Astronomy&Astrophysicsmanuscriptno.NGC5548-PaperI (cid:13)c ESO2016 August9,2016 Anatomy of the AGN in NGC 5548 I. A global model for the broadband spectral energy distribution M.Mehdipour1,2,J.S.Kaastra1,3,4,G.A.Kriss5,6,M.Cappi7,P.-O.Petrucci8,9,K.C.Steenbrugge10,11,N.Arav12,E. Behar13,S.Bianchi14,R.Boissay15,G.Branduardi-Raymont2,E.Costantini1,J.Ebrero16,1,L.DiGesu1,F.A. Harrison17,S.Kaspi13,B.DeMarco18,G.Matt14,S.Paltani15,B.M.Peterson19,20,G.Ponti18,F.PozoNun˜ez21,A.De Rosa22,F.Ursini8,9,C.P.deVries1,D.J.Walton23,17,andM.Whewell2 5 1 SRONNetherlandsInstituteforSpaceResearch,Sorbonnelaan2,3584CAUtrecht,theNetherlands 1 e-mail:[email protected] 0 2 MullardSpaceScienceLaboratory,UniversityCollegeLondon,HolmburySt.Mary,Dorking,Surrey,RH56NT,UK 2 3 DepartmentofPhysicsandAstronomy,UniversiteitUtrecht,P.O.Box80000,3508TAUtrecht,theNetherlands 4 LeidenObservatory,LeidenUniversity,POBox9513,2300RALeiden,theNetherlands n 5 SpaceTelescopeScienceInstitute,3700SanMartinDrive,Baltimore,MD21218,USA a 6 DepartmentofPhysicsandAstronomy,TheJohnsHopkinsUniversity,Baltimore,MD21218,USA J 7 INAF-IASFBologna,ViaGobetti101,I-40129Bologna,Italy 6 8 Univ.GrenobleAlpes,IPAG,F-38000Grenoble,France 9 CNRS,IPAG,F-38000Grenoble,France ] 10 InstitutodeAstronom´ıa,UniversidadCato´licadelNorte,AvenidaAngamos0610,Casilla1280,Antofagasta,Chile E 11 DepartmentofPhysics,UniversityofOxford,KebleRoad,Oxford,OX13RH,UK H 12 DepartmentofPhysics,VirginiaTech,Blacksburg,VA24061,USA . 13 DepartmentofPhysics,Technion-IsraelInstituteofTechnology,32000Haifa,Israel h 14 DipartimentodiMatematicaeFisica,Universita`degliStudiRomaTre,viadellaVascaNavale84,00146Roma,Italy p 15 DepartmentofAstronomy,UniversityofGeneva,16Ch.d’Ecogia,1290Versoix,Switzerland - o 16 EuropeanSpaceAstronomyCentre,P.O.Box78,E-28691VillanuevadelaCan˜ada,Madrid,Spain r 17 CahillCenterforAstronomyandAstrophysics,CaliforniaInstituteofTechnology,Pasadena,CA91125,USA t 18 Max-Planck-Institutfu¨rextraterrestrischePhysik,Giessenbachstrasse,D-85748Garching,Germany s a 19 DepartmentofAstronomy,TheOhioStateUniversity,140W18thAvenue,Columbus,OH43210,USA [ 20 CenterforCosmology&AstroParticlePhysics,TheOhioStateUniversity,191WestWoodruffAve.,Columbus,OH43210,USA 21 AstronomischesInstitut,Ruhr-Universita¨tBochum,Universita¨tsstraße150,44801,Bochum,Germany 1 22 INAF/IAPS-ViaFossodelCavaliere100,I-00133Roma,Italy v 23 JetPropulsionLaboratory,CaliforniaInstituteofTechnology,4800OakGroveDrive,Pasadena,CA91109,USA 8 8 Received20November2014/Accepted18December2014 1 1 ABSTRACT 0 . Anextensivemulti-satellitecampaign on NGC5548 hasrevealed thisarchetypal Seyfert-1galaxytobeinanexceptional stateof 1 persistentheavyabsorption.Ourobservationstakenin2013–2014withXMM-Newton,Swift,NuSTAR,INTEGRAL,Chandra,HST 0 andtwoground-basedobservatorieshavetogetherenabledustoestablishthatthisunexpectedphenomenoniscausedbyanoutflow- 5 ingstreamofweaklyionisedgas(calledtheobscurer),extendingfromthevicinityoftheaccretiondisktothebroad-lineregion.In 1 thisworkwepresent thedetailsofour campaign andthedataobtainedbyalltheobservatories. Wedeterminethespectralenergy : v distributionofNGC5548fromnear-infraredtohardX-raysbyestablishingthecontributionofvariousemissionandabsorptionpro- i cessestakingplacealongourlineofsighttowardsthecentralengine.Wethusuncovertheintrinsicemissionandproduceabroadband X continuummodelforbothobscured(averagesummer2013data)andunobscured(<2011)epochsofNGC5548.Ourresultssuggest r thattheintrinsicNIR/optical/UVcontinuumisasingleComptonisedcomponentwithitshigherenergytailcreatingthe‘softX-ray a excess’. Thiscomponent iscompatible with emissionfrom awarm, optically-thick corona as part of theinner accretion disk. We theninvestigatetheeffectsofthecontinuumontheionisationbalanceandthermalstabilityofphotoionisedgasforunobscuredand obscuredepochs. Keywords.X-rays:galaxies–galaxies:active–galaxies:Seyfert–galaxies:individual:NGC5548–techniques:spectroscopic 1. Introduction the feedback mechanisms between SMBHs and their environ- ments is important in AGN science, as well as in cosmology. The supermassive black holes (SMBHs) at the heart of active They can have significant impacts and implications for the galactic nuclei (AGN) grow through accretion of matter from evolution of SMBHs and star formation in their host galaxies their host galaxies. This accretion is accompanied by outflows (e.g. Silk & Rees 1998; King 