Space Telescope and Optical Reverberation Mapping Project. V. Optical Spectroscopic Campaign and Emission-line Analysis for NGC 5548

21 

Full text

(1)

©2017.TheAmericanAstronomicalSociety.Allrightsreserved.

Space

Telescope

and

Optical

Reverberation

Mapping

Project.

V.

Optical

Spectroscopic

Campaign

and

Emission-line

Analysis

for

NGC

5548

L.Pei1,2,M.M.Fausnaugh3,A.J. Barth1,B.M.Peterson3,4,5,M.C.Bentz6,G.De Rosa5,K. D.Denney3,4,88,M.R.Goad7, C.S.Kochanek3,4,K. T.Korista8,G.A. Kriss5,9,R.W.Pogge3,4,V. N.Bennert10,M.Brotherton11,K.I.Clubb12, E.DallaBontà13,14,A.V. Filippenko12,J. E.Greene15,C.J. Grier3,16,17,M.Vestergaard18,19,W.Zheng12,ScottM.Adams3,20, ThomasG. Beatty3,16,21,A. Bigley12,JacobE.Brown22,JonathanS.Brown3,G. Canalizo23,J.M.Comerford24,CarlT.Coker3,

E.M.Corsini13,14,S.Croft12,K.V. Croxall3,4,A.J. Deason25,MichaelEracleous16,17,26,27,O.D. Fox12,E.L.Gates28, C.B.Henderson3,29,89,E.Holmbeck30,T.W.-S.Holoien3,4,J. J.Jensen18,C.A.Johnson31,P.L.Kelly32,33,34,S.Kim3,4,

A.King35,M.W.Lau25,MiaoLi36,CassandraLochhaas3,ZhiyuanMa22,E.R.Manne-Nicholas6,J. C.Mauerhan12, M.A.Malkan30,R.McGurk25,37,L.Morelli13,14,AnaMosquera3,38,DaleMudd3,F.MullerSanchez24,M.L.Nguyen11, P.Ochner13,14,B.Ou-Yang6,A.Pancoast39,40,90,MatthewT.Penny3,91,A.Pizzella13,14,RadosławPoleski3,JessieRunnoe16,17,41,

B.Scott23,JadersonS.Schimoia3,42,B.J. Shappee43,92,I.Shivvers12,GregoryV. Simonian3,A. Siviero13,GarrettSomers3,44, Daniel J.Stevens3,M.A.Strauss15,Jamie Tayar3,N.Tejos45,46,T.Treu30,39,93,J.VanSaders43,L.Vican30,S.Villanueva,Jr.3,

H.Yuk12,N. L.Zakamska9,W.Zhu3,M.D. Anderson6,P.Arévalo47,C.Bazhaw6,S.Bisogni3,48,G.A. Borman49, M.C.Bottorff50,W.N.Brandt16,17,51,A.A. Breeveld52,E.M.Cackett53,M.T.Carini54,D.M.Crenshaw6,

A. DeLorenzo-Cáceres55,M.Dietrich56,57,R.Edelson58, N. V. Efimova59,J.Ely5,P.A. Evans7,G. J.Ferland60,K.Flatland61, N. Gehrels62,S.Geier63,64,65,J. M.Gelbord66,67,D.Grupe68,A.Gupta3,P.B.Hall69,S.Hicks54,D.Horenstein6,KeithHorne55,

T.Hutchison50, M. Im70,M.D.Joner71,J.Jones6,J.Kaastra72,73,74,S.Kaspi75,76,B.C.Kelly39,J.A. Kennea16,M.Kim77, S.C.Kim77,S.A.Klimanov60,J.C.Lee77,D.C.Leonard61,P.Lira78,F.MacInnis50,S.Mathur3,4,I.M.McHardy79, C.Montouri80,R.Musso50,S.V. Nazarov49,H. Netzer75,R.P.Norris6,J.A. Nousek16,D. N.Okhmat49,I.Papadakis81,82, J.R.Parks6,J.-U.Pott37,S.E.Rafter76,83,H.-W.Rix37,D.A.Saylor6,K.Schnülle37,S.G.Sergeev49,M.Siegel84,A.Skielboe18,

M.Spencer71,D. Starkey55,H.-I.Sung77,K. G.Teems6,C.S.Turner6,P.Uttley85,C.Villforth86,Y. Weiss76,J.-H.Woo70, H.Yan22,S.Young58,andY. Zu4,87

1

DepartmentofPhysicsandAstronomy,4129FrederickReinesHall,UniversityofCalifornia,Irvine,CA92697,USA 2

DepartmentofAstronomy,UniversityofIllinoisatUrbana-Champaign,Urbana,IL61801,USA 3

DepartmentofAstronomy,TheOhioStateUniversity,140W18thAvenue,Columbus,OH43210,USA 4

CenterforCosmologyandAstroParticlePhysics,TheOhioStateUniversity,191WestWoodruffAvenue,Columbus,OH43210,USA 5

SpaceTelescopeScienceInstitute,3700SanMartinDrive,Baltimore,MD21218,USA 6

DepartmentofPhysicsandAstronomy,GeorgiaStateUniversity,25ParkPlace,Suite605,Atlanta,GA30303,USA 7

DepartmentofPhysicsandAstronomy,UniversityofLeicester,Leicester,LE17RH,UK 8DepartmentofPhysics,WesternMichiganUniversity,1120EverettTower,Kalamazoo,MI49008,USA

9

DepartmentofPhysicsandAstronomy,JohnsHopkinsUniversity,Baltimore,MD21218,USA 10

PhysicsDepartment,CaliforniaPolytechnicStateUniversity,SanLuisObispo,CA93407,USA 11

DepartmentofPhysicsandAstronomy,UniversityofWyoming,1000E.UniversityAvenue,Laramie,WY82071,USA 12

DepartmentofAstronomy,UniversityofCalifornia,Berkeley,CA94720-3411,USA 13

DipartimentodiFisicaeAstronomia“G.Galilei,”UniversitàdiPadova,Vicolodell’Osservatorio3,I-35122Padova,Italy 14

INAF-OsservatorioAstronomicodiPadova,Vicolodell’Osservatorio5I-35122,Padova,Italy 15

DepartmentofAstrophysicalSciences,PrincetonUniversity,Princeton,NJ08544,USA 16

DepartmentofAstronomyandAstrophysics,EberlyCollegeofScience,ThePennsylvaniaStateUniversity,525DaveyLaboratory,UniversityPark, PA16802,USA

17

InstituteforGravitationandtheCosmos,ThePennsylvaniaStateUniversity,UniversityPark,PA16802,USA

18DarkCosmologyCentre,NielsBohrInstitute,UniversityofCopenhagen,JulianeMariesVej30,DK-2100CopenhagenØ,Denmark 19

StewardObservatory,UniversityofArizona,933NorthCherryAvenue,Tucson,AZ85721,USA 20

CahillCenterforAstrophysics,CaliforniaInstituteofTechnology,Pasadena,CA91125,USA 21

CenterforExoplanetsandHabitableWorlds,ThePennsylvaniaStateUniversity,UniversityPark,PA16802,USA 22

DepartmentofPhysicsandAstronomy,UniversityofMissouri,Columbia,MO65211,USA 23

DepartmentofAstronomy,UniversityofCalifornia,Riverside,CA92521,USA 24

DepartmentofAstrophysicalandPlanetarySciences,UniversityofColorado,Boulder,CO80309,USA 25

DepartmentofAstronomyandAstrophysics,UniversityofCaliforniaSantaCruz,1156HighStreet,SantaCruz,CA95064,USA 26

CenterforRelativisticAstrophysics,GeorgiaInstituteofTechnology,Atlanta,GA30332,USA 27DepartmentofAstronomy,UniversityofWashington,Box351580,Seattle,WA98195,USA

28

LickObservatory,P.O.Box85,Mt.Hamilton,CA95140,USA 29

JetPropulsionLaboratory,CaliforniaInstituteofTechnology,4800OakGroveDrive,Pasadena,CA91109,USA 30

DepartmentofPhysicsandAstronomy,UniversityofCalifornia,LosAngeles,CA90095,USA 31

SantaCruzInstituteforParticlePhysicsandDepartmentofPhysics,UniversityofCalifornia,SantaCruz,CA95064,USA 32

DepartmentofPhysics,StanfordUniversity,382ViaPuebloMall,Stanford,CA94305,USA 33

KavliInstituteforParticleAstrophysicsandCosmology,StanfordUniversity,Stanford,CA94305,USA 34

SLACNationalAcceleratorLaboratory,2575SandHillRoad,MenloPark,CA94025,USA 35

SchoolofPhysics,UniversityofMelbourne,Parkville,VIC3010,Australia 36

DepartmentofAstronomy,ColumbiaUniversity,550W.120thStreet,NewYork,NY10027,USA 37

MaxPlanckInstitutfürAstronomie,Königstuhl17,D-69117Heidelberg,Germany 38PhysicsDepartment,UnitedStatesNavalAcademy,Annapolis,MD21403,USA 39

DepartmentofPhysics,UniversityofCalifornia,SantaBarbara,CA93106,USA 40

Harvard-SmithsonianCenterforAstrophysics,60GardenStreet,Cambridge,MA02138,USA 41

(2)

42

InstitutodeFísica,UniversidadeFederaldoRiodoSul,CampusdoVale,PortoAlegre,Brazil 43CarnegieObservatories,813SantaBarbaraStreet,Pasadena,CA91101,USA 44

DepartmentofPhysicsandAstronomy,VanderbiltUniversity,6301StevensonCircle,Nashville,TN37235,USA 45

MillenniumInstituteofAstrophysics,Santiago,Chile 46

InstitutodeAstrofísica,PontificiaUniversidadCatólicadeChile,VicuñaMackenna4860,Santiago,Chile 47

InstitutodeFísicayAstronomía,FacultaddeCiencias,UniversidaddeValparaíso,GranBretanaN1111,PlayaAncha,Valparaíso,Chile 48

OsservatorioAstrofisicodiArcetri,largoE.Fermi5,I-50125,Firenze,Italy 49

CrimeanAstrophysicalObservatory,P/ONauchny,Crimea298409,Russia 50

FountainwoodObservatory,DepartmentofPhysicsFJS149,SouthwesternUniversity,1011E.UniversityAvenue,Georgetown,TX78626,USA 51

DepartmentofPhysics,104DaveyLaboratory,ThePennsylvaniaStateUniversity,UniversityPark,PA16802,USA 52MullardSpaceScienceLaboratory,UniversityCollegeLondon,HolmburySt.Mary,Dorking,SurreyRH56NT,UK

53

DepartmentofPhysicsandAstronomy,WayneStateUniversity,666W.HancockStreet,Detroit,MI48201,USA 54

