Follow-up Imaging of Disk Candidates from the Disk Detective Citizen Science Project: New Discoveries and False Positives in WISE Circumstellar Disk Surveys

With higher sensitivity than any previous full-sky infrared survey instrument, the Wide-field Infrared Survey Explorer (WISE; Wright et al. 2010) detected over 747 million sources in its all-sky survey. Many teams have searched for new circumstellar disks in the WISE data based on infrared excess at W4 (22 μm emission) compared to W1 (3.4 μm emission), discovering thousands of candidate debris disks (see, e.g., Kennedy & Wyatt 2013; Wu et al. 2013; Patel et al. 2014; Cotten & Song 2016; Patel et al. 2017) and young stellar object (YSO) disks (Liu et al. 2011, 2014; Rebull et al. 2011; Koenig & Leisawitz 2014; Koenig et al. 2015). In particular, because of its sensitivity and full-sky scope, the WISE mission is uniquely suited to the search for M-dwarf debris disks, which are of particular interest because of the relative lack of detected disks around these stars in comparison to higher-mass stars (e.g., Plavchan et al. 2005, 2009; Lestrade et al. 2009; Avenhaus et al. 2012; Morey & Lestrade 2014; Theissen & West 2014; Binks & Jeffries 2017).

However, confusion and contamination limit every search for disks with WISE. The point-spread function (PSF) at W4 has an FWHM of 12''. This wide PSF can allow emission from multiple point sources (e.g., background stars) or image artifacts to contribute to the W4 photometry, producing a false-positive [W1]–[W4] excess. Additionally, the color loci of debris disks and YSOs overlap the color loci of several other astronomical phenomena, including background galaxies and stars embedded in nebulosity (Koenig et al. 2012). To eliminate these false positives, objects must be examined in visible and near-infrared images along with the WISE images. Most published searches have utilized visual inspection of the WISE images (see, e.g., Debes et al. 2011; Kennedy & Wyatt 2012; Wu et al. 2013; Cruz-Saenz de Miera et al. 2014; Patel et al. 2014) to address these contamination and confusion problems.

Kennedy & Wyatt (2012) produced a study of these confusion and contamination issues, focusing on infrared excesses around stars observed by Kepler (Borucki et al. 2010), with the goal of expanding the number of known stars that host both planets and debris disks. They searched for infrared excesses in a cross-match of the Kepler Input Catalog (KIC) to 2MASS and WISE, finding 7965 disk candidates. However, they argued that all but 271 (3.4%) of these objects were coincident with Galactic dust emission, as identified by the IRAS 100 μm background, and therefore were false positives.

The Disk Detective citizen science/crowd-sourcing project (Kuchner et al. 2016, hereafter Paper I) uses citizen science to examine infrared excess candidates from WISE, beginning with a website, http://www.diskdetective.org, where volunteers examine images from WISE, the Two-Micron All-Sky Survey (2MASS), the Digitized Sky Survey (DSS), and the Sloan Digital Sky Survey (SDSS) to check for false positives. Since launch in 2014 January, over 30,000 users have made over 2.6 million classifications via this Zooniverse website. Shortly after project launch, a group of highly dedicated volunteers began their own email discussion group for the project. Since then, members of this "advanced user group" have helped train other users and research follow-up targets in the literature, and they now form a crucial extension of the Disk Detective science team (Paper I).

In this paper, we address two specific forms of false positive that occur in the Disk Detective input catalog and other searches for circumstellar disks with WISE:

• 1.  
  Confusion in the WISE images: the contribution to noise in an image due to superposed signals from faint unresolved sources that cluster on the scale of the observing beam.

  • (a)  
    We use data from the DSS, the SDSS, and the 2MASS to search for background sources via the Disk Detective website.
  • (b)  
    We use high-resolution imaging on small telescopes (the 1.5 m Telescope at Mt. Palomar and the du Pont Telescope at Las Campanas) to identify background sources that are too faint for SDSS and 2MASS or unresolved by these surveys.
• 2.  
  Contamination of the WISE images: the presence of image artifacts (e.g., diffraction spikes, latent images, or optical ghosts) within the WISE beam. We search for these contaminants by examining the WISE images via the Disk Detective website.

We present the first results from our follow-up imaging campaigns with the Robo-AO adaptive optics instrument on the 1.5 m telescope at Palomar Observatory (Baranec et al. 2014) and the RetroCam instrument on the du Pont Telescope at Las Campanas Observatory in Chile (e.g., Rheault et al. 2014). These instruments provided an angular resolution of 015 and <1'', respectively, improving on the 5'' effective resolution of the 2MASS Point Source Catalog.¹¹ We combine these observations with the results of the website-based evaluation to estimate the fraction of WISE excesses that are true disk candidates. In Section 2, we review our website classification procedure, analyze the distribution of clean sources and false positives, and consider the results of an advanced-user-driven literature review. In Section 3, we review our procedure for selecting these particular targets for further examination, and we discuss our methodology for collecting high-resolution images of these targets. In Section 4, we describe our method for identification of contaminated targets. In Section 5, we present the results of our follow-up observations, estimate the presence of further unresolved sources, and combine these with the results from website analysis to estimate the ultimate yield of the Disk Detective input catalog. We apply these findings to other surveys in Section 6, including a reanalysis of the M-dwarf disk search of Theissen & West (2014). In Section 7, we present our list of uncontaminated disk candidates. Finally, in Section 8 we summarize our results and discuss future plans for these targets and high-resolution follow-up imaging.

The online engine of our citizen science disk search is DiskDetective.org, where users view sets of images showing the same WISE point source in several bands. In this section, we review our online classification method (discussed in more depth in Paper I) and analyze the latest online classification results.

2.1. DiskDetective.org: Identification of WISE Debris Disk Candidates with Citizen Science

2.2. False-positive Rates in Website Classifications

Figure 1 shows the distribution of raw numbers of complete subjects broken down by category ("good," "multiple," or "other") as a function of Galactic latitude and the distribution of "good" objects alone as a function of Galactic latitude. Most of the complete subjects are false positives; only 9.80% (±0.08%) of the complete subjects are labeled "good" by a majority of volunteers. "Multiples" are the dominant form of false positive: they make up 68.87% (±0.12%) of all complete subjects.