2010), chemical enrichment of (powerful relativistic jets of plasma and/or winds of ionised their surrounding intergalactic medium (e.g. Oppenheimer & gas),whichtransportmatterandenergyawayfromthenucleus, Dave´ 2006),andthecoolingflowsatthecoreofgalaxyclusters thus linking the SMBHs to their host galaxies. Understanding 1 M.Mehdipouretal.:AnatomyoftheAGNinNGC5548.I. (e.g.Ciotti&Ostriker2001). optical/UV reverberation mapping studies of NGC 5548 (e.g. Peterson et al. 2002; Pancoast et al. 2014) have provided one The ionised outflows are an essential component of the of the most detailed pictures available of the broad-line region energy balance of AGN and are best studied through high- (BLR) size and structure in this AGN. Prior to our recent resolution X-ray and UV spectroscopy of their absorption line campaignonNGC5548,theX-raypropertiesandvariabilityof spectra. The Reflection Grating Spectrometer (RGS) of XMM- NGC5548wereknowntobetypicalofstandardSeyfert-1AGN Newton and the Low-Energy & High-Energy Transmission sharingcommoncharacteristics(e.g.Bianchiet al. 2009;Ponti Grating Spectrometers (LETGS and HETGS) of Chandra, etal.2012). together with the Cosmic Origins Spectrograph (COS) of the Hubble Space Telescope (HST) have vastly advanced our In2013–2014wecarriedoutanambitiousmulti-wavelength knowledgeoftheseoutflowsinrecentyears(seee.g.thereview campaignonNGC5548,whichwassimilarbutmoreextensive by Costantini 2010). With current instrumentation, the nearby than our campaign on Mrk 509. This remarkable campaign and bright Seyfert type AGN are the best laboratories for has utilised eight observatories to take simultaneous and studying these outflows. The majority of the absorbing gas is frequent observations of the AGN. It incorporates instruments detectable in the X-ray band. The outflows detected in the soft onboard five X-ray observatories: XMM-Newton (Jansen et al. X-rays (referred to as ‘warm absorbers’) are found to have 2001), Swift (Gehrels et al. 2004), NuSTAR (Harrison et al. column densities of 1020−1024 cm−2 and consist of multiple 2013), INTEGRAL (Winkler et al. 2003), Chandra’s LETGS phases of photoionised gas with temperatures of 104−106 K, (Brinkman et al. 2000), as well as the HST COS (Green et al. travelling at velocities of up to a few 103 km s−1 (see e.g. 2012), and two ground-based optical observatories: the Wise Blustin et al. 2005). Understanding the origin and launching Observatory (WO) and the Observatorio Cerro Armazones mechanismoftheionisedoutflowsinAGN isanactiveareaof (OCA). These observatories have collected over 2.4 Ms of research.Ithasbeensuggestedthattheseoutflowsareproduced X-rayand800ksofoptical/UVobservationtime.Aspreviously byirradiationofthe dustygastorusstructure,whichsurrounds reported by Kaastra et al. (2014), NGC 5548 was discovered theSMBHandaccretiondisk(e.g.Krolik&Kriss2001),orthat tobeobscuredinX-rayswithmainlynarrowemissionfeatures they originate as radiatively driven winds from the accretion imprinted on a heavily absorbed continuum. This obscuration disk (e.g. Begelman et al. 1983; Proga et al. 2000), or are in is thought to be caused by a stream of clumpy weakly-ionised the form of magnetohydrodynamic (MHD) winds from the gaslocated atdistances ofabout2–7lightdaysfromthe black accretiondisk(e.g.Ko¨nigl&Kartje1994;Bottorffetal.2000). holeandpartiallycoveringtheX-raysourceandtheBLR.From its associatedbroadUV absorptionlinesdetectedin HST COS Many aspects and physical properties of these outflows spectra,theobscurerisfoundtobeoutflowingwithvelocitiesof are still poorly understood. For instance, the location of the upto5000kms−1.TheintenseSwiftmonitoringonNGC5548 ionisedabsorbersneedstobeestablishedtodistinguishbetween shows the obscuration has been continuously present for a the different outflow mechanisms and to determine their mass few years (at least since Feb 2012). As the ionising UV/X-ray outflow rates and kinetic luminosities, which are essential radiationisbeingshieldedbytheobscurer,newweakly-ionised parameters in assessing their impact on their surroundingsand featuresofUVandX-rayabsorberoutflowshavebeendetected. their contribution to AGN feedback (e.g. Hopkins & Elvis Compared to normal warm absorber outflows commonly seen 2010).ThedistanceofanX-rayorUVabsorbingoutflowtothe in Seyfert-1s at pc scale distances, the remarkable obscurer in ionisingsourcecanbedeterminedfromestimatesofthedensity