DepartmentofPhysicsandAstronomy,WesternKentuckyUniversity,1906CollegeHeightsBoulevard#11077,BowlingGreen,KY42101,USA 55

SUPAPhysicsandAstronomy,UniversityofSt.Andrews,Fife,KY169SSScotland,UK 56

DepartmentofPhysicsandAstronomy,OhioUniversity,Athens,OH45701,USA 57

DepartmentofEarth,EnvironmentandPhysics,WorcesterStateUniversity,Worcester,MA01602,USA 58

DepartmentofAstronomy,UniversityofMaryland,CollegePark,MD20742,USA 59

PulkovoObservatory,196140St.Petersburg,Russia 60

DepartmentofPhysicsandAstronomy,TheUniversityofKentucky,Lexington,KY40506,USA 61

DepartmentofAstronomy,SanDiegoStateUniversity,SanDiego,CA92182,USA 62

AstrophysicsScienceDivision,NASAGoddardSpaceFlightCenter,MailCode661,Greenbelt,MD20771,USA 63

InstitutodeAstrofísicadeCanarias,E-38200LaLaguna,Tenerife,Spain 64

DepartamentodeAstrofísica,UniversidaddeLaLaguna,E-38206LaLaguna,Tenerife,Spain 65

GranTelescopioCanarias(GRANTECAN),E-38205SanCristóbaldeLaLaguna,Tenerife,Spain 66

SpectralSciencesInc.,4FourthAve.,Burlington,MA01803,USA 67

EurekaScientificInc.,2452DelmerSt.,Suite100,Oakland,CA94602,USA 68

SpaceScienceCenter,MoreheadStateUniversity,235MartindaleDr.,Morehead,KY40351,USA 69

DepartmentofPhysicsandAstronomy,YorkUniversity,Toronto,ONM3J1P3,Canada 70

AstronomyProgram,DepartmentofPhysics&Astronomy,SeoulNationalUniversity,Seoul,Korea 71

DepartmentofPhysicsandAstronomy,N283ESC,BrighamYoungUniversity,Provo,UT84602,USA 72

SRONNetherlandsInstituteforSpaceResearch,Sorbonnelaan2,3584CAUtrecht,TheNetherlands 73

DepartmentofPhysicsandAstronomy,UniversiteitUtrecht,P.O.Box80000,3508Utrecht,TheNetherlands 74

LeidenObservatory,LeidenUniversity,P.O.Box9513,2300RALeiden,TheNetherlands 75

SchoolofPhysicsandAstronomy,RaymondandBeverlySacklerFacultyofExactSciences,TelAvivUniversity,TelAviv69978,Israel 76

PhysicsDepartment,Technion,Haifa32000,Israel 77

KoreaAstronomyandSpaceScienceInstitute,Korea 78

DepartamentodeAstronomia,UniversidaddeChile,CaminodelObservatorio1515,Santiago,Chile 79

UniversityofSouthampton,Highfield,Southampton,SO171BJ,UK 80

DiSAT,Universitadell’Insubria,viaValleggio11,I-22100,Como,Italy

81DepartmentofPhysicsandInstituteofTheoreticalandComputationalPhysics,UniversityofCrete,GR-71003Heraklion,Greece 82

IESL,FoundationforResearchandTechnology,GR-71110Heraklion,Greece 83

DepartmentofPhysics,FacultyofNaturalSciences,UniversityofHaifa,Haifa31905,Israel 84

LasCumbresObservatoryGlobalTelescopeNetwork,6740CortonaDrive,Suite102,Goleta,CA93117,USA 85

AstronomicalInstitute“AntonPannekoek,”UniversityofAmsterdam,Postbus94249,NL-1090GEAmsterdam,TheNetherlands 86

UniversityofBath,DepartmentofPhysics,ClavertonDown,BA27AY,Bath,UK 87

DepartmentofPhysics,CarnegieMellonUniversity,5000ForbesAvenue,Pittsburgh,PA15213,USA

Received2016October20;revised2017January12;accepted2017February3;published2017March10

Abstract

Wepresent the results of anoptical spectroscopic monitoring program targeting NGC 5548 as partof alarger multiwavelengthreverberationmappingcampaign.Thecampaignspanned6monthsandachievedanalmostdaily cadencewithobservationsfromfiveground-basedtelescopes.TheHβ andHeIIλ4686broademission-line light

+0.36 +0.35

curveslag thatof the5100Åoptical continuumby4.17-0.36days and0.79-0.34days,respectively. TheHβ lag

relativetothe1158ÅultravioletcontinuumlightcurvemeasuredbytheHubbleSpaceTelescopeis∼50%longer than that measured against the optical continuum, and the lag difference is consistent with the observed lag betweentheopticalandultravioletcontinua.Thissuggeststhatthecharacteristicradiusofthebroad-lineregionis

∼50%larger thanthevalueinferred fromopticaldataalone.Wealsomeasuredvelocity-resolved emission-line lagsforHβandfoundacomplexvelocity-lag structurewithshorterlagsinthelinewings,indicativeofa broad-lineregiondominatedbyKeplerianmotion.TheresponsesofboththeHβandHeIIemissionlinestothedriving continuumchangedsignificantlyhalfwaythroughthecampaign,aphenomenonalsoobservedforCIV, Lyα, HeII (+OIII]),andSiIV(+OIV])duringthesamemonitoringperiod.Finally,giventheopticalluminosityofNGC5548 duringourcampaign,themeasuredHβlagisafactoroffiveshorterthantheexpectedvalueimpliedbytheRBLR– LAGN relationbasedonthepastbehaviorofNGC 5548.

88

NSFPostdoctoralResearchFellow. 89

NASAPostdoctoralProgramFellow. 90

EinsteinFellow. 91SaganFellow. 92

Carnegie-PrincetonFellow,HubbleFellow. PackardFellow.

(3)

Keywords:galaxies: active–galaxies: individual(NGC5548)–galaxies: nuclei–galaxies: Seyfert

Supportingmaterial:machine-readabletable

1. Introduction

Broademissionlinesareamongthemoststrikingfeaturesof quasars and active galactic nuclei (AGNs). These Doppler-broadened lines are emitted by gas occupying the broad-line region (BLR), which is located within several light-days to light-months of the central supermassiveblack hole (SMBH; e.g., Antonucci & Cohen 1983; Clavel et al. 1991; Peterson etal. 1998, 2004; Bentzetal. 2009b; Grieretal. 2013).The geometryandkinematicsoftheBLRplayasignificantrolein AGNresearchbecausethesepropertiescanbeusedtoinferthe mass of the central black hole (BH; e.g., Gaskell & Sparke 1986; Clavel et al. 1991; Kaspi et al. 2000; Denney et al. 2006, 2010; Pancoast et al. 2014). Additionally, it is possiblethatinfallingBLRgasmayfuelSMBHaccretion(e.g., Peterson 2006;Gaskell&Goosmann 2016)andoutflowinggas maybepartofdiskwindsthatcarryawayangularmomentum fromthediskand provideenergyand momentumfeedbackto the host galaxy (e.g., Emmering et al. 1992; Murray & Chiang 1997; Kollatschny 2003; Leighly & Moore 2004). Understanding thedynamicalstate and physicalconditions of gas in the BLR is of key importance in completing our understandingof theAGNphenomenon.

Owing to its small angular size, the BLR is currently impossibletoresolvespatiallyevenfortheclosestAGNs.An alternative method tostudy this region isto resolveit in the time domain using reverberation mapping (RM), atechnique that leverages the variable nature of quasars and Seyferts (Blandford & McKee 1982; Peterson 1993, 2014). AGNs exhibitstochasticfluxvariations,possiblybecauseof inhomo-geneousaccretionandthermalfluctuationsintheaccretiondisk (Czerny et al. 1999, 2003Collier & Peterson 2001; Kelly et al. 2009; Kozłowski et al. 2010; MacLeod et al. 2010). Photons from the central engine ionize the BLR gas, which thenechoescontinuum flux variationswithalight-travel time lag,τ.Theemission-linefluxL(vr,t)attimetandline-of-sight

velocityvrisrelatedtotheionizingcontinuumby

¥ L vr t =

ò

0

Y( vr,t) ( -t) dt, ( )1

( , ) C t

where C(t−τ) is the continuum emission at an earlier time t−τ, and Ψ(v, τ) is the transfer function that maps the continuum light curve to the time-variable line profile (Blandford&McKee 1982).

Thetransferfunction—alsoknownasthevelocity-delaymap —encodes important information about the BLR’s geometry and kinematics. There has been tremendous effort by many groups to recover velocity-delay maps (Rosenblatt & Mal-kan 1990; Horne et al. 1991; Krolik et al. 1991; Ulrich & Horne 1996; Bentz et al. 2010a; Pancoastet al. 2011, 2014; Grier et al. 2013; Li et al. 2013; Skielboe et al. 2015) and velocity-resolvedlinelags(e.g.,Kollatschny 2003;Bentzetal.

2009b;Denneyetal. 2010;Barthetal. 2011;Duetal. 2016a). In order to obtain Ψ(v, τ), RM campaigns must have a combinationof highcadence,long duration,highphotometric precision,andhighsignal-to-noiseratios(S/Ns),whichisoften notachievablebyground-basedprograms.Moretypically,RM campaignsareabletoonlymeasurethemeanemission-linelag

τ, which represents the response-weighted mean light-travel timefromtheionizingcontinuum totheBLR.

Assuming that the broad-line width is a result of the virialized motion of gas within the BH’s potential well, the emission-linelagandgasvelocitydispersioninferredfromthe linewidth( DV) canbeusedtoinfertheBHmassusing

ctDV2

=f . ( )

MBH 2

G

Here,cτ=RBLR isthecharacteristicradiusoftheBLR,andf is a dimensionless calibration factor of order unity that accounts for the unknown BLR geometry and kinematics.

Ground-based RM campaigns have produced BH mass

measurementsfor∼60localAGNstodate(seeBentz&Katz

2015, for references and a recent compilation). RM is also startingtobeusedforobjectsatcosmologicaldistances(Kaspi etal. 2007;Kingetal. 2015;Shenetal. 2016),withtheaimsof studyingtheUVcontinuumandemissionlinesandcalibrating BHmassesathighredshifts.

