Capability of Quasar Selection by Combining the SCUSS and SDSS Observations

2.7 MB·English

by  Hu Zou

Capability of Quasar Selection by Combining the SCUSS and
SDSS Observations
5
Hu Zou 1, Xue-bing Wu 2,3, Xu Zhou 1, Shu Wang 2, Linhua Jiang3, Xiaohui Fan 4,3, Zhou
1
0 Fan 1, Zhaoji Jiang 1, Yipeng Jing 5, Michael Lesser 4, Cheng Li 6, Jun Ma 1, Jundan Nie
2 1, Shiyin Shen 6, Jiali Wang 1, Zhenyu Wu 1, Tianmeng Zhang 1, Zhimin Zhou 1

n
a
J

4
1 ABSTRACT

] The South Galactic Cap u-band Sky Survey (SCUSS) provides a deep u-band imaging of
A about 5000 deg2 in south Galactic cap. It is about 1.5 mag deeper than the SDSS u-band. In
G thispaperweevaluatethecapabilityofquasarselectionusingbothSCUSSandSDSSdata,based
. onconsiderationsof the deep SCUSS u-bandimagingandtwo-epochu-bandvariability. We find
h
p that the combination of the SCUSS u-band and the SDSS griz band allows us to select more
- faint quasars and more quasars at redshift around 2.2 than the selection only with the SDSS
o
ugriz data. Quasars have significant u-band variabilities. The fraction of quasars with large
r
t two-epoch variability is much higher than that of stars. The selection by variability can select
s
a both low-redshift quasars with ultraviolet excess and mid-redshift (2 < z < 3.5) quasars where
[ quasarselectionby optical colorsis inefficient. The abovetwo selections are complementary and

make full use of the SCUSS u-band advantages.
1
v
Subject headings: galaxies: active quasars — methods: statistical
2
5
1. Introduction homogeneous data for studying the properties of
3
3 quasars and other subjects including connections
0 Quasars,identified ashigh-redshiftobjects,are to their host galaxies, IGM, Lyα forests, quasar
1. very distant according to the Hubble’s law. The clustering, central supermassive black holes, and
absorptionspectraofquasarsclosetothereioniza-
0 baryon acoustic oscillations (Dunlop et al. 2003;
5 tion epoch are the best probes to study the inter- Fanetal.2006;Rossetal.2009;Shen 2009;Daw-
1 galactic medium (IGM) and reionization (Fan et son et al. 2013), etc.
: al. 2002). In recent years, large-scale optical sur-
v Quasar targets for spectroscopic observations
i veys such as the Sloan Digital Sky Survey (SDSS;
X in the SDSS are mainly selected by optical col-
York et al. 2000) have provided numerous and
ors (Richards et al. 2002). Outliers away from
r
a the star sequence in color space are considered as
1KeyLaboratoryofOpticalAstronomy,NationalAstro-
spectroscopicfollow-upcandidates. Due to strong
nomical Observatories,ChineseAcademy ofSciences, Bei-
jing,100012, China;[email protected] ultravioletexcessandemissionlines, quasarswith
2DepartmentofAstronomy,PhysicsSchool,PekingUni- redshift lower than 2.2 or higher than 3.0 are ob-
versity,Beijing100871,China viously far from the star locus in color-color di-
3Kavli Institute for Astronomy and Astrophysics, agrams. A lack of or not so deep u band would
PekingUniversity,Beijing100871,China
significantly impact the ability to select the low-
4Steward Observatory, University of Arizona, Tucson,
redshift quasars. Optical colors of quasars in the
AZ85721
5Center for Astronomy and Astrophysics, Department redshiftrangeof2.2<z <3.0aresimilartothose
ofPhysicsandAstronomy,ShanghaiJiaoTongUniversity, of A-colored stars (blue horizontal branch stars
Shanghai 200240, China andblue stragglers)so that it is very difficult and
6ShanghaiAstronomicalObservatory,ChineseAcademy inefficient to select quasars only by simple color
ofSciences, Shanghai 200030,China cuts (Fan 1999; Richards et al. 2002; Wu et al.
