The SCUBA HAlf Degree Extragalactic Survey (SHADES) - III. I(12)

Determining an accurate position for a submm galaxy (SMG) is the crucial step that enables us to move from the basic properties of an SMG sample - source counts and 2-D clustering - to an assessment of their detailed, multi-wavelength properties, their con

12Ivisonetal.

thewholeSHADESSourceCatalogueandforthosesourceswithmultiplerobustcounterpartsat1.4GHzor24µm.Takingtheme-dianSHADES uxdensityasourthreshold,eightmultipleiden-ti cationslieaboveand11liebelow,respectively(sixapieceus-ingonlytheradio).However,asour uxdensitythresholdrisesto10mJysothefractionofsourceswithmultipleidenti cationsrisesfrom15/111to4/9(or8/111to4/9usingonlytheradio);evenig-noringthehighprobabilitythatoneoftheremaining vebrightsourcesmaybespurious(SXDF850.45)andthatanotherhassev-eralpossiblecounterparts(Lock850.34–Table6),thisisasignif-icanttrend.Itisplausiblethatthesesourcesareexamplesofcon-fusion(i.e.amalgamatedsources)butwenotethatthephysicallylinkedsystemsreportedtodateareoftensimilarlybright.

Weconcludethattheincidenceofveryhigh uxdensityandcounterpartmultiplicityareweaklylinkedandthatthecaseforapreferredseparationbetweenmultiplecounterpartsisplausiblebutnotproven.Inparticular,wenotethatalmosthalfofthebrightestnineSMGs–all>10mJy–havemultipleradiocounterpartsandthatallhaveseparationsintherange2–6arcsec,or20–70kpcattheirlikelyredshiftsandataninclinationof45 tothesky,per-hapsenablingef cientgasfuelingforcentralstarburstsorAGNviaoverlappinggalacticdisks—seethequalitativediscussionandillustrations(particularlyFigs11–13)inthemergersimulationsofSpringel,DiMatteo,&Hernquist(2005)whereaparticularlyin-tenseburstofactivityoccurson rstpassageforsystemsthatlackprominentbulges,withgalaxyseparationsof～30kpcforthesub-sequentfewtensofMyr.

6RADIOANDMID-IRIDENTIFICATIONTRENDSANDSUBMMSAMPLEREFINEMENTFollowingIvisonetal.(2002),weseektoexploittheclearpre-dictionthatspuriousSMGswilllackradioormid-IRcounterparts.Genuinesourcescan,ofcourse,evaderadioormid-IRdetection–becausetheylieatextremeredshift,forexample(seeIvisonetal.2005)–butgeneraltrendsintheidenti cationratemaybeevident.InthissectionwethereforeexplorewhatcanbelearnedabouttheSMGswithoutcounterparts.

Fig.6showsthecumulativeidenti cationrateforSMGsintheLHandSXDF eldsasafunctionofsubmmSNR(beforeandafter uxdeboosting),deboostedsubmm uxdensityandsubmm uxuncertainty.6.1Radiotrends

Lookingattheradioidenti cationtrendsasfunctionsofsubmm uxdensityandnoise,weseetherecoveryratetailingoffatthefaintest uxdensitylimits(<5mJy)inSXDF,whereastherateisremarkably atforfainter uxdensitiesintheLH eld.Both eldsshowimprovingidenti cationratesasthesubmmnoisedeclines,despitethedeboostingproceduresoutlinedinPaperII–aworryingtrend,thoughweshouldbearinmindthatsearchingforidenti -cationswithina xedradiusmustactasabiasagainstlow-SNRsources.Forthehighestvaluesofsubmm uxdensityandnoiseweseesimilaritieswithtrendsdiscussedbyIvisonetal.(2002)forthe8-mJySurvey,i.e.thebrightestsourceineach eldliesinaregionwithhighnoise,andneitherhasarobustradiocounterpart.

