ebook img

Precursors prior to Type IIn supernova explosions are common: precursor rates, properties, and correlations PDF

1.1 MB·
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Precursors prior to Type IIn supernova explosions are common: precursor rates, properties, and correlations

SUBMITTEDTOAPJ PreprinttypesetusingLATEXstyleemulateapjv.5/2/11 PRECURSORSPRIORTOTYPEIINSUPERNOVAEXPLOSIONSARECOMMON:PRECURSORRATES,PROPERTIES, ANDCORRELATIONS ERANO.OFEK1,MARKSULLIVAN2,NIRJ.SHAVIV3,AVIRAMSTEINBOK1,IAIRARCAVI1,AVISHAYGAL-YAM1,DAVIDTAL1, SHRINIVASR.KULKARNI4,PETERE.NUGENT5,6,SAGIBEN-AMI1,MANSIM.KASLIWAL7,S.BRADLEYCENKO8,RUSSLAHER9, JASONSURACE9,JOSHUAS.BLOOM6,ALEXEIV.FILIPPENKO6,ANDOFERYARON1 SubmittedtoApJ 4 ABSTRACT 1 There is a growing number of supernovae (SNe), mainly of Type IIn, which present an outburst prior to 0 theirpresumablyfinalexplosion. These precursorsmayaffectthe SN display,and arelikely relatedto some 2 poorlychartedphenomenainthefinalstagesofstellarevolution.Herewepresentasampleof16nearbyType n IInSNeforwhichwehavePalomarTransientFactory(PTF)observationsobtainedpriortotheSNexplosion. a By coaddingthese imagestaken priorto theexplosionin time bins, we searchforprecursorevents. We find J five Type IInSNe thatlikely have atleast one possible precursorevent(PTF10bjb; SN2010mc[PTF10tel]; 1 PTF10weh; SN2011ht; PTF12cxj), three of which are reported here for the first time. For each SN we 2 calculatethecontroltime(i.e.,theamountoftimeoursurveywasabletodetectaneventbrighterthanagiven luminosity). Based on this analysis we find that precursor events among SNe IIn are common: assuming a ] homogeneouspopulation,attheone-sided99%confidencelevel,morethan50%ofSNeIInhaveatleastone E pre-explosionoutburstthatisbrighterthan3 107L (absolutemagnitude 14)takingplaceupto1/3yrprior H to the SN explosion. The average rate of su×ch prec⊙ursorevents duringthe−year prior to the SN explosionis . likelylargerthanoneperyear(i.e.,multipleeventsperSNperyear),andfainterprecursorsarepossiblyeven h morecommon. Wealsofindpossiblecorrelationsbetweentheintegratedluminosityoftheprecursor,andthe p SNtotalradiatedenergy,peakluminosity,andrisetime. Thesecorrelationsareexpectediftheprecursorsare - o mass-ejectionevents, and the early-timelight curveof these SNe is poweredby interactionof the SN shock r andejectawithopticallythickcircumstellarmaterial. t s Subjectheadings: stars:mass-loss—supernovae:general—supernovae:individual:SN2010mc,PTF10bjb, a SN2011ht,PTF10weh,PTF12cxj,SN2009ip [ 1 v 1. INTRODUCTION eral theoretical mechanisms have been suggested to explain 8 Influx-limitedsynopticsurveys,afewpercentofalldiscov- thishighmasslossinthefinalstagesofstellarevolution(e.g., 6 ered supernovae (SNe) show narrow- to intermediate-width Rakavy et al. 1967; Woosley et al. 2007; Arnett & Meakin 4 ( 30–3000kms 1) hydrogen and helium emission lines. 2011; Chevalier 2012; Quataert & Shiode 2012; Shiode & 5 ∼ − Quataert2013;Soker&Kashi2013). These are dubbed Type IIn SNe (Schlegel 1990; Filippenko . Recently, five SNe with candidate pre-explosionoutbursts 1 1997;Kieweetal. 2012)andprobablyhavemassiveprogen- (precursors)have been detected a few months to years prior 0 itors (e.g., Gal-Yam & Leonard 2009). Their emission lines totheSNexplosion(Foleyetal. 2007;Pastorelloetal. 2007; 4 likely originatefrom relativelyabundantcircumstellarmate- Mauerhan et al. 2013a; Pastorello et al. 2013; Ofek et al. 1 rial (CSM) aroundthe SN progenitorstar (e.g., Chevalier & 2013b;Corsiet al. 2013; Fraser etal. 2013). In mostcases : Fransson1994;Chugai&Danziger1994;Chugaietal. 2003; v these precursors were detected from SNe IIn, or closely re- Ofek et al. 2007, 2010, 2014a; Smith et al. 2008, 2009), i latedevents. X ejectedonlyashorttime(oftheorderofmonthstodecades) Thefrequencyandpropertiesofthese precursorsarecriti- r prior to the SN explosion (Dessart et al. 2009; Gal-Yam & calforpinpointingtheeruptionmechanismsandunderstand- a Leonard 2009; Ofek et al. 2010; Ofek et al. 2013b). Sev- ing their effect on the eventualSN optical display, and may changeourviewofthefinalstagesofmassivestarevolution. 1Benoziyo Center for Astrophysics, Weizmann Institute of Science, Therefore,wehaveconductedasearchforprecursoreventsin 76100Rehovot,Israel asampleofnearbySNeIInforwhichwehavepre-explosion 2School of Physics and Astronomy, University of Southampton, observationsfromthePalomarTransientFactory(PTF10;Law SouthamptonSO171BJ,UK 3RacahInstituteofPhysics,TheHebrewUniversity,91904Jerusalem, etal. 2009;Rauetal. 2009).Oursamplecontains16SNeIIn Israel andwefoundprecursorseventsforfiveoftheSNeinoursam- 4CahillCenterforAstronomyandAstrophysics,CaliforniaInstituteof ple.Forthefirsttime,weestimatetherateofsucheventsand Technology,Pasadena,CA91125,USA showthattheyarecommon.Furthermore,weinvestigatepos- 5LawrenceBerkeleyNationalLaboratory,1CyclotronRoad,Berkeley, siblecorrelationsbetweenthepropertiesoftheprecursorsand CA94720,USA 6Department of Astronomy, University of California, Berkeley, CA theSNe. 