Zoom In Zoom Out Reset image size

Figure 2 shows the rate of multiples as a function of Galactic latitude. "Multiples" dominate especially in the Galactic plane; 74.84% (±0.16%) of subjects in the −5° < b < 5° range were classified as multiples, while only 4.74% (±0.30%) of subjects in that range are good, with over 100,000 complete subjects in this range. We also observe a large spike in the "multiple" rate in Galactic latitudes −35° < b < −30°. Figure 3 shows the density of "multiples" (i.e., the fraction of subjects that are multiples) as a function of Galactic latitude and longitude (redder indicates a larger fraction); this plot reveals that the spike in the −35° < b < −30° range in Figure 2 is associated with the LMC. Figure 3 also shows a higher density of "multiples" in the SMC than in its surroundings. Our statistics are poor in the 60° < b < 90° range because there are fewer complete subjects in this region of sky, which only recently became active on the website.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

2.3. Literature Review by Citizen Scientists

We have performed follow-up imaging of a subset of our DDOIs using the Robo-AO on the 1.5 m telescope at Mount Palomar and the RetroCam instrument on the Irenée R. du Pont Telescope at Las Campanas Observatory. Here we describe our target selection, observations, and reduction methodology for each telescope.

3.1. Robo-AO Observations

3.2. du Pont/RetroCam

A group of 10 citizen scientists examined the Robo-AO data with the SAOImage DS9 software package to visually examine the data to identify images with faint background objects, providing them with a set of images analyzed by the science team as a training set. We developed the following method for qualitative analysis:

• 1.  
  Display the image with 100% frame selection as black/white gray scale, min/max, and linear scaling to identify the target star.
• 2.  
  Shift to "zscale" to look for fainter objects. This allows fainter background objects to emerge more clearly than they would otherwise.
• 3.  
  Identify any faint background objects in the field of view, noting their positions.
• 4.  
  Note which objects have background objects within 12 arcsec of the target star.

Each volunteer independently analyzed a subset of the images using the above method. The group then discussed each image together to reach a consensus on each target. We followed a similar procedure with the du Pont data.

This visual inspection identified targets with evidence of background sources that could produce a false-positive excess. We then quantified whether these background sources significantly affected the excess at W4. To estimate the contribution each background object made at W4, we determined magnitudes for the background objects with aperture photometry using the IRAF DAOphot package. We assumed that background objects exhibited either an M-dwarf spectral energy distribution (SED) or a power-law SED with spectral index zero (corresponding to a YSO or heavily reddened early-type star) and determined the colors of these objects (calculated in Appendix B and presented in Table 8). Using the recorded photometry and these colors, we estimated the flux of the background object at W4 and subtracted this flux from the total W4 flux to yield the intrinsic flux of the target itself (including any contribution from circumstellar material). We then recalculated the target's [W1–W4] color using this corrected W4 flux to determine if a significant excess remained. The results are described in Section 5.

In this section, we determine false-positive rates based on the follow-up observations and unresolvable IR-bright background objects. We then combine these with the classification and literature review data to determine the overall false-positive rate for Disk Detective thus far, from which we estimate the eventual final yield of disks from AllWISE.

5.1. False-positive Rates from High-resolution Follow-up

The results of quantitative analysis of the 261 targets with reliable photometry observed with Palomar/Robo-AO and du Pont/Retrocam are presented in Figure 4. We combine the two samples without adjustment because (as described in Appendix B) both instruments are similarly sensitive to the same background objects. Overall, 244 of these 261 targets retain a significant infrared excess after the contribution of background objects has been removed, for a false-positive rate of 7% ± 1%. There is no detectable significant difference in contamination rate in the Galactic plane, due to the relatively small numbers involved (compared to the overall Disk Detective input catalog). Of the 39 objects in the −5° < b < 5° range, three are contaminated, leaving a false-positive rate of 8% (±4%). Out of the plane, 14 of 222 targets are contaminated, leaving a false-positive rate of 6% (±2%).

Zoom In Zoom Out Reset image size

We identified 16 objects as having insufficient excess at W4 once the estimated contributions from background objects were removed. We list these objects in Table 3, as well as any previous identifications as excess targets.

Table 3.  Targets with False-positive Excesses Due to Background Objects in High-resolution Follow-up Observations

Identifiers		Previous
Zooniverse	WISEA	citations
AWI0005w52	J000308.37+424452.4
AWI0005yiz	J010722.60+380143.9	1
AWI0002vbd	J020206.67+601741.4
AWI0003cs7	J021532.17+591424.4
AWI0004nfu	J032853.67+490412.8	1
AWI0000uj2	J153046.05+342756.4	1, 2, 3
AWI00006nb	J161808.08+104551.4
AWI0005bk2	J172912.43+005605.7
AWI0005brq	J181949.03+310841.7
AWI0005bud	J184141.31+313703.4
AWI0005lx9	J193040.05+350609.9
AWI00055by	J204443.79+425654.9	4
AWI0005vyx	J212959.78+413037.3
AWI0005a9r	J215305.45+682955.0	1
AWI0005w1h	J220503.97+444543.7
AWI0000kk6	J220601.14-020343.2

References. (1) Paper I. (2) Wu et al. (2013). (3) Cotten & Song (2016). (4) Clarke et al. (2005).

Download table as:  ASCIITypeset image

5.2. False-positive Excess from Unresolvable Infrared Galaxies

5.3. Expected Total Number of Disks in AllWISE

Given its large input catalog compared to other surveys, Disk Detective is well positioned to not only identify new warm debris disks in the WISE catalog, but also to inspect and reevaluate disks identified by other researchers. In this section, we estimate the numbers of published disks from other searches that are likely false positives, and we present a visual reinspection of the M-dwarf disk sample presented by Theissen & West (2014).

6.1. False Positives in Previous WISE Disk Searches

Because our sample encompasses the full 2MASS/WISE cross-match, we can apply our false-positive rates to other searches for warm debris disks with WISE. While we cannot assess individual objects in other surveys, due to limited overlap with our follow-up, we estimate the number of disks in each survey likely to be a false positive based on the rates we determined, depicted in Figure 5.

