The XXL Survey. XXVII. The 3XLSS point source catalogue

18 Pages·2017·1.9 MB·English

A&A620,A12(2018) Astronomy
https://doi.org/10.1051/0004-6361/201731880 &
©ESO2018 Astrophysics
The XXL Survey: second series Special issue
The XXL Survey
(cid:63),(cid:63)(cid:63)
XXVII. The 3XLSS point source catalogue
L.Chiappetti1,S.Fotopoulou2,C.Lidman3,L.Faccioli4,F.Pacaud5,A.Elyiv6,7,S.Paltani2,
M.Pierre4,M.Plionis8,9,C.Adami10,S.Alis11,B.Altieri12,I.Baldry24,M.Bolzonella21,A.Bongiorno13,
M.Brown25,S.Driver26,27,E.Elmer14,P.Franzetti1,M.Grootes28,V.Guglielmo10,15,A.Iovino16,E.Koulouridis4,
J.P.Lefèvre4,J.Liske29,S.Maurogordato17,O.Melnyk7,18,M.Owers3,19,B.Poggianti15,M.Polletta1,30,
E.Pompei20,T.Ponman31,A.Robotham26,27,T.Sadibekova4,R.Tuffs28,I.Valtchanov12,
C.Vignali6,21,andG.Wagner22,23
(Affiliationscanbefoundafterthereferences)
Received1September2017/Accepted4January2018
ABSTRACT
WepresenttheversionofthepointsourcecatalogueoftheXXLSurveythatwasused,inpart,inthefirstseriesofXXLpapers.Inthis
paperwerelease,inourdatabaseinMilanandatCDS:(i)theX-raysourcecataloguewith26056objectsintwoareasof25deg2with
afluxlimit(at3σ)of∼10−15ergs−1cm−2in[0.5–2]keV,and∼3 × 10−15ergs−1cm−2in[2–10]keV,yieldinga90%completeness
limitof5.8 × 10−15 and3.8 × 10−14 respectively;(ii)theassociatedmultiwavelengthcatalogueswithcandidatecounterpartsofthe
X-raysourcesintheinfrared,near-infrared,optical,andultraviolet(plusspectroscopicredshiftwhenavailable);and(iii)acatalogue
ofspectroscopicredshiftsrecentlyobtainedinthesouthernXXLarea.WealsopresentthebasicpropertiesoftheX-raypointsources
andtheircounterparts.OthercataloguesdescribedinthesecondseriesofXXLpaperswillbereleasedcontextually,andwillconstitute
thesecondXXLdatarelease.
Keywords. surveys–X-rays:general–catalogs
1.Introduction DR2,theXXLcollaborationplanstoreprocessthedatawithan
improved pipeline, described in Faccioli et al. (2018, hereafter
The XXL Survey is an X-ray survey carried out by the XMM-
XXLPaperXXIV).
Newtonsatellite,coveringtwo25deg2 areascalledXXL-Nand
The plan of the paper is as follows. Section 2 presents the
XXL-S,complementedbyobservationsatmultiplewavelengths.
X-ray data, the X-ray processing pipeline (Sect. 2.1), and basic
Initialresultsfromthesurveywerepublishedinadedicatedissue properties such as sky coverage, logN −logS, and flux dis-
of Astronomy & Astrophysics (volume 592). For the scientific
tribution (Sect. 2.2). Section 3 presents the multiwavelength
motivations of the survey, the characteristics of the observing
data, describes the counterpart association (Sect. 3.2), provides
programme, and the multiwavelength follow-up programmes,
some statistics (Sect. 3.4), presents the spectroscopic redshifts
refertoPierreetal.(2016,hereafterXXLPaperI).
(Sect.3.5),anddescribestheadditionalspectraobtainedwiththe
ThefirstXXLdatarelease(hereafterDR1),whichisassoci-
AAOmegaspectrographin2016(Sect.3.6),supplementingthose
atedwiththefirstseriesofpapers,includedthecatalogueofthe
published in Lidman et al. (2016, hereafter XXL Paper XIV).
100 brightest galaxy clusters, published in Pacaud et al. (2016,
Section 4 presents the Master Catalogue database site, the
hereafterXXLPaperII),andthecatalogueofthe1000brightest
X-raycataloguetables(Sect.4.1)withassociateddataproducts
point-like sources, published in Fotopoulou et al. (2016, here-
(Sect.4.2),andthemulti-λcataloguetables(Sect.4.3)withasso-
afterXXLPaperVI).Asecondrelease(DR2)willoccurjointly
ciateddataproducts(Sect.4.5).Section5summarisesthework.
with the publication of the present paper. A list of the contents
Appendix A compares the present catalogue with the previous
ofDR1andDR2(inclusiveofreferences)istabulatedinSect.4.
XMM-LSScatalogue(Chiappettietal.2013).
WepresentherethecatalogueofallthesourcesusedforDR1
and DR2, from which the above DR1 catalogues were drawn
2.X-raymaterial
(as well as a recent study on the environment and clustering of
AGN,Melnyketal.2018,alsoknownasXXLPaperXXI).After TheX-ray(XMM-Newton)observationsoftheXXLSurveywere
obtainedoverseveralyearsintwoskyregionsasacollectionof
(cid:63)BasedonobservationsobtainedwithXMM-Newton,anESAscience
uniformly spaced contiguous pointings with an exposure time
mission with instruments and contributions directly funded by ESA
of at least 10ks (complemented by a few GO pointings from
MemberStatesandNASA.
(cid:63)(cid:63)xxlpointings, 3XLSS, 3XLSSOPTN, 3XLSSOPTS, and the archives, and by pointings from earlier surveys, sometimes
XXL_AAOmega_16 database tables are also available at the CDS deeper). The list of the 622 XMM pointings (294 in the north-
viaanonymousftptocdsarc.u-strasbg.fr(130.79.128.5)orvia ernareaatδ∼−4.5◦,hereafterXXL-N;and328inthesouthern
http://cdsarc.u-strasbg.fr/viz-bin/qcat?J/A+A/620/A12 area at δ ∼ −55◦, hereafter XXL-S) is given in Appendix B of
ArticlepublishedbyEDPSciences A12,page1of18
A&A620,A12(2018)
Table1.XMM-LSSandXXLX-raycatalogues. Although validated extensively only for the soft band,
the recipe is nominally applied also to the hard band. A
detection that satisfies either C1 or C2 in a band is flagged
Catalogues XAMIN Area N.of Ref.
extended in that band (and by definition not spurious). The
Acronym version (deg2) sources
source extent, EXT, the so-called detection likelihoods1,
XLSS 3.1 5.5 3385 1 EXT_DET_STAT(fitasextended)andEXT_PNT_STAT(fitas
2XLSS 3.2 11.1 5572 2 pointlike),andthelikelihoodofbeingextended,EXT_STAT,
2XLSSd 3.2 11.1 6721 2 are XAMIN parameters for which we follow the notation
3XLSS 3.3.2 XXL-N 25 14168 3 usedinXXLPaperXXIV.Thecorrespondencetodatabase
XXL-S 25 11888 columnnamesisreportedinTable9.