94720-3411,USA Thepaperisorganizedasfollows.WedescribetheSNsam- 7ObservatoriesoftheCarnegieInstitutionforScience,813SantaBar- ple in 2, while the observations are presented in 3. The baraSt.,Pasadena,CA91101,USA § § 8AstrophysicsScienceDivision,NASA/GoddardSpaceFlightCenter, methodologyofprecursorselectionisdiscussedin 4andour § MailCode661,Greenbelt,MD,20771,USA 9SpitzerScienceCenter,CaliforniaInstituteofTechnology,M/S314-6, Pasadena,CA91125,USA 10http://ptf.caltech.edu/iptf/ 2 Ofeketal. candidatesarepresentedin 5. We giveourcontrol-timees- 4.1. DetectionMethods § timate in 6, and the precursor rate in 7. The CSM mass Ourcandidateprecursorselectionisbasedontwochannels. § § estimateisdiscussedin 8,thecorrelationsbetweentheSNe Thefirstchannelidentifiesprecursorsthatweredetectedina § andprecursorpropertiesin 9,andtheresultsin 10. singleimagepriortotheSNexplosion,atthe6s level,with- § § outtheneedtocoadddata.Thesecondchannelidentifiespre- 2. SAMPLE cursorsthat were detected in the coadd data at the 5s level, Our sample is based on SNe IIn that show intermediate- using15-daybinsandusingonlybinsthatcontainmorethan widthBalmeremission lines. FrominspectionofmanySNe fivefluxmeasurements. spectra obtained by PTF, we find that some SNe show this The second channel significantly increases the sensitivity hallmarkofSNeIInatearlytimes(afewdaysaftertheexplo- of PTF to precursor events by effectively coadding images sion),buttheselinesdisappearonatimescaleofaweek. Itis of a SN location in time bins of 15days. These time bins possiblethattheseSNealsosufferfromamoderatemass-loss often include a large number of observations, extending to ratepriorto theexplosion(e.g.,Gal-Yametal. 2014;Yaron depths beyond the nominal PTF survey limiting magnitude, et al. 2014). However, we excludefrom our sample objects andreachinganR-bandlimitingmagnitudeof<23.5. How- for which the spectra evolve into those of normal SNe II a ever,coaddingimagesthemselvesinarbitrarytimebins,and fewweeksafterexplosion.Examplesforsuchobjectsinclude thenconductingimage-subtractionanalysisoneachtrialbin, PTF11iqbandPTF10uls(Ofeketal. 2013a). isa relativelyexpensiveoperation. Instead,we carefullyap- Another important criterion for our SNe selection is that plyimagesubtractiontoindividualimages,andforeachim- they have a large number of pre-explosion images from the agewesavethefluxresidual(negativeorpositive)attheloca- PTF survey (Law et al. 2009; Rau et al. 2009). The exis- tionoftheSN.Wecanthencoaddthescalarfluxresidualsin tence of a large number of images is critical, as precursors anytemporalcombinationwe desire. Thismethodwas used mayberelativelyfaintanditisdesirabletocoaddimagesin inthecaseofSN2010mc(PTF10tel;Ofeketal. 2013b)and order to get a limiting magnitude that is deeper than that of PTF11qcj(Corsietal. 2013). the nominal survey. The SNe in our sample were found by Inthe firstchannel,theuncertainty(i.e.,s ) wasestimated thePTFaswellasamateurastronomers,theLickObservatory based on the Poisson noise propagated through the image- SNSearch(LOSS;Lietal.2000;Filippenkoetal. 2001),and subtractionpipeline,whileforthesecondchannel,wecalcu- theCatalinaReal-timeTransientSurvey(CRTS;Drakeetal. latederrorsin each binusing the bootstraptechnique(Efron 2009). WeselectedonlynearbySNe,foundwithin400Mpc, 1982).Wenotethatinmostcasesthebootstraperrorsarecon- forwhichwehaveadecentnumberofpre-explosionobserva- sistent with the uncertainties derived based on the standard tions. Table1listthe16SNeinoursample. deviation of the points in each bin, and the expected Pois- son noise. Therefore, our bootstrap error estimate suggests that the statistical uncertainties produced by the subtraction 3. OBSERVATIONS pipelinearerealistic. We usedPTFobservationsofthe SNe in oursample. The InordertokeepoursearchuniformweuseonlythePTFR- PTFdatareductionisdescribedbyLaheretal. (inprep.),and banddata for oursearch and analysis, as PTF was primarily thephotometriccalibrationisdiscussedbyOfeketal. (2012a, anR-bandsearch. However,wealsoshowtheg-banddatain 2012b).Oursearchisbasedonimagesubtraction,andtheflux thevariousplotswhereavailable;theamountofg-banddata residualsintheindividualimagesubtractionsforalltheSNe issmallincomparisontotheR-band. inoursamplearelistedinTable2. Candidateprecursoreventsdetectedviaoneofthechannels Figures 1–2 show the light curves before explosion (first arediscussedin 5. Intheinitialsearchwedonotattemptto and third columns) and after explosion (second and fourth § use different time bins. This decision was made in order to columns) of all the SNe in our sample. The pre-explosion limitthenumberofstatisticalexperiments,whichmayaffect light curve shows the median flux, relative to the reference the significance of our results. We note, however, that care- image flux, in 15-daysbins. Only binscontaining 6 mea- fulexaminationofspecificeventswithlongertimebinsmay ≥ surementsarepresented. The“+”signsshowsthelowerand containprecursoreventsthatarenotdiscussedhere. Forex- upper5s