28	−776 ± 27	809 ± 27	872.4	5.98	8	0.75
CWISEP J193518.58−154620.3	293.8277840	−15.7723504	29	25	58,482.28	305 ± 61	40 ± 64	308 ± 61	25.5	1.25	2	0.63
WISENF J193656.08+040801.2	294.2323616	4.1312231	31	20	58,507.54	−479 ± 36	−1155 ± 37	1250 ± 37	1130.2	⋯	0	⋯
CWISEP J194027.48−345650.6	295.1144888	−34.9475130	48	47	58,481.29	−19 ± 108	−168 ± 114	169 ± 114	2.2	5.31	2	2.66
CWISEP J194101.59+542335.9	295.2567040	54.3930038	30	33	58,348.47	8 ± 22	−214 ± 22	215 ± 22	92.0	3.58	8	0.45
CWISEP J194812.42−322334.9	297.0512077	−32.3929568	42	27	58,480.41	−368 ± 67	2 ± 70	368 ± 67	29.9	1.67	2	0.84
CWISEP J201146.45−481259.7	302.9439641	−48.2169094	42	33	58,439.13	134 ± 35	−324 ± 36	350 ± 36	94.6	2.37	4	0.59
CWISEP J201510.68−675005.6	303.7940891	−67.8350564	30	35	58,309.55	−62 ± 20	−194 ± 21	204 ± 21	91.7	1.73	2	0.86
CWISEP J203821.53−064930.9	309.5894424	−6.8257801	37	25	58,498.59	−201 ± 68	−401 ± 71	448 ± 70	40.6	0.97	2	0.49
CWISEP J205019.99−253652.8	312.5830506	−25.6149376	26	36	58,461.39	−182 ± 32	−276 ± 34	330 ± 34	96.5	4.49	8	0.56
CWISEP J205908.95+024105.6	314.7886251	2.6841705	30	30	58,527.32	658 ± 48	−513 ± 49	834 ± 49	294.0	4.55	2	2.28
CWISEP J210007.87−293139.8	315.0334875	−29.5278805	37	38	58,493.21	372 ± 71	−128 ± 74	394 ± 72	30.3	0.23	2	0.11
CWISEP J211909.29−192117.4	319.7890460	−19.3548643	31	26	58,509.55	160 ± 82	−50 ± 84	168 ± 82	4.2	1.98	2	0.99
CWISEP J212828.05+352912.4	322.1167327	35.4863939	74	74	58,377.15	−290 ± 83	−299 ± 84	416 ± 84	24.8	7.89	2	3.95
CWISEP J213249.05+690113.7	323.2051650	69.0207454	50	42	58,158.47	244 ± 24	159 ± 22	291 ± 23	157.2	1.56	2	0.78
CWISEP J213838.74−313808.5	324.6622036	−31.6361072	22	35	58,481.77	542 ± 34	−325 ± 35	632 ± 34	336.9	0.63	8	0.08
CWISEP J213930.45+042721.6	324.8770200	4.4553757	25	22	58,524.32	162 ± 67	−465 ± 69	493 ± 68	51.8	1.79	2	0.90
CWISEP J215841.50+732842.7	329.6733966	73.4785142	89	29	58,200.99	162 ± 23	−32 ± 18	165 ± 23	52.0	10.01	8	1.25
CWISEP J220452.02+063343.4	331.2163311	6.5618894	63	44	58,524.83	−299 ± 83	−223 ± 85	373 ± 84	19.8	6.60	2	3.30
CWISEP J221736.94−222647.6	334.4039704	−22.4465251	49	38	58,510.64	82 ± 89	121 ± 90	146 ± 90	2.7	4.50	2	2.25
CWISEP J222035.35−810322.6	335.1488953	−81.0565232	51	43	58,383.55	262 ± 50	−226 ± 49	346 ± 50	48.5	0.67	2	0.33
CWISEP J223022.60+254907.5	337.5932596	25.8180721	47	38	58,527.52	−535 ± 68	−491 ± 69	726 ± 68	113.4	5.55	4	1.39
CWISEP J223138.55−383057.2	337.9114974	−38.5164936	35	28	58,513.68	441 ± 72	−433 ± 74	618 ± 73	72.6	2.18	2	1.09
CWISEP J224747.42−004103.6	341.9474932	−0.6842782	57	64	56,336.39	34 ± 116	378 ± 117	380 ± 117	10.5	23.44	2	11.72