• 1.  
  Patel et al. (2014). This survey specifically avoided the Galactic plane in its cross-match with the Hipparcos catalog (van Leeuwen 2007), so we similarly apply an out-of-plane contamination rate. We will assume that the visual inspection described by the authors is comparable to the inspection and literature review by the Disk Detective team and will thus only apply the rate of contamination from follow-up imaging. Of the 113 new warm disks that were detected in this survey (Patel et al. 2015), we expect about seven to have previously undetected background objects that would contribute to a false-positive excess detection.
• 2.  
  Wu et al. (2013). This survey did not specifically avoid the Galactic plane as a selection criterion in its cross-match with Hipparcos, so we apply the statistics for the full sample to this paper's sample of 75 previously unidentified main-sequence stars with infrared excess indicative of a disk. We would therefore expect about five of these targets to be contaminated by one or more otherwise-undetected background objects.
• 3.  
  Cotten & Song (2016). This search identified ∼1750 debris disk candidates from a thorough review of pre-WISE disk literature and a cross-match of AllWISE with the Tycho-2 (Høg et al. 2000) survey, of which 1025 are new excess detections. Applying our recovery rate for high-resolution follow-up to this sample, we would expect that ∼64 targets would be contaminated by background objects only recoverable in high-resolution images that could produce a false-positive excess detection.
• 4.  
  McDonald et al. (2012, 2017). These searches cross-matched photometry data from Hipparcos, Tycho, and various other catalogs, using astrometry from, respectively, Tycho and Gaia. These works then used SED fitting of the cross-matched data to estimate stellar and excess parameters for >100,000 objects. However, neither search considered the source of the excess (e.g., self-produced dust around AGB stars rather than debris disks), and neither search included a robust visual inspection of the actual images (which eliminates ≈90% of excess candidates in the Disk Detective search), relying instead on visual inspection of the fit SEDs. Because these searches select for objects with star-like SEDs rather than preselecting sources with infrared excess the way our search does, we cannot directly apply our overall false-positive rates to their search. However, since we estimate there to be ∼21,600 disk candidates in AllWISE, we expect that >79,000 of the excesses identified by McDonald et al. (2012, 2017) are false positives.
• 5.  
  Marton et al. (2016). This search used support vector machines (SVMs), a class of supervised learning algorithm, to identify 133,980 YSO candidates in AllWISE, by identifying objects that were a given type of false positive in the data via SIMBAD and using these to train the SVM to avoid such objects. However, there was no visual inspection of the images, sources were not preselected for excess, the W4 PSF was not taken into account, and there is no acknowledgment of multiples, which make up the bulk of our false positives. If we assume that the SVM algorithm identifies nonmultiple false positives as effective as Disk Detective, we also assume that literature review would eliminate the SIMBAD-identified non-YSOs in their final sample (as per Table 1 of that paper), and we further assume that a similar number of objects would be false positives in follow-up imaging, this survey would have a lower limit false positive rate of 74%. While a search through WISE using machine-learning algorithms could prove valuable, the number of false positives that are only identified by visual inspection of images suggests that any such system would need to take the images themselves into account, rather than strictly learning based on photometric points and data quality flags.
• 6.  
  Kuchner et al. (2016). As Paper I was published before analysis of Robo-AO and du Pont photometry was complete, we also apply our rate analysis to this paper. Of the 37 new disk candidates presented in that paper, we expect that about two would be contaminated.

Zoom In Zoom Out Reset image size

The targets WISEA J153046.05+342756.4 (AWI0000uj2), WISEA J010722.60+380143.9 (AWI0005yiz), WISEA J032853.67+490412.8 (AWI0004nfu), and WISEA J215305.45+682955.0 (AWI0005a9r) present salient examples of the importance of imaging follow-up. AWI0000uj2, an A0 star, appears in Wu et al. (2013), in Cotten & Song (2016) as a "Reserved" excess candidate, and in Paper I. AWI0004nfu appears in Paper I and Zuckerman et al. (2012). AWI0005yiz and AWI0005a9r were both detected for the first time in Paper I.

Based on our follow-up observations with Robo-AO, all four of these objects have no excess once the contribution from background objects is subtracted at W4. AWI0000uj2 is likely significantly contaminated at W4 by two objects ∼4 mag fainter than it in the Sloan i band. AWI0004nfu exhibits seven background objects 3–9 mag fainter than it. AWI0005yiz exhibits several background objects ∼5–8 mag fainter than it, and AWI0005a9r exhibits several background objects 3–10 mag fainter than it in Sloan i. Further analysis and observations (e.g., additional wavelength coverage of the background objects to confirm the shape of their SED) are required to confirm that the observed excess is significantly affected by these targets. However, these cases illustrate that high-resolution follow-up can impact the quality of published infrared excesses.

6.2. A Reassessment of a Previous WISE-based M-dwarf Disk Search

Based on our follow-up observations with Palomar/Robo-AO and du Pont/Retrocam, we find that 244 of our observed targets, including 214 sources first identified by Disk Detective, have no significantly contaminating background objects within the 12'' radius of the W4 PSF, giving us the confidence to publish them as disk candidates. The candidates are listed in Table 4; previous surveys that have also identified these targets are listed in the Notes column. We briefly summarize characteristics of some objects of particular interest in Appendix C.

We used available photometry from SIMBAD, 2MASS, and WISE (including corrected W1, W2, and W4 photometry) to fit the SEDs of these systems. We initially fit the stellar component of the system with a blackbody to estimate the stellar temperature and ratio of stellar radius to distance. In the case where a blackbody fit yields a temperature less than 7000 K, we instead fit with a stellar model from the BT-Settl CIFIST package (Baraffe et al. 2015), also fitting for $\mathrm{log}(g)$. We initially fit the three bluest photometry points, then refit including the next bluest point if it is not in significant excess, repeating the last step iteratively until the next point is either W3 or in excess. We fit the remaining excess using a single-temperature blackbody to determine the dust temperature. We then find the best-fit parameters using the emcee package (Foreman-Mackey et al. 2013), using the previous parameter estimates as an initial guess. These SEDs, shown in Figure 6, provide a useful first-order estimate of these fundamental disk parameters. We list these parameters for likely debris disks in Table 5 and for likely YSO disks in Table 6. Table 5 consists of all candidates with only one point of excess (W4), as well as candidates with two points of excess and L_ir/L_⋆ < 10⁻³. Table 6 consists of all other objects. For objects with more than two points of excess, we also provide the spectral index α of a power-law fit to the object's WISE data as an estimate of YSO class.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

View figure set (245 panels)

Tarball of figure set images

Table 5.  Derived Parameters of Debris Disk Candidates

Zooniverse	WISEA	${T}_{\mathrm{eff}}$ (K)	Tdisk (K)	Lir/Lstar
AWI00062h7	J002133.47-661816.6	${11697}_{-113}^{+112}$	${216}_{-24}^{+27}$	(2.8 ± 0.4) × 10−5
AWI0000bs0	J003140.76-014737.3	${10319}_{-69}^{+64}$	202 ± 11	(8.6 ± 0.6) × 10−5
AWI0005mry	J003412.66+540359.0	9680 ± 72	>35	<0.12
AWI0005yiv	J005926.26+400918.2	${9165}_{-52}^{+58}$	>46	<0.005
AWI00055sx	J010652.55+743754.5	${10106}_{-85}^{+80}$	${182}_{-15}^{+16}$	(8.1 ± 1.1) × 10−5
AWI00055sz	J011636.23+740136.6	${9011}_{-89}^{+99}$	${685}_{-113}^{+125}$	$({4.5}_{-1.3}^{+2.0})\times {10}^{-4}$