errorrelativetothereferenceimage,whilethesolid ample, in Corsi et al. (2013) we report on a possible faint lines connect consecutive bins. The errors where calculated (absolutemagnitude 13), severalmonthslong, brightening using the bootstrap error on the mean. The vertical dashed − inthelightcurveoftheTypeIcPTF11qcjabout2.5yrprior lines show the estimated explosion time and time of maxi- toitsexplosion. mumlight(seeTable1). Wenotethatseveralobjectsshowpointsthataremarginally We obtained spectra of our SNe using various telescopes, belowthelower5s errorthreshold. Itispossiblethatthisis and the log of selected observationsis presented in Table 3. causedbyrealvariabilityoftheprogenitor(e.g.,Szczygiełet Some of the spectra are presented in this paper, while the al.2012),buthereweconcentrateontheoutburstsratherthan restareavailableelectronicallyfromtheWISeREPwebsite11 possibledimmings(seealso 5.6). (Yaron&Gal-Yam2012). § 4.2. Tests 4. PRECURSORCANDIDATESELECTION We performed several tests to verify the reliability of our Inthissectionwedescribethemethodsweusedtofindthe methodology,especiallyagainstfalsealarms. Inordertotest precursorcandidates. In 4.1wepresentthesearchmethods, detectionsmadebythefirstchannel(i.e.,precursorsdetected § while in 4.2 we discuss the reliability of our methodology insingleimages)weextractedthelightcurvesatrandompo- § andthefalse-alarmprobabilityoftheprecursorcandidates. sitions on top of the same host galaxy, but shifted in posi- tionrelativeto theSN.We foundthattypicallytheprobabil- 11http://www.weizmann.ac.il/astrophysics/wiserep/ ity to get a 6s detection is less than 0.1% per image. This Supernovaprecursors 3 400 16 150 13.4 100 2213.6 200 22 50 2313.8 2316.5 0 14 0 −50 14.2 17 −100 14.4 −200 −150 PTF10aaxf 14.6 PTF10aaxf −400PTF10aazn 17.5 PTF10aazn −200 −150 −100 −50 0 0 20 40 60 80 100 −300 −200 −100 0 0 20 40 60 80 100 17.5 300 600 19.5 18 200 400 100 2218.5 20 23 0 200 −100 19 222320.5 0 −200 19.5 −300PTF10achk PTF10achk −200PTF10bjb 21 PTF10bjb 20 −500 −400 −300 −200 −100 0 0 20 40 60 80 100 −400 −300 −200 −100 0 0 20 40 60 80 100 100 2217 18.5 300 e 19 d agnitu 50 231189 120000 2223192.05 M 0 0 x/ 20 −100 20.5 Flu −50 −200 21 21 21.5 −300 −100PTF10fjh 22 PTF10fjh PTF10gvf 22 PTF10gvf −300 −200 −100 0 0 20 40 60 80 100 −400 −300 −200 −100 0 0 20 40 60 80 100 200 17 18 22 100 2218.5 100 18 23 50 2319 19 0 0 19.5 20 −50 20 −100 −100 20.5 21 PTF10tel PTF10tel PTF10weh PTF10weh −200 21 −300 −200 −100 0 0 20 40 60 80 100 −400 −300 −200 −100 0 0 20 40 60 80 100 18.5 600 2000 18 400 19 1000 200 19 19.5 22 0 23 0 222320 −200 20 20.5 −1000 −400 21 21 −600PTF11fzz PTF11fzz −2000PTF12cxj PTF12cxj 21.5 −600 −400 −200 0 0 20 40 60 80 100 −800 −600 −400 −200 0 0 20 40 60 80 100 t−t [day] rise FIG.1.—ThelightcurvesoftenSNe(namesarelistedineachpanel). ForeachSNtwoplotsareshown(sidebysideinthefirstandsecondcolumnsand inthethirdandfourthcolumns). ColumnstwoandfourshowthelightcurveaftertheSNexplosion,whilethepanelsincolumnsoneandthreegiveonlythe coaddedfluxresidual(relativetothereferenceimage)priortotheSNexplosion(iftherearemorethanfiveobservationspertimebin).Thetwoverticaldashed linesshowtheassumedexplosiondate(trise)androughmaximum-luminositydate(tpeak;listedinTable1). Timeismeasuredrelativetotrise. Theredcircles representR-bandobservationswhilethebluesquaresshowg-banddata. AllofthemeasurementsareinthePTFmagnitudesystem(Ofeketal. 2012a;2012b). Thezeropointofthefluxresiduals(firstandthirdcolumns)is27,withtheexceptionofPTF10aazn(27.895),PTF10bjb(27.14),andPTF10tel(27.442). The tickmarksontheright-handaxesofthefirstandthirdcolumnsshowthefluxesthatcorrespondtoPTFmagnitudesof22and23. Inthefirstandthirdcolumns, filledsymbolsshowthefluxmeasurements,in15-daysbins.Onlybinscontainingsixormoremeasurementsareused.Theplussignsrepresentthe5s upperand lowerlimits,asestimatedfromthebootstraptechniqueineachbin(Efron1982).Ifplussignsarefoundinconsecutivebinstheyareconnectedbyasolidline. 4 Ofeketal. TABLE1 SNSAMPLE Name Type a (J2000) d (J2000) MR,peak z DM trise tpeak FAP (deg) (deg) (mag) (mag) (day) (day) SN2010jl IIn 145.722236 +09.495037 20.6 0.011 33.44 55474 55494 0.01 SN2010jj IIn 31.717743 +44.571558 −18.0: 0.016 34.22 55512 55525 0.02 PTF10achk IIn 46.489751 10.522491 −18.7 0.0327 35.77 55535 55550 0.00 PTF10bjb IIn? 192.424667 −10.800159 −< 16.4 0.026 35.27 55326 55434 ... SN2010bq IIn 251.730659 −+34.159641 1−8.5 0.032 35.73 55296 55311 0.00 PTF10gvf IIn 168.438496 +53.629126 −18.8 0.081 37.82 55321 55339 0.00 SN2010mc IIn 260.377817 +48.129834 −18.5 0.035 35.93 55427 55447 0.00 PTF10weh IIn 261.710251 +58.852064 −20.7 0.138 39.06 55453 55504 0.02 PTF11fzz IIn 167.694502 +54.105220 −20.3 0.082 37.85 55721 55783 0.00 PTF12cxj IIn? 198.161181 +46.485090 −17.3 0.036 35.96 56033 56048 0.01 SN2011cc IIn 248.456000 +39.263528 −18.1: 0.0319 35.72 55624 55741 0.02 SN2011fx IIn 4.498167 +24.562778 −17.1: 0.0193 34.61 55780: 55803: 0.00 SN2011ht IIn 152.044083 +51.849194 −16.8 0.004 30.96 55830: 55880: 0.00 SN2011hw IIn/Ibn 336.560583 +34.216417 −19.1: 0.023 34.99 55860 55884: ... SN2011iw IIn 353.700833 24.750444 −18.1: 0.023 34.99 55819 55894: 0.04 SN2012as IIn 231.285500 −+37.963722 −18.0: 0.0297 35.56 55968: 55976: 0.00 − NOTE.