CWISEP J224916.17+371551.4	342.3170746	37.2637442	32	31	58,545.41	−203 ± 84	−361 ± 85	414 ± 84	24.0	0.31	2	0.16
CWISEP J225059.28−432057.2	342.7475126	−43.3497456	38	73	58,503.31	225 ± 75	−434 ± 77	489 ± 76	40.9	0.73	2	0.36
CWISEP J225156.13+392408.4	342.9833992	39.4022511	25	29	58,399.63	−310 ± 48	−113 ± 48	330 ± 48	48.0	0.79	2	0.40
CWISEP J225109.50−074037.7	342.7905094	−7.6769274	40	36	58,493.01	696 ± 41	163 ± 41	715 ± 41	305.7	10.48	8	1.31
CWISEP J225511.56−191516.3	343.7980417	−19.2551834	130	130	58,505.94	−166 ± 186	−314 ± 190	355 ± 189	3.5	20.39	2	10.19
CWISEP J225628.97+400227.3	344.1221247	40.0407842	29	29	58,535.53	696 ± 43	−165 ± 43	716 ± 43	278.4	5.26	8	0.66
CWISEP J230158.30−645858.3	345.4933541	−64.9835116	34	21	58,442.51	107 ± 30	−441 ± 29	453 ± 29	246.9	24.25	18	1.35
CWISEP J231047.80+362004.6	347.6996243	36.3347544	37	34	58,563.06	286 ± 86	37 ± 87	288 ± 86	11.2	12.91	2	6.45
CWISEP J231114.50+135148.5	347.8100199	13.8635280	32	48	58,536.94	−295 ± 63	17 ± 64	296 ± 63	22.1	1.97	2	0.98
CWISEP J233216.39−290025.0	353.0683272	−29.0073009	29	22	58,532.57	72 ± 65	−360 ± 66	367 ± 66	31.3	1.53	2	0.76
CWISEP J235130.42−185800.2	357.8775955	−18.9668268	40	28	58,394.73	492 ± 47	−46 ± 48	494 ± 47	111.2	0.17	2	0.09
CWISEP J235547.99+380438.9	358.9511591	38.0775517	38	41	58,395.29	666 ± 64	51 ± 65	668 ± 64	108.7	6.49	8	0.81
CWISEP J235644.78−481456.3	359.1881598	−48.2492296	30	30	58,403.37	1048 ± 91	−133 ± 90	1056 ± 91	133.4	0.49	4	0.12

Note. ${\chi }_{\nu }^{2}$ is the reduced χ² of the linear motion fit, defined as ${\chi }_{\nu }^{2}={\chi }^{2}/$(no. of dof).

A machine-readable version of the table is available.

Download table as:  DataTypeset images: 1 2 3 4

Figure 6 shows a histogram that vertically stacks the number of non-motion-confirmed targets (black) on top of the number of motion-confirmed targets (blue). Targets with μ_tot > 380 mas yr⁻¹ are always motion-confirmed. The motion-confirmed fraction decreases to 50% at μ_tot ≈ 265 mas yr⁻¹. The minimum best-fit μ_tot of any motion-confirmed member of our sample is 139 mas yr⁻¹. The minimum value of ${\mu }_{\mathrm{tot}}/{\sigma }_{{\mu }_{\mathrm{tot}}}$ for any motion-confirmed target is 4.85, which suggests that, roughly speaking, our ${\chi }_{\mathrm{motion}}^{2}=23.03$ threshold is akin to a requirement of 5σ significant motion. Of our discoveries, 114 are motion-confirmed via the astrometric analysis presented in this work, as indicated in the Boolean "motion-confirmed" column of Table 1.

Zoom In Zoom Out Reset image size