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

Download table as:  DataTypeset image

Table 6.  Derived Parameters of YSO Disk Candidates

Zooniverse	WISEA	Teff (K)	αdisk	Tdisk (K)	Lir/Lstar
AWI0000nfp	J002155.14-672715.9	4700 ± 100	⋯	509 ± 21	(7.7 ± 0.05) × 10−3
AWI0000jvv	J003507.14+070625.0	3500 ± 100	⋯	${368}_{-11}^{+12}$	$({1.23}_{-0.04}^{+0.05})\times {10}^{-2}$
AWI00062m4	J004826.42+020753.0	4600 ± 100	⋯	${207}_{-4}^{+5}$	$({3.7}_{-0.1}^{+0.1})\times {10}^{-3}$
AWI00062mq	J011743.47-523330.8	6400 ± 100	−0.12 ± 0.01	436 ± 4	(9.0 ± 0.2) × 10−2
AWI0005aeg	J013833.77+780834.3	6700 ± 100	−2.11 ± 0.02	${1210}_{-25}^{+26}$	$({2.5}_{-0.1}^{+0.2})\times {10}^{-2}$

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

Download table as:  DataTypeset image

Of the 244 targets presented here, 223 have parallax measurements from Hipparcos or the second data release from Gaia, listed in Table 4. The parallax measurements indicate that 30 of these systems lie within 125 pc, making them prime candidates for follow-up observation. We list these candidates in Table 7.

Table 7.  Disk Candidates within 125 pc

Zooniverse	WISEA	Distance (pc)	V Magnitude	Teff (K)	Tdisk (K)	Lir/Lstar
AWI00062lo	J003140.76-014737.3	109 ± 6	7.07 ± 0.01	${10311}_{-62}^{+65}$	${202}_{-10}^{+11}$	$8.5{(}_{-0.58}^{+0.54})\times {10}^{-5}$
AWI0005yiv	J005926.26+400918.2	113 ± 4	7.532 ± 0.01	9169 ± 57	${59}_{-17}^{+29}$	$7.4{(}_{-6.4}^{+140})\times {10}^{-4}$
AWI000425z	J024755.37+553648.4	99 ± 3	6.918 ± 0.01	${9490}_{-67}^{+66}$	${131}_{-43}^{+38}$	$4.2{(}_{-0.82}^{+4.7})\times {10}^{-5}$
AWI0000phh	J025614.05+040254.2	125 ± 21	7.706 ± 0.014	${11304}_{-120}^{+122}$	${215}_{-14}^{+16}$	$6.4{(}_{-0.62}^{+0.60})\times {10}^{-5}$
AWI0005yk3	J030651.95+303136.8	73 ± 3	7.061 ± 0.011	${7640}_{-44}^{+46}$	243 ± 5	$7.5(\pm 0.20)\times {10}^{-4}$
AWI0005ykd	J032448.99+283908.6	106 ± 6	7.121 ± 0.01	${9537}_{-60}^{+71}$	118 ± 38	$5.5{(}_{-1.4}^{+11})\times {10}^{-5}$
AWI0005zy4	J032504.59+105835.0	116 ± 5	7.294 ± 0.011	${9493}_{-71}^{+69}$	${182}_{-51}^{+37}$	$3.4{(}_{-0.70}^{+0.85})\times {10}^{-5}$
AWI00062iw	J032555.87-355515.1	100 ± 6	6.385 ± 0.009	${10160}_{-81}^{+86}$	${216}_{-21}^{+22}$	$4.1{(}_{-0.39}^{+0.42})\times {10}^{-5}$
AWI0005ym9	J035157.43+255955.4	121 ± 9	7.68 ± 0.01	${9135}_{-48}^{+49}$	${158}_{-63}^{+44}$	$3.9{(}_{-1.1}^{+2.9})\times {10}^{-5}$
AWI0005ymc	J040040.65+202447.8	119 ± 5	8.05 ± 0.01	8410 ± 46	${131}_{-6}^{+5}$	$5.7{(}_{-0.35}^{+0.42})\times {10}^{-4}$
AWI0005zz5	J040238.47-004803.7	123 ± 8	6.93 ± 0.1	${11167}_{-80}^{+81}$	${176}_{-24}^{+20}$	$3.4{(}_{-0.43}^{+0.47})\times {10}^{-5}$
AWI0005ymi	J041249.03+193219.2	115 ± 5	7.783 ± 0.014	6100 ± 100	${175}_{-15}^{+14}$	$2.3{(}_{-0.19}^{+0.21})\times {10}^{-4}$
AWI0005wcl	J045519.57+163712.9	109 ± 8	7.17 ± 0.01	${10127}_{-80}^{+81}$	${214}_{-18}^{+21}$	$5.1(\pm 0.58)\times {10}^{-5}$
AWI0005d88	J083100.44+185806.0	91 ± 4	7.418 ± 0.012	${8319}_{-51}^{+47}$	${45}_{-10}^{+20}$	$1.0{(}_{-0.96}^{+20})\times {10}^{-2}$
AWI0000y1k	J111714.49-594610.8	76 ± 1	7.15 ± 0.01	${8048}_{-51}^{+52}$	${195}_{-16}^{+17}$	$1.3{(}_{-0.096}^{+0.099})\times {10}^{-4}$
AWI0005da8	J112256.98-203731.7	114 ± 5	7.466 ± 0.011	${8835}_{-59}^{+64}$	${132}_{-48}^{+42}$	$5.6{(}_{-1.3}^{+8.7})\times {10}^{-5}$
AWI00056ck	J121456.32-475654.6	117 ± 5	8.097 ± 0.011	${7889}_{-48}^{+47}$	${150}_{-8}^{+7}$	$3.3{(}_{-0.18}^{+0.20})\times {10}^{-4}$
AWI00056i5	J132026.77-491325.4	116 ± 4	7.948 ± 0.012	${8213}_{-55}^{+57}$	${172}_{-14}^{+13}$	$1.3{(}_{-0.093}^{+0.095})\times {10}^{-4}$
AWI0000uji	J151147.67+101259.8	117 ± 6	6.875 ± 0.012	${10812}_{-99}^{+100}$	${278}_{-29}^{+38}$	$3.7{(}_{-0.49}^{+0.50})\times {10}^{-5}$
AWI0000v1z	J152954.11+234901.6	114 ± 4	7.585 ± 0.011	${8576}_{-51}^{+49}$	${47}_{-13}^{+35}$	$3.1{(}_{-3.0}^{+160})\times {10}^{-3}$
AWI00002yt	J172007.53+354103.6	101 ± 2	8.272 ± 0.012	6800 ± 100	196 ± 30	$9.0(\pm 1.5)\times {10}^{-5}$
AWI0005igi	J172452.23-185133.5	98 ± 3	8.5 ± 0.015	6200 ± 100	${91}_{-17}^{+13}$	$2.3{(}_{-0.71}^{+3.0})\times {10}^{-3}$
AWI0005d5p	J180230.72+583738.4	89 ± 4	6.858 ± 0.01	9659 ± 75	${152}_{-45}^{+35}$	$3.3{(}_{-0.44}^{+1.3})\times {10}^{-5}$
AWI000621a	J205241.67-531624.8	63 ± 1	7.663 ± 0.011	6300 ± 100	${47}_{-10}^{+23}$	$1.3{(}_{-1.2}^{+21})\times {10}^{-2}$
AWI0006222	J210916.04-001405.6	103 ± 9	6.659 ± 0.01	${9118}_{-55}^{+60}$	${171}_{-32}^{+24}$	$5.4{(}_{-0.51}^{+0.65})\times {10}^{-5}$
AWI00019i2	J221055.01+575629.4	112 ± 4	7.42 ± 0.01	${8824}_{-54}^{+52}$	${261}_{-20}^{+21}$	$8.4{(}_{-0.79}^{+0.82})\times {10}^{-5}$
AWI0000gjb	J224206.62-032824.4	112 ± 12	7.159 ± 0.011	${11159}_{-98}^{+100}$	${180}_{-21}^{+19}$	$3.8{(}_{-0.40}^{+0.42})\times {10}^{-5}$
AWI00062gs	J230533.05+145732.5	124 ± 12	6.762 ± 0.01	11596 ± 102	${162}_{-24}^{+19}$	$3.3{(}_{-0.40}^{+0.48})\times {10}^{-5}$
AWI00062l3	J235537.71+081323.7	113 ± 8	6.818 ± 0.01	${11396}_{-105}^{+109}$	${269}_{-29}^{+37}$	$3.2{(}_{-0.40}^{+0.41})\times {10}^{-5}$
AWI00062h1	J235746.21+112827.6	102 ± 6	6.644 ± 0.01	${11092}_{-96}^{+92}$	205 ± 27	$2.6{(}_{-0.35}^{+0.36})\times {10}^{-5}$