– The application of the P1 recipe, which aims to define a
References. (1)Pierreetal.(2007);(2)Chiappettietal.(2013);(3)this sample of point sources with a high degree of purity and
paper. which is complete down to the lowest count rate possible
(XXL Paper XXIV), used for the first time in the 3XLSS
catalogue,characterisedby
XXLPaperI(whichalsoprovidesdetailsontheobservingstrat- PNT_DET_STAT>30AND(EXT<3(cid:48)(cid:48)OREXT_STAT=0).
egy), and is available at the Centre de Données astronomiques
Striclyspeaking,thesourceswhicharenotC1,C2,orP1
de Strasbourg (CDS) and in our database (see Sect. 4 and
aretobeconsideredundefined(i.e.therearenotenoughpho-
AppendixB).
tons to unambiguously characterise them). The undefined
sourcesareflaggedaspointlikeinthedatabase.Thosewith
PNT_DET_STAT < 15 are considered spurious in the band
2.1.TheX-raypipeline
(i.e.notconsideredforinclusioninthecatalogue).
We use version 3.3.2 of the XAMIN X-ray pipeline Pacaud – Flux conversion. A conventional mean flux
et al. (2006) to process the XMM X-ray data. Earlier versions (FLUX +FLUX )/2 in both bands is computed
MOS pn
of the pipeline were used to process data for the production from count rates via the usual conversion factors CF
of the XMM-LSS survey catalogues, which can be considered listed in Table 2 with a fixed spectral model (Γ = 1.7,
the precursors of the present catalogue on a smaller area (see N =2.6×1020cm−2).
H
Table 1). For this reason we prefix all sources in the current – Errorcomputation.XAMINcurrentlydoesnotprovideerrors
cataloguewith3XLSS. oncountrates,fluxes,orposition.Apositionalerroriscom-
With respect to the newer pipeline reference paper, puted as a function of count rate and off-axis angle (e.g.
(XXL Paper XXIV), all catalogues produced to date, including Table 4 of Chiappetti et al. 2013). Flux errors are com-
thepresentone,usepipelineversionscollectivelygroupedasthe puted by calculating the Poissonian error on gross counts2
basic version called XAminP06 in paper XXIV; in other words, according to the formula of Gehrels (1986), assuming that
detectionsaredoneoneachpointingseparately. the fractional error on rates is the same as that on counts,
Version3.3.2ofthepipelinediffersfromearlierversionsin and propagating them through the CF formula. In Fig. 1
relativelyminordetails,includinganoptimisationofthecodefor we compare the S/N using the computed flux errors to the
thedetectionofpointsources,especiallybrightones,withafix XAMIN detection statistic. The customary 3σ and 4σ lev-
forsomenumericproblems.Theywillberetainedinthenewer els seem to occur for likelihood levels ∼65 and 115, higher
XAminF18pipelinepresentedinXXLPaperXXIV. thanthosefoundinthecomparisonof XAMIN 3.2withthe
As described in Sect. 2.1 of XXL Paper XXIV, the basic XMDSpipelinegiveninChiappettietal.(2013),whichonly
pipeline starts from event lists filtered for soft proton flares, dealswithsomeofthelongestexposuresinXXL,anduses
produced by standard SAS tasks, to generate wavelet images adifferentanalysissoftware.
andendswithaSEXTRACTORsourcelist.XAMINcharacterises Additionalstepsareperformedaftertheingestionoftheband
the sources with a maximum likelihood fit with both a point- tablesinthedatabaseforthepreparationofthecatalogues.These
likeandanextended(β-profile)model,andprovidesa FITS file stepsaredescribednext.
for each pointing with basic parameters separately for the soft
[0.5–2]keV,orB,bandandforthehard[2–10]keV,orCD,band.
2.1.1.Bandmerging
The parameters produced by XAMIN (which can be sum-
marised as exposure times, statistics, raw source positions, Band merging of detections in the soft and hard bands occurs
sourceandbackgroundcounts,andsourcecountrates)arelisted in a 10(cid:48)(cid:48) radius following the procedure used for XMM-LSS
indetailinTable1ofXXLPaperXXIVandTable2ofPacaud and described in detail in Chiappetti et al. (2013). A merged
etal.(2006),andareflaggedinCol.4(“X”)ofTables9and10 sourcecanbeclassified(usingatwo-lettercode,Pforpointlike,
ofthispaper.Otherparametersarecomputedaposterioriduring E for extended, – for undetected, respectively, in the soft and
database ingestion on individual bands, as in Chiappetti et al. hard bands) as extended (and by definition non-spurious) if it
(2013).Theseparametersincludethefollowing:
– The application of the C1/C2 recipe to classify extended 1 The detection statistic used by XAMIN, although customarily but
sources(Pacaudetal.2006),whicharecharacterisedby improperly referred to as likelihood, is a modified Cash statistic, and
• theC1recipeforasampleofclustersuncontaminatedby the extension statistic EXT_STAT is function of the difference of the
detectionstatisticsaspointlikeandasextended.Thedetectionstatistic
point-likesources,
isalinearfunctionofthelogarithmofthetruelikelihood.Theformulae
EXT > 5(cid:48)(cid:48)AND EXT_STAT > 33 AND EXT_DET_STAT >
usedforcomputationarereportedinSect.2.3.1ofPacaudetal.(2006).
32,and
2 Gross counts are reconstructed by adding net counts and back-
• theC2recipeforclustersallowinga50%contamination groundcountsintheextractionarea,whichareparametersproducedby
bymisclassifiedpointsources, XAMIN.TheGehrelsformulaongrosscountswasused,e.g.inXMDS
EXT>5(cid:48)(cid:48)ANDEXT_STAT>15; (Chiappettietal.2005)andHELLAS2XMM(Baldietal.2002).