Theionizingcontinuum is emittedatwavelengths <912Å and isgenerally unobservable,due totheLyman limitof the host galaxy. Given this limitation, the far-UV continuum at λ≈1100–1500Åshouldbeusedtoderiveemission-linelags becauseitiscloseinwavelengthtotheionizingcontinuumand should therefore serve as an accurate proxy. However, wavelengths shorter than ∼3200Å are inaccessible fromthe ground,sotherest-frameopticalcontinuumisoftenused asa proxyfortheionizingsourceinlow-redshiftAGNs.Although the far-UV and optical continua have been shown to vary almostsimultaneouslyinsomecases(e.g., Claveletal. 1991; Reichertetal. 1994;Koristaetal. 1995;Wandersetal. 1997), more recent high-cadence studies have foundthat the optical continuum can lag the UV continuum by up to a few days (Collieretal. 1998;Sergeevetal. 2005;McHardyetal. 2014; Shappee et al. 2014; Edelson et al. 2015; Fausnaugh etal. 2016).Thiscansignificantlyaffect themeasured broad-linelagif theBLRhasacharacteristic radiusonthe orderof light-days.Thevariableopticalcontinuumhasalsobeenshown tohavesmootherfeatures andsmaller amplitudesthanitsUV counterpart (e.g., Peterson et al. 1991; Dietrich et al. 1993,

1998; Stirpe et al. 1994; Santos-Lleó et al. 1997; Shappee etal. 2014;Fausnaughetal. 2016).Thesedifferencesbetween theUVandopticalcontinuasuggestthattheopticalcontinuum is not fully interchangeable with the ionizing source for determiningreverberationlags.

(4)

Table1

InstrumentCharacteristicsandDataReductionParametersforAllTelescopes

Telescope Instrument Numberof Median Wavelength Wavelength Pixel Median [OIII]

Epochs Seeing

(arcsec)

Dispersion

(Åpixel−1)

Coverage

(Å) (arcsecScalepixel−1)

S/N Fvar

(%)

MDM Boller&ChivensCCDSpectrograph 143 1.7 1.25 4225−5775 0.75 118 0.62

Lick KastDoubleSpectrograph 35 1.5 1.02 3460−5500 0.43 194 0.32

Asiago Boller&ChivensCCDSpectrograph 21 4.0 1.00 3250−7920 1.00 160 0.27

APO DualImagingSpectrograph 13 1.4 1.00 4180−5400 0.41 160 0.28

WIRO WIROLongSlitSpectrograph 6 2.1 0.74 5599−4399 0.52 217 0.47

Note.ThewavelengthcoverageforLickreferstoonlytheKastblue-sidecamera.TheS/NvaluereferstothemedianS/Nperpixelovertherestwavelengthrange 5070–5130Å.The[OIII]Fvaristheamountofresidualvariationsinthe[OIII]lightcurveafterspectralscalingandgivesanindicationoftheflux-scalingaccuracy.

Since most RM campaigns use only optical data, it is imperative that we understand thesystematic effects of using theopticalratherthantheUVcontinuuminRMstudiesandthe relevant implicationsforBHmassestimates.

To this end, we present the results of a 6-month ground-basedRMprogrammonitoringthegalaxyNGC5548(redshift z=0.0172).Thispaperisthefifthinaseriesdescribingresults from the AGN Space Telescope and Optical Reverberation Mapping (AGN STORM) campaign, the most intensive multiwavelength AGN monitoring program to date. The campaign is centered around 171 epochs of daily cadence observations using the Cosmic Origins Spectrograph on the Hubble Space Telescope (HST). Concurrent with the HST programwere4monthsofSwiftobservationsand6monthsof ground-basedphotometricandspectroscopicobservations.First resultsoftheHST,Swift,andground-basedphotometryprograms werepresentedbyDeRosaetal.(2015),Edelsonetal.(2015), and Fausnaughet al. (2016) (Papers I–III, respectively).Goad etal.(2016) (PaperIV)exploretheanomalousbehaviorofthe UV continuum and broad emission-line light curves observed during a portion of this campaign. This paper focuses on the ground-basedspectroscopicdataandemission-lineanalysis.

NGC5548isoneofthebest-studiedSeyfertgalaxiesandhas beenthesubjectof manypastRM programs.Most notably,it was the target of a13 yr campaign carriedout by the AGN Watch consortium (Peterson et al. 2002, and references therein),whichwasinitiallydesignedtosupport UV monitor-ing ofNGC 5548 carriedoutbythe InternationalUltraviolet Explorer (IUE; Clavel et al. 1991). Individual years of this campaignachievedmediansamplingcadencesof1–3daysfor spectroscopic observations. Subsequently, NGC 5548 was monitored in programs described by Bentz et al. (2007), Denney et al. (2009), Bentz et al. (2009b), and G. De Rosa et al. (2017, in preparation) with campaign durations of 40, 135, 64,and 120days (respectively),andeachwith amedian sampling cadence of ∼1 day. A more recent RM program describedbyLuetal.(2016)monitoredthisAGNfor180days with amedianspectroscopic samplingof ∼3 days.The 2014 AGN STORM campaign’s combination of daily cadence, 6-month duration, and multiwavelength coverage makes it the mostintensiveRM campaignever conducted.

Therearetwoprimarygoalsofthepresentwork.Thefirstis to compare the Hβ emission-line lag measured against simultaneouslyobservedfar-UVand opticalcontinuainorder tounderstandtheeffectsofsubstitutingtheopticalcontinuum fortheionizingcontinuuminreverberationmeasurements.The secondgoalistoexamineindetailtheresponsesoftheoptical

emission lines tocontinuum variations and compare them to those of the UV lines, which will provide a more complete picture of the structure and kinematics of the BLR than previousstudiesthatusedonly opticaldata.

Wedescribe the spectroscopicobservationsandreductionsin Section 2.Section 3detailsourproceduresforfluxandlight-curve measurements.InSection 4,wepresentouranalysisof emission-line lags, line responses, line profiles, and BH mass measure-ments.Wediscusstheimplicationsofourresultsandcompareour measurementswiththosefrompreviouscampaignsinSection 5. Section 6summarizesourfindings.Wequotewavelengthsinthe restframeofNGC5548unlessotherwisestated.

2. Observations and Data Reduction

Spectroscopic datawereobtained from five telescopes:the McGraw–Hill 1.3m telescope atthe MDM Observatory,the Shane 3 m telescope at the Lick Observatory, the 1.22 m GalileotelescopeattheAsiagoAstrophysicalObservatory,the 3.5mtelescopeatApache Point Observatory(APO), andthe 2.3mtelescopeattheWyomingInfraredObservatory(WIRO). ObservationsatMDMwerecarriedoutwithaslitwidthof5″ oriented inthenorth–south direction,and spectra attheother telescopes were taken with a 5″-wide slit oriented at the parallacticangle (Filippenko 1982).Theoptical spectroscopic monitoring began on 2014 January 4 (UT dates are used throughoutthispaper)andcontinuedthrough2014July6with approximatelydailycadence.

Table 1liststhepropertiesofthetelescopesandinstruments usedtoobtainspectroscopicdata,andFigure 1showsthemean spectrumconstructedusing datafromAsiago,whichobtained the only spectrathat cover the full optical wavelength range. MDM contributed the largest number of spectra with 143 epochs.The35epochsofLickspectrawereobtainedbyseveral groups of observers who used slightly different setups and calibrations. The Kastspectrograph (Miller &Stone 1993) at LickObservatoryhasred-sideandblue-sidecameras,butsince thered-sidesetupwasverydifferentforeachgroup,wepresent only the blue-side data here. Asiago, APO, and WIRO contributed21,13,and6 epochsof spectra,respectively.Our analysis focuses primarily on the MDM data set for homogeneity.

Data reduction procedures included bias subtraction, fl

(5)

Figure 1.Mean spectrum of NGC 5548from the Asiago dataset, which includes21epochsofspectrawithspectralresolutionof1.0Åpixel−1andhas a medianS/Nof160.Labeledarethe HeII λ4686,Hβ λ4861,and[OIII]

λλ4959,5007emissionlines.

(Horne 1986)butunweightedextractionsfortheAGNspectra. This is because the optimal extraction method requires the spatial profile of the target to be a smooth function of wavelengthand tendstotruncatethepeaksofstrongemission linessuchas[OIII]thathavedifferentspatialextentsfromthe surroundingcontinuum.

The data were wavelength-calibrated using night-sky lines and flux-calibrated using standard stars. Our most frequently usedfluxstandardstarswereFeige34,BD332642,andHZ44. Fornightswhenmultipleexposuresweretaken,wealignedthe

flux-calibrated one-dimensional spectra by applying small wavelength shifts to each spectrum before combining them. Wedonotexpectsignificantdifferentialatmosphericrefraction (Filippenko 1982)becauseof thelargeslitwidthusedforour observations.

FortheMDMdata,thefirst133epochswereflux-calibrated usingFeige34,whilethelast10epochs,takenfrom2014June 20to2014June30,wereflux-calibratedwithBD332642.This caused spurious changes in the shape of some emission-line features, so we use only the first 133 MDM epochs for our presentanalysis.

2.1. SpectralFluxCalibrations

Toplacetheinstrumentalfluxesonanabsolutefluxscale,we measuredthenarrow[OIII]λ5007linefluxfromspectrataken underphotometricconditionsandscaledallothernightlyspectra tohavethesame[OIII]flux.Therewere21epochsidentifiedas having been observed under photometric conditions by the MDM observers. We determined the flux of the [OIII] line (λobserved=5093Å)byfirstsubtractingalinearfittocontinuum windows oneithersideof thelineand thenintegrating overa

fixed wavelength range. We used the rest-frame wavelength ranges 4976.5–4948.0Å and 5027.7–5031.6Å to fit the continuum and integratedover the range 4980.5–5026.7Åfor the line flux. The 2σ outliers from this set of [OIII] flux measurements were discarded, the meanwas recomputed, and thisprocess wasrepeateduntiltherewerenomore2σoutliers, which resulted in atotal of 16 final photometric spectra.The meanspectrumofthese16epochshasan[OIII]λ5007lineflux

−1 −2

of (5.01±0.11)×10−13 ergs cm , which represents our bestestimate ofthetrue [OIII]fluxforNGC 5548duringthis

campaignandisnotexpectedtovaryovera6-monthperiod.For comparison,Petersonetal.(2013)foundthe[OIII]fluxinNGC

−1 −2

5548 to be (4.77±0.14)×10−13 ergs cm in their 2012 monitoringcampaign, andthedifferenceiswithintherangeof total[OIII]variabilityobservedforNGC5548overthecourseof 21yr(seePetersonetal. 2013).

In addition to the intrinsic variability of the AGN, many other factors contribute to nightly variations in the spectra. Theseincludechangesintransparencydue toclouds, changes in seeing conditions, inconsistent instrument focus, and miscentering of the AGN inthe slit during observations.We used the flux-scalingmethod described by van Groningen & Wanders(1992)toalignthenightlyspectraandplacethemon aconsistent fluxscale. Foreach spectruminthe dataset,the algorithm looks for a combination of wavelength shift, multiplicative scale factor, and Gaussian kernel convolution thatminimizes the residualbetweeneach individualspectrum and areferencespectrumover aregioncontainingthe narrow

[OIII]line.

