The GALEX Time Domain Survey I. Selection and Classification of Over a Thousand UV Variable Sources
S. Gezari, D.C. Martin, K. Forster, J.D. Neill, M. Huber, T. Heckman, L. Bianchi, P. Morrissey, S.G. Neff, M. Seibert, D. Schiminovich, T.K. Wyder, W. S. Burgett, K. C. Chambers, N. Kaiser, E. A. Magnier, P. A. Price, J.L. Tonry
aa r X i v : . [ a s t r o - ph . C O ] F e b Accepted for publication in ApJ
Preprint typeset using L A TEX style emulateapj v. 02/07/07
THE
GALEX
TIME DOMAIN SURVEY I. SELECTION AND CLASSIFICATION OF OVER A THOUSAND UVVARIABLE SOURCES
S. Gezari , D.C. Martin , K. Forster , J.D. Neill , M. Huber , T. Heckman , L. Bianchi , P. Morrissey , S.G.Neff , M. Seibert , D. Schiminovich , Wyder, T.K. , W. S. Burgett, K. C. Chambers, N. Kaiser, E. A.Magnier, P. A. Price, and J.L. Tonry Accepted for publication in ApJ
ABSTRACTWe present the selection and classification of over a thousand ultraviolet (UV) variable sourcesdiscovered in ∼
40 deg of GALEX
Time Domain Survey (TDS)
N U V images observed with acadence of 2 days and a baseline of observations of ∼ GALEX
TDS fields were designedto be in spatial and temporal coordination with the Pan-STARRS1 Medium Deep Survey, whichprovides deep optical imaging and simultaneous optical transient detections via image differencing.We characterize the
GALEX photometric errors empirically as a function of mean magnitude, andselect sources that vary at the 5 σ level in at least one epoch. We measure the statistical propertiesof the UV variability, including the structure function on timescales of days and years. We reportclassifications for the GALEX
TDS sample using a combination of optical host colors and morphology,UV light curve characteristics, and matches to archival X-ray, and spectroscopy catalogs. We classify62% of the sources as active galaxies (358 quasars and 305 active galactic nuclei), and 10% as variablestars (including 37 RR Lyrae, 53 M dwarf flare stars, and 2 cataclysmic variables). We detect a large-amplitude tail in the UV variability distribution for M-dwarf flare stars and RR Lyrae, reaching up to | ∆ m | = 4 . N U V < | ∆ m | > . ∼ − for quasars, AGNs, and RR Lyrae stars,respectively. We also calculate a surface density rate in the UV for transient sources, using theeffective survey time at the cadence appropriate to each class, of ∼
15 and 52 deg − yr − for Mdwarfs and extragalactic transients, respectively. Subject headings: ultraviolet: general — surveys INTRODUCTION
Unlike the optical, X-ray, and γ -ray sky, which havebeen systematically studied in the time domain in thesearch for supernovae (SNe) and gamma-ray bursts(GRBs), the wide-field UV time domain is a relativelyunexplored parameter space. The launch of the GALEX satellite with its 1.25 deg diameter field of view, and lim-iting sensitivity per 1.5 ks visit of ∼
23 mag in the
F U V ( λ eff = 1539 ˚A) and N U V ( λ eff = 2316 ˚A) (Martin et al.2005; Morrissey et al. 2007), enabled the discovery of UVvariable sources in repeated observations over hundredsof square degrees for the first time.The UV waveband is particularly sensitive to hot Department of Astronomy, University of Maryland, CollegePark, MD 20742-2421, USA California Institute of Technology, MC 249-17, 1200 East Cal-ifornia Boulevard, Pasadena, CA 91125, USA Institute for Astronomy, University of Hawaii, 2680 WoodlawnDrive, Honolulu HI 96822, USA Department of Physics and Astronomy, Johns Hopkins Univer-sity, 3400 North Charles Street, Baltimore, MD 21218, USA Laboratory for Astronomy and Solar Physics, NASA GoddardSpace Flight Center, Greenbelt, MD 20771, USA Observatoires of the Carnegie Institute of Washington,Pasadena, CA 90095, USA Department of Astronomy, Columbia University, New York,NY 10027, USA Department of Astrophysical Sciences, Princeton University,Princeton, NJ 08544, USA ( ≈ K) thermal emission from such transient andvariable phenomena as young core-collapse SNe, the in-ner regions of the accretion flow around accreting su-permassive black holes (SMBHs), and the flaring statesof variable stars. The characterstic timescales of vari-ables and transients in the UV range from minutes toyears. M dwarf flare stars have strong magnetic activ-ity that manifests itself in flares of thermal UV emissionon the timescale of minutes (Kowalski et al. 2009). RRLyrae stars have periodic pulsations which drive tem-perature fluctuations from ∼ ∼ . GALEX data observed aspart of the All-Sky, Medium, and Deep Imaging base-line mission surveys (AIS, MIS, DIS) from 2003 to 2007,yielded the detection of M-dwarf flare stars (Welsh et al.2007), RR Lyrae stars, AGN, and quasars (Welsh et al.2005; Wheatley et al. 2008; Welsh et al. 2011), and flares Gezari et al.from the tidal disruption of stars around dormant su-permassive black holes (Gezari et al. 2006, 2008a, 2009).Serendipitous overlap of 4
GALEX
DIS fields with theoptical CFHT Supernova Legacy Survey, enabled the ex-traction of simultaneous optical light curves from imagedifferencing for 2 of the tidal disruption event (TDE) can-didates (Gezari et al. 2008a), and enabled the associationof transient UV emission with two Type IIP supernovae(SNe) within hours of shock breakout (Schawinski et al.2008; Gezari et al. 2008b). Chance overlap of
GALEX observations with the survey area of the optical PalomarTransient Factory (PTF) detected a Type IIn SN whoserising UV emission over a few days was interpreted asa delayed shock breakout through a dense circumstellarmedium (Ofek et al. 2010).Motivated by the promising results from the analy-sis of random repeated
GALEX observations, and thedemonstrated value of overlap with optical time domainsurveys, we initiated a dedicated
GALEX
Time DomainSurvey (TDS) to systematically study UV variabilityon timescales of days to years with multiple epochs of
N U V images observed with a regular cadence of 2 days.The
GALEX
TDS fields were selected to overlap withthe Pan-STARRS1 Medium Deep Survey (PS1 MDS)(Kaiser et al. 2010).