1
2011). There are two alternatives that can help usefulfor quasarselection. Inthis paper,we com-
to pick out those quasar targets more efficiently. parethequasarselectionbycombiningtheSCUSS
One is combining optical and near-infrared(NIR) and SDSS data with that only by the SDSS data
photometry, e.g. the Two Micron All Sky Survey and try to understand the quasar targeting po-
(2MASS; Skrutskie et al. 2006), UKIRT Infrared tential and capability when the SCUSS u band is
DeepSkySurvey(UKIDSS;Lawrenceetal.2007), involved. In Section 2, we give brief descriptions
and Wide-field Infrared Survey Explorer (WISE; of the SCUSS and our quasar and star samples.
Wright et al. 2010). The decreasing rate of a stel- Section 3 presents some advantagesof the SCUSS
lar continuum from optical to NIR wavelength is u-bandphotometricdata,whicharefavourablefor
larger than that of a quasar. Optical colors to- the quasar selection. Quasar selections by com-
gether with infrared colors are utilized to distin- bining SCUSS and SDSS data are analyzed and
guish quasarsat z >2.2 (Maddox et al. 2008;Wu discussedinSection4. Section5istheconclusion.
et al. 2011, 2012).
The optical variability is the other way to se- 2. Data
lect quasar candidates. Its role will become more
2.1. SCUSS data
and more prominent with the ongoing or up-
coming time-domain surveys such as the Palo- The SCUSS is undertaken by the National As-
mar Transient Factory (PTF; Law et al. 2009), tronomical Observatories of China. The adopted
Panoramic Survey Telescope & Rapid Response telescope is the 2.3 m Bok telescope located on
System (PAN-STARRS; Kaiser 2004), and Large Kitt Peak, Arizona. The camera deployed at the
Synoptic Survey Telescope (LSST; LSST Science prime focus provides a field of view of about 1
Collaboration et al. 2009). Light curves have square degree. The photometric system is the
been parameterized by a structure function or SDSS-like u filter, whose effective wavelength is
a damped random walk model (Schmidt et al. about 3538˚A andFWHM is about 520˚A (Zou et
2010;MacLeodet al. 2011;Butler & Bloom 2011; al. 2014). ThefilterisalittlebluerthantheSDSS
Palanque-Delabrouilleet al. 2011). Quasars,vari- u band. The SCUSS coversanareaof about 5000
ables and non-variable stars can be successfully square degrees in the SGP. The exposure time is
separatedwithagreatcompletenessandpurityin 5 minutes and the 5-σ magnitude limit is deeper
the parameter space. However, quasar targeting than 23.0 mag.
bylightcurvesneedstensoftimesamplingpoints,
The detailed image processing and photome-
and only limited sky areas have deep time-series
try for the SCUSS can be referred to the paper
photometric observations. The u-band variabil-
of Zou et al. (2014). In general,the global astro-
ity is the largest among the five SDSS bands and ′′
metric accuracy is about 0.13 . The catalogs in-
it becomes larger as the time lag is longer. The
cludebothphotometryforstackedimagesandco-
magnitude change in u band might be helpful for
added photometry for single-epoch images. The
quasar targeting selection.
co-added PSF magnitudes and co-added aperture
The SCUSS is an imaging survey in the south magnitudes, if a proper aperture radius is cho-
Galactic cap (SGP) with an SDSS-like u fil- sen,arethebestbrightnessmeasurementsofpoint
ter(Zhou et al. 2014). The survey is about 1.5 sources. In this paper, we use the co-added PSF
mag deeper thanthe SDSS u band. Itis expected magnitude. The SCUSS magnitude is converted
that, comparing with target selections only by to the SDSS photometric system using the trans-
SDSS photometry, one can select more quasars formation equation of
with less star contamination by using both deep
2
SCUSS u band and SDSS other bands if objects uSDSS =uSCUSS+0.0586(uSCUSS−g)−0.0207(uSCUSS−g) −0.0377,
present no variability. The SCUSS observations (1)
started in 2010 and ended in 2013. The average where 0.8<uSCUSS−g <2.8. The corresponding
time lag between SCUSS and SDSS observations error of the transformed SCUSS u-band magni-
is about 2–3 years. We can obtain the two-epoch tude is estimated by error transfer. This trans-
u-bandvariabilitybetweenthesetwosurveys. The formationisderivedbypointsourceswithSCUSS
variability difference between quasars and stars is and SDSS photometric errors less than 0.05 mag.