We constructed a separate reference spectrum for each telescopebyaveragingthehighest-S/Nspectraineachdataset and thenbroadened the referencespectrum so that the [OIII] linewidthmatchesthebroadest[OIII]linewidthinthedataset. Thisextrabroadeningofthereferencespectrumhelpstoreduce the [OIII] residuals from spectral scaling (Fausnaugh 2016). Wethenscaledeachspectrumtohavethesame[OIII]fluxas the photometrically calibrated mean MDM spectrum. This bringsallspectratoacommonfluxscaleafterspectralscaling. Toassesstheaccuracyofspectralscaling,weestimatedthe intrinsicfractionalvariabilityoftheresidual[OIII]λ5007light curveaftercorrectingforrandom measurementerrors,

s2- ád2ñ

Fvar = , ( ) 3

á ñf

where σ2 is the [OIII] flux variance,ád2ñis the mean-square

value of the measurement uncertainties determined from the nightlyerror spectra producedby thedata reductionpipeline, andáfñis the unweighted mean flux.The Fvar for the [OIII] λ5007 light curvegives agood estimateof the residualfl ux-scaling errors (Barth & Bentz 2016), and the value for each telescopeislistedinthelastcolumnofTable 1.WefoundFvar to be between 0.27% and 0.62% for all telescopes, which meansthatthereisanadditionalscatteroflessthan1%inthe

[OIII] light curve above the measurement errors. These Fvar values are consistent with or better than the best values typically obtained in ground-based campaigns. For example, Barth et al. (2015) found Fvar values ranging from 0.5% to 3.3%forindividualAGNsinthe2011LickAGNMonitoring Project.

(6)

Figure2.Meanandexcessrms (Equation (4)) spectrafromtheMDMdataset areshowninblack,andthermsspectrumwiththeAGNandstellarcontinuum removedisshowninred(seeSection3.1).

maximum-likelihood method to obtain the rms. We adopt a simpler approach that uses the excess variance as a way to exclude variations that are not intrinsic to the AGN. This “excessrms”valueateachwavelengthisdefined as

N

2 2

rms = 1

å

[( F - á ñ) -dl ,i] , 4

e- l

N-1i=1 l ,i Fl ( )

whereNisthetotalnumberofspectrainthedataset,áFlñisthe

mean flux at each wavelength, and Fλ,i and λ,i are the

wavelength-specificfluxesandassociatedmeasurement uncer-tainties from individual epochs, respectively. This method estimatesthedegreeofvariabilityabovewhatisexpectedgiven themeasurementuncertainties andpixel-to-pixelnoise.

3. Spectroscopic Flux Measurements

The 5100Å continuum flux density was determined by averaging the flux over the rest-frame wavelength range 5070–5130Å. The Hβ line fluxes were measured from the scaled spectra using the same method as for [OIII] λ5007, where wesubtractedalinear fittothesurroundingcontinuum (wavelengthwindows4483.0–4542.0Åand5033.5–5092.5Å) and integratedacross the line profile(4748.4–4945.1Å). The uncertainty ineach measurementisacombination of Poisson noise and residuals from spectral scaling. We computed the spectralscalinguncertaintybymultiplyingeachflux measure-mentbythe[OIII]Fvar valueforthatdatasetandthenadding thisvalueinquadraturetothePoissonnoisetoobtainthefinal

fluxuncertainty foreach measurement. Thereisan additional sourceofspectralscalinguncertaintyfromslightdifferencesin the overall spectral shape from night to night. This effect is likelysmallforHβbecauseitisveryclosetothe[OIII]λ5007 line thatanchorsthespectralscaling.

−2) Figure3.Continuum (10−15ergs−1cm−2Å−1) andHβ (10−15ergs−1cm lightcurves(THJD=HJD−2,450,000).TheLick,APO,Asiago,andWIRO lightcurveswerescaledandshiftedtomatchtheMDMlightcurve,whichhas the longest temporal coverage and highest sampling cadence. The plotted uncertaintiesincludePoissonnoiseandthenormalizedexcessvarianceofthe [OIII]lightcurve(Section2.1).

Spectrophotometric calibrationsof the referencespectra, as described in the previous section, converted all instrumental

fluxestoabsolutefluxes,whichmeansthatmeasurementsfrom alltelescopesshouldnowbeonthesamefluxscale.However, lightcurvesfromdifferentobservingsites maybeoffset from each other owing to aperture effects (Peterson et al. 1995,

1999).While our observationswere standardizedto havethe same 5″×15″ aperture size,significant differences inimage qualitybetweenobserving sitescouldstillcausefluxoffsets.

To intercalibrate the Hβ light curves, we used data points from each non-MDM telescope (FHβ,t) that are nearly contemporaneous with MDM observations (FHβ,MDM) and performedaleast-squaresfittotheequation

FH ,MDMb =fFHb ,t ( ) 5

tofindthescalefactorfthatputseachlinelightcurveonthe same flux scale as the MDM data. For the continuum intercalibration,wealsoincludeanadditiveshiftGtoaccount forthedifferencesinthehost-galaxyfluxadmittedbydifferent apertures:

F5100,MDM=fF5100,t+G. ( ) 6

Thescale factorsfor theLick, Asiago, APO,and WIRO light curves are f=[0.961,0.963,1.037,0.918], and the shift constants are G=[−0.155, −0.640, −0.041, 0.024] in units of10−15ergs−1cm−2Å−1.ThecombinedcontinuumandHβ lightcurvesareshowninFigure 3(THJD=HJD−2,450,000), andthe5100ÅcontinuumandHβfluxesarelistedinTable 2. We attempted to measure the HeIIλ4686 flux from the nightly spectra. However, this line is very weak and also heavilyblendedwiththebroadHβ,asshowninFigure 2.Thus, wewereunabletoobtain anHeIIlightcurveusing thelinear interpolationmethod toremove thecontinuum.

3.1. SpectralDecomposition

(7)

Table2

FluxMeasurementsforContinuumandEmissionLines

HJD−2,450,000 Telescope F5100 FHβ FHβ,SD FHeII,SD

6663.00 MDM 10.766±0.075 726.012±4.985 710.187±4.897 21.720±2.606

6663.65 Asiago 10.921±0.040 741.771±3.586 K K

6664.03 MDM 11.154±0.075 732.511±5.156 715.057±5.061 28.154±3.222

6665.02 MDM 10.788±0.075 724.537±4.946 709.473±4.860 25.135±2.451

6667.02 MDM 10.872±0.076 735.001±5.393 711.347±5.288 37.008±3.964

6668.00 MDM 10.966±0.075 727.261±4.946 708.252±4.859 35.227±2.450

6669.01 MDM 10.956±0.075 733.312±4.941 714.465±4.855 40.589±2.430

6669.65 Asiago 11.147±0.035 724.604±2.449 K K

6670.02 MDM 11.008±0.076 734.810±5.179 712.584±5.083 42.710±3.308

6670.65 Asiago 10.806±0.035 728.863±2.626 K K

Note.The5100Åcontinuumfluxdensity(10−15ergs−1cm−2Å−1)includescontributionsfromboththeAGNandthehostgalaxy.TheHβandHβSDfluxeswere obtainedusingalinearcontinuummodelandthespectraldecompositionmethod,respectively.TheHeII fluxisbasedonthespectraldecompositionmodel.All emission-linefluxesareinunitsof10−15ergs−1cm−2andincludecontributionsfrombothbroad- andnarrow-linecomponents.

(Thistableisavailableinitsentiretyinmachine-readableform.)

and narrowHeII, FeIIemissionblends,thestellarcontinuum, and the AGN continuum. The host-galaxy starlight was modeled with an 11Gyr, solar-metallicity, single-burst spec-trum from Bruzual & Charlot (2003). For the FeII model component,wetestedthreedifferenttemplatesfromBoroson& Green (1992),Véron-Cettyetal. (2004),and Kovačevićetal. (2010). The FeII templates were broadened by convolution with aGaussian kernelinvelocity.Thefreefitparametersfor FeII include the velocity shift relative to broad Hβ, the broadening kernel width, and the flux normalization of the broadened template spectrum. The Boroson & Green (1992)

and Véron-Cetty et al. (2004) templates are monolithic and require only one flux normalization parameter, whereas the Kovačevićetal.(2010)templatehasfivecomponentsthatcan vary independently in flux. The Kovačević et al. (2010)

templateachievesthebestfittothenightlyspectra,presumably aresultof thelarger number offree fitparametersdue tothe multicomponentKovačevićetal. (2010)template.

We made several modifications to the spectral fitting procedures used by Barth etal. (2015). First, because of the complex line profiles, we used sixth-order Gauss–Hermite functions (van der Marel &Franx 1993) tofit the broadand narrow Hβ and narrow [OIII] lines instead of fourth-order functions.Second, thereissignificantdegeneracy betweenthe weak FeII blend and the continuum flux in the nightly fits. Since theFeIIfitis poorlyconstrainedand sometimes varied drastically fromnight tonight,thecontinuum modelfluxalso variedsignificantlyas aresult, whichinturnintroduced noise to the broad Hβ fit component. To address this issue, we constrained theFeII fluxtoliewithin10% ofthe valuefrom thefittothemeanspectrum(Barthetal. 2013).Wealsofixed theFeIIredshifttothatofthemeanspectrumandconstrained the FeIIbroadeningkerneltobe within5%of itsvaluefrom themeanspectrumfit. The HeIλ4922andλ5016linesarevery weak and are heavily blended with broad Hβ, making it impossibletoconstraintheirfitparameters.Wethereforedonot

fitforthesecomponentsinourmodel.

The broad HeIIλ4686 component hasvery low amplitude comparedtotheotherfitcomponents,anditisblendedwiththe blue wing of broad Hβ. It is also highly variable, as demonstrated by the broad bump in the rms spectrum. This made it difficult to fit the HeII broad-line profile accurately, and the width varied significantly from night to night when

fitted as a free parameter. Since the HeII λλ1640 and 4868

linesareexpectedtoformunderthesamephysical conditions andshouldthushavesimilarwidths,weusedfitstotheλ1640 line in concurrent HST spectra to constrain the λ4686 line width.