GALEX
TDS and PS1 MDS arewell-matched in field of view, sensitivity, and cadence(shown in Table 1). Here we present the analysis of 42
GALEX
TDS fields which intersect with the PS1 MDSfootprint, for a total area on the sky of 39.91 deg , whichwere monitored over a baseline of 3.32 years (February2008 − June 2011). In this paper we use PS1 MDSdeep stack catalogs to characterize the optical hosts of
GALEX
TDS sources. Simultaneous UV and opticalvariability of
GALEX
TDS sources culled from matcheswith the PS1 transient alerts (Huber et al. 2011) will bepresented in future papers.The paper is structured as follows. In § GALEX
TDS survey design, and in § § § GALEX
TDS sources, in § § GALEX
TDS OBSERVATIONS
GALEX
TDS monitored 6 out of 10 total PS1MDS fields, with 7
GALEX
TDS pointings (labeledPS fieldname
MOS pointing ) at a time to cover the PS17 deg field of view. During the window of observingvisibility of each GALEX
TDS field (from 2 − − σ point-source limit of m AB ∼ . N U V detector devel-oped a problem on 2010 May 4 during observations ofPS ELAISN1, and so we do not include epochs observedbetween this time and when the the instrument was fixedon 2010 June 23 in our analysis. Figure 1 shows the posi-tion of the
GALEX
TDS fields relative to the PS1 MDS
Fig. 1.—
GALEX
TDS 1 . . fields, and Table 2 lists the R.A. and Dec of their cen-ters, the Galactic extinction ( E ( B − V )) for each fieldfrom the Schlegel et al. (1998) dust maps, and the num-ber of epochs per field. The median number of epochsper field is 24. PS CDFS MOS00 is an exception with114 epochs, because it was monitored with a rapid ca-dence ( | ∆ t | ∼ GALEX pointings fromthe footprint of the PS1 MDS fields were necessary inorder to avoid UV-bright stars in the field of view thatwould violate the detector’s bright-source count limits.Figure 2 shows the temporal sampling of
GALEX
TDSobservations in the
N U V in comparison to the PS1 MDSobservations in the g P1 , r P1 , i P1 , z P1 , and y P1 bandsfrom February 2008 to June 2011. PS1 began takingcommissioning data of the MDS fields in May 2009, butdid not begin full survey operations until a year later.The GALEX
F U V detector became non-operational inMay 2009, and so we only include
N U V images in ourstudy. STATISTICAL MEASUREMENTS
Selection of Variable Sources
Since most galaxies are unresolved by the
GALEX
N U V
GALEX of 0 . ′′
5, theBayesian probability that the match is real is largerthan the Bayesian probability that the match is spurious(Budav´ari & Szalay 2008). The final master list includes
ALEX
Time Domain Survey 3
Fig. 2.—
Dates of
GALEX
TDS
NUV observations comparedto PS1 MDS observations in the g P1 , r P1 , i P1 , z P1 , and y P1 bands.Dotted lines show yearly intervals. GALEX images are Poisson-limited, the Poisson error underestimates the total errorin the
GALEX catalog magnitudes by a factor of ∼ empirically by calculating the standard devi-ation of aperture magnitudes in bins of mean magnitude, h m i . We only include objects in the pipeline-generatedcatalogs that are detected in all or ≥
10 epochs. In eachbin of N objects with h m i i = ( P nk =1 m i,k ) /n (each bintypically has N = 50 to 1000 sources), we calculate foreach epoch k of a total of n epochs, σ ( h m i , k ) = vuut N − N X i =1 ( m i,k − h m i i ) (1)where m i,k is the magnitude (in the AB system) givenby m i,k = − . f ) + zp + C ap − . E ( B − V ), f is the background-subtracted flux in a 6 arcsec radiusaperture, zp = 20.08, the aperture correction is C ap = − .
23 mag (Morrissey et al. 2007), and we correct forGalactic extinction using the values for E ( B − V ) listedin Table 2. We use 3 σ clipping to remove outliers in thecalculation of σ ( h m i , k ) which can arise from artifacts.The astrometric precision depends on the signal-to-noise of the source, thus we also empirically measure amagnitude-dependent clustering radius. We do so bymeasuring the cumulative distribution of spatial sep-arations between the position in each epoch and themean position for sources in bins of h m i , and record d ( h m i , k ), the value for which 95% of sources have aseparation less than or equal to that value. The re-sulting value for d is a strong function magnitude,increasing from ∼ h m i = 18 . ∼ h m i = 23 . σ ( h m i , k ) and d ( h m i , k ) for an example GALEX
TDS
Fig. 3.—
Empirical determination of 1 σ photometric errorsas a function of mean magnitude from the standard deviationof sources detected by the pipeline catalogs in ≥
10 epochs forone of the
GALEX
TDS fields PS COSMOS MOS23. Solid greenline shows a quadratic fit to the error function for the fieldPS COSMOS MOS23, and dashed blue line shows a quadratic fitto the median error function for all of the
GALEX
TDS fields.Solid red line shows the expected 1 σ Poisson error, which underes-timates the total photometric error by a factor of ∼ Fig. 4.—
Maximum spatial separation from the mean for 95% ofthe sources as a function of mean magnitude from the cumulativedistribution of sources detected by the pipeline catalogs in ≥ GALEX
TDS fields PS COSMOS MOS23.Solid green line shows a quadratic fit to the error function for thefield PS COSMOS MOS23, and dashed blue line shows a quadraticfit to the median distance function for all of the
GALEX
TDSfields. Note that due to systematic differences in the PSF be-tween fields, the distance function radius for PS COSMOS MOS23is up to ∼ h m i < ∼
20 mag, d ∼ d ∼ field PS COSMOS MOS23, a quadratic fit to the me-dian function for all epochs in that field, and the medianfunction fit over all fields.In our master list of source positions, we include allsources detected, including sources detected in only one Gezari et al.epoch, and fix the centroid to the epoch for whichthe source is detected with maximum flux. We mea-sure forced aperture magnitudes at the positions of eachsource in epochs where the source was not detected bythe pipeline or the spatial separation of the matchedsource is greater than d ( h m i , k ). When the aper-ture magnitude is fainter than m lim in an epoch, it isflagged as an upper limit and replaced with m lim , where m lim = − . p ( B sky N pix /T exp ) + zp + C ap , where B sky = 3 × − counts s − pixel − , N pix = 16 π and T exp is the exposure time of that epoch in seconds.We select sources that have at least one epoch for which | m k − h m i| > σ ( h m i , k ), where h m i is calculated onlyfrom epochs that have a magnitude above the detectionlimit of that epoch. We use this selection method to besensitive to short-term and long-term variability, as wellas transients. This 5 σ selection is quite conservative, andrequires variability amplitudes increasing from | ∆ m | > . h m i ∼
18 mag up to | ∆ m | > . h m i ∼
23 mag.We make the following cuts to the 5 σ variable sourcesample to remove artifacts:i) We remove sources with pipeline artifact flagsindicating window bevel reflections or ghostsfrom the dichroic beam splitter.ii) We remove the brightest objects, with h m i < .