2
It is applied to point sources here and the max-
imum systematic difference between SCUSS and
SDSS is about 0.036 mag. The SDSS magnitudes
are PSF magnitudes. In the rest of this paper, all
magnitudes are corrected by the Galactic extinc-
tion map of Schlegel et al. (1998).
2.2. Quasar and star samples
The spectroscopically confirmed quasars and
stars are obtained by matching objects from the
SDSS DR10 with SCUSS catalogs. We require
thatSCUSSobjectsarenotsaturatedandnotpol-
lutedbyothersourcesandtheSDSSclassifications
arepoint-like. Thei-bandmagnitudeislimitedto
the range between 18 and 22.5 mag. There are
42418 quasars and 56518 stars in total. Figure Fig. 1.—(a)and(b)aredistributionsoftheuSDSS
1 shows some properties of these two samples in- magnitudeanduSDSS−g colorofthe starsample.
cluding the distributions of the uSDSS magnitude, (c), (d), and (e) are distributions of the uSDSS
uSDSS −g color, and redshift (only for quasars). magnitude, uSDSS − g color, and redshift of the
Therearetwopeaksinthe redshiftdistributionof quasar sample, respectively.
quasars, locating in the ultraviolet-excess region
and at intermediate redshift, respectively. Most
mid-redshift quasars are obtained by the Baryon
Oscillation Spectroscopic Survey (BOSS; Dawson
et al. 2013), which plans to map the large-scale
structure traced by the Ly-α forest.
3. SCUSS Advantages for Quasar Selec-
tion
3.1. Deeper SCUSS photometry
The SCUSS u band is reported to be 1.5 mag
deeper than the SDSS u band. We plot the mag-
nitude error as a function of magnitude in Figure
2. The 5-σ magnitude limits at σ = 0.2 for the
SCUSS and SDSS u bands are 23.45 and 22.03
mag, respectively. The magnitude error of the
SDSSu-bandat22.0mag,officiallydefinedasthe
limiting magnitude of95%detectionrepeatability
Fig. 2.— Median magnitude error as a function
for point sources, is about 0.2 mag , while that of
of magnitude. The data come from a random
the SCUSS u-band is about 0.05 mag.
SCUSS region with typical imaging depth. Here
Another intuitive comparisonof the data qual-
magnitudes are the ones without Galactic extinc-
ity between these two surveys is to plot color-
tioncorrections. The blue curveisforthe SDSSu
colordiagramsandseethecolordispersionoffaint
band, while the red one is for the SCUSS u-band.
main-sequence stars. Figure 3 shows the distri-
Thedashedhorizontalandverticallinesarecorre-
butions of both quasar and star samples in the
sponding to the photometric error of 0.2 mag and
color-color plane of uSDSS −g or uSCUSS −g vs. the magnitude of 22.0 mag.
g−r. The colordispersionofstarswith SCUSSu
is obviously smaller than that with SDSS u. The
uSCUSS−g dispersion at g−r =1.4 (mostly faint
3
M stars) is about 0.55, while the uSDSS −g dis- distributions for stars and quasars are about 0.06
persion is about 0.76. In addition, in the region mag and 0.3 mag, respectively. The scatter of
occupied by quasars with z >2.5, there are many stars mainly comes from the photometric error,
starsinitiallyselectedasquasarcandidatesbythe while the scatterfor quasarsmightcome fromthe
SDSS as shown in the left of Figure 3. They are photometricerror,intrinsicvariability,andphoto-
mostly faint stars with large SDSS u-band pho- metric system difference between the SCUSS and
tometric errors. However, most of these stars are SDSS. Anyhow, the difference of the distributions
stillinthemainsequenceduetodeeperSCUSSu- between quasars and stars are distinct. It can be
band as seen in the rightpanel of this figure. The used to separate these two kinds of objects. The
deeperSCUSSubandwithSDSSotherbandswill large two-epoch magnitude differences of quasars
evidently improve the quasar selection if quasars are also implied in the broader uSCUSS −g color
present no variability. distributions in the right of Figure 3.