Download table as:  ASCIITypeset image

We can also use these parallaxes and 2MASS photometry to construct a Hertzsprung–Russell (HR) diagram for our disk candidates (shown in Figure 7). On this figure, point color corresponds to disk temperature, while point size corresponds to L_ir/L_⋆. A gray curve shows the zero-age main sequence (Pecaut & Mamajek 2013). The point in gray in the lower left indicates typical uncertainties, ∼0.04 mag in $(J-{K}_{s})$ color and ∼0.15 mag in ${M}_{{{\rm{K}}}_{s}}$. Although some stars have K-band excesses that shift their points to the right in this diagram, Figure 7 reveals that none of the host stars with parallaxes has a main-sequence type later than roughly K7. Even with our "drain-the-lake" approach to searching for debris disks, the M-dwarf disks remain hidden, at least in this sample.

Zoom In Zoom Out Reset image size

The diagram also reveals some potentially interesting outliers. The yellow point near the center of the figure is WISEA J191845.28+371449.2. This star, listed in SIMBAD as an A2 at a distance of 573 pc, has excess flux in all three 2MASS bands and all four WISE bands, which is well modeled by a single-temperature blackbody at 1535 K with L_ir/L_⋆ = 0.0647 (Figure 8). The star lies roughly 0.7 mag to the right of the zero-age main sequence primarily because of its K_s excess of 0.623 mag. If this star is indeed a main-sequence star with a debris disk as its SED fit suggests, this system would be an example of an "extreme debris disk" (Meng et al. 2015; Theissen & West 2017), that is, a signpost of a recent giant impact. Given the lack of an additional mid-IR-driven excess, this object could also be an example of a precursor to a two-component system with a hot debris disk (Akeson et al. 2009), where a gap has not yet been cleared between the two components. However, Gaia DR2 indicates that this object could be superluminous for a spectral type of ∼A2 (the expected spectral type for the SED-fit stellar temperature), and optical spectroscopy indicates that this object could instead be a weak-lined Herbig Ae star (A. S. Bans et al. 2018, in preparation). Additional follow-up with radio/submillimeter observations would be necessary to confirm whether this object hosts a warm debris disk, or instead has the expected substantial cold material reservoir of a Herbig Ae system.

Zoom In Zoom Out Reset image size

Zuckerman (2015) noted the high frequency of warm dust disks occurring around stars that were members of binary systems. Oh et al. (2017) specifically searched for new comoving pairs and systems in the Tycho-Gaia Astrometric Solution (Gaia Collaboration et al. 2016a; Lindegren et al. 2016). Out of 619,618 stars searched, they identified 8472 stars as members of comoving pairs, and a further 2134 as members of larger comoving systems. We note that 27 of our disk candidates presented here appear as members of comoving systems in Oh et al. (2017; listed in Table 4), out of 105 with parallax data sufficient for inclusion in the Oh et al. (2017) survey. Twelve of these 105, or 11% ± 3%, are members of comoving pairs only, significantly higher than the overall rate of 1.37% found by Oh et al. (2017). This significantly higher rate contributes further support to the hypothesis of Zuckerman (2015).

We tested the likelihood of moving group membership for each of the new disk candidates we present here using BANYAN Σ (Gagné et al. 2018), and we compared our targets with previous moving group membership determinations. We present the first 22 μm excess detection around J111714.49-594610.8, a known member of the Lower Centaurus Crux (LCC). Testing its kinematics with BANYAN Σ, however, indicates a 44.9% probability of membership in LCC, and a 40.5% probability of membership in Carina. We also note the first detection of an infrared excess around J140353.79-534628.3, a known Sco-Cen member (Hoogerwerf 2000) for which BANYAN Σ yields an 88.9% chance of membership in the Upper Centaurus-Lepus complex (UCL). We find that J164540.79-310226.6, a star with infrared excess previously detected by Cruz-Saenz de Miera et al. (2014), has a 96.9% probability of membership in the UCL. These determinations give us age benchmarks for these systems, allowing us to compare them to disk evolutionary models. Identification of disks in moving groups has yielded unexpected and valuable results with regard to theories of disk evolution, such as the identification of a primordial disk around an M-dwarf in the Carina association, at an age more than nine times greater than the e-folding time for primordial disk dissipation around solar-type stars (Silverberg et al. 2016; Murphy et al. 2018).