0, due to the large area subtended by thePSF which causes uncertainty in the back-ground subtraction.iii) We select sources within a radius of 0 .
55 degof the center of the field, in order to avoidglints and PSF distortions, which are moreprominent on the edges of the image, fromun-corrected spatial distortions of photonsrecorded by the detectors.iv) We do not include objects that are within1.5 arcmin of a m <
17 mag source, to avoidregions affected by the bright source’s PSFand ghost artifacts. Ghost artifacts can ap-pear within 30 −
60 arcsec above and belowa bright source in the Y detector direction.Ghosts are point-like, and thus can only beidentified from their Y detector position rel-ative to a bright source. While ghosts donot usually appear in the
GALEX pipelinecatalogs, we apply this cut since our forcedaperture photometry could mistake ghosts fortransient sources.v) We veto objects for which in the epoch ofmaximum | m k − h m i| /σ ( h m i , k ) or maximumflux, the aperture flux ratio of the object has f /f . > R , where f is the 6.0 arcsecradius aperture flux, f . is the 3.8 arcsecradius aperture flux, and R is the max-imum cumulative aperture flux ratio mea-sured for 90% of the sources in the referencesource sample used to calculate σ ( h m i , k ) inthat epoch. This cut removes fluctuations inthe background due to reflections from brightstars just outside the field-of-view, as well Fig. 5.—
Left : Gallery of good and vetoed variable sources dur-ing their epochs of minimum (“low”) and maximum (“high”) flux.Green circle shows a 6 arcsec (4 pixel) aperture radius. Sourcesvetoed for f /f . > R shown in panels a and b , a source vetoedas a likely ghost within 1.5 arcmin of a m <
17 mag source shownin panel c , and a manually vetoed source shown in panel d . Foreach source the grayscale is linear, and is scaled to the peak of thesource in its “high” state. as epochs where the PSF is distorted due toa degradation in resolution which sometimesoccurs in the Y detector direction. This alsovetoes cases when the pipeline shreds a sourceinto multiple sources, and the source is de-tected as variable because the center of theaperture is off-center from the peak sourceflux.vi) Finally, we visually inspect all of the remain-ing variable sources to remove any remainingartifacts that passed through the cuts above.Figure 5 shows a gallery of good sources and vetoedvariable sources from our automated cuts (for a brightreflection in panel a, an off-center source in panel b, anda likely ghost in panel c) and manual cuts (for a diffusereflection in panel d). Our final GALEX
TDS 5 σ variablesample after the cuts listed above has a total of 1078sources. Variability Statistics
We characterize the variability of each 5 σ variable UVsource using several statistical measures. We measure thestructure function (following di Clemente et al. (1996)), V (∆ t ) = r π h| ∆ m ij |i t − h σ i + σ j i ∆ t (2)where brackets denote averages for all pairs of points onthe light curve of an individual source with i < j and t j − t i = ∆ t . The 2 day cadence of the observations combinedwith the seasonal visibility of the fields results in a dis-tribution of time intervals between observations (shownin Figure 6) that fall into 6 characteristic timescale bins:∆ t = 2 ± . t = 4 ± . t = 6 ± . t = 8 ± . t = 0 . ± .
14 yr, and ∆ t = 1 . ± . ALEX
Time Domain Survey 5
Fig. 6.—
Histogram of the time intervals between all pairs of ob-servations for the 42 fields in
GALEX
TDS. Hatched regions showthe time intervals over which the structure function is measuredfor all of the sources. and define S d to be the maximum value of the struc-ture function evaluated for ∆ t , ∆ t , ∆ t , and ∆ t ,and S yr to be the maximum value of the structure func-tion evaluated for ∆ t and ∆ t . We also measurethe intrinsic variability as defined by Sesar et al. (2007), σ int = p Σ − ξ , where Σ = n − P nk =1 ( m k − h m i ) and ξ = n P nk =1 σ ( h m i , k ) , and the maximum ampli-tude of variability, max( | ∆ m | ). UV Light Curve
In order to flag possible transient UV events that maybe associated with a SN or TDE, we differentiate be-tween stochastic variability and flaring variability. Weidentify flaring UV variability as sources that show aconstant flux ≥
10 days before the peak of the lightcurve, and do not fade more than 2 σ below the faintestpre-peak magnitude (measured ≥
10 days before thepeak). This selection criteria is tailored to the
N U V risetimes observed in SNe (Gezari et al. 2008b; Brown et al.2009; Gezari et al. 2010; Milne et al. 2010; Ofek et al.2010) and TDE candidates (Gezari et al. 2006, 2008a,2009). We define constant pre-peak flux as a lightcurve with a reduced χ ν / < χ ν = P pk =1 ( m k −h m i ) / ( σ ( h m i , k ) / ( n − n is the number ofepochs ≥
10 days before the peak). For those sources forwhich there are only upper limits ≥
10 days before thepeak, χ ν is set to 1. Flaring sources with no detectionsbefore the peak are further labeled as transients. Sourceswith χ ν / ≥
3, or that fade below 2 σ of the faintest mag-nitude measured ≥
10 days before the peak are labeledas stochastically variable. We flag 116 flares, 145 tran-sients, 595 stochastically variable sources, and remainwith 222 sources with neither light-curve classificationflag. In Figure 7 we show example light curves of sourcesflagged as stochastically variable (’V’), flares (’F’), andtransients (’T’). HOST PROPERTIES
Archival Optical Imaging Catalogs
We first characterize the host properties of the UV vari-able sources using archival optical u , g , r , i , and z pho-tometry and morphology from matches to the SDSS Pho-tometric Catalog, Release 8 (Aihara et al. 2011) ( m lim ∼
22 mag), the CFHTLS Deep Fields D1, D2, and D3( m lim ∼ . Fig. 7.—
Example light curves of sources flagged as stochasticallyvariable (’V’), flares (’F’), and transient flares (’T’). Epochs ≥ ≥
10 days before the peak.
W4 ( m lim ∼
25 mag) merged catalogs version T0005 ,and the SWIRE ELAIS N1 and CDFS Region catalogs( m lim ∼
24 mag) (Surace et al. 2004). For sources withmatches in multiple catalogs, we use the match from thedeepest catalog. We convert the CFHTLS magnitudes tothe SDSS system using the conversions in Regnault et al.(2009), and the SWIRE Vega magnitudes to the SDSSsystem using the transformations measured for stellar ob-jects available at the INT WFS web site . We thencorrect for Galactic extinction using the Schlegel et al.(1998) dust map values for E ( B − V ) listed in Table 2.Figure 1 shows the overlap of the GALEX
TDS fieldswith the available archival optical catalogs.