A12,page2of18
L.Chiappettietal.:TheXXLSurvey.XXVII.
Fig.1. Cross-calibrationoftheS/N(fluxdividedbycomputedfluxerror)vs.thedetectionstatistic PNT_DET_STAT forthesoft(left)andhard
(right)X-raybands.Theredandgreendotscorrespond,respectively,topoint-likesourcesinXXL-NandXXL-S.Thesolidthicklineisalinearfit
inlog–logspacetotheS/Naveragedinpseudo-logarithmic(thepseudo-logarithmicbinningisaspacingof1inPNT_DET_STATupto100,then5
upto1000,50upto10000,and500above10000)likelihoodbins.Thetwopalegreyhorizontallinesarefiducialmarksforthe3σand4σlevels.
TheequivalentdottedverticallinesareforPNT_DET_STATof65and115.
Table 2. Conversion factor CF from count rate to flux in units of The number of such ambiguous cases in the final catalogue
1012ergs−1cm−2forarateofonecounts−1. (i.e. after the next step of overlap handling) is very limited: 74
out of 14168 sources in XXL-N and 69 out of 11888 sources
inXXL-S,i.e.approximately0.5%ofthetotal.Forcomparison,
EPICcamera Soft(B)band Hard(CD)band
0.7% of the sources in the 2XLSS catalogue (Chiappetti et al.
MOS 5.0 23.0 2013)wereambiguous.
pn 1.5 7.9
2.1.2.Pointingoverlapremoval
Notes. Aphoton-indexpowerlawwithΓ=1.7andameanN valueof
H
2.6×1020cm−2areassumed.ThetwoMOScamerasareassumedtobe Accounting for duplicate detections in pointing overlap regions
identical. (“overlap removal”) is the final stage of catalogue generation.
Consistently with the selection criteria defined in Pacaud et al.
(2006), aimed at the best balance between contamination and
is extended in both bands (EE) or in the single one where it is completeness, and already used in Pierre et al. (2007) and
detected (E- and -E), or if it is extended in the soft band (EP). Chiappettietal.(2013),detectionswithDET_STAT<15aredis-
Itisclassifiedaspointlikeinalltheothercases(PP,P-,-P,and carded at this stage and only non-spurious sources are brought
PE). forward as catalogue sources (for an object to be kept a non-
Band merging acts on all detections in all pointings, spuriousdetectioninonebandissufficient).
including those considered as spurious (i.e. detection statis- Detections occurring in a single pointing are always kept
tic DET_STAT<15 in both bands), which are not included irrespective of the pointing quality; instead, in all cases where
in the released catalogue. We have 26555 merged detections two or more detections occur within 10(cid:48)(cid:48) of one another in
in XXL-N, of which 17398 are non-spurious. The respective different pointings, the same procedure used for XMM-LSS
numbersforXXL-Sare27173and18145. and described in Chiappetti et al. (2013) is applied, i.e. if one
A band-merged detection in two bands that is spurious detection is in a pointing with a better bad-field flag (column
in one band (flagged by the boolean flags Bspurious=1 or Xbadfieldinthedatabase),thenitiskeptinpreferencetoany
CDspurious=1) still provides some usable information (rate, other. Otherwise if Xbadfield is the same, the source closest
flux,etc.)andisretainedinthecatalogue. to the XMM pointing centre (Cols. Boffaxis and CDoffaxis
Fordetectionsintwobands,thebandinwhichthedetection inthedatabase)iskept.Thereis,however,adifferencewiththe
likelihood of the source is the highest is the band from which XMM-LSScase:Xbadfieldisnotabooleanflag,butcanhave
thecoordinatesaretaken.Ifwecomparetheseparationbetween arangeofvalues:
B and CD positions (column Xmaxdist in the database) with – 0: pointing belongs to XXL, has good quality, and is the
thepositionaluncertaintiesinthetwobandsaddedinquadrature singleorbestpointingofasequenceofrepeats;
(σ)wefindthatXmaxdist<σin33%ofthecases,within2σin – 1: pointing does not belong to XXL (archival or deep
71%ofthecasesandwithin3σin77%,whichlooksreasonable. followuppointing)andhasgoodquality;
Itispossibletohaveanambiguousbandmergingcasewhen – 2:anotherXXLpointinginasequenceofrepeatedpointings
a detection in a band happens to be associated with two differ- (seebelow)thathasgoodquality;
ent objects in the other band (i.e. it gives rise to two entries in – 3: quality is bad (which may occur if the exposure is too
themergedtable).IfthetwoentriesbothhaveXmaxdist<6(cid:48)(cid:48)or shallow,orbackgroundistoohigh,orboth).
Xmaxdist>6(cid:48)(cid:48), they are both retained as intrinsically ambigu- We note that 94% of the sources have Xbadfield=0 and only
ous. If one is below 6(cid:48)(cid:48)and one above, the lower-distance 2%haveXbadfield=3.
entry remains a merged two-band detection, while the other is One should remember that a given pointing might have
“divorced”asahard-onlyorsoft-onlysource. been re-observed (repeats) if previous observation(s) did not
ThenamingofambiguoussourcesisdescribedinSect.4.1. achieve the scheduled exposure. Repeats are characterised by
A12,page3of18
A&A620,A12(2018)
Fig.2.DistributionoftheX-rayfluxandhardnessratio(HR)forXXL-N(toppanels)andXXL-S(bottompanels).Theleftcolumnshowsthesoft
band,thecentrecolumnthehardband,andtherightcolumnthefluxhardnessratio(customarilydefinedas[hard−soft]/[hard+soft]),computed
forsourcesdetectedinbothbands.Theblackhistogramsinthefourleftmostplotscorrespondtoallsourcesdetectedineachband;thethinred
histogramstosourcesdetectedonlyinoneband(hencebydefinitionnotspuriousinit);thegreendashedhistogramstosourcesalsodetectedin
theotherband(thesumofthegreenandredhistogramscorrespondstothetotal);thesubsetofthesourcesdetectedinbothbandsbutnominally
spuriousinthebandofrelevanceareshownbythedottedbluehistogram.IntheHRdistribution,thethinblackhistogramcorrespondstoallsources
detectedinbothbandsunderanycondition,whichinturnfallintothefollowingthreecategories:thethickbluehistogramrepresentssourcesthat
arenon-spuriousinbothbands;thedashedmagentaonerepresentsthosethatarespuriousinthehardband;andthetinythickgreenoneatthe
extremerightthosethatarespuriousinthesoftband.Somehistogramsareslightlydisplacedalongthex-axisforclarity.
a XFieldName of the form XXLsmmm-ppc with a different let- noted above. The offsets per pointing, available in web pages
ter c=a,b,c,... in the rightmost position. In some cases it reachable from the “table help pages” in the database, have
has been possible to combine the event files of repeats (called been added as new columns in the pointing list table, already
z-pointings because the last letter in the XFieldName is con- published in XXL Paper I; the updated table is presented in
ventionallyz,e.g.XXLn000-04z):z-pointingsarebydefinition AppendixB.
consideredsuperiortotheindividualrepeatsinthesequence.