—TheSNesample. TyperefertoSNtype,a (J2000)andd (J2000)aretheJ2000.0rightascensionanddeclination,respectively. MR,peakisthepeak absoluteR-bandmagnitude,zistheSNredshift,DMisdistancemodulus,triseistheMJDoftheestimatedstartoftheSNrise,andtpeakistheMJDofthelight curvepeak.Colonsignindicatesanuncertainvalue. ThetriseforSN2011iwisbasedonadetectionincoaddedPTFdataanditappearsasapartofthefastrise followingthisdetection.FAPisthefalse-alarmprobabilitytodetectaprecursorbycoaddingimagesin15-daybinsasestimatedusingthebootstrapmethod(see §4.2). Thevaluesarebasedon100bootstrapsimulationsandthereforetruncatedtotwofiguresafterthedecimalpoint. SNewithnodataarethoseinwhich theprecursorisclearlydetectedinmanyindividualimages,andtherefore,thebootstrapanalysisonthecoaddeddataisineffective(see§4.2). SNebelowthe horizontallinewerenotdetectedbyPTF,althoughPTFhavepre-explosionimagesoftheirlocations. References: SN2010jl:PTF10aaxf;Newtonetal.(2010);Stolletal.(2011);Smithetal.(2011);Chandraetal.(2012);Ofeketal.(2013a;2014a). SN2010jj:PTF10aazn;Rich(2010);Silvermanetal.(2010). PTF10achk:reportedhereforthefirsttime. PTF10bjb:reportedhereforthefirsttime. SN2010bq:PTF10fjh;Duszanowicz(2010);Challisetal.(2010);Ofeketal.(2013a). PTF10gvf:reportedhereforthefirsttime. PTF10tel:Ofek(2012);Ofeketal.(2013a;2013b). PTF10weh:reportedhereforthefirsttime. PTF11fzz:reportedhereforthefirsttime. PTF12cxj:reportedhereforthefirsttime. SN2011cc:Masonetal.(2011). SN2011fx:Ciabattarietal.(2011). SN2011ht:Bolesetal.(2011);Prietoetal.(2011);Romingetal(2012);Mauerhanetal.(2013b) SN2011hw:Dintinjanaetal.(2011). SN2011iw:Mahabaletal.(2011). SN2012as:Jinetal.(2012). TABLE2 SNOBSERVATIONS Name Band MJD-trise MJD Mag MagErr LimMag Flux FluxErr (day) (day) (mag) (mag) (mag) (counts) (counts) PTF10aaxf R 232.84900 55241.15100 81.555 0.913 21.441 66.4 55.8 PTF10aaxf R −232.49900 55241.50100 82.323 −0.985 20.590 −134.7 122.2 PTF10aaxf R −229.61100 55244.38900 21.890 −0.392 21.802 −110.7 40.0 PTF10aaxf R −229.56800 55244.43200 82.112 0.653 21.247 110.9 66.7 PTF10aaxf R −223.68700 55250.31300 81.913 −1.045 20.936 − 92.3 88.8 − − − NOTE. —PhotometricmeasurementsandfluxresidualsofthetheSNeinoursample. Magnitudearecalculatedinlaptitudes,andtheyhavemeaningonly whensmallerthanthelimitingmagnitude. ThistableispublishedinitsentiretyintheelectronicversionofApJ.Aportionofthefulltableisshownherefor guidanceregardingitsformandcontent. Supernovaprecursors 5 16.5 150 19 1500 17 100 22 1000 19.1 500 17.5 50 23 0 222318 0 19.2 −500 18.5 −50 −1000 19.3 19 −100 −1500SN2011cc 19.5 SN2011cc −150SN2011fx 19.4 SN2011fx −400 −300 −200 −100 0 0 20 40 60 80 100 −300 −200 −100 0 0 20 40 60 80 100 19 19.5 1000 e d 19.2 Magnitu 5000 222319.4 5000 222320 x/ 19.6 Flu −500 19.8 −500 −1000SN2011ht SN2011ht SN2011hw SN2011hw 20 20.5 −300 −200 −100 0 0 20 40 60 80 100 −600 −400 −200 0 0 20 40 60 80 100 20.5 18.3 1000 100 22 20.6 18.4 50 2320.7 500 18.5 0 20.8 0 222318.6 20.9 18.7 −50 21 −500 18.8 21.1 18.9 −100 SN2011iw SN2011iw −1000SN2012as SN2012as −400 −300 −200 −100 0 0 20 40 60 80 100 −400 −300 −200 −100 0 0 20 40 60 80 100 t−t [day] rise FIG.2.—LikeFigure1butforsixadditionalSNe. 1100 TABLE3 1000 LOGOFSPECTROSCOPICOBSERVATIONS 900 SNname MJD Telescope Instrument 800 (day) ns bi 700 SN2010jl 55505 Keck-I LRIS s1 SN2010jj 55505 Keck-I LRIS 0. 600 PTF10achk 5555554474 ULiHck883m SKNasItFS er in 500 b PTF10bjb 55262 Keck-I LRIS m 400 u SN2010bq 55300 Gemini-N GMOS N 300 PTF10gvf 55322 Keck-I LRIS SN2010mc 55434 Gemini-N GMOS 200 55442 Lick3m Kast 55449 KPNO4m RCSpec 100 55455 Lick3m Kast 0 PTF10weh 55479 KPNO4m RCSpec −6 −4 −2 0 2 4 6 Residuals [s ] 55502 Lick3m Kast 55503 KPNO4m RCSpec FIG.3.—Thedistributionoffluxresidualsins units,atalltheepochsand PTF11fzz 55736 Hale5m DBSP at16randompositionsaroundthepositionofPTF12cxj,onthesamehost 55873 WHT ISIS galaxy. PTF12cxj 56035 Gemini-N GMOS NOTE.—MJDistheobservationmodifiedJulianday. (coadding flux residuals in 15-day time bins), for each SN werun100bootstrapsimulationsinwhichwemixedtheflux is because the noise is not distributed normally (e.g., there residualsandtimes,andbinnedthedataagainin15-daybins are outliers) owing to occasional subtraction artefacts, cos- with the same selection criteria. The probability to have a mic rays, or asteroids in the field of view. For some SNe single detection, based on these simulations, is listed in col- wehavehundredsofimages,andthereforetheprobabilityfor umn FAP (false-alarm probability) in Table 1. This proba- a detection is on the order of a few percent per SN. There- bility may approach several percent for the entire sequence fore,weconsiderasgoodcandidatesonlyeventswhichhave ofimages. However,afterpassingtheselectioncriteria,each two consecutive detections (see 