NGC 5548 is a new breed of weakly-ionised, higher-velocity of the absorbinggas. The density can be measured fromeither outflowing gas, which is much closer to the black hole and density-sensitive UV lines (e.g. Gabel et al. 2005) or from the extends to the BLR. As reported in Kaastra et al. (2014) the recombinationtimescale of the ionised absorber(Kaastra et al. outflowing obscurer is likely to originate from the accretion 2012).IntheX-rayband,thelatterdeliversmorerobustdensity disk. Based on the high outflow velocity of the obscurer, its measurements than the use of density-sensitive X-ray lines. In short-timescaleabsorptionvariability,anditscoveringfractions therecombinationtimescalemethod,astheintrinsicluminosity ofthecontinuumandtheBLR,theobscurerisincloseproximity oftheAGNvariesovertime,theionisationstateoftheabsorber to the central source, and its geometry extends from near the changes with a time delay; by measuring this lag, the electron disktooutsidetheBLR. density and hence the distance of the absorber to the ionising source can be obtained. Such a study was first done as part of our 2009 multi-wavelength campaign on the Seyfert-1/QSO Inthisworkwepresentabroadbandspectralanalysisofthe Mrk 509 (Kaastra et al. 2012). From the response of three NGC5548data.Thestructureofthepaperisasfollows.Section absorber components with the highest ionisation in the soft 2 gives an overview of our multi-satellite campaign. In Sect. X-rays,individualdistancesof5–100pcwerederived,pointing 3 we present lightcurves of NGC 5548 constructed at various to an originin the narrow-lineregion(NLR) or the AGN torus energiesfromnear-infrared(NIR)tohardX-rays.InSect.4we region. explain the required steps in determining the spectral energy distribution(SED)ofNGC5548.InSect.5weexaminethesoft The archetypal Seyfert-1 galaxy NGC 5548 is one of the X-ray excess in NGC 5548 and present an appropriate model most widely studied nearby active galaxies. It was one of the for it. In Sec. 6 we describe the modelling of the broadband 12classifiedobjectsintheseminalworkofSeyfert(1943),and continuumin unobscuredandobscuredepochsandpresentour since the sixties it has been the target of variousAGN studies. results. The thermal stability curves, corresponding to various In more recent times it was the first object in which narrow ionising SEDs, are presented in Sect. 7. We discuss all our X-rayabsorptionlinesfromwarmabsorberswerediscoveredby findingsinSect. 8andgiveconcludingremarksinSect.9.The Kaastra et al. (2000) using a high-resolution Chandra LETGS processing of the data from all the instruments is described in spectrum. Later on, a detailed study of its warm absorber was AppendixA. carriedoutbySteenbruggeetal.(2005).Furthermore,extensive 2 M.Mehdipouretal.:AnatomyoftheAGNinNGC5548.I. Thespectralanalysisandmodelling,presentedinthiswork, FortheXMM-Newtonobservations,thelowestX-rayfluxes were done using the SPEX1 package (Kaastra et al. 1996) ver- wereobservedinObs.1:F =1.40×10−12ergcm−2s−1 0.3−2keV sion 2.05.02.We also made use of tools in NASA’s HEASOFT2 and F = 1.51 × 10−11 ergcm−2 s−1. These fluxes, 2−10keV v6.14package.Thespectrashowninthisreportarebackground- compared to those from the unobscured 2000 and 2001 ob- subtracted and are displayed in the observed frame, unless servations, are smaller by factors ranging from 17–27 (for otherwise stated in the text. We use C-statistics (Cash 1979) 0.3–2 keV) and 2–3 (for 2–10 keV). Later on, in the summer for spectral fitting (unless otherwise stated) and give errors XMM-Newton campaign, the X-ray fluxes increased from their at 1σ (68%) confidence level. The redshift of NGC 5548 is minimum in Obs. 1 by a factor of 2.8 (0.3–2 keV) and 2.3 set to 0.017175 (de Vaucouleurs et al. 1991) as given in the (2–10 keV), peaking at XMM-Newton Obs. 4. The X-ray flux NASA/IPAC Extragalactic Database (NED). The adopted cos- during the summer XMM-Newton campaign was low relative mologicalparametersforluminositycomputationsin ourmod- to other obscured epochs, as seen from the long-term Swift ellingareH =70kms−1Mpc−1,Ω =0.70andΩ =0.30. monitoring. The average fluxes of the summer 2013 XMM- 0 Λ m Newton observations (F = 3.02 × 10−12 ergcm−2 s−1 0.3−2keV and F = 2.77 × 10−11 ergcm−2 s−1) are smaller by a 2−10keV 2. Multi-wavelengthcampaignonNGC5548 factor of 1.6 (for 0.3–2 keV) and 1.1 (for 2–10 