2.2.Basicproperties
2.1.3.Astrometriccorrection
Table 3 provides the number of pointlike and extended sources
Following the procedure used for XMM-LSS, an astromet- in the catalogue, split by the C1, C2, and P1 recipes described
ric correction was applied to all sources using the SAS task above.
EPOSCORR,applyingtoallpositionsinapointingaglobalrigid In Fig. 2 we show the X-ray flux distributions for both the
shift(norotationwasapplied).Theoffsetswerecomputedusing softandhardbands,andthehardnessratio(customarilydefined
referenceopticalcatalogues.ForXXL-NtheCFHTLegacySur- as[hard−soft]/[hard+soft])distribution(forsourcesdetectedin
vey T007 version from Vizier (catalogue II/317, Hudelot et al. bothbands)forXXL-NandXXL-S.
2012)wasused,exceptforonepointingwheretheUSNOA2.0 The relationship between flux and detection statistic
catalogue was used and 16 pointings for which no correction PNT_DET_STAT wasreportedinFig.4ofXXLPaperI.There-
could be computed (bad pointings, with few X-ray detections, fore,wefeltitunnecessarytorepeattheinformationhere.
which are usually ignored by the overlap removal procedure The cross-calibration of the S/N using the computed flux
in favour of better repeats of the same pointing). For XXL-S errors versus the XAMIN detection statistic has been reported
the reference catalogue was from the Blanco Cosmology Sur- aboveinFig.1.
vey (BCS, Menanteau et al. 2009), except for 11 pointings for The photometric accuracy of the present pipeline is pre-
which no correction could be computed for the same reasons sented in Fig. 4 of Chiappetti et al. (2013), to which we refer
A12,page4of18
L.Chiappettietal.:TheXXLSurvey.XXVII.
Table3.SourcenumbersfortheX-raycatalogues.
N.ofsources XXL-N XXL-S
Total 14168 11888
Pointlike 13770 11413
...andP1 7246 5565
PP(detectedin2bands) 4597 3479
...andP1 3917 2907
PE(detectedin2bands) 14 21
...andP1 10 19
P-(onlysoft) 7389 6247
...andP1 3085 2451
-P(onlyhard) 1770 1666
...andP1 234 188
ExtendedforXAMIN 398 475
SoftC1(Bc1c2=1indatabase) 136 118
SoftC2(Bc1c2=2indatabase) 188 208
Hard-onlyC1orC2(-E) 74 149
XLSSCclustersa 186 142
Notes. (a) All of these are present in the 365 cluster catalogue of
XXLPaperXX,buttwosouthernobjects(XLSSC613and630)flagged
“tentative”;17clustersfromXXLPaperXX(allbutoneinTableG.1
of such paper) are not listed in our pointlike catalogues because the
originalX-raydetectionwasflaggedpoint-likeandspurious;22clus-
tersinTableG.2ofXXLPaperXXhavenoXLSSCnumberassigned,
butcouldbematchedtoourpointlikecataloguesusingtheXcatname
databasecolumn.
the reader since it is unaffected by the changes in the pipeline
version.
2.2.1.Skycoverage Fig. 3. Sky coverage plots in the soft (top panel) and hard (bottom
panel)bands.Thedashedanddash-dottedlinescorrespondtoXXL-N
The sky coverage for XXL-N and XXL-S is shown in Fig. 3. andXXL-S,andthesolidlinetothecombinationofthetwo.
Removing bad fields, i.e. keeping only (Xbadfield<3), the
sky coverage is 95% of that shown. The process of removing
AppendixA)reportedbyElyivetal.(2012),plottedasreddia-
duplicateskeepssourcesin273northernand321southernpoint-
monds. We also plot earlier data shown in the latter paper: a
ings,ofwhichrespectively242and288aregood.However,since
green solid line for the fit from XMDS (a subset of XMM-
theremovaloperatesonindividualsources,98%ofthecatalogue
LSS analysed with a different pipeline; Chiappetti et al. 2005),
entriesareingoodpointings(only313XXL-Nand261XXL-S
andresultsfromtwoindependentsurveys,COSMOS(violetstar
sourcesareinbadpointings).
datapoints;Cappellutietal.2007)and2XMM(magentadashed
The 90% of the coverage is achieved for a flux
line; Ebrero et al. 2009). Among the more recent publications
of 5.8×10−15ergs−1cm−2 in the soft band, and of
weshowdatafromNuSTAR(bluedottedline,forthehardband
3.8×10−14ergs−1cm−2inthehardband.
only, rescaled from 3 to 8keV; Harrison et al. 2016), Stripe82
(observed by XMM, green squares; LaMassa et al. 2016), and
2.2.2.Sourcecounts ChandraCDFS7Ms(bluetriangles;Luoetal.2017).Thehard-
band points for CDFS were rescaled3 from 2 to 7keV. Our
ThelogN–logS relationhasbeencalculated(inthetwoenergy
logN–logS isconsistentwiththeresultofearliersurveys,with
bands) as described in Elyiv et al. (2012), taking into account
COSMOS being the most deviant. The slight difference with
the numerically calculated probabilities to detect sources with
Elyivetal.(2012)inthehardbandcouldbeduetothedifferent
a certain flux, an off-axis distance in a pointing with effective
pipeline version or to the lower average exposure of the entire
exposure, and a particle background level. The sources in the
XXLversustheXMM-LSSpointings.
overlaps between pointings were taken into account according
totheVoronoitessellationtechniquedescribedindetailinElyiv
etal.(2012),andonlygoodpointingswithexposuresbelow15ks 3.Multiwavelengthmaterial
were considered in order to have uniformity of coverage. It is
ForthepurposeofidentifyingtheX-raydetectionsatotherwave-
showninFig.4fortheentiresurvey;thecurvescalculatedsep-
lengths,weuseoptical,near-IR(NIR),IR,andUVphotometric
arately for the XXL-N and XXL-S areas are extremely similar,
whichindicatesnoeffectofcosmicvariance.