Pan-STARRS1 Medium Deep Survey
The
GALEX
TDS fields overlap with the PS1MDS fields MD01 (PS XMMLSS), MD02 (PS CDFS),MD04 (PS COSMOS), MD07 (PS GROTH), MD08(PS ELAISN1), and MD09 (PS VVDS22H). ThePan-STARRS1 observations are obtained through a setof five broadband filters, ( g P1 , r P1 , i P1 , z P1 , and y P1 ).Further information on the passband shapes is describedin Stubbs et al. (2010). The PS1 MD fields are observedwith a typical cadence in a given filter of 3 days, withan observation in the g P1 and r P1 bands on night one,in the i P1 band on night two, and the z P1 band onnight three, with y P1 -band observations during each ofthree nights on either side of the Full Moon. Imagedifferencing is performed on the nightly stacked images,reaching a typical 5 σ detection limit of ∼ . g P1 , r P1 , i P1 bands and ∼ . y P1 band. Image difference detections from thePS1 Image Processing Pipeline (IPP; Magnier (2006))and an independent pipeline hosted by Harvard/CfA(Rest et al. 2005) are internally distributed to thePS1 Science Consortium as transient alerts for visualinspection and classification. http://terapix.iap.fr/rubrique.php?id rubrique=252 ∼ wfcsur/technical/photom/colours Gezari et al.
Fig. 8.—
Comparison of the star/galaxy separator used in thePS1 MD catalogs (thick dashed lines) to matches in the SDSS ( left )and CFHT ( right ) archival catalogs in the PS GROTH field as afunction of i -band magnitude. Deep stacks of the multi-epoch observations weregenerated to provide deep imaging with a 5 σ point-source limiting magnitude of ∼ g P1 , r P1 , i P1 , z P1 , and y P1 bands,respectively, and typical seeing (PSF FWHM) of ∼ . , . , . , . , . m = − . m ′ , with a relative zeropointadjustment m ′ made in each band for each individualepoch (Schlafly et al. 2012) before stacking to conformto the absolute flux calibration in the AB magnitudesystem (Tonry et al. 2012). We convert the PS1 mag-nitudes to the SDSS system using the bandpass trans-formations measured for stellar SEDs in Tonry et al.(2012), and correct for Galactic extinction using theSchlegel et al. (1998) dust map values for E ( B − V )listed in Table 2. We obtain morphology informationfrom the PS1 IPP output parameters in the i P1 filterfor the PSF magnitude ( PSF INST MAG ), the aperturemagnitude (
PSF AP MAG ), and the PSF-weighted frac-tion of unmasked pixels
PSF QF , to define a point sourceor extended source as:
IF PSF INST MAG − PSF AP MAG < .
04 mag
AND PSF QF > . THEN class = pt
IFPSF INST MAG − PSF AP MAG > .
04 mag
AND PSF QF > . THEN class = ext We calibrated these parameter cutsby comparing sources detected in both the PS1 MDSand archival optical catalogs. Figure 8 shows the PS1star/galaxy separation criteria for 110,804 sources de-tected in both PS1 and SDSS catalogs in the PS GROTHfield, and for 169,461 sources detected in both PS1 andCFHT catalog in the PS GROTH field, with i <
22 mag,the faintest magnitude for 96% of the optical hosts of the
GALEX
TDS sources, and the magnitude limit whereall three catalogs are complete. Even though the CFHTcatalogs are deeper than SDSS, they do not attempt toseparate stars and galaxies for i > ∼
21 mag, and classifyall sources fainter than this magnitude as point sources.However, it is clear from both comparison plots, that thePS1 criterion of
PSF INST MAG − PSF AP MAG < .
04 magdoes an even better job of separating the locus of starsfrom galaxies than both catalogs down to i ∼
22 mag.
Archival Redshift and X-ray Catalogs
We also take advantage of the many archival X-ray and spectroscopic catalogs available from the over-lap of the
GALEX
TDS survey with legacy surveyfields. In the PS CDFS field we us X-ray catalogsfrom the 0.3 deg Chandra Extended CDFS survey(Giacconi et al. 2002; Lehmer et al. 2005; Virani et al.2006), and redshift catalogs from the VIMOS VLT Deep Survey (VVDS) (Le F`evre et al. 2004), and a compi-lation of redshift catalogs from GOODS and SWIRE . In the PS XMMLSS field we use X-ray catalogsfrom the 5.5 deg XMM-LSS survey (Chiappetti et al.2005; Pierre et al. 2007), and redshift catalogs fromVVDS (Le F`evre et al. 2005). In the PS COSMOSfield, we use X-ray catalogs from the 1.9 deg XMM-Newton Wide-Field Survey (Hasinger et al. 2007) andthe 0.9 deg Chandra COSMOS survey (Elvis et al.2009), and redshifts from the Magellan COSMOS AGNsurvey (Trump et al. 2007, 2009), the VLT zCOSMOSbright catalog (Lilly et al. 2007, 2009), and the ChandraCOSMOS Survey catalog (Civano et al. 2012). In thePS GROTH field we use X-ray catalogs from the 0.67deg Chandra Extended Groth Strip (Nandra et al. 2005;Laird et al. 2009) and redshift catalogs from the DEEP2Galaxy Redshift Survey (Newman et al. 2012). ForPS ELAISN1 we use the X-ray catalog from the 0.08 deg Chandra ELAIS-N1 deep X-ray survey (Manners et al.2003). Figure 1 shows the overlap of the
GALEX
TDS fields with the archival X-ray surveys. Finally,we also match the sources with the ROSAT All-SkyBright Source and All-Sky Survey Faint Source catalogs(Voges et al. 1999, 2000). CLASSIFICATION
We classify the
GALEX
TDS sources using a combi-nation of optical host photometry and morphology, UVvariability statistics, and matches with archival X-rayand redshift catalogs. Table 3 summarizes the sequenceof steps we use to classify the sources, which we describein detail below.