3.2. Two-epoch variability 3.3. Larger photometric system difference
for quasars
As seen in Figure 3, many quasars with 2 <
z <3 are mixed with A-colored stars. The target The SCUSS u filter is very similar to the SDSS
selections only based on optical colors are ineffi- u filter, but it is a little bluer. The central wave-
cient for these quasars. However, quasars usually lengthoftheSCUSSubandisabout3538˚A,while
present light variabilities. In principle, the vari- that of the SDSS u is about 3562 ˚A (Zou et al.
abilitywouldbelargerasthewavelengthisshorter 2014). Stellar spectra are dominated by the con-
andtheobservationlagislonger. Theu-bandvari- tinuum, so their photometric differences between
abilityislargestamongallSDSSbands. Thereisa the SCUSS andSDSS asshowninEquation(1)is
typical observation time lag of 2–3 years between small. Thephotometriceffectduetodifferentpho-
the SCUSS and SDSS. Thus, the two-epoch u- tometric systems is less than 0.036 mag for main-
bandvariabilitybetweenthesetwosurveysshould sequence stars. However, quasar spectra present
be large enough as a useful tool to select quasar strongultravioletexcessandemissionlines. Their
targets. Thequasarselectionbyvariabilityshould photometric differences should be much bigger.
also be complementary to other selection meth- We get 4000 quasar spectra with different red-
ods based on colors, especially for quasars with shifts from the SDSS DR10. The wavelength of
2<z <3. SDSS spectra ranges from 3800 to 9200 ˚A, which
We compare the two-epoch magnitude differ- is out of the u-band wavelength coverage. These
ences of both stars and quasars in Figure 4. Only quasar spectra are blue-shifted by z =0.3 so that
objectswithphotometricerrorslessthan0.05mag they can be convolved with two u filter responses
are considered. The standard deviations of the to generate synthetic magnitudes. Here, both
SCUSS and SDSS filter responses are the ones
withatmosphericextinctionatthetypicalairmass
of 1.3 (Zou et al. 2014). The resulting magni-
tude differencebetweenthesetwofiltersisplotted
as a function of redshift in Figure 5. In another
way,weredshiftthequasarcompositespectrumby
different redshift values and check the systematic
variation with redshift as also shown in Figure 5.
Thesetwodatasetspresentacoincidentvariation.
Most quasars lie in a narrow band along the red-
shift. The wave-shapevariationalongthe redshift
is because of different emission lines entering and
Fig. 3.—QuasarandstarsamplesintheuSDSS−g
departingfromtheuband. Themagnitudediffer-
anduSCUSS−g vs. g−r diagrams. Blackdotsare
ence is less than 0.03 when z <2.0. It goes up to
starsandpointswithcolorsarequasarscoloredby
about 0.1 mag when z > 2.0, which is caused by
redshift.
the strongest Lyα line in quasar spectra. In con-
4
trast with stars, it is evident that quasars present
much more difference induced by different photo-
metric filters. This kindofdifference ishelpful for
discriminating quasars from stars. Thus, we do
not need to correct it and assume the magnitude
differencemainly comefromintrinsicvariabilityif
the photometric error is ignorable.
3.4. Quasar variability independent on
redshift
Figure 6 shows photometric magnitude differ-
ences of quasars between the SCUSS and SDSS u
bands as a function of redshift. The photometric
errorsarelimitedto0.1mag,whichapproximately
corresponds to the quasar redshift up to about 3.
We find that the average quasar variability varies
at a levelof 0.04mag in the redshift range of0–3.
Thus,thetwo-epochvariabilityisalmostindepen-
dent on the redshift, which can help to discover
morequasarswith0<z <3.5wherevariabilityis Fig. 4.— Distributions of magnitude differences
still available. The independence of variability on between the SCUSS and SDSS for quasars (solid
redshiftingribandswasalsoconfirmedbyZuoet line) and stars (dot-dashed line), which are nor-
al. (2012) who divided the quasars into different malized to their maximums. The uSCUSS and
subsamples with different physical parameters. uSDSS magnitudeerrorsarelimitedtobelessthan
0.05 mag.