3 Bandrescalingentailsshiftingthefluxtotheright,multiplyingbya
We plot representative data from the literature. We show factorequaltotheratioofthefluxinthe2–10keVbandtothefluxin
the XMM-LSS data (which are a subset of XXL-N analysed theoriginalband,computedforanunabsorbedpowerlawwithΓ=1.7,
with a previous version of our pipeline, compare Table 1 and consistentlywiththeCFsinTable2.
A12,page5of18
A&A620,A12(2018)
Table4.Photometricsurveysandfiltersusedforcounterpartassocia-
tion.
Survey Sub-surveys Mnemonic Filters
North
SDSS – sdss ugriz
CFHT D1W1WAWBWC cfht ugriyz
VISTA VHSVIDEOVIKING vista zY J H K
UKIDSS UDSDXS ukidss J H K
WIRcam – wircam K
IRAC – irac 3645
GALEX DISMISAISGI galex fuvnuv
WISE – wise w1 w2 w3
w4
South
BCS – bcs griz
DECam – decam griz
VISTA VHS vista J H K
SSDF – irac 3645
GALEX MISAISGI galex fuvnuv
WISE – wise w1 w2 w3
w4
Notes. Themnemonicforsurveys(alllowercase)andfilters(capitalisa-
tionasshown)isusedtonamethecolumnslistedinTable11.TheIRAC
filter mnemonics “36” and “45” correspond to the 3.6µm and 4.5µm
filters.TheCFHTfilterlabelled“y”isani-bandreplacementfilter.
(CFHT), using the MegaCam wide field optical imaging
Fig. 4. logN–logS relation for both XXL areas in the soft and hard facility. The northernmost part was observed at CFHT in
X-ray bands (black crosses with 1σ error bars), compared with some the g,r,z bands (PI: M. Pierre) and the relevant fields are
literaturemeasurements(seetextfordetails).
labelledWA,WB,WCinTable4.
– The Sloan Digital Sky Survey (SDSS) also provides shal-
dataasdescribedinTables3and4ofXXLPaperVI,towhich loweru,g,r,i,zobservationsinthewholeXXL-Narea.We
the reader is referred in particular for details about the limiting usedtheDR10datarelease(Ahnetal.2014).
magnitudesinthevariousbands,andforthedataarchivesused. – XXL-N has been observed with the WIRcam camera on
Intotal,thereareeightsurveyscoveringXXL-Nandsixsurveys CFHTintheK band(2.2µm);seeMoutardetal.(2016).
s
coveringXXL-Sinvariousfilters,assummarisedinTable4and – The UKIRT infrared deep sky survey (UKIDSS, Dye et al.
inSect.3.1. 2006) has two fields (in the J, H, K bands) targeted on the
Rather than using publicly available catalogues, we per- XMM-LSSarea,enclosedinXXL-N:theultradeepsurvey
formed our own homogeneous photometric extraction per filter (UDS), and the deep extragalactic survey (DXS). We used
and per tile, handling tile overlaps. All magnitudes for all theDR10datarelease.
surveys are in the AB system, corrected to 3(cid:48)(cid:48) aperture mag- TheXXL-Sareaiscoveredbythefollowingsurveys:
nitudes, and corrected for galactic extinction as described in – The Blanco Cosmology Survey (BCS), using the Mosaic2
XXLPaperVI. imager at the Cerro Tololo Inter-American Observatory
The full photometric catalogues, including the photomet- (CTIO)a4-mtelescope,observedalargesouthernarea,giv-
ric redshift computed as described in XXL Paper VI, and all ingfullcoverageofXXL-Sintheg,r,i,zbands(Menanteau
sources (not just the counterparts of the X-ray sources) will be etal.2009;Desaietal.2012).
published in a forthcoming paper (Fotopoulou, in prep.). The – The Dark Energy Camera (DECam, Flaugher et al. 2015),
association of published spectroscopic redshifts with our coun- alsoontheCTIOBlancotelescope,provideddeeperg,r,i,z
terpartsisdescribedinSect.3.5,whilemoredetailaboutthenew observationsofXXL-S(Desaietal.2015).
AAOmegaspectroscopicobservationsforwhichwepublishthe Thefollowingsurveyscoverbothareas:
completecatalogueisgiveninSect.3.6. – TheVistaHemisphereSurvey(VHS;McMahonetal.2013),
usingtheVisibleandInfraredSurveyTelescopeforAstron-
omy(VISTA)coveredXXL-NandXXL-Sinthe J, H,and
3.1.Summaryofphotometricsurveys
K bands.AdditionalcoverageofXXL-Nissuppliedbythe
TheXXL-Nareaiscoveredbythefollowingsurveys: VISTAKilo-degreeInfraredGalaxySurvey(VIKING;Edge
– Most of XXL-N was covered by the CFHT Legacy Sur- et al. 2013) and VISTA Deep Extragalactic Observations
vey (CFHTLS) Wide1 (W1) and Deep1 (D1) fields in Survey(VIDEO;Jarvisetal.2013).
five filters (u,g,r,i,z); we used the T0007 data release – The Spitzer satellite has observed the large Spitzer South
(Veillet 2007; Hudelot et al. 2012). Observations were Pole Telescope Deep Field (SSDF), with its Infrared Array
obtained with the 3.6m Canada–France–Hawaii Telescope Camera (IRAC) in channel 1, 3.6µm, and 2, 4.5µm. This
A12,page6of18
L.Chiappettietal.:TheXXLSurvey.XXVII.
field fully covers XXL-S. See Ashby et al. (2014). Similar or exceptionally the best of the rejects, which is “recovered”.
observations(PI:M.Bremer)wereacquiredoverXXL-N. When several candidates above the thresholds exist, the one
– The Wide-field Infrared Survey Explorer (WISE) mission with best estimator is considered the primary counterpart, and
observed the whole sky in four mid-IR (MIR) bands: all the others are secondaries. If the estimator ratio between
W1=3.4µm, W2 = 4.6µm, W3 = 12µm, W4 = 22µm the primary and the best secondary is above 10, the primary
(Wrightetal.2010).WeusedtheALLWISEdatarelease. should be definitely preferred and the secondaries are just
– The GALEX satellite surveyed the entire sky (Morrissey nominally included. Otherwise the primary is only nominally
et al. 2005) in two ultraviolet bands, 1344–1786Å (far-UV, better than the secondaries, but it is indicative of an intrinsic
FUV) and 1771–2831Å (near-UV, NUV), with various sur- ambiguity.
veys:all-skyimagingsurvey(AIS),mediumimagingsurvey Finally, we tie everything together by comparing the ranks
(MIS),andguestinvestigator(GI)programmesinbothXXL assignedbythedifferentmethods,cleaningupambiguouscoun-
areas,andthedeepimagingsurvey(DIS)inXXL-Nonly. terpart sets sharing a counterpart in more than one survey,
demoting to secondaries the detection in just the surveys with
poorerpositionresolution(WISE,IRAC,GALEX)andpromot-
3.2.Counterpartassociation
ingthecaseswithcounterpartsinseveralsurveys.