Most of our published targets do not yet have spectroscopically determined spectral types or luminosity classes, and many do not have measured distances. An ongoing spectroscopic follow-up campaign of DDOIs (A. S. Bans et al. 2018, in preparation) will present a more complete and more detailed analysis of the distribution of spectral types of our objects.

Between this paper, Paper I, and Silverberg et al. (2016), Disk Detective has now published 215 previously unidentified disk candidates. Of the 144 with either published spectral types or known parallaxes, the majority (110) are early-type main-sequence stars. A total of 125 of the 215 have disk temperatures T_disk < 300 K, and disk temperatures range up to 1800 K. A majority of these disks (114) have fractional infrared luminosities L_ir/L_star > 10⁻³, suggesting that these are likely primordial disks, per the criterion suggested by Williams & Cieza (2011); The characteristics of this sample overall suggest that while designed to identify debris disks, Disk Detective also effectively locates new primordial disks.

In this paper, we presented the results of follow-up imaging of 261 Disk Detective Objects of Interest, determining whether background contaminants appeared and whether they significantly impacted the infrared excess around these objects observed with WISE. We find that background objects, while apparent in the images, significantly affect the observed excess at a rate of ∼6%. Combining these data with false-positive rates from classifications and literature review, we find that AllWISE should yield ∼21,600 high-quality disk candidates based on excess at 22 μm. Applying this result to other surveys, we estimate that 4%–8% of published disk candidates from high-quality surveys may have background objects in follow-up high-resolution imaging bright enough to significantly affect the detected excess. Based on our expected yield of disks from AllWISE, we found that the searches of McDonald et al. (2012, 2017) and Marton et al. (2016) have lower-limit false-positive rates greater than 70%. We considered the 175 disk candidates of Theissen & West (2014) and found that the vast majority of these candidates would not be detected by Disk Detective due to insufficient S/N at W4. All 13 targets in the Theissen & West (2014) search with W4 S/N sufficient for the Disk Detective methodology to apply were false-positive identifications after visual inspection.

We presented a sample of 244 disk candidates, vetted through visual inspection by citizen scientists and high-resolution follow-up imaging to refine the observed excess. Disk Detective has now published 215 newly identified disk systems, of which 114 have fractional infrared luminosities indicative of primordial disks (either full protoplanetary or transitional). We find 12 of our disk candidates to be in comoving pairs, providing further support for the hypothesis of Zuckerman (2015) that there is a causal relationship between a distant companion and a warm, dusty debris disk. We identified one system, WISEA J191845.28+371449.2, as a likely "extreme" debris disk, based on its high fractional infrared luminosity. Thirty of these systems lie within 125 pc, including 26 debris disks. These nearby disk systems are good targets for adaptive optics and coronagraphic imaging to directly image exoplanets in orbit around the host star.

High-resolution follow-up imaging can eliminate many false positives, but it will not eliminate not-yet-identified spectroscopic false-positive detections of primordial and debris disks (e.g., previously unidentified dust-producing M giants, AGNs, classical Be stars). A spectroscopic follow-up program to identify these sorts of false positives is ongoing. The results presented here should be used in conjunction with the results of that program (A. S. Bans et al. 2018, in preparation) to determine the expected yield of AllWISE.

We thank the anonymous reviewer for providing comments that helped to improve the content and clarity of this paper, We acknowledge support from grant 14-ADAP14-0161 from the NASA Astrophysics Data Analysis Program and grant 16-XRP16_2-0127 from the NASA Exoplanets Research Program. M.J.K. acknowledges funding from the NASA Astrobiology Program via the Goddard Center for Astrobiology.

This publication makes use of data products from the Wide-field Infrared Survey Explorer, which is a joint project of the University of California, Los Angeles, and the Jet Propulsion Laboratory (JPL)/California Institute of Technology (Caltech), and NEOWISE, which is a project of JPL/Caltech. WISE and NEOWISE are funded by NASA.

2MASS is a joint project of the University of Massachusetts and the Infrared Processing and Analysis Center (IPAC) at Caltech, funded by NASA and the NSF.

The Digitized Sky Survey was produced at the Space Telescope Science Institute under U.S. Government grant NAG W-2166. The images of these surveys are based on photographic data obtained using the Oschin Schmidt Telescope on Palomar Mountain and the UK Schmidt Telescope. The plates were processed into the present compressed digital form with the permission of these institutions.

Funding for the SDSS and SDSS-II has been provided by the Alfred P. Sloan Foundation, the Participating Institutions, the National Science Foundation, the U.S. Department of Energy, the National Aeronautics and Space Administration, the Japanese Monbukagakusho, the Max Planck Society, and the Higher Education Funding Council for England. The SDSS website is http://www.sdss.org/.

The SDSS is managed by the Astrophysical Research Consortium for the Participating Institutions. The Participating Institutions are the American Museum of Natural History, Astrophysical Institute Potsdam, University of Basel, University of Cambridge, Case Western Reserve University, University of Chicago, Drexel University, Fermilab, the Institute for Advanced Study, the Japan Participation Group, Johns Hopkins University, the Joint Institute for Nuclear Astrophysics, the Kavli Institute for Particle Astrophysics and Cosmology, the Korean Scientist Group, the Chinese Academy of Sciences (LAMOST), Los Alamos National Laboratory, the Max-Planck-Institute for Astronomy (MPIA), the Max-Planck-Institute for Astrophysics (MPA), New Mexico State University, Ohio State University, University of Pittsburgh, University of Portsmouth, Princeton University, the United States Naval Observatory, and the University of Washington.

This research has made use of the SIMBAD database, operated at CDS, Strasbourg, France. Some of the data presented in this paper were obtained from the Mikulski Archive for Space Telescopes (MAST). STScI is operated by the Association of Universities for Research in Astronomy, Inc., under NASA contract NAS5-26555. Support for MAST for non-HST data is provided by the NASA Office of Space Science via grant NNX13AC07G and by other grants and offices. This research has made use of the VizieR catalog access tool, CDS, Strasbourg, France.

This work has made use of data from the European Space Agency (ESA) mission Gaia (https://www.cosmos.esa.int/gaia), processed by the Gaia Data Processing and Analysis Consortium (DPAC; https://www.cosmos.esa.int/web/gaia/dpac/consortium). Funding for the DPAC has been provided by national institutions, in particular the institutions participating in the Gaia Multilateral Agreement.