4. Quasar Selection and Analysis
4.1. Ignorable variability
Quasar target selections in this section are in-
vestigated based on the consideration of whether
objects presenting variability. Those objects with
two-epoch variability less than three times the
photometric errors are regarded as sources with
ignorable or small variabilities, which can be ex-
pressed as
∆m=|uSCUSS−uSDSS|≤3σ, (2)
where σ = pσu2SCUSS +σu2SDSS and σuSCUSS and
σuSDSS are the SCUSS and SDSS u-band magni-
tudeerrors,respectively. Thereare68.4%quasars Fig. 5.— Synthetic magnitude difference between
and96.9%starsoftheirtotalsampleswith∆m≤ the SCUSS and SDSS as a function of redshift.
3σ. TheseobjectshavedeeperSCUSSubandand The magnitudes are calculated by convolving the
hencedeeperintrinsiccolors. Itisimaginablethat quasarspectrablue-shiftedbyz =0.3intheSDSS
the SCUSS u band instead of the SDSS u band DR10 with corresponding filter responses. The
would undoubtedly improve the efficiency of the red solid line with error bars present the means
quasar selection when combined with SDSS other and standard deviations in different redshift bins.
bands. The green dashed curve shows the magnitude dif-
ference derived by the composite quasar spectra.
In order to show the advantage of the quasar
selection based on SCUSS data, we introduce
5
a flux-based quasar target selection algorithm,
XDQSO,whichappliestheextreme-deconvolution
method to evaluate the probabilitythat anobject
is a quasar (Bovy et al. 2011). By applying the
XDQSO method to the above non-variable sam-
ples,we canknow howmanymorequasarscanbe
selected with SCUSS u and SDSS griz data than
with SDSS-only data. The probability of an ob-
ject that selected as a quasar in XDQSO is set to
be larger than 0.9.
AftertheXDQSOmethodisappliedtothenon-
variable samples with SCUSS u and SDSS griz
photometric data, about 46.6% quasars and 2.9%
stars are selected. When it is applied to SDSS
ugriz data, about 41.4% quasars and 2.4% stars
are selected. There are about 24.0% quasars that
are exclusively selected by XDQSO with SCUSS
Fig. 6.— Photometric magnitude differences of
u and SDSS griz data. These numbers are sum-
quasars between the SCUSS and SDSS as a func-
marized in Table 1. Figure 7 shows the magni-
tionof the spectroscopicredshift. The overlapped
tude andredshiftdistributions ofquasarsselected
redline witherrorbarsshowtheaverageswith1σ
by these two methods. The distributions of to-
scatters in different redshift bins.
tal samples with ignorable variabilityand quasars
exclusively selected by the XDQSO with SCUSS
data are also overlaid in this figure. The tar-
get selection with SCUSS data can select more
faint quasars with magnitude peak at 21.5 mag
and more quasars with redshift between 2 and
3 (peak at z = 2.2). We also check the quasar
selection efficiencies within two different redshift
ranges: 0<z <2 and 2<z <3.5, which are also
presented in Table 1. As a result, there are re-
spectively15.6%and31.2%morequasarsthatare
selected by the XDQSO with SCUSS data than
that with SDSS-only data. The selection here is
based on known quasar samples, most of which
are selected through SDSS optical colors. The
SCUSS u band is deeper than that of the SDSS,
whichimpliesthatapartofpotentialquasarsclose
to the SDSS u-band magnitude limit are miss-
ing. The SCUSS data can help us find more faint Fig. 7.—Magnitude (a)andredshift(b)distribu-
quasars than the above experiment only based on tions of quasars selected by the XDQSO method
the known samples. respectively applied to SCUSS u plus SDSS griz
data (green) and SDSS-only data (blue). Distri-
4.2. Obvious variability butionsofthetotalquasarsampleswithignorable
variability(black)andquasarsexclusivelyselected
Objects with ∆m > 3σ are considered as
byXDQSOwithSCUSSplusSDSSdata(red)are
sources with obvious or large variability. About
also overlapped.