The HeII λ1640 line was modeled with five Gaussian components(G. DeRosaetal. 2017,inpreparation),and we tookthethreebroadestcomponentstorepresentthebroadHeII λ1640line profile.ForeachMDMspectrum,the HeIIλ4686 broad-lineFWHMwasallowedtovarywithin3ÅoftheHeII λ1640FWHMmeasuredfromtheclosestHSTepoch.Thefirst 23epochsfromtheMDMcampaigndonothavecorresponding HSTspectra,soforeachofthese“pre-HST”epochs,wefound the three epochs from later in the campaign with the closest matching 5100Å continuum flux density. We then used the weighted mean of the broad HeII λ1640 widths from these three nights as the width constraint for the pre-HST epoch, wheretheweightsweredeterminedbyhowcloselythe5100Å

fluxes ofthelaterepochs matchedthatof thepre-HSTepoch. TheHeIIλ1640linewidthwashighlyvariableduringtheHST campaign,andthemodelFWHMwidths usedtoconstrainthe spectraldecompositionhaveameanof48Å,withaminimum of28Åandmaximumof59Å.

We applied spectral decomposition to the data from all telescopes, butsincethe MDMdataset isthelargest and has thehighestdataqualityandconsistency,weusethisdatasetfor allsubsequentanalysis.Figure 4showsthefitcomponentsfor themeanMDMspectrum,wheretheblackspectrumisthedata andtheredspectrumisthesumofallthemodelcomponents. Themodeldoes notfitthedetailedstructure of thebroadHβ line well, especially in the line core. To prevent this from impactingourmeasuredHβfluxes,wesubtractedalltheother well-modeledfitcomponentsexceptthebroadand narrowHβ componentsfromthefull spectrum and thenobtained theHβ line flux by integrating over the same wavelength range used to measure the flux without spectral decomposition. The HeIIλ4686 flux was taken to be the total flux in the broad- andnarrow-linemodelsforeachnight.ThenarrowHβ andHeIIλ4686linefluxesfromfitstothemeanspectrumare

−1 −2 −1 −2

(8)

Figure 4. Top:spectral decomposition components of the mean MDM spectrum. Theredspectrumisthesumofallmodelcomponents andtraces the data (black spectrum) well over most of the spectral range. Bottom:residualsfromthefullmodelfit.

The red spectrum inFigure2 shows the rms of the MDM spectra after subtracting the AGN continuum and stellar continuum models from individual spectra so that only the emission-line components remain. This rms spectrum is expected to be a more accurate representation of the emission-line variabilitythanthermsofthefullspectra(Barth et al. 2015). We show the rms here and not the excess rms defined by Equation (4) because, for parts of the spectra dominated by continuum emission, the continuum-subtracted

flux could be lower than the total flux uncertainties and the e-rms would beundefined. Thus, we usetheexcessrms only for thefullspectrumand notforindividualfitcomponents.

Panels (a)and (c)of Figure5 show the Hβ mean and rms

fluxesas afunction of line-of-sightvelocity(vr)afterspectral

decomposition,andpanel(b)showsthedifferencebetweenthe T1 and T2 mean fluxes. The rms flux has a statistical uncertainty of ∼12% andthe [OIII]residuals aremuchlower comparedtothecasewithnospectraldecomposition(Figure 2). The rms profilestill has jagged features, which likelyreflect realvariability acrossthebroademissionline.

Figure 6 showsthe1158ÅUVcontinuum lightcurve from Paper I,theMDMoptical5100Åcontinuumlightcurve,the V-bandphotometriclightcurvefromPaper III,andtheMDMHβ andHeIIλ4686emission-linelightcurves.TheHeIIlightcurve reaches aflat-bottomed minimum near THJD=6720. This is becausetheHeIIfluxincludescontributionsfromthebroad- and narrow-line components, so when the broad-line flux is near zero,thetotalHeIIlinefluxstaysataminimumvalueequalto the narrow-line flux. Light-curve statistics that quantify the variabilityofNGC5548duringthemonitoringperiodaregiven inTable 3. Fvar isas defined inEquation (3), and Rmax isthe ratio betweenthemaximumandminimumfluxes.

3.2. Host-galaxyFluxRemoval

Wemeasuredthehost-galaxycontribution tothe continuum usingan“AGN-free”imageofNGC5548generatedbyBentz etal.(2013)afterperformingtwo-dimensionalsurfacebrightness

Figure 5. (a) MDM mean spectrum, (b) difference between the T1

(THJD<6747)andT2(THJD>6747)meanspectra,and(c–e)rmsspectra; forHβaftersubtractingallotherfitcomponentsfromspectraldecomposition. Thecolorsareforthefullcampaign(black), T1(gray),andT2(orange;see Section3.3).ZerovelocityisdeterminedbythepeakofthenarrowHβlinein themeanspectrum,andthegraybandsindicateregionscontaminatedby[OIII] residuals.Thermsspectrahavestatisticaluncertaintiesof∼12%.

decompositiononHSTimagesofthegalaxy.Wefoundthatthe amountofstarlightexpectedthrougha5″×15″aperturewitha slitpositionangleof0°isF5100,gal =(4.52±0.45)×10−15 erg s−1 cm−2 Å−1.Subtractingthisfromthemeancontinuum flux

−2 Å−1 density of F5100 = (11.96±0.07)×10−15 erg s−1 cm givesameanAGNfluxof F5100,AGN =(7.44±0.50)×10−15 ergs−1 cm−2 Å−1,whichisconsistentwiththevalue ofF5100, AGN =(7.82±0.02)×10−15 ergs−1 cm−2 Å−1 measuredfrom thepower-lawcomponentofthespectraldecompositionforthe meanMDMspectrum.

3.3. AnomalousEmission-lineLight-curveBehavior

(9)

Figure6.Left:lightcurvesfortheUV1158Åcontinuum,optical5100Åcontinuum,V-bandcontinuum,Hβ,andHeII λ4686.Thecontinuumlightcurvesarein unitsof10−15ergs−1cm−2Å−1,andthelinelightcurvesareinunitsof10−15ergs−1cm−2.TheHβandHe

II fluxesincludecontributionsfrombothbroad- and narrow-linecomponents.Thesolidgreenverticallineindicateswheretheemission-linelightcurvesweretruncatedforthelaganalysis (seetext),andthedashedblack verticallineshowsthedivisionbetweentheT1andT2periods.Right:cross-correlationfunctionsforeachlightcurvemeasuredagainstthe1158Åcontinuum.The toprightpanelshowstheautocorrelationofthe1158Ålightcurve.Theblack,gray,andorangesolidlinesrepresenttheCCFsforthefullcampaign,T1,andT2, respectively,andthedottedverticallinesdenoteτcenforthefullcampaign.

scaled, and smoothed version of the continuum light curve. However, thisdoes notappear tobethe casefor aportionof our campaign. As described in Papers I and IV, there are significant differences between the UV continuum and emission-linelightcurvesafterTHJD=6780.Thecontinuum

fluxincreased whilethe emission-line fluxes either decreased or remained roughly constant in a suppressed state for the remainderofthecampaign.

TheHβemissionlineshowssimilarbehaviortothatofLyα and CIV, in that there is a marked difference in the line responsebetweenthefirstandsecondhalvesofthecampaign. This “decorrelation”phenomenonisillustratedinFigure 7(a), wheretheHβandUVcontinuumlightcurvestraceeachother well inthe firsthalf ofthe campaign, butthe continuum flux continuestotrendupwardbeyondTHJD=6740,whiletheHβ

fluxbeginstofall.Fortheremainderofthecampaign,theHβ

fluxremainsinasuppressedstateandthelightcurvedoesnot

follow the continuum light curve well in that prominent features in the continuum light curves (e.g., THJD≈6770, 6785) are not present in the emission-line light curve. The HeIIλ4686lightcurvebehavessimilarlyanddecorrelatesfrom the continuum at around THJD=6760 (Figure 7(b)). This phenomenon is also apparent when comparing the Hβ light curve to the 5100Å continuum, as shown in Figure 7(c). Thoughtherearesmalldifferencesbetweenthelightcurvesin T1(suchasaroundTHJD6685),theoverallcorrelationinT1is significantlybetter thaninT2.

(10)

Table3

StatisticsforHSTandMDMLightCurves

EmissionComponent Epochs MeanFlux rmsFlux Fvar Rmax

Fλ(1158Å) 171 43.48±0.86 11.14±1.21 0.255 4.07±0.18

Fλ(5100Å) 133 11.96±0.07 0.80±0.09 0.066 1.33±0.01

Hβ 133 738.49±2.40 28.29±3.38 0.038 1.22±0.01

HeII λ4686 133 78.71±2.95 35.14±4.17 0.444 7.48±0.91

Fλ(1158Å, T1) 51 35.85±1.79 12.61±2.54 0.351 3.31±0.15

Fλ(1158Å, T2) 120 46.72±0.80 8.66±1.13 0.184 2.36±0.08

Fλ(5100Å, T1) 67 11.31±0.08 0.67±0.12 0.059 1.27±0.01

Fλ(5100Å, T2) 67 12.51±0.04 0.37±0.06 0.029 1.13±0.01

Hβ(T1) 67 725.10±3.80 30.47±5.36 0.041 1.20±0.01

Hβ(T2) 67 750.14±2.38 20.11±3.33 0.026 1.13±0.01

HeII λ4686(T1) 67 65.84±4.16 33.61±5.89 0.507 5.95±0.73

HeII λ4686(T2) 67 89.90±3.75 32.71±5.34 0.362 4.07±0.37

Note.Continuumfluxdensitiesareinunitsof10−15ergs−1cm−2Å−1,andemission-linefluxesareinunitsof10−15ergs−1cm−2.T1andT2denotethefirstand secondhalvesofthecampaign(respectively)dividedatTHJD=6747.

Figure7.1158Åand5100Åcontinuumlightcurves(black)comparedwith scaled and shifted emission-line light curves (blue). The vertical line at THJD=6747indicatestheepochseparatingtheT1andT2segments.Ineach ofthepanels,theemission-linelightcurvecloselytracksthecontinuumlight curveinT1,butappearstocorrelatelesscloselywiththecontinuumvariations inT2.

separatedintotwosegmentsatTHJD=6747.Notethatthisis a differentdividing epoch fromthe one used inPaper I.The characteristics of the half-campaign light curvesare given in thebottomportionofTable 3.Figure 5showstheMDMmean andrmsspectraforT1andT2ingrayandorange,respectively. Whilethethreemeanspectralookalmostidentical,theT1and T2 rms spectra are significantly different, which indicates changesintheamountofvariabilitybetweenthetwocampaign halves.

4. Data Analysis

Inthefollowingsections,weexaminepropertiesoftheBLR by measuring the emission-line responses to continuum

variations. We also discuss the anomalous behavior of the emission-line lightcurves observedduringthis campaignand BHmassmeasurementsusing thisdataset.