keV) than the At the core of our campaign in summer 2013 (22 June to 1 average fluxes for the whole obscured epoch observed with August), there were 12 XMM-Newton observations, of which Swift between February 2012 and July 2014 (Mehdipour et al. five were taken simultaneously with HST COS, four with inprep). INTEGRAL and two with NuSTAR observations. Throughout our campaign and beyond, Swift monitored NGC 5548 on a After the end of the summer XMM-Newton observations daily basis. There were also optical monitoringswith WO and (Obs.1–12),SwiftkeptonmonitoringNGC5548untilthemid- OCA. The summer XMM-Newton observations were followed dle of September 2013, when it passed out of Swift’s visibility by three Chandra LETGS observations taken in the first half window. However,prior to that a sudden increase in the X-ray of September 2013, one of which was taken simultaneously flux was observedwith Swiftwhich started on 24 August2013 withNuSTAR.TheSeptemberobservationsweretriggeredupon (56528inMJD)andlastedaboutoneweek.Comparedtotheav- observing a large jump in the X-ray flux from our Swift mon- eragefluxofthesummerXMM-Newtonobservations,thefluxin- itoring.However,dueto schedulingconstraintsbythe time the creasedbyafactorof3.4(for0.3–2keV)and1.9(for2–10keV). triggeredChandraobservationsweremade,theweek-longpeak ThisresultedinthetriggeringoftheChandraLETGSobserva- of high X-ray flux was just missed and the tail end of the flare tions. Later on, the two final XMM-Newton observations(Obs. was caught. Nonetheless, the X-ray flux was still higher than 13 and 14) were taken in December 2013 and February 2014 during the XMM-Newton observations and improved LETGS (theseareoutsidetherangeofFig.2lightcurves).Comparedto spectra were obtained. During this autumn period (Sep-Nov the averageXMM-Newton flux during the summer 2013 obser- 2013),NGC5548wasnotvisibletoXMM-Newtonandthusno vations, the Obs. 13 flux was slightly lower by a factor of 1.2 XMM-Newtonobservationscouldbetriggered. (for0.3–2keV)and1.1(for2–10keV),andforObs.14theflux washigherbyafactorof1.5for0.3–2keVandaboutthesame A few months later, two more XMM-Newton observations for 2–10 keV. A comprehensive analysis of the XMM-Newton were taken, one in December 2013 and the other in February lightcurves and their properties is reported in Cappi et al. (in 2014.TheformerobservationwassimultaneouswithHSTCOS prep). and NuSTAR observations, and the latter close in time to an INTEGRAL observation. The timeline of all the observations inourcampaignisdisplayedinFig.1andtheobservationlogs 3.2.NIR/optical/UVlightcurves areprovidedinTable1.InAppendixA,wedescribetheobser- vations made by the instruments of each observatory and give In Fig. 3 we show the optical/UV lightcurves from Swift detailsontheirdatareductionandprocessing. UVOT, XMM-Newton OM, WO and OCA, taken in the I, R, V, B, U, UVW1, UVM2 and UVW2 filters. Figure 4 shows the HST COS fluxes variability at six narrow UV energy 3. LightcurvesofNGC5548 bands (which are free of spectral features and represented by their central wavelengths) between 1160 and 1793 Å for the Here we present lightcurves of NGC 5548, from NIR to hard summer 2013 observations. To calculate the flux in broadband X-rays, obtained from the eight observatories used during our photometricfilters, onerequiresconvolutionof theresponseof summer2013campaign. the instrument and the transmission of the filter as a function of wavelength, with the shape of the optical/UV spectrum, to 3.1.X-raylightcurves convert from instrumental units to energy flux at an effective wavelength. For the shape of the optical/UV spectrum, rather Figure 2 shows the X-ray lightcurves of NGC 5548, from than using default spectral shapes commonly incorporated in the Swift, XMM-Newton, NuSTAR, INTEGRAL and Chandra the data reduction softwares (such as standard stars spectra, observations, with the average energy flux of each observation or a power-law), we used the optical/UV spectrum model that calculated over four X-ray energy bands from soft to hard we derived for NGC 5548from modelling the OM and UVOT X-rays. The X-ray fluxes have been calculated by fitting each Optical and UV grism spectra and the HST COS continuum datasetovertherequiredenergybandwiththemodeldescribed data (see Sect. 4 and Fig. 5). This approach allows for a more lateroninSect.6. accurate flux calculation in the photometric