Cross-Match with Optical Catalogs
We first cross-matched our 1078
GALEX
TDS sourceswith the archival u, g, r, i, z catalogs described in § GALEX and ground-based optical catalogs (Budav´ari & Szalay 2008), andcorresponds to a spurious match rate of only 1 − GALEX
TDS fields(Bianchi et al. 2011). However, we found that there wasa population of “orphans” (no optical match within 3arcsec) that were detected in their
N U V low-state, andhad a match between 3 − u band) to1057/1078 (98%). Figure 9 shows a histogram of the r -band magnitude of the optical hosts, and the N U V − r colors of the GALEX
TDS sources in their low-state.The optical hosts have a distribution that peaks at r ∼ ∼ arettura/CDFS master/index.html ALEX
Time Domain Survey 7
Fig. 9.—
Left : Histogram of r magnitudes of optical matches. Right : Histogram of
NUV − r colors of optical matches for thosesources detected in the NUV during their low-state (solid lines),and those sources with upper limits during their low-state (dashedlines). of PS1 MDS, and
N U V − r ∼ N U V low-state are shown as upperlimits in the
N U V − r color histogram, and peak at N U V − r > Orphans
We visually inspected the PS1 stack images at the lo-cations of the 21 sources with no optical matches, andconfirm that they are true orphan events. Furthermore,all of the orphans are undetected in their low-state in the
N U V , with upper limits of
N U V > (22 . − .
1) mag.Thus the orphan hosts are likely distant stars or faintgalaxies (i.e., dwarf galaxies or high-redshift galaxies)that are undetected during their low-state in the opti-cal and
N U V . Color and Morphology Cuts
We first use the color and morphology of the opti-cal hosts to classify the
GALEX
TDS 5 σ UV variablesources. We define quasars as sources with optical point-source hosts with u − g < . − . < g − r < . | ∆ m | > Fig. 10.—
Maximum
NUV variability amplitude as a function oflow-state
NUV magnitude for sources classified as RR Lyrae (red),M dwarfs (yellow), quasars (blue) AGN (green), and CVs (cyan).Dashed line shows the median 5 σ error selection function used toselect the variable sources. . < u − g < .
45 (4) − . < g − r < . − . < r − i < . − . < i − z < . u -band, which is not availablefor sources with PS1-only matches. However, we defineM dwarf stars as point sources with r − i > .
42 (5) i − z > . g < . r < . i < . u -band data.We classify stars on the main stellar locus as those with1 . < u − g < .
25 (7)0 . < g − r < . − . < r − i < . − . < i − z < . u -band data, and their classifica-tions as quasars, RR Lyrae, M dwarfs, and stars. UV Variability Cuts
Figure 12 shows the structure function on timescalesof days and years described in § § N U V structure function on Gezari et al.
Fig. 11.—
Colors of archival optical matches to UV variablesources with point-like optical hosts (black points). Dashed blueline shows the region in color-color space used to define quasarsfrom optical colors and morphology alone. Sources with classifica-tion are color coded as quasars in blue, RR Lyrae in red, and Mdwarf stars in yellow, and main stellar locus stars in cyan. Sourceswith archival X-ray matches are circled in purple. timescales of years to days ( S yr /S d ). While quasarsdemonstrate a wide range of S yr /S d , all RR Lyrae have S yr /S d <
3. We use this UV variability property to relaxour color constraints, and increase our photometric sam-ple of quasars to all sources with optical point-sourcehosts with S yr /S d ≥
3. This is equivalent to a struc-ture function power-law exponent cut of γ > .
2, where S (∆ t ) ∝ ∆ t γ (Hook et al. 1994; Vanden Berk et al.2004; Schmidt et al. 2010). This structure-function ra-tio selection results in the classification of another 30quasars. Two additional sources with optical point-source hosts have archival quasar spectra, resulting in afinal quasar sample of 358 . We define active galactic nu-clei (AGN), as sources with optically extended hosts thatshow stochastic UV variability (see § X-ray Sources
The archival X-ray catalogs overlap with ∼ .
45 deg ofthe GALEX
TDS survey area. Within this area, 81/89quasars, 92/105 AGN, and 8/9 M-dwarf stars are de-tected in the X-rays. In addition, there are 9 opticalpoint sources with X-ray matches that are likely quasarsand M dwarfs just outside the quasar and M-dwarf color-color selection regions. UV variability selection appearsto be selecting a similar population of active galaxiesand M dwarfs as X-ray detection, since ∼
90% of the UVvariability-selected active galaxy and M dwarf sample isalso detected in the X-rays. However, only 2% of all theX-ray sources (the majority of which are active galaxies)are detected as UV variable at the selection threshold of
Fig. 12.—
Histogram of the
NUV structure function of sourcesclassified as RR Lyrae (red) and quasars (blue) on a timescale ofdays and years. Red arrow and Blue arrow indicate the meanof S yr measured in the SDSS r -band for RR Lyrae and quasars,respectively from Schmidt et al. (2010). Fig. 13.—
Histogram of the ratio of the
NUV structure functionon timescales of years and days for sources classified as RR Lyrae(red) and quasars (blue). Dotted line shows the selection criteria of S y /S d ≥ the GALEX
TDS catalog.
GUVV Catalog
We also cross-match our
GALEX
TDS sample with thefirst and second
GALEX
Ultraviolet Variability Cata-logs (GUVV-1 & GUVV-2) from Welsh et al. (2005) andWheatley et al. (2008). These catalogs include 894 UVvariable sources (∆ m > . N U V ) selectedfrom an analysis of archival
GALEX
AIS, MIS, DIS,and Guest Investigator (GI) fields with repeated observa-tions. With a cross-matching radius of 4 arcsec, we finda match with 36 GUVV sources. For the 15 matches thathave GUVV classifications, are all classified by GUVV asactive galaxies (AGN or quasars), which are in agreementwith our
GALEX
TDS classifications. Of the 21 matches
ALEX
Time Domain Survey 9
Fig. 14.—
Colors of archival optical matches to UV variableswith extended optical hosts. Sources classified as AGN (by eitherstochastic UV variability, archival spectra, and/or an X-ray match)are color coded in green. Sources with matches with archival X-raymatches are circled in purple. Note that all galaxy sources with anarchival X-ray match are defined as AGNs. without GUVV classifications, we find 5 sources classi-fied by
GALEX
TDS as RR Lyrae, 9 classified as activegalaxies (AGN or quasars), 6 with optical point-sourcehosts, and 1 with a galaxy host.