5). Figure 3 illustrates the candidate is tested using various binning schemes, and only § histogramof“random”-positionfluxresidualsinunitsofthe sourcesthatshowtwoconsecutiveindependentdetectionsare flux-residualerrors(i.e., s )forthecaseofrandompositions consideredasgoodcandidates(see 5). § aroundPTF12cxj. It is apparentthat even thoughthere is a We also search for correlationsbetween the flux residuals smallexcessaroundpositiveresiduals,theprobabilityofget- and airmass, and flux residuals and the amplitude of the ex- tingresidualswhicharebiggerthan6s issmall. pected atmospheric refraction. Indeed, we find marginally In order to test detections made using the second channel significant correlations between these properties; however, 6 Ofeketal. theyaretoosmalltoaffectourresults. PTF observations detected the outburst in individual To summarize, tests of our methodology suggest that the images, suggesting that the outburst was already active false-alarmprobabilityforasingle-pointprecursordetection, 11monthspriortotheexplosion. AllofthePTFphotometric inanentireSNdataset,is 5%perobject. Inordertoavoid measurementsarelistedinTable2. WenotethatPTFdidnot ∼ false detections, we consider as precursor candidates only observetheSNitself. Assumingtherewasasingleprecursor, cases which show at least two detections clustered in time thepropertiesandenergeticsofthiseventarelistedinTable4. (i.e.,assumingthenoiseisnotcorrelated). 5.3. PTF10bjb 5. CANDIDATEPRECURSOREVENTS PTF10bjb was discovered by PTF on 2010 Feb. 16, and OursearchyieldedseveralSNewithcandidateprecursors. the onlyspectrumobtainedon 2010Mar. 7 resemblesthose The precursors of PTF10bjb, SN2010mc, and SN2011ht ofluminousbluevariablesandSNe IIn. Close inspectionof were clearly detected in individual images (first channel), thelightcurvesuggeststhattheSNwasdiscoveredwhileina whilePTF12cxjandPTF10wehareweaklydetectedincoad- pre-explosionhighstate, andthatthespectrumwasobtained ded images (second channel). Although these two weak priortotheexplosion. ThespectrumshowsBalmeremission eventspassallourtestsandweconsiderthemtobereal,given lines, the broadercomponentof which has a velocity (s ) of thelowsignal-to-noiseratioofthedetections,wealsodiscuss about 600kms 1. Also detected are emission lines of HeI − ourresultsassumingthesetwoeventsarenotreal. Imagesof andCaII.TheHeIl 5876lineshowsanarrowP-Cygnipro- alltheprecursorsarepresentedinFigure4. filewithavelocityof 600kms 1,whiletheHa lineshows − aP-Cygniprofilewith∼avelocityof 300kms 1. Thespec- − 5.1. PTF10tel trumofthiseventispresentedinFig∼ure6. PTF10telisdiscussedindetailbyOfeketal. (2013b). For The light curve of PTF10bjb is shown in Figure 7. Be- completeness,herewesummarizethemainpropertiesofthis tween 80daysupto0dayspriortotheassumedexplosion ∼ event. TheobservationsofPTF10telshowanoutburstabout time of this SN, which is based on the fast rise in the SN 40dayspriortotheprobableexplosion. Thisprecursorisde- light curve, we detect a significant excess in the flux resid- tectedviaboththefirstandsecondchannels. Ourphotomet- uals. This excess had a peak absolute magnitude of 15.1 − ric and spectroscopic observations suggest that this event is and it lasted for 110days. Two years after trise, the SN ∼ produced by an energetic outburst releasing 10 2M at is detected at a flux level which is a factor of 5 dimmer − ∼ typical velocities12 of 2000kms 1, and powe∼red by at l⊙east than the pre-explosion outburst. Assuming zero bolometric − correction,thepeakbolometricluminosityofthisoutburstis 6 1047ergofenergy. ×For completeness we show in Figure 5 the light curve of ∼3×1041ergs−1,anditsradiatedenergyis∼2.6×1048erg. PTF10telfromOfeketal. (2013b). Thephysicalparameters In addition, there is a possible detection of the progenitor ofthisSNandprecursorarelistedinTable4. at an absolute magnitudeof roughly 14 about a year prior − to the SN explosion. However, given that there is a single 5.2. SN2011ht detection at this time, we do not consider this to be a good SN2011ht was discovered by Boles (2011) on 2011 Sep. precursor candidate (see 4.2). The physical parameters of 29.2(UTCdatesareusedthroughoutthispaper),atapparent theSNandtheprecursora§relistedinTable4. magnitude17.0. Basedonaspectrumobtainedon2011Sep. 30, Pastorello et al. (2011) suggested that it is a SN impos- 5.4. SN2011hw tor sharing some similarities with the eruption of the lumi- SN2011hw was discovered by Dintinjana et al. (2011) nous blue variable UGC2773-OT (Smith et al. 2010). The on 2011 Nov. 18.7, at apparent magnitude 15.7. Valenti et spectrumshowsnarrowlinesofH,CaII(H&Kandthenear- al. (2011) reported that a spectrum obtained on 2011 Nov. infrared triplet, with P-Cyg profiles), and a forest of narrow 19.8 shows it to be similar to the transitional Type IIn/Ibn FeIIabsorptionfeatures. Also, prominentNaI D, ScII,and SN2005la (Pastorello et al. 2008). The spectrum is blue BaIIfeaturesaredetectedinabsorption. Prietoetal. (2011) and exhibits emission lines of H and