filters, taking into account the presence of strong AGN emission lines with a 1 http://www.sron.nl/spex realisticcontinuummodel.Forinstance,forthesixSwiftUVOT 2 http://heasarc.nasa.gov/lheasoft filters, the custom-fit spectral model results in 5 to 15% flux 3 M.Mehdipouretal.:AnatomyoftheAGNinNGC5548.I. Summer/autumn 2013 campaign on NGC 5548 15−Jun−13 1−Jul−13 15−Jul−13 1−Aug−13 15−Aug−13 1−Sep−13 15−Sep−13 OCA WO Chandra INTEGRAL NuSTAR HST/COS XMM−Newton Swift 20 40 60 80 Days (MJD) − 56455 Winter 2013/2014 campaign on NGC 5548 15−Dec−13 1−Jan−14 15−Jan−14 1−Feb−14 WO INTEGRAL NuSTAR HST/COS XMM−Newton Swift 190 200 210 220 230 240 Days (MJD) − 56455 Fig.1. Timeline of our multi-wavelengthcampaign on NGC 5548.The thickness of each rectangularsymbol on the time axis is indicativeofthelengthofthatobservation.ThedaysinwhichSwift,WOandOCAmadeobservationsareindicatedbycrosses. Table 1.ObservationlogoftheNGC5548campaign.FortheSwift, OCA andWO monitorings,theObs.numberscorrespondto daysinwhichobservationsweretaken.ForSwift,onlyrecentobservationstakenin2013–2014arereported,includingallmonitoring programsduringthisperiod.TheSwiftlengthsinksarethetotallengthoftheobservationsineachyear.ForHSTCOSthespanof eachobservationinksisgiven. Starttime(UTC) Length Starttime(UTC) Length Observatory Obs. ID yyyy-mm-dd hh:mm (ks) Observatory Obs. ID yyyy-mm-dd hh:mm (ks) XMM-Newton 1 0720110301 2013-06-22 03:53 50.5 HSTCOS 1 lc7001 2013-06-22 13:25 13.0 XMM-Newton 2 0720110401 2013-06-29 23:50 55.5 HSTCOS 2 lc7002 2013-07-12 02:23 14.2 XMM-Newton 3 0720110501 2013-07-07 23:28 50.9 HSTCOS 3 lc7003 2013-07-24 16:43 8.9 XMM-Newton 4 0720110601 2013-07-11 23:11 55.5 HSTCOS 4 lc7004 2013-07-30 15:15 16.0 XMM-Newton 5 0720110701 2013-07-15 22:56 55.5 HSTCOS 5 lc7005 2013-08-01 03:21 12.3 XMM-Newton 6 0720110801 2013-07-19 22:40 56.5 HSTCOS 6 lc7006 2013-12-20 02:49 13.0 XMM-Newton 7 0720110901 2013-07-21 22:32 55.5 NuSTAR 1 60002044002 2013-07-11 09:50 51.6 XMM-Newton 8 0720111001 2013-07-23 22:24 55.5 60002044003 2013-07-12 00:10 52.2 XMM-Newton 9 0720111101 2013-07-25 22:15 55.5 NuSTAR 2 60002044005 2013-07-23 14:25 97.2 XMM-Newton 10 0720111201 2013-07-27 22:06 55.5 NuSTAR 3 60002044006 2013-09-10 21:25 97.5 XMM-Newton 11 0720111301 2013-07-29 21:58 50.4 NuSTAR 4 60002044008 2013-12-20 08:30 98.1 XMM-Newton 12 0720111401 2013-07-31 21:49 55.5 Chandra 1 16369 2013-09-01 00:01 29.7 XMM-Newton 13 0720111501 2013-12-20 14:01 55.3 Chandra 2 16368 2013-09-02 10:33 67.5 XMM-Newton 14 0720111601 2014-02-04 09:33 55.5 Chandra 3 16314 2013-09-10 08:17 122.0 INTEGRAL 1 10700010001 2013-06-29 21:34 100.0 Swift(2013)a 1–160 b 2013-01-04 00:24 326.6 INTEGRAL 2 10700010002 2013-07-11 21:13 102.0 Swift(2014)c 161–291 d 2014-01-02 14:53 182.5 INTEGRAL 3 10700010003 2013-07-15 23:31 106.5 OCAe 1–27 - 2013-05-20 03:17 e INTEGRAL 4 10700010004 2013-07-23 03:54 128.9 WO(2013)f 1–93 - 2013-06-02 07:22 f INTEGRAL 5 11200110003 2014-02-09 10:00 38.2 WO(2014)g 94–150 - 2013-12-16 14:38 g a TheSwiftmonitoringin2013endedon2013-12-3118:19. b TheSwifttargetIDsofNGC5548in2013:30022,80131,91404,91711,91737,91739,91744,91964. c TheSwiftmonitoringupto2014-07-0100:00isreportedhere.TheSwiftmonitoringofNGC5548iscurrentlyongoingin2014–2015. d TheSwifttargetIDsofNGC5548in2014:30022,33204,91964. e TheOCAmonitoringendedon2013-07-2501:13.ObservationstakenintheB,V,Rfilters,with150sexposureineachfilter. f TheWOmonitoringendedon2013-09-2404:59.ObservationstakenintheB,V,R,Ifilters,with300sexposureineachfilter. g TheWOmonitoringendedon2014-04-1408:25.ObservationstakenintheB,V,R,Ifilters,with300sexposureineachfilter. 4 M.Mehdipouretal.:AnatomyoftheAGNinNGC5548.I. 7.0 1.2 0.3−2.0 keV 2.0−10 keV 6.0 −2−1m s 01..80 XELERPTITCG−Spn 45..00 XELERPTITCG−Spn c g 0.6 er 3.0 −110 0.4 2.0 1 0.2 1.0 0.0 0.0 8.0 10−30 keV 8.0 30−80 keV −2−1m s 6.0 NuSTAR 6.0 NINuTSETGARRAL c g 4.0 4.0 er −110 2.0 2.0 1 0.0 0.0 0 20 40 60 80 0 20 40 60 80 Days (MJD) − 56450 Days (MJD) − 56450 Fig.2.X-raylightcurvesofNGC5548fromoursummer2013campaign.Thelightcurvesshowtheobservedfluxandaredisplayed between 7 June and 14 Sep 2013. The XRT data are shown as blue circles, EPIC-pn as magenta stars, LETGS as dark yellow squares,NuSTARasreddiamondsandINTEGRALdataasgreensquares.Theverticaldashedlinesindicatethe intervalbetween XMM-NewtonObs.1and12. −1AÅ 0.8 I WO R WO −1s OCA −2m 0.7 c g er 0.6 −140 1 0.5 1.6 −1AÅ V UVOT B UVOT −2−1m s 1.4 OWOMCOA OWOMCOA g c 1.2 er −140 1.0 1 4.0 −1AÅ 3.5 U UVW1 −2−1m s 3.0 UOVMOT UOVMOT c 2.5 g er 2.0 −140 1.5 1 4.0 −1AÅ 3.5 UVM2 UVW2 −2−1m s 3.0 UOVMOT UOVMOT c 2.5 g er 2.0 −140 1.5 1 0 20 40 60 80 0 20 40 60 80 Days (MJD) − 56450 Days (MJD) − 56450 Fig.3.NIR,opticalandUVlightcurvesofNGC5548takenwithvariousfiltersfromoursummer2013campaign.Thelightcurves show the observed energy flux and are displayed between 7 June and 14 Sep 2013. The UVOT data are shown as blue circles, OM asmagentastars, WO asredcircles,andOCA dataasgreensquares.The verticaldashedlinesindicatetheintervalbetween XMM-NewtonObs.1and12. 