31.6% quasars and only 3.1% stars have obvious
two-epoch variability. The fraction of stars hav-
ing large variability is much smaller than that of
quasars,so the objects with ∆m>3σ is regarded
6
Table 1: Quasar selection for objects with ignorable variability
Star Quasar 0<z <2 2<z <3.5
Total 56518 42418 20074 21366
∆m≤3σ 54744 (96.9%) 29011 (68.4%) 11458 (57.1%) 16610 (77.7%)
SCUSS u + SDSS griz 1570 (2.9%) 13511 (46.6%) 5998 (52.3%) 7274 (43.8%)
SDSS ugriz 1335 (2.4%) 12024 (41.4%) 5548 (48.4%) 6266 (37.7%)
Exclusive ··· 2884 (24.0%) 865 (15.6%) 1958 (31.2%)
Note.—The 2nd and 3rd columns show the selection efficiencies with the total star and quasar samples. The 4th and 5th
columnsshowthequasarselectionefficiencieswithintwodifferentredshiftranges. The2ndrowgivessamplenumbers. The3rd
rowgivescorrespondingsampleswithignorablevariability(∆m≤3σ). The3rdand4throwsrespectivelypresenttheselection
efficiencies oftwotargetselections: XDQSOwithSCUSSuplusSDSSgriz data andXDQSOwithSDSS-onlydata. Thelast
rowgives thequasarsexclusivelyselectedbyXDQSOwithSCUSSuandSDSSgriz datarelativetoXDQSOwithSDSS-only
data.
as quasar candidates. We compared the selec- strate the capability of our quasar selection with
tionby the two-epochvariabilitywith that by the fullpointsourcesfromtheSDSS.Thequasarprob-
XDQSO method applied to the SDSS ugriz data abilityinXDQSOis stillsettobe largerthan0.9.
of large-variability samples. About 80% quasars The SCUSS u-band apparent magnitude is limit-
and11.2%starsareselectedbyXDQSO.Thereare ted to 23.5 mag. The i-band magnitude is be-
25.7% quasars exclusively selected by variability. tween18 and22.5mag. Isolatedsourcesclassified
ThesenumbersaresummarizedinTable2. Figure as point-like by the SDSS are considered. When
8 shows the magnitude and redshift distributions objects present ignorable variability, the selection
of quasars obtained by these two selections. The with SCUSS and SDSS data can find 29% more
selection by two-epoch variability can select more quasar targets than the one only with the SDSS
faint quasars with magnitude peak at 20.5 mag. data. Ontheotherhand,theselectionbyvariabil-
It also can select more low-redshift quasars with ity can find 168.7% additional targets, although
ultravioletexcess(peakatz =0.7)andmoremid- star comtaminationwould be more serious. In to-
redshift quasarswith 2<z <3 (peak at z =2.5). tal, we find that our selection can select 87.7%
There are 17.3% and 43.3% quasars exclusively additional quasar targets. However,its actual ap-
selected by variability for the two redshift ranges plication with alterable parameters, such as the
asspecifiedinthe previoussection,whicharealso quasar probability in the XDQSO and the vari-
shown in Table 2. ability amplitude (several times larger than the
The SDSS u-band photometric error is larger photometric error), should be dependent on the
than the SCUSS u band, which is more promi- requirements of the density and homogeneity of
nent close to the SDSS magnitude limit. So the quasar targets, the selection efficiency, and mag-
selection by variability is limited to a brighter nitude range, etc.
magnitude range than the XDQSO selection with
SCUSS u plus SDSS griz data. The XDQSO se- 5. Conclusion
lection with the SCUSS plus SDSS data can se-
The SCUSS provides a deep and wide imaging
lectmorefainterquasarsandquasarswithredshift
survey in an about 5000 deg2 area of the south
peakedat2.2,whiletheselectionbyvariabilitycan
Galactic cap with u band. The SCUSS u-band is
selectmore low-redshiftandmid-redshift quasars.
deeperthantheSDSSubandandtheobservation
Thesetwoselectionsarecomplementaryandmake
time lag between the SCUSS and SDSS provides
the most of the SCUSS u-band advantages.