4.1.Emission-lineLags

Wemeasuredthe Hβ and HeIIλ4686 lagsrelative toboth the 5100Å continuum and the 1158Å continuum. All light curvesweredetrendedbysubtractingalinear least-squaresfit to the data to remove long-term trends that may bias lag calculations (Welsh 1999). In this case, we found very weak trendsforallthelightcurves,and detrendinghasavery small effect (∼0.01 days) on the measured lags.We computed the cross-correlation coefficient r for lags between −20 and 40 days inincrementsof 0.25 days using the interpolated cross-correlationfunction(ICCF;White&Peterson 1994).Twolag estimates were made for each light-curve pair—the value correspondingtormax (τpeak)andthecentroidofallvalueswith r>0.8rmax (τcen).Estimatesforthefinalτpeak andτcen values and their uncertainties were obtained using Monte Carlo bootstrapping analysis (Peterson et al. 2004), where many realizations of the continuum and emission-line light curves were created by randomly choosing n data points with replacement from the observed light curves, where n is the totalnumberofpointsinthedataset.Ifadatapointispickedm times,thentheuncertaintyonthatpointisdecreasedbyafactor

1/2

of m . Each value is then varied by a random Gaussian deviate scaled by the measured flux uncertainty. We con-structed103 realizationsofeachlightcurveandcomputedthe cross-correlation function (CCF) for each pair of line and continuum light-curve realizations to create a distribution of τpeak andτcen values.Themedianvaluefromeachdistribution andthe central68%intervalarethentakentobethefinallag anditsuncertainty.

(11)

Table4

Rest-frameEmission-lineLags

LightCurves τpeak τcen τcen,T1 τcen,T2 τJAVELIN τJAVELIN,T1 τJAVELIN,T2

0.74 0.39 + 0.49 + 0.71 0.48 0.64 + 0.97

HβversusFλ(1158Å) 6.14-+ 0.98 6.23-+ 0.44 7.62-0.49 5.99-0.75 6.56-+ 0.49 6.91-+ 0.63 7.42-1.07

+ 0.25 + 0.37 + 0.49 + 0.76 0.46 0.60 + 1.03

HβversusFλ(1367Å) 5.90-0.74 5.89-0.37 7.24-0.48 5.99-0.82 6.12-+ 0.47 6.52-+ 0.57 7.11-1.06

+ 0.98 + 0.36 + 0.40 + 0.77 + 0.57 + 0.68 + 1.13

HβversusFλ(5100Å) 4.42-0.25 4.17-0.36 4.99-0.47 3.10-0.80 3.84-0.59 5.15-0.69 4.78-1.17

+ 0.98 + 0.37 + 0.57 + 0.55 + 0.45 + 0.66 + 0.93

HβversusVband 3.93-0.98 3.79-0.34 3.82-0.47 4.13-0.58 3.54-0.46 4.89-0.71 4.05-0.78

+ 0.49 + 0.24 + 0.39 + 0.36 + 0.27 + 0.35 + 0.25

HeII versusFλ(1158Å) 2.46-0.25 2.69-0.25 3.71-0.38 3.19-0.35 2.65-0.27 3.27-0.35 2.99-0.26

+ 0.25 + 0.25 + 0.36 + 0.29 + 0.25 + 0.35 + 0.25

HeII versusFλ(1367Å) 2.21-0.25 2.45-0.24 3.43-0.43 3.16-0.33 2.41-0.26 3.04-0.36 2.79-0.25

0.25 0.35 + 0.28 0.36 0.37 + 0.51 0.38

HeII versusFλ(5100Å) 0.49-+ 0.25 0.79-+ 0.34 1.21-0.36 0.85-+ 0.36 0.16+ -0.37 1.13-0.48 0.85-+ 0.35

0.25 0.34 0.43 + 0.34 0.23 0.37 0.23

HeII versusVband 0.49-+ 0.49 0.50-+ 0.26 0.40-+ 0.39 1.46-0.27 0.44-+ 0.23 0.63-+ 0.36 0.91-+ 0.21

+ 0.25 + 0.31 + 0.28 + 0.41

Fλ(5100Å)versusFλ(1158Å) 1.97-0.49 2.23-0.26 2.55-0.33 2.77-0.45 L L L

+ 0.74 0.38

Hβfull,MDMversusFλ(1158Å) 5.90-0.49 6.26-+ 0.37 L L L L L

+ 0.74 0.37

Hβfull,allsitesversusFλ(1158Å) 5.65-0.25 6.49-+ 0.37 L L L L L

Note.Rest-frameHβandHeII λ4686lags(days)forthefullcampaignandfortheT1(THJD<6747)andT2(THJD>6747)subsets,measuredusingbothICCF andJAVELIN.ThelasttwolinesshowtheHβlagsmeasuredusingthelightcurvederivedwithoutspectraldecompositionuptoTHJD=6828.75.

Figure8.τcen(top)andJAVELIN(bottom)lagprobabilitydistributionsfor Hβ(left)andHeII (right)measuredagainstthe1158Åcontinuum.Blacksolid linesareforthefullcampaign,graydashedlinesareforT1,andorange dot-dashedlinesareforT2.

WealsocomputedHβandHeIIλ4686lagsfortimeperiods T1 andT2 andfoundtheT1 lagstobeconsistentlylongerby about2σ.ForHβ,theT1andT2lagsbracketthefull-campaign lag,whileforHeIIλ4686,thefull-campaignlagisshorterthan both T1and T2lags.Forcomparison,Paper IfoundtheLyα, SiIV, CIV,and HeIIλ1640 lagstobe longerforT2 thanfor T1, whichistheoppositeofwhatwefindfor Hβ.

To illustrate the effects of spectraldecomposition, we also includeinTable 4theICCFlagsfortheHβlightcurvewhere the fluxes were measured using the straight-line continuum-subtraction method and without spectral decomposition. We

Figure 9. JAVELIN light curves from simultaneously modelingtheHβand calculated lags for both the MDM-only and themultisite Hβ

HeII λ4686emissionlineswiththeUV1158Åcontinuum.Thedatapointsare light curves, which were truncated at THJD=6828.75 to measuredfromobservations,theblacksolidlinesaretheweightedmeansofthe excludethelast 10epochsofMDMdata(seeSection 2).The modellightcurvesconsistentwiththedata,andthethicknessoftheshadedregions indicatesthe1σspreadofthoselightcurves.Thetopthreepanelsshowtheobserved lags measuredwithand without usingspectral decomposition

dataandJAVELINmodellightcurveswithoutanyerrorscaling,andthebottom areconsistenttowithin1σ. threepanelsshowthedataandlight-curvemodelswithscaleduncertainties.

InadditiontotheICCFmethod,wecomputedthe

emission-line lags using the JAVELIN suite of Python codes (Zu randomwalkprocess(DRW;Kellyetal. 2009;Zuetal. 2013).

et al. 2011). Weused JAVELIN tolinearly detrend the light JAVELIN explicitlymodels the emission-line light curvesas

(12)

Figure 10.Velocity-resolvedHβlightcurvesfor1500kms−1bins(black) withthecentralvelocityforeachbinshownatthetopleftofeachpanel.HST

1158Ålightcurvesthathavebeenscaledandshiftedtomatchthefirsthalfof theHβlightcurvesareshowninred,andthedashedlineindicatestheepoch thatseparatesT1andT2forthisanalysis.TheHβresponsetothecontinuumis distinctlyvelocitydependentduringthesecondhalfofthecampaign.

curve.Sincethedecorrelation oftheline andcontinuum light curves during the latter half of the campaign clearly violates theseassumptions,itisofinteresttoexaminetheconsequences

fortheJAVELINmodels.Forthesemodels,wesimultaneously

fit the Hβ and HeII light curves using either the 1158Å, 5100Å, orV-bandcontinuum lightcurve.TheAGNSTORM light curves are too short to accurately determine the DRW damping timescale, so this value was fixed to τDRW≈164 daysasderived byZuetal.(2011)fromfitstothe13yrlight curveofNGC5548(Petersonetal. 2002).Theprecisevalueof τDRW isnotcriticalforthealgorithmtoworkprovidedthatitis approximately correct.

Thetop three panelsof Figure 9 show theresults of using

JAVELIN directly on the observed data. Despite the long

DRW timescale, the light-curve models show rapid fl uctua-tions.Thealgorithm tries tomatch thesuppressed fluxinthe linelightcurvetothecontinuumlightcurve,andbecauseofthe smalluncertainties,it stronglypreferslag timesforwhichthe continuum and line light curves have minimal temporal overlap. This caused the posterior lag distribution to have multiple narrow peaks corresponding to lags that best desynchronizethelightcurves.

Wecanattempttocompensateforthisproblembyincreasing theuncertaintiestoencompasstheamplitudeofthe decorrela-tion. This requires scaling up the full-campaign light-curve uncertaintiesbyfactorsof 5and 3forthecontinuum andline lightcurves, respectively. ThelowerpanelsinFigure 9 show theJAVELINresultsfromfittingtothelightcurveswithscaled errors,where the broaderuncertainties allowthe algorithm to construct smooth light-curve models. Since there is more statistical weight from fitting both lines simultaneously, the models track the line light curves best and show a smooth systematic offset for the continuum where the line and continuum lightcurvesaredecorrelated. When computingthe half-campaignlags,JAVELINfavorsasmaller linefluxscale factorfor T2 toaccountforthe suppressedline fluxes,which beginsneartheepochseparatingT1andT2.Wethereforedid not need to scale the flux errors by as much as for the full campaigntoaccountforthedecorrelation,andweusedanerror scaling factor of 3 for both continuum and line light curves. TheresultingJAVELINlags,showninTable 4,areconsistent with those measured using the ICCF method. Similar to the ICCF lags, the JAVELIN lags for the full-campaign light curves are also shorter than those for T1 and T2. However, since the JAVELIN assumptions of the relationship between the line and continuum light curves are not valid for this campaignand thefluxerrors werescaledforthe solepurpose of producing convergent solutions, we do not use the

JAVELINlagsforsubsequentanalysis.

We examinedthe velocity-resolved emission-line response bydividingtheHβlineprofileintobinswithvelocitywidthof 500kms−1 andsetzerovelocityusingthepeakofthenarrow Hβ component in the mean spectrum. We constructed light curvesfor each velocitybin separately, and Figure 10shows the light curvesfor 1500kms−1 velocitybins acrossthe Hβ lineprofile(black),withthe velocityatthecenterofeachbin showninthetop leftofeach panel.There weresix epochsof spectra94 that produced outlying Hβ fluxes and significantly higher than average flux uncertainties for individual velocity bins, so we removed them from the velocity-resolved light curvesinordertoimprovethelagmeasurements.