5 M.Mehdipouretal.:AnatomyoftheAGNinNGC5548.I. 4.0 HST COS 4.0 HST COS CCOOSS OObbss.. 11 CCOOSS OObbss.. 22 11116600 AÅAÅ CCOOSS OObbss.. 33 −1AÅ 3.5 11336677 AÅAÅ −1AÅ 3.5 CCCCOOOOSSSS OOOObbbbssss.... 4545 −1s 11448800 AÅAÅ −1s −2m 3.0 11773355 AÅAÅ −2m 3.0 c 11775500 AÅAÅ c g g er 2.5 11779933 AÅAÅ er 2.5 −140 −140 1 1 2.0 2.0 1.5 1.5 10 20 30 40 50 60 1200 1300 1400 1500 1600 1700 1800 Days (MJD) − 56450 Observed Wavelength (ÅA) Fig.4. HST COS fluxesof NGC 5548at differentUV wavelengthsfromsummer2013observations.Thelightcurves(leftpanel) and spectra (right panel) show the observed energy flux obtained from narrow energy bands which are free of spectral features. Thedisplayedrangeofthelightcurvesisbetween16Juneand6August2013,withtheverticaldashedlinesindicatingtheinterval betweenXMM-NewtonObs.1and12. improvement in our knowledge of the continuum flux in each and the modelling of the X-ray components (6–9) in Sects. filter. The calculation of the COS fluxes at the narrow UV 4.6–4.9. We have modelled the grism spectra from OM and energy bands is described in Appendix A.6. In the lightcurves UVOT together with simultaneous HST COS continuum mea- of Fig. 3, the UVOT and OM fluxes have been calculated at surements to establish the contributionof each componentand thefollowingwavelengthsforeachfilter:V (5402Å),B (4329 consequently correct the data taken in the photometric filters. Å), U (3501 Å), UVW1 (2634 Å), UVM2 (2231 Å), UVW2 For fitting the grism spectra in SPEX we used χ2 minimisa- (2030 Å). For OCA and WO the fluxes have been given at the tion instead of C-statistics (used for X-ray spectra), as the op- tical/UV spectra have sufficient counts per bin. In Fig. 5 the followingwavelengthsforeachfilter:I(8060Å),R(7000Å),V best-fitmodelto thestackedOM andUVOTgrismspectra and (5500Å),B(4330Å). HST COS continua obtained from our simultaneous 2013 ob- servations are shown. In Fig. 6 we display all the stacked data As illustrated in the Fig. 3 lightcurves,the NIR/optical/UV fromoursummer2013campaign.Thefigureshowshowmuch fluxes were more or less continuously increasing throughout theaforementionedNIR/optical/UVcorrectionsmodifythedata our summer XMM-Newton campaign (Obs. 1–12) and beyond, fromobserved(leftpanel)tocorrected(rightpanel);theemerg- until the end of our Swift monitoring in middle of September ing shape of the NIR/optical/UV SED after the corrections is 2013, when the source was not visible to Swift for a couple of remarkableand exhibits a thermal disk spectrum. In Fig. 6 the months. In fact the optical/UV fluxes recordedby Swift at mid displayedX-raydataareonlycorrectedforGalacticinterstellar SeptemberwerethehighestobservedinalloftheSwiftmonitor- absorptionandstillincludetheeffectsofheavyabsorptionbythe ingofNGC5548spanningfrom2005to2014,whichalsocoin- obscurer and the ‘de-ionised’ warm absorber (i.e. less ionised cidedwithourtriggeredChandraLETGSobservations.Figure4 warmabsorberduetotheshieldingoftheionisingradiationby showsthattheCOSUVfluxesincreasedfromtheirminimumat theobscurer).Similarly,inFig.7thecorrectedoptical/UV(OM) XMM-NewtonObs.1(COSObs.1)andpeakedatXMM-Newton and X-ray (EPIC-pn and RGS) data from 2000, 2001 and av- Obs. 8 (COS Obs. 3), followed by a decrease afterwards. The eragesummer2013are displayed,demonstratingthatwhilst in COS flux changefromthe first observationgets largertowards 2013 the optical/UV continuum is higher than during the un- higher UV energies:factor of 1.8 at 1793 Å to factor of 2.3 at obscuredepochs,thesoftX-raysare heavilysuppressedbythe 1160Å. obscurer. 