The Robo-AO system was developed by collaborating partner institutions, the California Institute of Technology, and the Inter-University Centre for Astronomy and Astrophysics, and with the support of the National Science Foundation under grants AST-0906060, AST-0960343 and AST-1207891, the Mt. Cuba Astronomical Foundation, and by a gift from Samuel Oschin. C.B. acknowledges support from the Alfred P. Sloan Foundation.

IRAF is distributed by the National Optical Astronomy Observatory, which is operated by the Association of Universities for Research in Astronomy (AURA) under a cooperative agreement with the National Science Foundation. PyRAF is a product of the Space Telescope Science Institute, which is operated by AURA for NASA. This research made use of ds9, a tool for data visualization supported by the Chandra X-ray Science Center (CXC) and the High Energy Astrophysics Science Archive Center (HEASARC) with support from the JWST Mission office at the Space Telescope Science Institute for 3D visualization. Resources supporting this work were provided by the NASA High-End Computing (HEC) Program through the NASA Center for Climate Simulation (NCCS) at Goddard Space Flight Center.

Facilities: FLWO:2MASS - 2MASS Telescope at Fred Lawrence Whipple Observatory, CTIO:2MASS - , WISE - , PO:1.5 m (Robo-AO) - , du Pont (RetroCam) - .

Software: IRAF (Tody 1993), PyRAF, AstroPy (Astropy Collaboration et al. 2013), NumPy (Van Der Walt et al. 2011), SciPy (Jones et al. 2001), Matplotlib (Hunter 2007), pandas (McKinney 2010), emcee (Foreman-Mackey et al. 2013).

Since Paper I, we have made two key changes to the website classification setup: we changed the scheme for retiring subjects, and we corrected the online photometry to account for saturation effects at W1 and W2.

A.1. New Retirement Scheme

A.2. Removal of WISE 1 Dropouts

In this section, we derive a minimum brightness for background objects to produce a false-positive excess detection in our system. We also show how these brightnesses propagate to the i, Y, and H bands we use in our follow-up observations.

B.1. Minimum Contaminant Brightness in W4

Let us suppose that we have a target star with no circumstellar dust, whose spectrum is accurately approximated by the Rayleigh–Jeans law in the WISE bands. We will denote this target star's intrinsic magnitudes as m_t and its intrinsic flux as f_m,t. Let us also suppose a background contaminant whose intrinsic magnitudes will be denoted m_c and whose flux will be denoted f_m,c. This background contaminant lies substantially inside the W4 PSF HWHM, such that the observed W4 magnitude [W4]_obs is the magnitude of the combined light from the target and contaminant,

$$\begin{eqnarray}&&{f}_{W4,\mathrm{obs}}={f}_{W4,t}+{f}_{W4,c},\end{eqnarray} \tag{ 1 }$$

but outside the W1 PSF HWHM, such that [W1]_obs = [W1]_t.

Assuming that our target star has $[W{1}_{t}]-[W{4}_{t}]=0$, we wish to know the minimum brightness W4_c such that [W1_obs] − [W4_obs] ≥ 0.25.

Making substitutions suggested by the above equalities, we have

$$\begin{eqnarray*}&&W{4}_{t}-W{4}_{\mathrm{obs}}\geqslant 0.25.\end{eqnarray*}$$

Using the equation for flux/magnitude conversion, this becomes

$$\begin{eqnarray}&&{f}_{W4,c}\geqslant 0.258925{f}_{W4,t}.\end{eqnarray} \tag{ 2 }$$

Converting this back into magnitude differences, we have

$$\begin{eqnarray}&&W{4}_{c}-W{4}_{t}\leqslant 1.467.\end{eqnarray} \tag{ 3 }$$

We thus show that contaminants more than ∼1.5 mag dimmer than the target star in W4 will not produce a significant enough excess for the star to become a false-positive entry in the Disk Detective Input Catalog. As such, with the greater depth probed by high-resolution imaging, we must quantitatively assess whether a detected background contaminant will produce a false positive, rather than qualitatively assessing it, as was done with the website-based classifications (Binks 2016).

B.2. Applying the Minimum Brightness to the i, Y, and H Bands

To determine if a potential background object is bright enough to produce a false-positive excess in W4, we must know the difference in magnitude between such a contaminant and the target in the bands in which we conducted high-resolution follow-up observations. We assume for this exercise that the target's optical and near-IR SED is approximately identical to an idealized Vega; that is, it is of zeroth magnitude in all bands. This sets the contaminating magnitude limit of 1.467. Below, we give examples of three typical contaminants, each with a W4 magnitude of 1.467: a background M-dwarf and two different power-law spectra. Table 8 lists the i, Y, and H magnitudes for these objects.

Table 8.  W4 Colors of Selected Background Object Types

	Colors
Characteristics	i − W4	YC − W4	HC − W4
M-dwarf: ${T}_{\mathrm{eff}}=3300\,{\rm{K}},\mathrm{log}(g)=5.0$	3.545	1.896	0.801
Class I YSO: λfλ ∼ λ0	2.94	2.05	1.82
ULIRG: $\lambda {f}_{\lambda }\sim {\lambda }^{2}$	10.25	8.67	7.48

Download table as:  ASCIITypeset image

B.2.1. False Positive That Is Due to a Background M-dwarf

The initial mass function of the neighborhood peaks at a spectral type of M2-M3.5, which corresponds to a stellar effective temperature of T_eff ≃ 3250–3400 K. Accordingly, we select a model M-dwarf atmosphere with T_eff = 3300 K and $\mathrm{log}(g)=5.0$ from the BT-Settl package of model atmospheres (Baraffe et al. 2015) as our contaminating M-dwarf. The BT-Settl models have precomputed magnitudes for many filter systems, which we use here to compute colors. This model has a [H]–[W4] value of 0.801, corresponding to an H-band contaminant delta magnitude of 2.268 mag. The [i]–[W4] value for this model is 3.545 mag, yielding a magnitude difference in Robo-AO data of 5.012 mag.