HeI. The most promi- obtainedafurtherspectrumon2011Nov.11.5.Theyreported nent HeI lines compete in strength with Ha . The FWHM substantialevolutionwithrespectto theinitialclassification, velocity of Ha is 2700kms 1, while that of HeI l 5876 is withstrongBalmerandweakerHeIandFeIIemissionlines − superposedonabluecontinuum. Basedonthespectrumand 2000kms−1. We notethatthe apparentmagnitudeofthis ∼ the SN absolute magnitude,they suggestedthat it is likely a SNreportedbyDintinjana&Mikuz(2011)correspondstoan SN IIn. Thisis supportedby Rominget al. (2012),who re- absolute magnitude of about 19.3. Our initial search sug- − portedontheSwift-UVOTobservationsofthisSN,detecting gestedthatthereisadetectionofanoutburst 3monthsprior ∼ a 7mag rise in the UVW2 band over 40days, peaking at a totheSNexplosion. However,sincethisisbasedonasingle u-bandmagnitudeofabout 18. detection,wedonotconsiderthistobeagoodcandidate. − Fraseretal. (2013)reportedonanoutburstpeakingataz- bandabsolutemagnitudeof 11.8detected 9monthsprior 5.5. PTF10weh − ∼ totheexplosion,andwiththelastdetection 4monthsbefore PTF10weh was discovered by PTF on 2010 Sep. 22.14. ∼ theexplosion.TheeventwasdetectedbybothPanSTARRS-1 The SN brightened to a peak absolute magnitude of about andthe CatalinaSkySurvey. Thedurationofthe outburstis 20.7over 40days. Aspectrumtakenon2010Oct. 10ex- not well constrained, and it can be either a single event that −hibitsa blue∼continuumtogetherwithBalmer emissionlines lastedformorethan4monthsormultipleshorterevents. as well as HeI and HeII. The widest componentof the Ha linehasavelocity(s )of 1000kms 1. Thefirsttwospec- − 12Thisisthelargers ofatwo-componentGaussianfit. traofthissourceareshow∼ninFigure8. Supernovaprecursors 7 FIG.4.—Imagesoftheprecursors,SN,andnondetectionsfortheSNePTF10weh,PTF10tel,SN2011ht,PTF12cxj,andPTF10bjb(fromtoptobottom).In eachrow,thefiveimagespresent(lefttoright)thereferenceimage,thepre-SNexplosionnondetection,thebrightestprecursorforeachobject(wheremorethan onewasdetected),theSN,andapost-SNimage(ifavailable). Thecontoursfromthereferenceimagearealsooverplotted. Alloftheprecursorscanbeclearly seen,exceptPTF10wehwhichismarginallydetected. AdetailedinspectionoftheimagesofPTF10wehinwhichtheprecursorisdetecteddonotrevealany cosmeticproblems(e.g.,cosmicrays)neartheprecursorlocation,andthebalanceofevidenceisthatthisprecursorisreal(seediscussionregardingfalse-alarm probabilityin§4.2).However,inFigure19wealsoshowthecorrelationsbetweentheSNandprecursorproperties,excludingPTF10weh. ThelightcurveofthisSNandthecandidateprecursorevent zero bolometric correction, this correspondsto a luminosity isshowninFigure9.Thereisapossibleprecursor 3months of 1.9 1042ergs 1.Thelowerlimitontheradiatedenergy priortotheSNrise,lastingfor 40days.Aclose∼-upviewof of∼thepr×ecursoris −5 1048erg.Thephysicalparametersof theprecursorlightcurveispres∼entedinFigure10. Figure11 theSNanditsoutb∼urst×arelistedinTable4. givesanotherversionoftheprecursorlightcurve,showingall Inaddition,thepre-explosionlightcurveofthisSNreveals ofitsindividualfluxmeasurements,alongwithfluxmeasure- additional possible detections 11–14months prior to the SN mentsatrandompositionsinthehostgalaxyandothernearby explosion. However,giventhatthesedetectionsarenotcon- galaxies. Inspectionofthedatarevealsthatthedetectionde- secutive in time, we do not regard them as good precursor pendson the binningscheme. However,giventhatthere are candidates. fourtemporallyadjacentpointsthatdeviatebymorethan5s , we consider this to be a good precursor candidate. The ab- solute magnitude of this precursor is about 17. Assuming 5.6. PTF12cxj − 8 Ofeketal. TABLE4 PRECURSORPROPERTIES Name D t d t LSN,peak ESN Lprec,peak Eprec Ha narrow Ha wide e−1MCSM (day) (day) (ergs 1) (erg) (ergs 1) (erg) (kms 1) (kms 1) (M ) − − − − ⊙ SN2010mc 20 30 6.9 1042 2 1049 2.6 1041 6 1047 500 2000 0.015 PTF10bjb −85 110 >8× 1041 >×2.0 1049 3.3×1041 2.×4 1048 ∼300 ∼600 0.56 SN2011ht −205 210 1.5×1042 2.5 1×049 1.5×1040 2 ×1047 ∼260 ∼1800 0.06 PTF10weh −80 40 5.3×1043 7.2×1050 1.9×1042 5×1048 100 1000 0.5 PTF12cxj-A −10 7 2.3×1042 8 ×1048 2.2×1041 9×1046 ∼200 ∼1200 0.014 PTF12cxj-B −700 15 × × 4.3×1040 6×1046 ∼ ∼ SN2009ip −25 50 4.6 1042 3 1049 2 ×1041 8×1047 200 2000 0.02 −660 ... × × 1×1041 ...× ∼ ∼ ... −710 60 1×1041 6 1047 0.02 −1060 60 2×1041 6×1047 0.02 − × × NOTE. —PropertiesoftheprecursorsandSNe. Ha narrowandHa widearethelinewidthofthenarrowandwidecomponentsoftheHa linesinunitsofthe Gaussians (multiplyby2.35togetFWHM).D tistheaveragetimeoftheprecursor.Allofthevaluesareorder-of-magnitudeestimates.TheHa linewidthsfor SN2011htareadoptedfromRomingetal.(2012).ThevalueslistedforSN2009ipareadoptedfromFoleyetal.(2011),Smithetal.(2010),Drakeetal.(2010), Ofeketal.(2013c),andMarguttietal.(2014).SNelistedabovethehorizontallineareinoursampleandusedtoestimatetheprecursorrates.Eventsbelowthe horizontallinearenotinoursample,butusedinthecorrelationanalysis(Figs.19and20).WenotethatSN2006jc(Pastorelloetal.2007;Foleyetal.2007)was excludedfromthecorrelationanalysissincetheCSMinthisSNishydrogendeficient. 