an opportunity to investigate the two-epoch vari-
ability. Throughcombiningthe SCUSSandSDSS
4.3. Application of our selection
data, we evaluate the capability of the quasar se-
We randomly select a region of about 10 deg2 lection in consideration of the advantages of the
withconsistentSCUSSimagingqualitytodemon-
7
Table 2: Quasar selection for objects with obvious variability
Star Quasar 0<z <2 2<z <3.5
Total 56518 42418 20074 21366
∆m>3σ 1774 (3.1%) 13407 (31.6%) 8616 (42.9%) 4756 (22.3%)
SDSS ugriz 199 (11.2%) 10669 (79.6%) 7346 (85.3%) 3318 (69.8%)
Exclusive ··· 2738 (25.7%) 1270 (17.3%) 1438 (43.3%)
Note.—Similar to Table 1, but for quasars with large variability. Objects with the two-epoch variability ∆m > 3σ are
considered as quasar targets. The last row gives the quasars exclusively selected by variability relative to the selection by
XDQSOwithSDSS-onlydata.
SCUSS u band data. exclusively selected by variability. The above two
The deeper SCUSS u-band photometry makes quasar selections are complementary, which make
stars and quasars more tightly distributed in the fulluseoftheSCUSSadvantages. Whenapplying
color-color diagrams if SDSS other bands are in- to the full SDSS point sources, our method can
cluded and objects present no variability. It helps also select about 87.7%additional quasar targets.
to select more fainter quasars. Quasars have
power-law continua and many strong emission This work is supported by the National Nat-
lines. The photometric difference for quasars due ural Science Foundation of China (NSFC, Nos.
to different photometric systems is larger than 11203031,11433005,11073032,11373035,11203034,
that of stars, which is helpful for the separation 11303038,11303043,11033001,11373008),by the
of these two objects. The average variability of National Basic Research Program of China (973
quasars is about 0.3 mag. The distribution of Program, Nos. 2014CB845704, 2013CB834902,
the two-epochvariabilitybetweenthe SCUSSand 2014CB845702, and 2014CB845705), and by the
SDSS for quasars is quite different from that of Strategic Priority Research Program ”The Emer-
stars. Besides, the quasar two-epoch variability gence of Cosmological Structures” of the Chinese
is independent on redshift at a level of 0.04 mag Academy of Sciences(No. XDB09000000). Z. Y.
(0<z <3). Wu wassupportedby the Chinese NationalNatu-
Basedonthe aboveadvantages,weanalyzethe ral Science Foundation grant No. 11373033. This
quasarselectionintwoways: oneiscombiningthe work was also supported by the joint fund of As-
SCUSS deeper u-band data and SDSS griz data tronomy of the National Nature Science Founda-
when objects present ignorable variability; the tionofChinaandtheChineseAcademyofScience,
other is utilizing the two-epoch variability. The underGrantsU1231113. TheSCUSSisfundedby
XDQSO method, which gives the quasar proba- the Main Direction Program of Knowledge In-
bility,isintroducedforcomparisons. Wefindthat novation of Chinese Academy of Sciences (No.
theXDQSOmethodwithSCUSSuandSDSSgriz KJCX2-EW-T06). It is also an international co-
data can select more faint quasars and quasars operative project between National Astronomical
with redshift around 2.2 than the XDQSO with Observatories, Chinese Academy of Sciences, and
SDSS ugriz data. There areabout 24.0%quasars StewardObservatory,UniversityofArizona,USA.
exclusively selected by the former. The SCUSS Technical supports and observational assistances
data can also help us find the missing quasars ofthe Bok telescope areprovidedby StewardOb-
close to the SDSS u-band magnitude limit. We servatory. TheprojectismanagedbytheNational
regard objects with ∆m > 3σ as quasar candi- AstronomicalObservatoryofChinaandShanghai
dates with obvious variability, because there are Astronomical Observatory.
relatively small fraction of stars with such large SDSS-III is managed by the Astrophysical Re-
variability. Thequasarselectionbyvariabilitycan search Consortium for the Participating Insti-
select both low-redshift quasars and mid-redshift tutions of the SDSS-III Collaboration including
quasars with 2<z <3. There are 25.7% quasars the University of Arizona, the Brazilian Partici-
8
pation Group, Brookhaven National Laboratory, Palanque-Delabrouille, N., Yeche, C., Myers,
CarnegieMellonUniversity,UniversityofFlorida, A. D., et al. 2011, A&A, 530, A122
the FrenchParticipationGroup,the GermanPar-
Richards, G. T., Fan, X., Newberg, H. J., et al.