In order to associate X-ray sources with potential counter- We assign a final rank (numbers provided in Table 5) so
parts at other wavelengths, we first generated multiwavelength that each X-ray source has a single primary counterpart, and
counterpart sets, matching the individual primary photometric zero, one, or more secondaries. Rejected counterparts will not
catalogues (one per survey per band: 28 in XXL-N and 19 in be listed in the multiwavelength catalogues. The identification
XXL-S; see Table 4) among them (using a matching radius of rank is coded by a three-character string. There will be always
0.7(cid:48)(cid:48) for the optical and IR catalogues where the image PSF is one primary counterpart with rank between 0 and 1, and there
small, and 2(cid:48)(cid:48) for GALEX, IRAC, and WISE) as described in maybeoneormoresecondarycounterpartswithrankabove2.
XXL Paper VI. For the further association with X-ray sources, Thecodesareasfollows:
wealsoconsideredthecaseofcounterpartsinonesingleprimary – 0.0: physically single counterpart within 10(cid:48)(cid:48) of the X-ray
catalogue. position;
FortheassociationofentirecounterpartsetswiththeX-ray – 0.1:logicallysinglecounterpart(allotherswererejected);
sourceswithinasearchradiusof10(cid:48)(cid:48),weusetwoestimators.In – 0.2:“recovered”(thesinglecounterparthasbadestimators);
one case, we computed the simple probability of chance coin- – 0.4: “blank field” (no catalogued counterpart within 10(cid:48)(cid:48) of
cidence according to Downes et al. (1986; done purely as a theX-rayposition;
cross-check and for reference since it is what was used for the – 0.9: the primary counterpart is much better than all the
XLSSand2XLSScatalogues): secondaries(ranked2.2);
– 1.0: the primary counterpart is just marginally better than
P=1−exp(−πn(<m)d2). (1) oneofthesecondaries;
– 2.1:thissecondarycounterpartisjustmarginallyworsethan
In the other case, we used the more robust likelihood ratio
thecorrespondingrank1.0primary;
estimator (Sutherland & Saunders 1992), which has been used
– 2.2:anyothersecondarycounterpart.
forothersurveys,e.g.Brusaetal.(2007):
Ranks below 0.2 correspond to single primaries; a rank of 0.9
corresponds to a preferred primary which is better than all its
LR=(q(m)exp(−0.5d2/σ2)/2πσ2)/Kn(m). (2)
secondaries; a rank of 1 to a nominal primary which is only
In both formulae, d is the distance in (cid:48)(cid:48) between the X-ray marginallybetterthanoneofthesecondaries.
position and the candidate counterpart position, the positional
error σ is assumed for simplicity equal to 1(cid:48)(cid:48), while the depen- 3.3.Brightstarclean-up
dencyonthemagnitudemofthecounterpartiseitherviathesky
We have associated our counterparts with sources in the
density (sources per square arcsec) n(< m) of objects brighter
SIMBAD and NED databases, and are providing an identifier
than m, or the sky density per magnitude bin n(m). The term
for the relevant cases. We used this information as one way of
q(m)inEq.(2)isdefinedas
partially screening out stars, which are not of interest for AGN
q(m)=g(m)−Kn(m), (3) studies.WealsousedtheUSNOA2.0cataloguetoscreenstars,
aswellasourownspectroscopicredshifts,describedinSect.3.5.
where g(m) is the sky density of putative true counterparts, i.e. We assign a star flag of 1 to the counterparts with
uniqueobjectswithinasuitableradius(inourcase3(cid:48)(cid:48))ofeach z <0.003, i.e. stars confirmed spectroscopically, a flag of 2
spec
X-raysource,and K istheratio N /N betweenthenumberof toobjectsthatlooklikeastaruponvisualinspection,andaflag
U tot
such putative counterparts N and the total number N of all of 3 when both conditions occur. Visual inspection of thumb-
U tot
(alsonon-unique)objectsinthebandclosetotheX-raysources nailopticalimages(seeSect.4.5)wasdonesystematicallyforall
withinthesameradius,i.e.inclusiveofbackgroundobjects. counterpartswithamatchtoasourceinaknownstarcatalogue
We computed these estimators for each survey and band inSIMBADorNED.Wealsoaddedobjectsinothercatalogues
where a potential counterpart is present, and assigned the best (like 2MASS) which were inspected visually and found to be
value (highest LR) to the counterpart set. We then ordered the star-like.Wealsosystematicallyinspectedthecounterpartswith
counterpart sets by decreasing LR, and also divided them into a match in USNO A2.0 with a magnitude brighter than R < 14
threebroadgroups:good,fair,andbadaccordingtoLR>0.25, (usuallyalltheunnamedobjects,i.e.nototherwiseinSIMBAD
0.05<LR<0.25,LR<0.05usingPasacross-check. orNED,arestarsandtherestaregalaxies).
We then assign a preliminary rank, rejecting most of the Of the 14168 primary counterparts in XXL-N we find 163
cases with bad scores. A primary single counterpart is either “spectroscopic” stars, and visually identify 177 more; 23 were
a physical solitary association, a single non-bad association, flaggedasstarsbybothmethods.InXXL-S,thecorresponding
A12,page7of18
A&A620,A12(2018)
InXXL-Sabout61%ofthecounterpartsareobservedinat
least four surveys; 90% of the primaries have a DECam coun-
terpart;80%aBCScounterpart;78%haveIRAC;63%VISTA;
44%WISE;and26%haveaGALEXcounterpart.
ForeachX-raysourceprimarycounterpart,wecomputethe
distance between the astrometrically corrected X-ray position
and the counterpart in the survey that is nearest. A histogram
ofthesedistancesisshowninFig.5.