Wedetermined theICCFlagfor eachofthese binnedlight curveswith respecttoboth theUV and optical continua,and weshowtheselagsasafunctionofline-of-sightvelocityinthe top left panel of Figure11. The middle left panel shows the velocity-resolvedHβ–UVlagforT1andT2,andthemaximum cross-correlation coefficients (rmax) are shown in the right panels.Thebottom panelof Figure 11shows the MDM full-campaignmean spectrum for reference, and Table 5 liststhe velocity-resolved lags. Emission-line variations have lower amplitudesduringT2 comparedwith T1,which ledtopoorly constrained ICCF lags with large uncertainties for velocity

94

(13)

Figure 11.Topleft:ICCFHβlags(τcen)for500kms− 1

binsmeasuredagainstthe1158Åand5100Åcontinua.Middleleft:lagsforT1andT2measured against the 1158Åcontinuum. Bottomleft:MDMmeanspectrumfor thefull campaign. Right:maximum cross-correlationcoefficients (rmax)forindividual velocitybins.

bins withlow lineflux.Wethereforereduced theupperlimit of the CCF lag range from 40 to 20 days when computing theT2 lags.

The velocity-resolved Hβ–UV lag for the full campaign is shortest (τcen≈2days) in the line wings where vr≈±7000kms−1. The lag increases as vr approaches zero

from both sides of the lag profile, reaches local maxima of τcen≈10daysataboutvr≈±3000kms−1,andthensteadily

decreases until it reaches a local minimumof τcen≈4 days nearthelineprofilecenter.Thelagprofilemeasuredagainstthe optical continuumhasasimilarshape, butwith alllags∼2–3 daysshorter,aswewouldexpectfromthe∼2-daylagbetween thetwocontinua(Table 4).Asimilardouble-peakedlagprofile is also observed for Lyα (see Paper I). The T1 lag profile closely resembles that of the full campaign, but the T2 lag profileshowsaslightlydifferentstructure.Thebinswherethe T1andT2lagsaremostdiscrepantarealsowheretheT2light curves are least correlated with the continuum (rmax<0.4). However, even excluding these outliers, there are still discernible differences between the T1 and T2 lag profiles.

Additionally,theT2rmax valuesarelowerthanthoseofT1in everyvelocitybin,whichclearlydemonstratesthatthelineand continuumlightcurvesarelesscorrelatedinT2 thaninT1.

The shape of the velocity-resolved lag profilecan provide qualitative information about the kinematics of the line-emitting gas (e.g., Kollatschny 2003; Bentz et al. 2009b; Denney et al. 2010; Barth et al. 2011; Du et al. 2016a). In simple modelsof theBLR (Ulrichetal. 1984; Gaskell 1988; Welsh & Horne 1991; Horne et al. 2004; Goad et al. 2012; Gaskell & Goosmann 2013; Grier et al. 2013), pure infall motionwould leadtolongerlagson thebluesideof theline profile,andforoutflow,themostredshiftedgaswouldhavethe longestlag.ForgasinKeplerianorbits,theshortestlagswould beinthe line wings,since gas with highervr is closertothe

centralBH. Gaswithvery lowvrcouldhave awiderangeof

(14)

Table5

Rest-frameHβVelocity-resolvedLags

Wavelength

(Å) (kmvrs−1)

τFull

(days)

rmax,Full τT1

(days)

rmax,T1 τT2

(days)

rmax,T2

4743.82−4751.92 7000 1.74-+ 0.600.59 0.61 2.05+ -0.941.04 0.79 2.13-+ 0.660.59 0.42

4751.92−4760.03 6500 2.34-+ 0.490.49 0.72 2.57-+ 0.710.60 0.88 3.08-+ 0.770.98 0.52

4760.03−4768.13 6000 2.94-+ 0.480.45 0.73 3.41+ -0.700.63 0.88 3.48-+ 0.660.71 0.62

4768.13−4776.23 5500 3.18-+ 0.460.38 0.75 3.25+ -0.540.57 0.83 3.46-+ 0.620.72 0.62

4776.23−4784.33 5000 3.92-+ 0.470.50 0.73 3.12+ -0.530.57 0.85 4.95-+ 0.941.04 0.56

4784.33−4792.43 4500 4.69-+ 0.490.56 0.73 3.94+ -0.490.63 0.85 5.88-+ 0.700.63 0.65

4792.43−4800.53 4000 6.19-+ 0.680.63 0.63 7.00+ -0.850.74 0.83 6.26-+ 1.291.34 0.53

4800.53−4808.63 3500 7.26-+ 0.610.62 0.68 8.45-+ 0.870.94 0.79 6.93-+ 1.572.34 0.51

4808.63−4816.73 3000 8.69-+ 0.670.64 0.63 10.90+ -0.890.89 0.81 6.40-+ 1.482.71 0.42

4816.73−4824.84 2500 8.70-+ 1.131.29 0.53 10.54+ -0.961.55 0.77 4.92-+ 0.861.65 0.42

4824.84−4832.94 2000 7.63-+ 1.010.87 0.58 10.46+ -1.071.24 0.77 4.31-+ 0.500.62 0.66

4832.94−4841.05 1500 5.91-+ 0.530.58 0.68 8.71-+ 0.870.78 0.80 5.36-+ 0.740.62 0.70

4841.05−4849.15 1000 4.65-+ 0.450.48 0.71 5.79+ -0.530.63 0.89 5.36-+ 0.600.55 0.74

4849.15−4857.25 500 5.27-+ 0.490.38 0.74 5.80+ -0.650.61 0.90 6.11-+ 0.670.51 0.76

4857.25−4865.35 0 6.01-+ 0.490.37 0.79 6.50-+ 0.600.52 0.90 6.59-+ 0.760.57 0.71

4865.35−4873.45 500 6.40-+ 0.390.47 0.75 8.06-+ 0.700.67 0.84 6.43-+ 0.710.64 0.67

4873.45−4881.56 1000 6.99-+ 0.460.38 0.77 8.86-+ 0.390.55 0.90 6.87-+ 0.590.53 0.74

4881.56−4889.66 1500 7.77-+ 0.370.44 0.80 9.80+ -0.460.41 0.92 7.05-+ 0.520.57 0.73

4889.66−4897.76 2000 8.59-+ 0.390.40 0.81 10.65+ -0.550.53 0.90 6.38-+ 0.680.75 0.65

4897.76−4905.86 2500 10.18-+ 0.670.66 0.74 11.13+ -0.680.67 0.87 6.47-+ 2.382.01 0.27

4905.86−4913.96 3000 10.19-+ 0.920.98 0.67 11.18+ -0.610.71 0.86 2.58-+ 0.981.21 0.21

4913.96−4922.06 3500 8.23-+ 0.961.03 0.66 9.82+ -0.960.86 0.81 3.78-+ 1.462.74 0.27

4922.06−4930.16 4000 7.25-+ 0.910.87 0.65 8.80-+ 0.700.71 0.87 4.06-+ 0.971.57 0.50

4930.16−4938.27 4500 6.01-+ 0.590.55 0.62 7.01+ -0.720.81 0.88 5.52-+ 0.700.64 0.62

4938.27−4946.38 5000 5.04-+ 0.530.59 0.47 6.63-+ 0.560.61 0.88 4.85-+ 0.680.69 0.58

4946.38−4954.48 5500 4.00-+ 0.570.56 0.40 5.22+ -0.830.84 0.80 4.55-+ 0.640.72 0.59

4954.48−4962.58 6000 3.41-+ 0.600.63 0.43 3.16+ -1.301.05 0.83 4.89-+ 0.700.65 0.62

4962.58−4970.68 6500 3.20-+ 0.530.61 0.53 3.46+ -0.720.74 0.85 4.30-+ 0.960.87 0.53

Note.ICCFHβlags (τcen) forthefullcampaignandfortheT1 (THJD<6747) andT2 (THJD>6747) segments.Thefirstcolumngivestherest-framewavelength rangeforeachbin,andthesecondcolumngivestheline-of-sightvelocityatthecenterofeachbin.Thermaxvaluesgivethemaximaofthecross-correlationfunctions betweenthelightcurveforeachvelocitybinandtheUVcontinuum.

Previous studiesof theUV and optical linesinNGC 5548 have inferred either Keplerian orbits (Horne et al. 1991; Wandersetal. 1995;Denneyetal. 2009;Bentzetal. 2010b)or infalling motion (Crenshaw & Blackwell 1990; Done & Krolik 1996;Welshetal. 2007;Pancoastetal. 2014;Gaskell &Goosmann 2016)fortheBLRgas.Fromourdata,theshape of the Hβ velocity-resolved lag profile suggests a BLR dominated by Keplerian motion. The discrepancy between the T1 and T2 lag profiles may suggest a change in the distributionordynamicsoftheBLRgas,thoughsuchchanges typicallyoccurontimescalesmuchlongerthanourcampaign. More detailedinterpretation requires comparisonwithtransfer functionsgeneratedforvariousdynamicalmodelsoftheBLR, which will be the subject of future work in this series (A. Pancoastetal.2017,inpreparation).

Figure 10 also shows modified versions of the 1158Å continuumlightcurveinred.Theselightcurveswereshiftedin timebytheaverageT1lagofthebinsincorporatedineachHβ lightcurve,whichareshowninthebottomrightofeachpanel. The fluxes wereroughly scaled and shifted tomatchthe first halfoftheHβlightcurves.Comparingthelineandcontinuum lightcurves,itisevidentthattheHβresponseinT2isheavily dependent ontheline-of-sightvelocity.

4.2.AnomalousEmission-lineResponsetoContinuum

Our data show that previous assumptions about the relationship between the continuum and emission-line light curves—namely, that the emission-line light curves are smoothed, scaled,and time-shifted versions ofthe continuum lightcurve—arenot alwaysvalid, aswe observedalltheUV and optical emission-line light curves decorrelating from the UV continuum about halfwaythrough the monitoringperiod. Paper IV examined this effect for the UV emission lines by measuringchangesintheemission-lineequivalentwidth(EW),

line

= F , ( )

EW 7

Fcont

and theresponsivity (ηeff), whichis thepower-law indexthat relatesthedrivingcontinuumfluxtotheresponding emission-linefluxes,

logFline =A+heff[ logFcont] . ( )8

(15)

(Baldwin 1977),whichisdescribedby

line=B+b[ logFcont] , ( )

log EW 9

wherethechoiceforFcont isassumedtobeareasonableproxy fortheionizingcontinuum.Thus,βisalsoknownastheslope of theBaldwinrelation.