Unclassified Sources
The remaining 302 unclassified sources include 91 withoptical point-source hosts, which are likely stars, quasarswith non-standard colors (high-redshift or reddened), orunresolved galaxies, 190 with galaxy hosts, and 21 or-phans. The 190 galaxy hosts may either be faint AGNwith poorly constrained UV light curves, or hosts of UV-bright extragalactic transients. In Figure 15 we show themaximum | ∆ m | in the N U V as a function of low-state
N U V magnitude of the remaining unclassified sources.The unclassified UV source with the most extreme am-plitude, ELAISN1 MOS15-09 with | ∆ m | > . z = 0 . GALEX
TDS and PS1 MDS was attributedto the tidal disruption of a star around a supermassiveblack hole (Gezari et al. 2012). Also in this sample is aUV transient spectroscopically confirmed to be a TypeIIP SN 2010aq at z=0.086 (COSMOS MOS26-29), whoseUV/optical light curve from
GALEX
TDS and PS1 MDSwas fitted with early emission following SN shock break-out in a red supergiant star (Gezari et al. 2010). Both ofthese spectroscopically classified extragalactic transientsare labled in Figure 15.Our 5 σ selection criteria translates to a limiting sen-sitivity to transients in a host galaxy with a magnitude m host of a magnitude of m trans = − . m host − σ ( m host) − . − m host − . ) , (8)which ranges from m trans ∼ . m host = 18 magto m trans ∼ . m host = 23 mag. Thus, our vari-ability selection threshold is less sensitive to transientsin host galaxies with bright N U V fluxes. On the red se-quence of galaxies, where M NUV ≈ − . Fig. 15.—
Maximum
NUV variability amplitude for sourceswithout a classification. Dashed blue line shows the median 5 σ variability selection function as a function of mean magnitude. Thenature of the optical host is indicated (point source, galaxy, ororphan), and flaring sources (including transients) are marked inyellow. z > .
05 one gets m host >
22 mag.However, star-forming galaxies on the blue sequence are2.5 mag brighter in the
N U V , and thus the host galaxybrightness can be a factor in reducing the senstivity tofaint transients. For example, our
GALEX
TDS 5 σ sam-ple does not include SN 2009kf, a luminous Type IIP SNin a star-forming galaxy at z = 0 .
182 which we reportedour
GALEX
TDS detection of in Botticella et al. (2010).This source varied at only the 4.25 σ level in the N U V during its peak. However, because this transient wasselected from a spatial and temporal coincidence witha PS1 transient alert, we could lower our threshold forvariability selection in the UV. The systematic selectionof SN and TDE candidates from the joint
GALEX
TDSand PS1 MDS transient detections will be presented infuture papers. DISCUSSION
Classification Demographics
Figure 16 shows a pie diagram of the source classifica-tions. Out of the total of 1078
GALEX
TDS sources,62% are classified as actively accreting supermassiveblack holes (quasars or AGN), and 10% as variable andflaring stars (including RR Lyrae, M dwarfs, and CVs).Note that the relative fraction of the different classes ofsources is sensitive to both their intrinsic magnitude dis-tribution, and the magnitude-dependent variability selec-tion function of the sample. Table 4 gives the
GALEX
TDS catalog, sorted by decreasing
N U V amplitude, withthe
GALEX
ID, R.A., Dec, low-state
N U V magnitude,maximum amplitude of
N U V variability ( | ∆ m max | ), in-trinsic variability ( σ int ), the structure function on day( S d ) and year ( S y ) timescales, the characteristics of the N U V light curve: flaring (F) or stochastically variable(V), the morphology of the matching optical host: point-source (pt) or extended (ext), the color classification ofthe matching optical host: RR Lyrae (RRL), M dwarfstar (Mdw), star (Star), or quasar (QSO), the archivalredshift, an X mark if there is a match with an archivalX-ray source, and finally the
GALEX
TDS classifica-tion: RR Lyrae (RRL), M dwarf star (Mdw), star (star),quasar (QSO), AGN, UV flaring source or UV transientsource with galaxy host (Galaxy Flare or Galaxy Trans),0 Gezari et al.
Fig. 16.—
Pie chart of
GALEX
TDS classifications: sources clas-sified as stars (Star), M dwarfs (Mdw), RR Lyrae (RRL), quasars(QSO), and active galactic nuclei (AGN), and sources with no clas-sification with galaxy hosts (Galaxy), no host (Orphan), and point-source hosts (Point).
UV flaring source or transient source with point-sourceoptical host (Point Flare or Point Trans) or UV flar-ing source or transient source with orphan optical host(Orphan Flare or Orphan Trans), stochastically vari-able source with a point-source optical host (Point Var),stochastically variable orphan (Orphan Var), or none ofthe above (?).In Figure 17 we show the cummulative surface densitydistribution of classified UV variable sources as a func-tion of peak magnitude (high(NUV)) and maximum am-plitude (max( | ∆ N U V | )). For the variable UV sources,these correspond to total surface densities of 8 . ± . . ± .
8, and 1 . ± . − for quasars, AGNs, andRR Lyrae, respectively. For the transient source, we cancalculate a total surface density rate, / ( area × t eff ),where t eff is the effective survey time at the cadence thatmatches the characteristic timescale of the transient. Forextragalactic transients such as young SNe and TDEs,which vary on a timescale of days, we use the time in-tervals for which the fields were observed with a cadenceof 2 . ± . GALEX
TDS sources with a galaxy host, this yields asurface density rate of 52 ±
38 deg − yr − for extragalac-tic transients. For M dwarfs which vary on timescalesshorter than an individual observation, we use the totalexposure time for each epoch. If we assume a survey witha cadence of 2 days and t exp = 1 . ±
10 deg − yr − . UV Variability Properties of Classified Sources
Various optical studies of rest-frame UV variabil-ity in high redshift quasars have demonstrated thatthe variability of AGN increases with decreasing restwavelength (di Clemente et al. 1996; Vanden Berk et al.2004; Wilhite et al. 2005). In Figure 18, we show his-tograms of σ int for the UV variable sources with clas-sifications. Quasars show a distribution of σ int with amean that is ∼ Fig. 17.—
Cummulative distribution of surface density ofUV variable sources with
GALEX
TDS classifications: M dwarfs(Mdw), RR Lyrae (RRL), quasars (QSO), active galactic nuclei(AGN), and extragalactic transients (GAL-Flare) as a functionof peak magnitude (high(NUV)) ( left ) and maximum amplitude(max( | ∆ NUV | )) ( right ). cal wavelengths from the SDSS Stripe 82 sample fromSesar et al. (2007). This effect is even more pronouncedin the magnitude of the structure function on yearstimescales (S y ), which has a mean that is 5 times largerthan the mean measured in the r -band ( λ eff = 6231)from Schmidt et al. (2010). This trend is consistent withthe wavelength dependent rise in variability amplitudeobserved in the structure function for quasars in the rest-frame UV (Vanden Berk et al. 2004) and observed UV(Welsh et al. 2011).The fact that AGN become bluer during high states offlux (Giveon et al. 1999; Geha et al. 2003; Gezari et al.2008a) has been attributed to increases in the charac-teristic temperature of the accretion disk in response toincreases in the mass accretion rate (Pereyra et al. 2006;Li & Cao 2008). However, Schmidt et al. (2012) arguethat the color variability observed in individual quasarsin their SDSS Stripe 82 sample is stronger than expectedfrom just varying the accretion rate ( ˙ M ) in accretiondisk models. In a future study, we will use simultaneousUV and optical light curves from GALEX
TDS and PS1MDS for our 358 individual quasars to test this resultwith a larger dynamic range in wavelength.For the subsample of 95 quasars with archival red-shifts ( z mean = 1 . σ z = 0 . σ int vs. the low-state N U V absolute magnitude, andfind a steep negative correlation fitted by log( σ int ) =(1 . ± .