4. Determinationofthecontinuumemission In order to establish the actual intrinsic continuum in the 4.1.Galacticinterstellarreddening NIR/optical/UV part of the SED, the following absorption effects and emission contributions must be taken into account: (1)Galacticinterstellarreddening;(2)Hostgalaxystellaremis- sion; (3) Emission lines from the broad-line region (BLR) and Tocorrecttheoptical/UVfluxesforinterstellarreddeninginour narrow-line region (NLR); (4) Feii blended emission feature; Galaxy,the reddeningcurveof Cardelliet al.(1989)wasused, (5) Balmer continuum emission. Moreover, in the X-ray band including the update for near-UV given by O’Donnell (1994). the following must be taken into account: (6) Galactic X-ray TheebvmodelinSPEXappliesthisde-reddeningtothedata.For absorption; (7) Absorption by the obscurer; (8) Absorption by NGC5548thecolourexcessisE(B−V) = 0.02mag(Schlegel thewarmabsorber;(9)SoftX-rayemissionlines. etal. 1998).Thescalar specifyingthe ratioof totalto selective extinctionR ≡ A /E(B−V)wassetto3.1.InFig.5,thered- V V In the following subsections, we describe the modelling of dened continuum model is shown in dotted black line and the the above NIR/optical/UV components (1–5) in Sects. 4.1–4.5 de-reddenedoneindashedblackline. 6 M.Mehdipouretal.:AnatomyoftheAGNinNGC5548.I. 4.2.Hostgalaxystellaremission 4.5.Balmercontinuumfeature To take into accountthe contributionof starlightfrom the host ApartfromFeiithereisanothersignificantemissioncomponent galaxy of NGC 5548 in our NIR/optical/UV data, we pro- in the ∼2000–4000 Å region of AGN spectra: the Balmer duced an appropriate model for inclusion in our spectral mod- continuum (see e.g. Grandi 1982). The Balmer continuum for elling.Bentzetal.(2009,2013)havedeterminedthehostgalaxy an optically-thin emission with constant electron temperature flux at an optical wavelength for a sample of AGN (includ- is given by FBC =FBEe−h(ν−νBE)/kTe (Grandi 1982), where FBC ν ν ν ing NGC 5548) using HST. However, for our observations of is the Balmer continuumflux at frequencyν, FBE is the fluxat ν NGC 5548,this host galaxy flux was re-calculatedto take into the Balmer edge ν , h the Planck constant, k the Boltzmann BE account the 5 arcsec radius circular aperture used in our pro- constantandT theelectrontemperature.TheBalmeredgeisat e cessings. For the HST F550M medium-band V filter, the host theoretical wavelength of 3646 Å. We used this model to take galaxyfluxina5arcsecradiuscircularaperturewasdetermined into account the contribution of the Balmer continuum in our tobe6.2×10−15ergcm−2s−1Å−1withanuncertaintyof∼10% opticaldata.ThemodelwasconvolvedwithanintrinsicDoppler (MistyBentz,privatecommunication).Theeffectivewavelength velocitybroadeningσ andtheresolutionofthegrismsinSPEX. v of thisflux measurementis at an observedwavelengthof5580 Wefittedthismodeltothegrismdataforvariouscombinations Å. Then, in order to calculate the host galaxy spectrum at the of T and σ values and found the best-fit Balmer continuum e v otherwavelengths,weusedatemplatemodelspectrumandnor- componenthasTe ∼ 8000K andσv ∼ 10000kms−1.We note malisedittoNGC5548hostgalaxyfluxat5580Å.A5arcsec thatthepurposeoffittingtheBalmercontinuuminthisworkis aperturetakesinonlytheinnermostfewkpcofthehostgalaxy, to obtain its total flux contribution to the optical data in order and so the galaxy bulge template of Kinney et al. (1996) was to establish the underlying continuum component, so we do adopted. In Fig. 5, the contribution of the host galaxy stellar notdelveintothe Te and σv parametershere whichcarrylarge modelisdisplayedasthesolidgreenline. uncertainties. The flux contribution by the Balmer continuum wasfoundtobeabout5.75×10−12ergcm−2s−1. 4.3.EmissionlinesfromBLRandNLR TheBalmercontinuum,togetherwithFeiiemission,forma broad and blended feature in AGN spectra dubbed the ‘small- TotakeintoaccountemissionlinesproducedfromtheBLRand blue-bump’. This is displayed in Fig. 5, in which our best-fit NLR,wemodelledtheOMOpticalgrismandUVOTUVgrism model to the grism spectra and HST COS continua points are spectra,coveringarangefrom1895to6850Å.Thesimultane- shown. The small-blue-bump is the feature in magenta colour ity of the grism spectra with COS is important as it enables between ∼2000–4000 Å; beneath that the Balmer continuum us to model the grism spectra and COS continuum points to- model(withoutFeiiemission)isdisplayedindarkyellow. gether. This wider energy band in the optical/UV helps in es- tablishingthe underlyingoptical/UVcontinuumand modelling theBLRandNLRlines.Fortheunderlyingoptical/UVcontin- 4.6.GalacticinterstellarX-rayabsorption uum, the Comptonisationcomponentcomt in SPEX (explained later in Sects. 