TheSXDFradioidenti cationrateversusrawsubmmSNRshowsasteepdeclinebelowanSNRof4;after uxdeboostingthiseffectismitigatedsomewhat,withmatchingtrendsintheSXDFandLH elds.Itisnoteworthythattheoverallradiorecoveryrate

inSXDFisover10percenthigherthanintheLH eld,despitetheshallowerdepthoftheSXDFradioimaging.Weattributethistothreeeffects,eachofwhichwebelievecontributestotheunexpect-edlylowLHidenti cationrate: rst,theLHradioimageisasinglepointing,designedoriginallytoidentifySMGsinthesmall8-mJySurvey eld(cf.amosaicofthreeinSXDF),sotheperniciousef-fectofbandwidthsmearingwillbeevidentforasigni cantlylargerfractionoftheSHADES eldinLHthaninSXDF;second,al-thoughitisclearlyusefultoworkwiththebestpossibleradiodata,deepimaginginevitablyyieldsmorefaint,unrelated,backgroundsources,causingPvaluesforrelativelybrightcounterpartstoriserelativetothosecalculatedforalowersourcedensity;third,itispossible(thoughithasyettobeshownunambiguously–Ivisonetal.2002;Chapmanetal.2004;Muxlowetal.2005)thatasignif-icantfractionoftheemissioninsomeSMGsisresolvedbyhigh-resolutionradiodata.Thattheseeffectsaresigni cant,collectively,isdemonstratedbythesigni cantlyhigherSMGidenti cationrateintheshallower,lowerresolutionSXDFdata;inaddition,sevenLHSMGs(LOCK850.10,.34,.37,.38,.40,.77and.100)aredetectedrobustlyonlyinthenoisier,low-resolutionradioimage,thoughwenotethatinseveralcasesthe4.2-arcsecFWHMimagealonedoesnotallowustodifferentiatebetweenplausiblespectroscopictar-gets.Thereareseverallessonshere:ensureinterferometricdatacontainanadequatefractionofshortspacings–asynthesisedbeamwith1.5–2arcsecFWHMprovidesagoodcompromiseforidenti -cationofFIR-luminousgalaxies;wherenecessary,i.e.whentheareaofinterestissimilartothatoftheradiointerferometer’spri-marybeamandthespectralresolutionispoor(δλ/λ<1000),obtaindatainacompactmosaicofpointingsratherthanasingle,deeppointing.

6.2Mid-IRtrends

Thetrendofoverallrecoveryrateisreversedinthemid-IR,theLHyieldinga20percenthigheridenti cationratethantheSXDF.Thereasonisobvious,however:itisduetothesubstantialextradepthoftheLHSpitzer24-µmdata(σ=11versus47µJy).OnlyoneSMGisidenti edsolelyonthebasisofitsmid-IRemissioninSXDFcomparedwithtenintheLH.Forboth eldsthedeclineatlowdeboostedSNRislessmarkedthantheradiotrend.Againstsubmm uxdensityandnoise,the24-µmidenti cationtrendsforboth eldsmatchthoseatradiowavelengths(withtheaforemen-tioned20percentoffsetfortheSXDFsources);theverybrightestsourcesagainlackrobustcounterparts.

6.3Overalltrends

ThelowerrowofplotsinFig.6showtheoverallidenti cationtrends–thefractionofsourcesidenti edat1.4GHzand/or24µm,includingthethreecasesmentionedin§3whereweakradioand24-µmcounterpartsarecoincident(oneofwhichisthebrightestLHsource,LOCK850.34).

Theidenti cationtrendsaresimilarforthetwoSHADES elds:identi cationisessentiallycompleteaboveadeboostedsubmmSNRof～4withanabruptstepdownto60–70percentthereafter;also,successratesimproveasthesubmmnoisedeclines.TheSXDFidenti cationratetailsoffbelowadeboostedsubmmSNRof2.5andatsubmm uxdensitiesbelow5mJy.ThismaybeduetothelimiteddepthoftheSXDFradioand24-µmimagingratherthananyde ciencyoftheSXDFcatalogue,butwenotethatitisastrongtendency.

c0000RAS,MNRAS000,000–000