1) + β . M NUV , where β = 0 . ± .
04, in ex-cellent agreement with the trend for increased variabil-ity in lower luminosity quasars seen from optical ob-servations with β = 0 . ± .
005 (Vanden Berk et al.2004), and shallower than expected for a Poissonian pro-cess which has β = 0 . z mean = 0 . σ z = 0 . σ int and low-state N U V absolute mag-nitude. This is most likely a result of dilution of thevariability amplitude from the contribution of the hostgalaxy in the
N U V .The largest values of | ∆ m | (plotted in Figure 10) arefound in RR Lyrae and M dwarfs, with a tail of largeamplitude variations reaching up to | ∆ m | = 2 . | ∆ m | = 4 . . − . N U V structure functionalso shows a weak dependence of amplitude on timescalewhen comparing the structure function on days to years
ALEX
Time Domain Survey 11
Fig. 18.—
Intrinsic
NUV variability as a function of low-state
NUV magnitude for sources classified as RR Lyrae (red), M dwarfs(yellow), quasars (blue) and AGN (green). Arrows shows the me-dian σ int measured in the optical for quasars (blue arrow), RRLyrae (red arrow), and M dwarfs (orange arrow) from Sesar et al.(2007). Fig. 19.—
Intrinsic
NUV variability as a function of low-state
NUV absolute magnitude, M NUV = low(
NUV ) − DM where DM is the distance modulus, for the subsample of quasars (blue dots)and AGN (green circles) with catalog redshifts. Dashed blue lineshows the fit to the quasars to log( σ int ) ∝ β . M . The spectro-scopically classified extragalactic transients (TDE PS1-10jh andSN 2010aq) are labeled. timescales, however, with an amplitude that is ∼ N U V than in the optical. This wavelengthdependence on variability amplitude can be explained ifthe variability is driven by variations in surface tempera-ture from pulsations of the stellar envelope, where higherstates of flux are associated with higher temperatures(Sesar et al. 2007). CONCLUSIONS
We provide a catalog of over a thousand UV variablesources and their classifications based on optical hostproperties, UV variability behavior, and cross-matches with archival X-ray and redshift catalogs. This yields asample of 53 M dwarfs, 37 RR Lyraes, 358 quasars, and305 AGN. We find median intrinsic UV variability am-plitudes in RR Lyrae and quasars that are factors of > GALEX
TDS enables us to systematically discover persistent andtransient (i.e. tidal disruption of a star) accreting super-massive black holes over wide fields of view, study thecontemporaneous UV and optical variability of variablestars, and catch young core-collapse supernovae withinthe first days after explosion. The overlap of the
GALEX
TDS with PS1 MDS and multiwavelength legacy surveyfields will continue to be helpful for classifying transientsources in these heavily observed fields. We also measurethe surface densities of variable sources and the surfacedensity rates for transients as a function of class in theUV for the first time.With
GALEX
TDS we are only scratching the surfaceof UV variability. Our 5 σ sample of 1078 sources is lessthan 0.3% of the 419,152 UV sources in the field (withan average density of 1 . × UV sources per squaredegree down to m lim = 23 mag). Looking to the future,the discovery rate for UV variable sources and UV tran-sients could increase by several orders of magnitude withthe launch of a space-based UV mission with a wide fieldof view (several deg ), a survey strategy of daily cadenceobservations over ∼
100 deg , and detectors with an or-der of magnitude improved photometric precision relativeto GALEX . In the optical sky, 90% of quasars vary with σ int > .
03 mag on the timescales of years (Sesar et al.2007). Given the factor of ∼ σ int observed forquasars in the N U V , one could achieve a nearly completesample of low-redshift quasars with photometric errors of σ ( m ) ∼ .
01 mag. In coordination with a ground-basedoptical survey, such as Pan-STARRS2 (Burgett 2012) orLSST , this could yield the simultaneous UV and op-tical detection of ≈ variable quasars and ≈ RRLyrae and M dwarfs, as well as increase the discoveryrate of UV-bright extragalactic transients (young SNeand TDEs) by a factor of ≈ GALEX (GalaxyEvolution Explorer) is a NASA Small Explorer, launchedin 2003 April. We gratefully acknowledge NASAs sup-port for construction, operation, and science analysis forthe
GALEX mission, developed in cooperation with theCentre National dEtudes Spatiales of France and theKorean Ministry of Science and Technology. The Pan-STARRS1 survey has been made possible through con-tributions of the Institute for Astronomy, the Universityof Hawaii, the Pan-STARRS Project Office, the Max-Planck Society and its participating institutes, the MaxPlanck Institute for Astronomy, Heidelberg and the MaxPlanck Institute for Extraterrestrial Physics, Garching,The Johns Hopkins University, Durham University, theUniversity of Edinburgh, Queen’s University Belfast, theHarvard-Smithsonian Center for Astrophysics, and theLas Cumbres Observatory Global Telescope Network, In-corporated, the National Central University of Taiwan, lsst.org/lsst/science/overview REFERENCESAihara, H., et al. 2011, ApJS, 193, 29Bianchi, L., Efremova, B., Herald, J., Girardi, L., Zabot, A.,Marigo, P., & Martin, C. 2011, MNRAS, 411, 2770Botticella, M. T., et al. 2010, ArXiv e-prints, 1001.5427Brown, P. J., et al. 2009, AJ, 137, 4517Budav´ari, T., & Szalay, A. S. 2008, ApJ, 679, 301Burgett, W. S. 2012, ArXiv e-prints, 1207.3460Chevalier, R. A., & Irwin, C. M. 2011, ApJ, 729, L6Chiappetti, L., et al. 2005, A&A, 439, 413Cid Fernandes, R., Sodr´e, Jr., L., & Vieira da Silva, Jr., L. 2000,ApJ, 544, 123Civano, F., et al. 2012, ArXiv e-prints, 1205.5030di Clemente, A., Giallongo, E., Natali, G., Trevese, D., & Vagnetti,F. 1996, ApJ, 463, 466Elvis, M., et al. 2009, ApJS, 184, 158Geha, M., et al. 2003, AJ, 125, 1Gezari, S., et al. 2008a, ApJ, 676, 944——. 