InXXL-N(XXL-S)85%(81%)ofallprimarycounterparts
arewithin4(cid:48)(cid:48).Thisoccursfor94%(93%)ofthebestsinglepri-
maries (ranks 0.0 and 0.1) or for 91% (91%) for the bona fide
primaries(ranks0.0,0.1,and0.9).Largerdistancesoccurforthe
marginalprimaries(rank1.0),with74%(69%)within4(cid:48)(cid:48),orfor
the“badsingles”(rank0.4)with57%(54%).
IntheleftcolumnofFig.6weplot,asanexample,thedis-
Fig.5.DistancebetweentheprimarycounterpartsandtheX-raysource tributionofthemagnitudeofthecounterpartsinther(cid:48)filter(the
forsourcesinXXL-N(inblack)andinXXL-S(inred).
otherfiltersaresimilar)inXXL-NandXXL-S.Thisiscompared
withtheoverallsourcedensityn(m)(usedinSect.3.2)fromthe
Table5.Basicstatisticsofthemultiwavelengthcatalogues. corresponding surveys, shown in the right column of the same
figure. We note that the spectroscopic coverage (green curves)
has a sharper cut-off in XXL-S (and is more regular under the
Rank Explanation XXL-N XXL-S
cut-off),wherethedataismainlybasedonpointedobservations
All DistinctX-raysources 14168 11888 around our X-ray targets (see Sect. 3.6), while in XXL-N the
spectroscopic data comes from a variety of observations (see
Primarycounterparts
Sect.3.5).
0.0 Physicallysingle 77 57 In order to compare the X-ray and optical fluxes, in Fig. 7
0.1 Logicallysingle 3462 2789 we plot the r(cid:48) magnitude versus the soft (B) broad-band flux.
0.2 “Recovered”bad 764 468 In absence of a systematic X-ray spectral fit or SED analy-
0.4 “Blankfield” 18 21 sis (currently available only for the 1000 brightest sources, see
0.9 Definiteprimary 4138 3768 XXLPaperVI),weprefertoshowrawparameters.Theoptical
1.0 Marginalprimary 5709 4785 magnitudeistakenfromanyavailablecatalogue(i.e.CFHTand
SDSS in XXL-N, and BCS and DECam in XXL-S), preferring
Secondaries
CFHT or BCS when two measurements are present for a given
2.1 Marginalsecondary 5709 4785 counterpart.Thedifferentcoveragecharacteristicsofthevarious
2.2 Anyothersecondary 11708 10850 surveysisapparentfromtheplots.Withreferencetothestarflag
described in Sect. 3.3, we see that the stars identified spectro-
scopically(flag=1,greenasterisksinthefigure)covertheentire
space, while those identified visually (flag=2, red diamonds in
the figure) are usually concentrated at the brighter magnitudes.
numbersare384,174,and7.Wenotethatthisscreeningisnotat
InXXL-Ntherearesixexceptionswithr(cid:48) >18:fourofthemare
allcomplete.
inthePBfaintbluestarcatalogue(Berger&Fringant1984),one
isinUSNOA2.0,andoneisLP649-93.Thelasthasadefinite
star-like appearance, the others are unconspicuous, and two of
3.4.Statistics
themarereportedbySIMBADaspossiblequasars.Thisgivesan
Table5liststhenumberofcounterpartssortedbyrank.For60% ideaofthelimitedcontaminationofthemethodusedinSect.3.3.
ofthecases(rank<1)thereisasinglecounterpartoracounter- AnalternateviewispresentedinFig.8,wherewecompute
partthatisclearlypreferred.Intheremainderofthecasesthere a proxy for the optical-to-X-ray spectral index α , using the
ox
issomeambiguityintheidentityoftherealcounterpart. broad-band [0.5–2]keV flux, instead of a monochromatic flux
The number of X-ray sources that have no apparent cat- fromafit,andthestandardspectralmodelofTable2.
alogued counterpart at all (“blank fields”) is very small: 18
in XXL-N and 21 in XXL-S. Not all blank fields are really
3.5.Associatingredshiftswiththecounterparts
empty (11 in XXL-N and 4 in XXL-S), as they might have
been affected by contamination from a nearby source which For the X-ray sources whose counterparts have spectroscopic
prevents a detection, or the potential counterpart could actu- observations, we report the redshift z . The redshifts may
spec
allybeabrightsourcethatwentundetectedbythephotometric come from large spectroscopic surveys with which we have
extraction. collaborative agreements, like VIPERS (Guzzo et al. 2014) or
Tobeconservative,onecouldexcludetherankof0.2coun- GAMA (Liske et al. 2015; Baldry et al. 2018) in the XXL-N
terparts(i.e.thosewherethereisasingleobjectneartotheX-ray area; or, in XXL-S, from large campaigns of our own (see
sourcebutwithbadassociationestimators). XXL Paper XIV and Sect. 3.6); or from the compilation held
In XXL-N about 67% of the counterparts are observed in inMarseilleatCESAM4,whichincludesbothpublisheddataof
at least four surveys; 89% of the primaries have a CFHT coun- externaloriginandsmallercampaignsbyXXLPIs.Theorigins
terpart in some band; 77% have IRAC; 70% have a VISTA, are briefly listed in the footnote to Table 11; a full list with
WIRcam,orWISEdetection;39%haveGALEX;and28%have
aUKIDSScounterpart. 4 http://cesam.oamp.fr/xmm-lss/
A12,page8of18
L.Chiappettietal.:TheXXLSurvey.XXVII.
Fig.6.r(cid:48)-bandmagnitudedistributionforobjectsinXXL-N(toppanels)andXXL-S(bottompanels).Leftcolumn:histogramsforcounterparts:
greyreferstoallcandidatecounterparts(primariesandsecondaries);blacktotheprimarycounterparts;andgreentothesubsetoftheprimaries
withaspectroscopicredshift.Rightcolumn:densitiesn(m)forallobjectsinspecificsurveys.ForCFHTthisamountstomorethan5millionobjects
in∼43deg2;forBCS(bottompanel,blacksolidcurve)toapproximately2millionobjectsin∼47deg2;andforthedeeperDECamsurvey(dashed
bluecurve)toapproximately2millioninjust∼16deg2.