17 s s s s Hg Hb HeI Ha CaII 18 −18 19 −17 −1] ]gamR[ FTP2201 50LEddfor50M⊙star 0.1M⊙Ni56+Co56 −−1165]gam[ R .sbFTP ˚−−12scmA10−16 A g 22 −14 er [ x u 23 −13 Fl 24 −12 −60 −40 −20 0 20 40 60 JD − 2455428.722 [day] FIG.5.—ThelightcurveofSN2010mc(PTF10tel)asobtainedwiththe 4000 5000 6000 7000 8000 9000 Palomar48-inchtelescope. Theredcirclesarebasedonindividualimages, ˚ Observedwavelength[A] thesquaresarebasedoncoaddedimages,whiletheemptytrianglesrepresent the3s upperlimitsderivedfromcoaddedimages. Theerrorbarsrepresent FIG.6.—ThespectrumofPTF10bjbasobtainedbyKeck-I/LRISon2010 the1s errors.TheobjectmagnitudesaregiveninthePTFmagnitudesystem. Mar.7. Thedashedlineshowstheexpectedluminosityfromtheradioactivedecayof anejected massof0.1M ofNi56, assumingthatatlate timestheoptical depthissufficientlylarge⊙toconverttheradioactive energytoopticallumi- nosity,butnotsolargethatitgoesintoPdV work. Thislinerepresentsan upperlimitonthetotalamountofNi56 intheejecta; itwassettocoincide 18.5 s withthelatestobservationoftheSNatJD 2,455,758.Thedottedlinerep- −16.5 resentsabolometricluminosityequalto50≈timestheEddingtonluminosity 19 fora50M star(anorderofmagnitudeestimateofthemassoftheprogenitor −16 assumingi⊙tisamassivestar). Therightedgesofthe“S”symbolsabovethe 19.5 lfirgohmtcOufrevkeeintdailc.a2t0e1t3hbe)e.pochsatwhichweobtainedspectra(Figureadopted B mag] 20 −−1155.5Mag PTF12cxjwasdiscoveredbyPTFon2012Apr. 16.14.The R [APTF202.15 −14.5Abs. SNrosetoanR-bandabsolutemagnitudeof 17.2overtwo − −14 weeks. ThespectrumoftheSN,presentedinFigure12,was 21.5 obtained on 2012 Apr. 18 during the early rise of the SN. −13.5 ThespectrumhasBalmerandHeIlinesinemission. Fitting 22 a two Gaussian modelto the Ha line shows that, the widest −400 −200 0 200 400 600 −13 componenthasavelocitywidth(s )of1200kms 1. MJD−55326.0 [day] − ThisSNhasalargenumberofpre-explosionobservations, FIG.7.— Thelight curve of PTF10bjb exhibits a clear outburst lasting and it shows a single detection 384 days prior to its t . for 4monthspriortotheexplosion(identifiedbythefastriseinthelight ∼ rise curv∼e). Anadditionalpossibledetectionisseen 400dayspriortotheex- Closer inspection reveals that this detection depends on the plosion,butwedonotconsiderittobeagoodca∼ndidatesinceitisbasedon binningscheme,andthatwithshorterbinsoffourdays,mul- asinglebinneddetection(seethediscussionaboutfalse-alarmprobabilityin tiple detections at an absolute magnitude of about 13.5 §4.2).Theupperlimitscorrespondsto5s boundscalculatedin15-daybins. − are seen. Moreover, inspection of the g-band data shows a Supernovaprecursors 9 Hg HeIIHb HeI Ha 2010 Nov 03 −1] ˚A −2m 2010 Oct 10 c10−16 −1 s g r e [ x u Fl FIG.11.— Thecounts-space light curve ofSNPTF10weh. Eachblack circlerepresentsapoint-spreadfunction(PSF)fluxmeasurementonanindi- vidualPTFexposureafterimagesubtraction. Theprecursorislabelled,and 4000 4500 5000 5500 6000 6500 7000 7500 8000 ˚ themeanprecursorfluxisshownasagreenhorizontalline,withthehashed Observedwavelength[A] arearepresentingtheuncertainty.Thebluelineshowstheaverageofrandom measurementsinothergalaxiesintheimage,illustratingthetypicalfluxun- FIG.8.—SpectraofPTF10wehasobtainedusingtheKPNO4mtelescope certaintyandresidualinthefluxmeasurementsafterimagesubtraction. The equippedwiththeRCspectrograph. redcirclesshowtheaverageofrandommeasurementsinthesamehostasthe SN(wherethehostisextendedenoughtoallowthistest),trackingtheran- domsubtractionnoisethatmayaffectthefluxmeasurements.Thehorizontal redlinedenotesthemeanoftheredcircles,andisconsistentwithzeroflux 18 −21 asexpectedafterimagesubtraction. Theabsolutemagnitudescaleisshown 18.5 −20.5 ontheright-handaxisforreference,andtheabscissashowsthetimeindays 19 −20 relativetotheSNtrise. 19.5 −19.5 Hg Hb HeI Ha g] 20 −19 B ma 20.5 −18.5Mag R [APTF212.15 −−1187.5Abs. ˚−1A]1e−16 22 −17 −2 m 22.5 −16.5 c 23 −16 −1s 23−.5500 0 500 1000 erg MJD−55453.0 [day] [ x u FIG.9.—ThePTFR-bandlightcurveofPTF10weh.Filledcirclesshowin- Fl dividualmeasurements,whilefilledboxesmark3-daybinnedmeasurements withatleasttwomeasurementsperbin.Theemptytrianglesdenote5s upper limits. Thebinsizeis3dayspriortotheSNexplosionand15daysafterthe SNexplosion. 1e−17 4000 5000 6000 7000 8000 9000 ˚ Observedwavelength[A] 18 −21 FIG.12.—SpectrumofPTF12cxjasobtainedwiththeGemini-Ntelescope 18.5 −20.5 equippedwiththeGMOSspectrograph. 19 −20 19.5 −19.5 marginaldetectionaround 435dayspriortotrise. ThefullR- − g] 20 −19 bandlightcurveisillustratedinFigure13. Themostconvinc- B ma 20.5 −18.5Mag ingprecursorcandidateeventsaredetectedabouttwoweeks [ATF 21 −18 Abs. atinodns7o0f0tdhaisysSpNrifioerltdo,ttrhisee.liGghivtecnurthveelianrtgheisnpulmotbuertiolifzeodbs2e-rdvaay- RP21.5 −17.5 binsand4s upperlimits. Aclose-upviewofthelightcurve 22 −17 ofthetwoprecursorsisshowninFigure14. TheR-bandpeak 22.5 −16.5 absolutemagnitudesofthesecandidateprecursorsare 14.8 − 23 −16 and 13. However,thecandidateprecursortwoweeksprior − 23.5 to the explosionis veryclose in time to the SN fastrise and −100 −50 0 50 100 150 200 250 300 350 400 MJD−55453.0 [day] thereforemayberegardedaspartoftheSN rise. Asseenin FIG.10.