2012, Nature, 485, 217——. 2008b, ApJ, 683, L131——. 2009, ApJ, 698, 1367——. 2006, ApJ, 653, L25——. 2010, ApJ, 720, L77Giacconi, R., et al. 2002, ApJS, 139, 369Giveon, U., Maoz, D., Kaspi, S., Netzer, H., & Smith, P. S. 1999,MNRAS, 306, 637Hasinger, G., et al. 2007, ApJS, 172, 29Hook, I. M., McMahon, R. G., Boyle, B. J., & Irwin, M. J. 1994,MNRAS, 268, 305Huber, M., et al. 2011, in American Astronomical Society MeetingAbstracts 218, 328.12Kaiser, N., et al. 2010, in Society of Photo-Optical InstrumentationEngineers (SPIE) Conference Series, Vol. 7733, Society of Photo-Optical Instrumentation Engineers (SPIE) Conference SeriesKowalski, A. F., Hawley, S. L., Hilton, E. J., Becker, A. C., West,A. A., Bochanski, J. J., & Sesar, B. 2009, AJ, 138, 633Laird, E. S., et al. 2009, ApJS, 180, 102Le F`evre, O., et al. 2005, A&A, 439, 845——. 2004, A&A, 428, 1043Lehmer, B. D., et al. 2005, ApJS, 161, 21Li, S.-L., & Cao, X. 2008, MNRAS, 387, L41Lilly, S. J., et al. 2009, ApJS, 184, 218——. 2007, ApJS, 172, 70Magnier, E. 2006, in The Advanced Maui Optical and SpaceSurveillance Technologies ConferenceManners, J. C., et al. 2003, MNRAS, 343, 293Martin, D. C., et al. 2005, ApJ, 619, L1Milne, P. A., et al. 2010, ApJ, 721, 1627Morrissey, P., et al. 2007, ApJS, 173, 682Nakar, E., & Sari, R. 2010, ApJ, 725, 904Nandra, K., et al. 2005, MNRAS, 356, 568 Newman, J. A., et al. 2012, ArXiv e-prints, 1203.3192Ofek, E. O., et al. 2010, ApJ, 724, 1396Pereyra, N. A., Vanden Berk, D. E., Turnshek, D. A., Hillier, D. J.,Wilhite, B. C., Kron, R. G., Schneider, D. P., & Brinkmann, J.2006, ApJ, 642, 87Pierre, M., et al. 2007, MNRAS, 382, 279Rabinak, I., & Waxman, E. 2011, ApJ, 728, 63Regnault, N., et al. 2009, A&A, 506, 999Rest, A., et al. 2005, ApJ, 634, 1103Richards, G. T., et al. 2002, AJ, 123, 2945Schawinski, K., et al. 2008, Science, 321, 223Schlafly, E. F., et al. 2012, ApJ, 756, 158Schlegel, D. J., Finkbeiner, D. P., & Davis, M. 1998, ApJ, 500, 525Schmidt, K. B., Marshall, P. J., Rix, H.-W., Jester, S., Hennawi,J. F., & Dobler, G. 2010, ApJ, 714, 1194Schmidt, K. B., Rix, H.-W., Shields, J. C., Knecht, M., Hogg,D. W., Maoz, D., & Bovy, J. 2012, ApJ, 744, 147Sesar, B., et al. 2010, ApJ, 708, 717——. 2007, AJ, 134, 2236Stubbs, C. W., Doherty, P., Cramer, C., Narayan, G., Brown, Y. J.,Lykke, K. R., Woodward, J. T., & Tonry, J. L. 2010, ApJS, 191,376Surace, J. A., et al. 2004, VizieR Online Data Catalog, 2255, 0Svirski, G., Nakar, E., & Sari, R. 2012, ApJ, 759, 108Szkody, P., et al. 2011, AJ, 142, 181Tonry, J. L., et al. 2012, ApJ, 750, 99Trammell, G. B., Vanden Berk, D. E., Schneider, D. P., Richards,G. T., Hall, P. B., Anderson, S. F., & Brinkmann, J. 2007, AJ,133, 1780Trump, J. R., et al. 2009, ApJ, 696, 1195——. 2007, ApJS, 172, 383Vanden Berk, D. E., et al. 2004, ApJ, 601, 692Virani, S. N., Treister, E., Urry, C. M., & Gawiser, E. 2006, AJ,131, 2373Voges, W., et al. 1999, VizieR Online Data Catalog, 9010, 0——. 2000, VizieR Online Data Catalog, 9029, 0Welsh, B. Y., et al. 2005, AJ, 130, 825Welsh, B. Y., Wheatley, J. M., & Neil, J. D. 2011, A&A, 527, A15Welsh, B. Y., et al. 2007, ApJS, 173, 673West, A. A., et al. 2011, AJ, 141, 97Wheatley, J., Welsh, B. Y., & Browne, S. E. 2012, PASP, 124, 552Wheatley, J. M., Welsh, B. Y., & Browne, S. E. 2008, AJ, 136, 259Wilhite, B. C., Vanden Berk, D. E., Kron, R. G., Schneider, D. P.,Pereyra, N., Brunner, R. J., Richards, G. T., & Brinkmann, J. V.2005, ApJ, 633, 638Wyder, T. K., et al. 2007, ApJS, 173, 293Yanny, B., et al. 2009, AJ, 137, 4377
ALEX
Time Domain Survey 13
TABLE 1
GALEX
TDS and PS1 MDS
Survey Field of View Plate Xcale PSF FWHM m lim per Epoch Cadence Seasonal Visibility(deg) (arcsec/pixel) (arcsec) (mag) (days) (months) GALEX
TDS 1.1 1.5 5.3 23.3 2 ∼ ∼ TABLE 2
GALEX
TDS Fields
Name R.A. (J2000) Dec (J2000) E ( B − V ) Epochs(deg) (deg) (mag)PS XMMLSS MOS00 35.580 − − − − − − − − − − − − − − − − − TABLE 3
GALEX
TDS Classifications
Archive PS1 ClassificationStep N unclass pt ext pt ext orphan QSO RRL Mdw Star AGN pt galoptical match 1078 487 391 76 103 21QSO color cut 1078 326 326RRL color cut 753 37 37Mdw color cut 716 44 9 53stellar locus color cut 663 17 17 S yr /S d ≥ TABLE 4
GALEX
TDS Catalog a NUV Optical X-ray ClassID R.A. Dec m low | ∆ m max | σ int S d S yr LC Morph r AB Color z GROTH MOS01-21 216.1622 54.0911 22.54 4.60 1.04 0.80 1.13 V pt 14.92 Mdw MdwVVDS22H MOS05-05 333.8326 -0.5491 21.14 4.47 0.99 0.90 0.74 F pt 21.24 QSO CVELAISN1 MOS15-02 242.0397 54.3586 > > > > > > > > · · · > > · · · · · · > > < aa