Fig.7.Opticalvs.X-rayfluxforXXL-N(left)andXXL-S(right).Ther(cid:48)magnitudeoftheprimarycounterpartisplottedagainstthebroad-band
softflux[0.5–2keV].Symbolsandcolour-codesareasfollows:pinkdotscorrespondtomagnitudesavailableonlyfromeitherCFHT(preferred)or
SDSS(inXXL-N),oreitherBCS(preferred)orDECam(inXXL-S);greenasteriskstostarsidentifiedassuchbytheirspectroscopicredshift(star
flag1);reddiamondstostarsidentifiedvisually(starflag2);blueasteriskstospectroscopicandvisualstars(starflag3).Thedashedanddotted
linesmarktheAGNlocusofX/O=0±1(e.g.Brandt&Hasinger2005).
references is provided in Table 2 of Guglielmo et al. (2018), priority into three classes, and inside the same priority group
hereafterXXLPaperXXII.Moredetailsaboutthecollectionand it takes the spectrum with the best quality flag, dividing the
reductionofspectroscopicinformationatCESAMisprovidedin survey-providedredshiftflagsintofouruniformclasses.
Sect. 3 of Adami et al. (2018, hereafter XXL Paper XX). For In XXL-N we actually used the very same sample used in
thecasewherethesamecounterparthasmorethanoneredshift XXLPaperXXII,towhichwere-addedthestars(z <0.003)
spec
measurement, the procedure described in XXL Paper XXII excluded there, and applied exactly the same quality choice.
is adopted. Namely, it first groups the originating surveys by For XXL-S we once more applied an analogous recipe: the
A12,page9of18
A&A620,A12(2018)
Fig.8. Proxyfortheα indexforXXL-N(leftpanel)andXXL-S(rightpanel)asafunctionofopticalflux.Theoptical-to-X-rayspectralindex
ox
isdefinedasα =log(f /f )/log(ν /ν )wheretheopticalmonochromaticfluxiscomputedfromther(cid:48)ABmagnitude2.5logf =−r(cid:48)−48.6,and
ox x o x o o
thereferencefrequencyνo derivedfromtheλeff ofthefilter(6258Å forCFHT,6185Å forSDSS,6266Å forBCS,and6433Å forDECam);νx
correspondsto2keV;themonochromatic2keVflux f isextrapolatedfromtheobserved[0.5–2keV]softfluxF usingthespectralmodelof
x soft
Table2,i.e. f =7.626176×10−27×2×1.47115×108F .OnlysourcesdetectedinthesoftX-raybandwithaprimarycounterpartinther(cid:48)filter
x soft
areconsidered.Symbolsandcolour-codesareasinFig.7.
highest origin flag was assigned to the recent AAOmega cam-
paign(Sect.3.6),thentothepreviousonefromXXLPaperXIV,
then to the other campaigns in the CESAM compilation, and
finally to NED; the quality flags are identical to the ones in
XXL Paper XXII, to which we refer for a complete descrip-
tion, exceptfor the“star” flag6 listedin Sect.3.6.4, ignored in
XXLPaperXXII,butconsideredequivalenttothe“best”flag4.
Finallyweassociatedthespectroscopictargetswithouropti-
cal and NIR photometric potential counterparts within 1(cid:48)(cid:48). In
XXL-N we dealt with a total of 5521 redshifts, of which 3235
refer to primary counterparts, 1098 to secondaries, and 1188 to
rejectedcandidates.InXXL-S,excluding540sourcesobserved
with no valid redshift, we have a total of 4745 redshifts (3873
primarycounterparts,661secondariesand211rejects).Figure9
givestheredshiftdistributionoftheprimarycounterparts.
Fig.9.RedshiftsoftheprimarycounterpartsintheXXL-Narea(black)
In XXL-N 33% of the redshifts of primary counterparts
andintheXXL-Sarea(red).
derivefromGAMA,30%fromVIPERS,14%fromSDSSDR10,
and the rest from miscellaneous origins. In XXL-S 60% of the
redshifts come from XXL Paper XIV, 38% from our own 2016 theUCAC4catalogue.ThereweretwocategoriesofAGN,those
observations described in the next section, and only a handful detectedintheBCSsurvey,andalltherest.
fromotherorigins. The radio sources were selected from the ATCA
2.1GHz survey of the XXL-S (Butler et al. 2018, here-
after XXL Paper XVIII). Only radio sources brighter than
3.6.The2016AAOmegaspectra
r < 22.0 (equivalent to r < 21.8) were targeted, with a
AB vega
In XXL Paper XIV, we noted that we used the AAOmega maximumof120radiosourcesper2dFpointing.
spectrograph (Smith et al. 2004) in conjunction with the two- Candidate cluster galaxies were selected according to their
degreefield(2dF)fibrepositionerontheAAT(Lewisetal.2002) position and magnitude. For a given cluster, a galaxy was
to measure the redshifts of 3660 sources in the XXL-S field. selected if it was within 1Mpc of the cluster centre and had an
Herewereportonnewobservationswiththesameinstrument. r-band magnitude within the range m −3.0 < r < m +1.75,
(cid:63) (cid:63)
where m is the r-band magnitude of an L galaxy at the red-
(cid:63) (cid:63)
shiftofthecluster.Onlyclustergalaxycandidatesbrighterthan
3.6.1.Targetselection
r =22.5weretargeted.Therewerethreecategoriesofclusters.
AB
Thetargetsweresplitintoseveralcategories,asexplainedbelow. Clusters that lacked spectroscopic confirmation had the highest
ThereweretwocategoriesforAGN,oneforradiogalaxies,three priority.Fortheseclusters,weusedphotometricredshifttoesti-
forclustergalaxies,oneforfieldgalaxies,andoneforobjectsof matem andtheangulardistancecorrespondingto1Mpc.Next
(cid:63)
specialinterest(e.g.gravitationallensesandfossilgroups).The came clusters that were spectroscopically confirmed, followed
highestprioritywasforAGN,thenexthighestforradiosources, by clusters that were optically identified. We also assigned 17
followedbyclustergalaxiesandfieldgalaxies. fibres to candidate fossil group members, as part of the search
TheAGNwereselectedfromourX-raysourcelist,targeting for fossil groups described in XXL Paper XX; 13 of them had
onlyobjectsbrighterthanr =21.8.Obviousbrightstarswere redshiftswithqualityflagsabove3,asdefinedinSect.3.6.4.Of
AB
removedbyexcludingobjectsthatwere3(cid:48)(cid:48) fromabrightstarin these,twowerefoundtobestars.
A12,page10of18