— Close-up view ofthe PTF R-bandlight curve ofPTF10weh Figure14,thetwodetectionsarefollowedbyanondetection aroundtheprecursorcandidatetime(seealsoFig.9).Filledcirclesdenotein- which is about a magnitude deeper than the possible detec- dividualmeasurements,whilefilledboxesmark3-daybinnedmeasurements tionat 8dayspriortot . Furthermore,at2 extrapolation withatleasttwomeasurementsperbin. Theemptytrianglesare5s upper − rise to the SN rise flux suggests that the SN started its rise after limits. Thebinsizeis3dayspriortotheSNexplosionand15daysafterthe SNexplosion. this candidate precursor. We cannot rule out the possibility thatthisprecursoris partof the SN lightcurve(e.g., similar 10 Ofeketal. 18 18.5 −17.5 19 −17 19.5 −16.5 g] 20 −16 B ma 20.5 −15.5Mag [ATF 21 −15 Abs. RP21.5 −14.5 22 −14 22.5 −13.5 23 −13 23.5 −12.5 −1000 −800 −600 −400 −200 0 200 400 MJD−56033.0 [day] FIG.15.—LikeFigure11,butforPTF12cxj. FIG.13.—ThePTFR-bandlightcurveofPTF12cxj.Filledcirclesdenote individualmeasurements,whilefilledboxesmarkbinnedmeasurements.The emptytrianglesshow4s upperlimits. Thebinsizeis2dayspriortotheSN 0.8 explosion.Theminimumnumberofmeasurementsineachbinis4. 0.7 0.6 18 18.5 −17.5 0.5 n 19 −17 o ati 0.4 19.5 −16.5 el mag] 202.05 −−1165.5Mag uto corr 00..23 [TF 21 −15 bs. A 0.1 3s RP A 21.5 −14.5 0 22 −14 −0.1 22.5 −13.5 23 −13 −0.2 0 10 20 30 40 50 60 70 80 23.5 −12.5 Time lag [days] −750 −700 −650 −20 −10 0 10 20 30 40 50 MJD−56033.0 [day] FIG.16.—Thediscreteautocorrelationfunction(Edelson&Krolik1988) FIG.14.— The light curve of PTF12cxj shows two possible precursors ofthefluxresiduals ofPTF12cxj, taken beforetrise. Inordertocalculate (based on multiple consecutive detections, where single detections are ig- theautocorrelation function weusetimebins of3days. Thedotted hori- nored). Theupperlimitscorrespondsto4s boundscalculatedin2daybins. zontallines representthelowerandupper3s bounds, estimated assuming Extrapolationofthelightcurve,basedonat2law(grayline),suggeststhat s 1/√N,whereNisthenumberofmeasurementsineachtimebin. ≈ thedetectionsvisibleataround 10daystookplacebeforetheSNexplosion (seealsoFigure4).Furthermore−,thereisadeepnondetectionbetweenthese signal is caused by the lunar synodic period (i.e., the limit- detectionsandtheassumedtimeoftheSNexplosion.Wenotethatevenifwe removethiseventfromoursample,therateofprecursorsdoesnotchanged ingmagnitudeisbetterduringdarktime). Nevertheless,this bymuch,andthecorrelations(Fig.19)arestilldetected. means that the progenitor of PTF12cxj is likely detected in several binned images, and that its absolute R-band magni- tudeisabout 13,brightereventhanthepossibleprogenitor toSN2011dh;Arcavietal. 2011). − of SN2010jl/PTF10aaxf(Smith et al. 2011). Given the de- As before,Figure15 showsanotherversionofthe precur- tectionofasignalintheautocorrelation,weconsiderthetwo sorlightcurve,withalloftheindividualfluxmeasurements, precursoreventsasreal,andtheirpropertiesarelistedinTa- alongwithfluxmeasurementsinrandompositionsonnearby ble4. galaxiesandthesamehostgalaxy. The second candidate precursor is detected 700days ∼ 6. CONTROLTIME priortotheSNexplosion,anditisalsobasedontwomarginal detectionsseparatedbyabouttwoweeks. Ifreal,thepeakab- InordertocalculatetherateofSNprecursors,weneedto soluteR-bandmagnitudeofthisprecursorisabout 13. estimatethe“controltime” —thatis, forhowlongeachSN − ForalloftheSNeinoursample,wealsocalculatedtheau- locationwasobserved(priortoitsexplosion)toagivenlim- tocorrelationfunctionof the flux residualsprior tot . The itingmagnitude. Table5lists,foreachSN,thetimebinwin- rise only SNe which exhibited significant (at the 3s level) auto- dows(of15days)priortotheSNexplosionandthe5s sensi- correlation at lag one (i.e., corresponding to two successive tivitydepthateachwindowforbinswithmorethanfivemea- measurements) are PTF12cxj (4.7s ) and PTF10tel (3.2s ). surements(secondchannel),orthemedian6s limitingmag- Figure16presentsthediscreteautocorrelationfunction(Edel- nitude at windows with fewer than six measurements (first son & Krolik 1988) of all the flux residuals of PTF12cxj, channel). measured before t . The figure shows significant autocor- Tocalculatethecontroltimeasafunctionofabsolutemag- rise relation on time scales of a few days to ten days. This may nitude,wesumoveralltheSNe thenumberoftimebins(in indicatethatthefluxresidualsarenotpureuncorrelatednoise, Table 5) in which the limiting absolute magnitude is deeper butcontainafractionoftheprogenitorlight. Moreover,itis than the absolute magnitudeof interest, and multiply by the possiblethattheprogenitorisvariableontimescalesofafew binsize(i.e.,15days).Figure17displaysthesamplecumula- days. Analternativeexplanationtothevariabilityisthatthis tivecontroltimeasafunctionofabsoluteR-bandmagnitude.

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.