FollowingthesameproceduresasinPaper IV,wecompute the responsivity ηeff and EW for the portion of the Hβ light curvethatcorrelateswiththeUVcontinuumandthenexamine how these values change in different segments of the light curves. The valuesFcont and Fline refer tothecontinuum and emission-line fluxes after removing nonvariable components such as host-galaxy and narrow-line flux contributions, and aftercorrectingforthemeantimedelaybetweenthecontinuum and line light curves (Pogge & Peterson 1992; Gilbert & Peterson 2003; Goad et al. 2004). There is very little host-galaxyfluxinthe1158Åcontinuum,which isdominatedby the variable AGN. For the line fluxes, we took Hβ fluxes measured after linear continuum removal (without spectral decomposition) and subtracted a constant narrow Hβ flux measured from the MDM mean spectrum fit to remove the nonvariable line component.To correct for the emission-line time delay, we shifted the Hβ light curve by 8 days, which corresponds to the lag for the portion of the line light curve closelycorrelatedwiththeUVcontinuum.Figure 12(a)shows the1158Åcontinuumlightcurve(black)withthetime-shifted andflux-scaledHβlightcurve,whichhasbeentruncatedatthe beginning tomatchthefirstepochofcontinuumobservations. ToshowtheHβlightcurve’sgeneralbehaviortowardtheend of the campaign, wehave shownhere the full 143 epochs of Hβfluxmeasurementsinsteadofthe133-epochlightcurvewe usedintheHβlaganalysis.However,sincethelast10epochs ofspectrasufferfrominconsistentspectralfluxcalibration(see Section 2),wedonotusethesepointsincalculatingηeff orβ. WedividedtheHβlightcurveintofivesegments.Thefirst segment corresponds tothe period when the line light curve closely follows the continuum light curve (blue points); the secondandthirdsegmentscorrespondtoperiodswhentheline light curve decouples from the continuum (cyan points) and remains inastateof depressed flux(redpoints); the last two segmentscorrespondtothelinelightcurverecoveringfromthe depressed state (magenta points) and correlating once again withthecontinuumlightcurve(greenpoints).Theepochsthat dividethese segmentsareTHJD=[6743,6772,6812,6827]. Figures 12(d)and(e)showtheHβbroad-linefluxand EW asafunctionofthe1158Åcontinuumfluxdensitydetermined fromtheHSTepochclosesttoeachMDMepoch.Theredlines representlinearleast-squaresfitstoEquations(8)and(9)using onlythebluepoints.Wefoundthatthecyan,red,andmagenta points—corresponding to when the light curves are not well correlated—liewellbelowthebestfitsofEquations(8)and(9)

tothebluepoints.Furthermore,theepochsduringtheanomaly (redpoints)are characterizedby anηeff value consistent with zero (ηeff=0.02±0.03; black dotted line in Figure 12(d)), whichshowsthattheemission-linestrengthremainedconstant independentofcontinuumstrength.Similarresultswerefound for the CIV, Lyα, HeII(+OIII]), and SiIV(+OIV]) emission linesinPaper IV.Table 6summarizestheηeff andβvaluesfor allbroademissionlinesmeasuredduringthis campaign.Note that the values from Paper IV were computed using epochs both before and after the anomaly (blue and green points),

Figure12.(a)1158Åcontinuumlightcurvewiththetime-shiftedHβlight curve(seetextforcolorscheme).(b,c)ReconstructedHβlightcurve(gray dots)andthepercentoffluxlostduringtheanomaly,respectively.(d,e)Hβ broad-lineflux(10−15ergs−1cm−2)andEW(Å)asafunctionofthe1158Å continuum flux density (10−15ergs−1cm−2Å−1). The solid red lines are linearleast-squaresfitstothebluepoints,whicharefromtheperiodwhenthe lineandcontinuumlightcurvesarewellcorrelated.Theslopesofthesefitsare shownineachpanel.Thedottedblacklineinpanel (d) istheleast-squaresfit to theredpoints(anomaly)andhasaslopeofηeff=0.02±0.03.

Table6

Responsivity,SlopeoftheBaldwinRelation,andFluxDeficitduringthe AnomalyforAllEmissionLines

LineID ηeff β Flost

Lyα SiIV+OIV] CIV

0.30±0.01 0.45±0.01 0.25±0.01

−0.73±0.02

−0.58±0.03

−0.75±0.01

9% 23% 18% HeII+OIII]

0.58±0.04 0.15±0.01

−0.48±0.04

−0.85±0.02

21% 6%

Note.Allvaluesweremeasuredusingthe1158Åcontinuumfluxesfromthis campaign.ThefirstfourrowsshowvaluesfromPaperIV.

whereas ourvalues werecalculatedusing only epochs before theanomaly(bluepoints).

(16)

Table7

HβResponsivityandAGNOpticalContinuumFluxDensityforNGC5548 over25yr

Year ηeff MeanFcont rmsFcont

1989 0.56±0.04 6.54 1.27

1990 0.84±0.03 3.79 0.91

1991 0.95±0.09 6.06 0.92

1992 0.94±0.05 3.34 1.17

1993 0.43±0.04 5.69 0.87

1994 0.74±0.04 6.40 1.11

1995 0.68±0.04 8.71 1.01

1996 0.54±0.03 7.07 1.52

1997 0.80±0.07 4.73 0.89

1998 0.51±0.02 10.05 1.44

1999 0.41±0.04 8.48 1.82

2000 0.65±0.06 3.59 1.20

2001 1.00±0.12 3.65 0.86

2014 0.59±0.01 7.44 0.50

Note.Fluxdensitiesareinunitsof10−15ergs−1cm−2Å−1.Therst13rows showvaluesfromGoadetal.(2004),andthelastrowshowsvaluesfromthis campaign. Allηeffvalues listedwere measuredwith respectto the5100Å continuum.

when the lightcurves are wellcorrelated (blue points). This simulatedlightcurveisshowninFigure 12(b)withgraydots, and comparing it with the observed data (colored points) shows that the Hβ line had lower-than-expected variability amplitudeandmeanfluxduringT2.Ifwedefinethefractional

flux loss as Flost=(Fsim−Fobs)/Fsim (Figure 12(c)), then this implies an Hβ flux deficit of ∼6% duringthe anomaly (red points),which is close tothe deficit for Lyα but much smallerthanthatoftheotherUVemissionlines,asshownin Table 6.

Table 7summarizestheHβresponsivitiesandAGNoptical continuum flux densitiesmeasured by Goad etal. (2004) for everyyearofthe13yrmonitoringcampaigncarriedoutbythe AGNWatchconsortium(Petersonetal. 2002),95alongwiththe valuesofηeff=0.59±0.01andáF5100,AGNñ =( 7.440.50) ´

-15 -1 -2 -1

10 erg s cm Å calculatedfromthisdataset.Boththe responsivity and optical continuum flux density for this campaign are close to those measured from the 1998 data. Using the mean Hβ flux without narrow-linecontributions of

-15 -1 -2

áFHb ñ = ( 6902.40) ´10 erg s cm , wefindamean

EWforthiscampaignof92.75±2.45Å,whichislowerthan values measured by Goad & Korista (2014) for previous campaigns.

The Hβ light curve appears to decorrelate from the UV

continuum light curve at a somewhat earlier time

(THJD≈6742) than CIV (THJD≈6765 as found in Paper IV). Ifwe assume thatthe Hβ light curve decorrelates and recorrelates with the continuum light curve at the same timesasCIVandusethesamedividingepochsasthoseusedin Paper IV (THJD=[6766, 6777, 6814, 6830]), then ηeff=0.13±0.01andβ=−0.88±0.01forthebluepoints. While HeIIλ4686 also shows anomalous behavior during theT2period(Figure 7),itsbroad-linecomponentisveryweak and the fitted profile is poorly constrained in the spectral decomposition process. The HeII light curve is thus noisier than that of Hβ, and the Fline and EW values are poorly

95

ThecontinuumfluxdensitiesfromPetersonetal.(2002)havebeenupdated byBentzetal.(2013)andKilerciEseretal.(2015).

correlatedwith Fcont.Wetherefore do notperform adetailed analysisof theresponsivityforHeII.

4.3.Line WidthandMBHEstimate

TheBH massinNGC 5548 has beenestimated for several previousRMcampaigns,includingtheAGNWatchconsortium (Petersonetal. 2002),Bentz etal.(2007, 2009b),andDenney etal.(2010).TheAGNWatchgroupdeterminedtheBHmass using data from each of the program’s monitoring years, and subsequentcampaignseachproducedanindependentBHmass value. Here we compute the BH mass using data from this campaignandcompareitwithpreviousresults.

We measured the line widths from the Hβ mean and rms spectraas showninFigure5.The narrowHβ lineessentially disappearsinthermsspectrumbutisstillpresentinthemean spectrum.Weremovedthisemissioncomponentbysubtracting a Gaussian fit to the Hβ narrow line in the mean MDM spectrum, and then linearly interpolated over the narrow Hβ residualsand the[OIII]λλ4959and 5007residuals.

Two emission-line width values are typically measured in RM:theFWHMand thelinedispersion,definedby

2 2

c ⎞ ⎛ål Si 2⎞

2 i

sline= ⎝ ⎜

⎠ ⎝⎜ -l0

⎠, (10)

l0 åSi

whereSiisthefluxdensityatwavelengthbinλiand λ0 isthe

flux-weighted centroid wavelength of the line profile. We measuredσline andFWHMforthemeanandrmsspectrausing the line profile within the rest-frame wavelength range 4669.8–5063.0Å. We treated the mean profile as double peaked and followed the procedures described by Peterson etal.(2004)tomeasureitsFWHM.Fromeachofthetwopeaks at4827.1and4886.1Å,wetracedthelineprofileoutwarduntil thefluxreached0.5Fmax,andthentracedthelineprofileinward fromthecontinuum untilthefluxagain reached0.5Fmax.The twowavelengthsat0.5Fmax —whichgenerallyagreewellfora smooth line profile—are then averaged to obtain the wave-length athalf-maximum oneach side of theprofile. Therms profileismorecomplicatedbecauseithasmorethantwopeaks andthetroughsbetweenthemcanreachwellbelowhalfofthe peak fluxes. We therefore identified a single maximum flux Fmax and traced the profile from the continuum toward the center on both sides until the flux reached 0.5Fmax. The separationbetweenthetwowavelengthsat0.5Fmax istakento betheFWHMoftherms spectrum.

TheHβ linewidthsandtheiruncertaintiesweredetermined usingMonteCarlobootstrapanalysis.Withntotalspectra,we randomlyselectednspectrafromthedatasetwithreplacement, constructedmeanandrmslineprofiles,and measuredtheline dispersionandFWHM.Themedianandstandarddeviationof 104 bootstraprealizationsareusedfortheσline andFWHMand theirestimateduncertainties.

Figure

Updating...

Related subjects :