B.2.2. False Positive That Is Due to Background Sources with Power-law Spectra

Some objects are reasonably represented with a power-law spectrum λf_λ ∼ λ^α. For these objects, colors can to first order be approximated as

$$\begin{eqnarray*}&&[{m}_{{\lambda }_{1}}]-[{m}_{{\lambda }_{2}}]\sim 2.5{\mathrm{log}}_{10}\left(\displaystyle \frac{{f}_{0,{\lambda }_{1}}}{{f}_{0,{\lambda }_{2}}}\right)-2.5(\alpha -1){\mathrm{log}}_{10}\displaystyle \frac{{\lambda }_{1}}{{\lambda }_{2}}.\end{eqnarray*}$$

For the colors of interest in this paper, this corresponds to

$$\begin{eqnarray*}&&[H]-[W4]\sim 1.817\,+\,2.831\alpha ,\end{eqnarray*}$$

$$\begin{eqnarray*}&&[Y]-[W4]\sim 2.048\,+\,3.313\alpha ,\end{eqnarray*}$$

and

$$\begin{eqnarray*}&&[i]-[W4]\sim 2.940\,+\,3.654\alpha .\end{eqnarray*}$$

Young stellar objects are defined by the value of α; a Class I YSO has slope α = 0 (Lada 1987; Kenyon & Hartmann 1995). This spectrum also roughly approximates a heavily reddened early-type star. If such an object were in the background of our images and bright enough to produce a false positive, the delta magnitude in the H band would be 3.284, and the Y-band delta magnitude would be 3.515. These could both likely be detected in our images. The i-band delta magnitude would be 4.407, also likely detectable.

Also of interest is the case where α = 2. This roughly corresponds to the SED of a (U)LIRG, which has an AGN component (Vega et al. 2008). In this case, however, the H-band delta magnitude would be 8.946 and the Y-band delta magnitude would be 10.141, while the i-band delta magnitude would be 11.715. These are all clearly undetectable in our follow-up image data and are thus treated instead with the prescription of Papovich et al. (2004) in Section 5.2.

Below are brief comments on selected disk candidates presented in Section 7. Having noted previous identifications as disk candidates in Section 7, we primarily discuss characteristics of the subject's appearance, either in the images used on the Disk Detective website or in the follow-up images.

• 1.  
  J021327.01+421923.3. This system, which was previously identified by Cotten & Song (2016) and McDonald et al. (2017), exhibits slight extension in the W4 image.
• 2.  
  J023720.84+395345.8. This system is a known spectroscopic binary (Hube 1981). The companion, an early-G main-sequence star based on the binary mass function, has a projected angular separation of ≪1'', making it undetectable as a separate component in follow-up data. The SED indicates no significant effect on the 22 μm excess.
• 3.  
  J025926.83+593531.6. We recover this system, a spectroscopic binary (Abt 2009), that was previously identified as a source in the W5 region by Koenig et al. (2008). The secondary component has a minimum mass of 1.7326 R_⊙ based on the binary mass function, suggesting a possible A7V star. The projected angular separation is ≪1'', making it undetectable as a separate component in follow-up data. The SED indicates no significant effect on the 22 μm excess.
• 4.  
  J030854.20-185809.1. This system, a known A0V system, is best fit as a two-stellar-component system with a K dwarf component, as well as a 349 K disk. Further observation is necessary to confirm this additional component.
• 5.  
  J034400.28+243324.6. This δ Scuti variable, which appears in Cotten & Song (2016) and McDonald et al. (2017), exhibits slight extension at W4.
• 6.  
  J041517.47+505124.0. This system, which appears in Marton et al. (2016), has been identified as a Be star previously (Merrill & Burwell 1943) and more recently as having Hα in emission (Kohoutek & Wehmeyer 1999). Further spectroscopic follow-up is necessary to determine the nature of this emission (i.e., whether the target is a classical Be star, rather than a debris disk host).
• 7.  
  J043521.12-081730.0. This object exhibits slight extension at W4.
• 8.  
  J051143.75+122012.5. This system exhibits slight extension at W4.
• 9.  
  J051328.63-043910.6. This object is a component of a binary system. Its companion is 35'' away from the source, too distant to affect the observed excess.
• 10.  
  J052331.01-010423.6. This star exhibits eclipses from circumstellar material, per Osborn et al. (2017).
• 11.  
  J053707.15+603636.4. This object, which appears in Cotten & Song (2016), exhibits slight extension at W4.
• 12.  
  J054330.38+251724.4. This system has previously been identified as an Hα emitter (Kohoutek & Wehmeyer 1999). Further spectroscopic follow-up is necessary to determine the nature of this emission.
• 13.  
  J054733.26+521144.5. This system has previously been identified as an Hα emitter (Merrill & Burwell 1950). Further spectroscopic follow-up is necessary to determine the nature of this emission.
• 14.  
  J111714.49-594610.8. This target is a member of the LCC of the Sco-Cen OB2 association, per Hoogerwerf (2000). Testing its kinematics with BANYAN Σ yields a 44.9% membership probability in LCC and a 40.5% membership probability in Carina.
• 15.  
  J111925.92-301922.9 and J165204.85+145827.2. We report the first detection of a disk around the α² CVn variable AWI0005cwg. We also note that AWI00004o8, identified in Paper I, is also an α² CVn variable.
• 16.  
  J114336.83-802900.5. This target exhibits slight extension at W4.
• 17.  
  J132026.77-491325.4. This target is a known disk-hosting member of Sco-Cen (Chen et al. 2012). Reevaluation with BANYAN Σ yields an 88.7% probability of membership in LCC.
• 18.  
  J134909.18-541342.3. This target is a known disk-hosting member of Sco-Cen (Chen et al. 2012). Reevaluation with BANYAN Σ yields a 42.1% membership probability in LCC and a 49.5% membership probability in UCL.
• 19.  
  J144458.63-280251.9. This target is a known multiple system (Horch et al. 2011) with an excess first detected by Cruz-Saenz de Miera et al. (2014). We do not resolve both components of the binary system.
• 20.  
  J173254.69+404312.3. This target exhibits very slight extension at W4.
• 21.  
  J173832.90+425112.9. This target is a known Cepheid variable.
• 22.  
  J183311.41+025439.0. This object shows a faint background object in DSS2 survey data that does not appear in Robo-AO observations.
• 23.  
  J185211.39+102422.6. This object shows a faint background object in DSS2 survey data that does not appear in Robo-AO observations.
• 24.  
  J190901.24+110641.3. This target exhibits Hα in emission (Kohoutek & Wehmeyer 1999).
• 25.  
  J192136.46+220744.7. This object shows a faint background object in DSS2 survey data that does not appear in Robo-AO observations.
• 26.  
  J192437.52+563454.9. This object shows a very faint background object in DSS2 survey data, which does not appear in the Robo-AO data for the system.
• 27.  
  J210144.07+521717.6. This object is an emission-line star (Kohoutek & Wehmeyer 1999).
• 28.  
  J212952.96+525601.9. This object exhibits a slight asymmetrical extension at W4.
• 29.  
  J215947.70-593411.9. This target is a known δ Scuti variable (Rodríguez et al. 2000).
• 30.  
  J230112.67-585821.9. We note slight extension at W4.