ticipation Group, Harvard University, the Insti-
2002,AJ, 123, 2945
tuto de Astrofisica de Canarias, the Michigan
State/Notre Dame/JINA Participation Group,
Ross, N. P., Shen, Y., Strauss, M. A., et al. 2009,
JohnsHopkins University,LawrenceBerkeleyNa-
ApJ, 697, 1634
tional Laboratory, Max Planck Institute for As-
trophysics, Max Planck Institute for Extraterres- Schlegel, D. J., Finkbeiner, D. P., & Davis, M.
trial Physics, New Mexico State University, New 1998,ApJ, 500, 525
York University, Ohio State University, Pennsyl-
Schmidt, K.B., Marshall,P.J., Rix, H.-W., etal.
vania State University, University of Portsmouth,
2010,ApJ, 714, 1194
Princeton University, the Spanish Participation
Group, University of Tokyo, University of Utah,
Shen, Y. 2009, ApJ, 704, 89
VanderbiltUniversity,UniversityofVirginia,Uni-
versity of Washington, and Yale University. Skrutskie, M. F., Cutri, R. M., Stiening, R., et al.
2006,AJ, 131, 1163
REFERENCES
York, D. G., Adelman, J., Anderson, J. E., Jr., et
Bovy,J.,Hennawi,J.F.,Hogg,D.W.,etal.2011, al. 2000, AJ, 120, 1579
ApJ, 729, 141
Wright, E. L., Eisenhardt, P. R. M., Mainzer,
Butler, N. R., & Bloom, J. S. 2011, AJ, 141, 93 A. K., et al. 2010, AJ, 140, 1868
Dawson, K. S., Schlegel, D. J., Ahn, C. P., et al. Wu, X.-B., Wang, R., Schmidt, K. B., et al. 2011,
2013,AJ, 145, 10 AJ, 142, 78
Dunlop,J.S.,McLure,R.J.,Kukula,M.J.,etal. Wu, X.-B., Hao, G., Jia, Z., Zhang, Y., & Peng,
2003,MNRAS, 340, 1095 N. 2012,AJ, 144, 49
Fan, X. 1999,AJ, 117, 2528 Zhou, X., Zhao, Z.-J, Zou, H., et al. 2014, in pre-
paresion.
Fan, X., Narayanan, V. K., Strauss, M. A., et al.
2002,AJ, 123, 1247 Zou, H., Zhao, Z.-J, Zhou, X., et al. 2014, in pre-
paresion.
Fan,X.,Strauss,M.A.,Becker,R.H.,etal.2006,
AJ, 132, 117 Zuo, W., Wu, X.-B., Liu, Y.-Q., & Jiao, C.-L.
2012,ApJ, 758, 104
Kaiser, N. 2004, Proc. SPIE, 5489, 11
Law, N. M., Kulkarni, S. R., Dekany, R. G., et al.
2009,PASP, 121, 1395
Lawrence, A., Warren, S. J., Almaini, O., et al.
2007,MNRAS, 379, 1599
LSSTScienceCollaboration,Abell,P.A.,Allison,
J., et al. 2009, arXiv:0912.0201
MacLeod,C.L.,Brooks,K.,Ivezi´c,Zˇ.,etal.2011,
ApJ, 728, 26
Maddox, N., Hewett, P. C., Warren, S. J., &
Croom, S. M. 2008, MNRAS, 386, 1605
This2-columnpreprintwaspreparedwiththeAASLATEX
macrosv5.2.
9
Fig. 8.—Magnitude (a)andredshift(b)distribu-
tionsofquasarsselectedbythetwo-epochvariabil-
ity (black) and XDQSO method with SDSS-only
data(blue). Distributionsofthetotalquasarsam-
ples with obvious variability (black) and quasars
exclusively selected by variability (red) are also
plotted.
10