5 and 6) was used. The broad and narrow emis- The effects of the Galactic neutral absorption in the interstel- sionlinesweremodelledusingGaussianlineprofiles.Forbroad lar mediumare includedin ourmodellingby applyingthe hot lines, multiple Gaussian componentswith different widths, but modelinSPEX.Thismodelcalculatesthetransmissionofgasin the same wavelength were added until a good fit was obtained collisional ionisation equilibrium.For a given temperature and foreach line. The best fit obtainedfor the BLR and NLR lines set of abundances, the model calculates the ionisation balance was saved as a spectral model componentwhich was included and then determines all the ionic column densities by scaling in all ourbroadbandmodellinglater in Sect. 6, to take into ac- totheprescribedtotalhydrogencolumndensity N .Thetrans- H count the contribution of these lines in the photometric filters. missionincludesbothcontinuumandlineopacity.Tomimicthe ThefluxcontributionofalltheBLR/NLRlinesinthegrismen- transmissionofaneutralplasmaincollisionalionisationequilib- ergy range shown in Fig. 5 (solid black lines) was found to be rium(suchastheinterstellarmediumofourGalaxy),thetemper- about7.94×10−12ergcm−2s−1. atureoftheplasmaissetto0.5eV.InourmodellingtheGalactic HicolumndensityinthelineofsighttoNGC5548wasfixedto N =1.45×1020cm−2 (Wakkeretal.2011)withLoddersetal. H 4.4.Feiiblendedemissionfeature (2009)abundances. The Feii in the BLR produces several thousands of transi- tions,whichresultinablendedandcomplexspectrumbetween 4.7.Absorptionbytheobscurer ∼2000–4000Å (seee.g.Netzer&Wills1983;Willsetal.1985). In order to take into account the Feii contribution in our opti- Totakeintoaccountthenewly-discoveredX-rayobscurationin cal data we took the model calculated by Wills et al. (1985), NGC5548weadoptedthesamemodelfoundbyKaastraetal. which is convolved to an intrinsic width of 2500 kms−1. This (2014). In this model the obscurer consists of two ionisation wasthenimportedintoSPEXasaspectralmodelcomponentand phases, each modelled with an xabs component in SPEX. The convolvedwith the resolutionofthe grisms. The normalisation xabs model calculates the transmission through a slab of pho- scaling factor of the componentwas left as a free parameterto toionised gas where all ionic column densities are linked in a allowforfittingthetotalFeiifluxinNGC5548.Thefluxcontri- physically consistent fashion through the CLOUDY photoionisa- butionbyFeiiinNGC5548wasfoundtobeabout2.79×10−12 tionmodel.Thefittedparametersofanxabsmodelcomponent ergcm−2 s−1. Thiscomponentis shown in Fig. 5 as partof the aretheionisationparameter(ξ),theequivalenthydrogencolumn magentacolouredfeature. density (N ), the coveringfractionC of the absorber,its flow H f 7 M.Mehdipouretal.:AnatomyoftheAGNinNGC5548.I. ] 4.0 HST C III g II NGC 5548 COS M (Summer 2013) UV I] I I −1AÅ 3.0 grism Optical [O 1 α −s grism H 2 − m β c H g r 2.0 e 4 1 − 0 1 1.0 0.0 1000 2000 3000 4000 5000 6000 7000 Observed Wavelength (AÅ) Fig.5.De-reddenedoptical/UVspectrumofNGC5548.ThedisplayeddataarefromsimultaneousUVOTUVgrismspectra(shown inlightblue)at1896–3414Å,OMOpticalgrismspectra(shownindarkblue)at3410–6836Åandthesix HSTCOScontinuum points(shownaspurplesquares)between1160–1793Å.ThedisplayeddataaretheaverageofthecontemporaneousOM,UVOT andHSTCOSobservationstakeninsummer2013andarecorrectedforGalacticreddeningusingthemodeldescribedinSect.4.1. Thebest-fitmodel(describedinSect.4)isshowninredandvariouscomponentscontributingtothemodelarealsodisplayed.The underlyingcontinuummodel(comt)is the dashedblack curve.The dotted blackcurveis the reddenedversionof the continuum toillustratethecorrectionforreddening.Thebroadandnarrowemissionlinecomponents(Sect.4.3)areshownatthebottomand someoftheprominentemissionlinesarelabelled.Thebroadfeatureinmagentaisthe‘small-blue-bump’:blendedFeiiemission (Sect. 4.4) with Balmer continuum (Sect. 4.5). The model for the Balmer continuum alone is shown in dark yellow below the small-blue-bump.ThecontributionfromthehostgalaxyofNGC5548(Sect.4.2)isshowningreen. 1013 NGC 5548 1013 NGC 5548 (Summer 2013) (Summer 2013) Uncorrected Corrected Hz) 1012 OM Optical grism 1012 y UVOT UV grism J νF (ν IU RV WV 1B U I R V B U UVM2 XMM RGS UVW1 XMM RGS UVW2 XMM EPIC−pn UVM2 XMM EPIC−pn HST COS NuSTAR UVW2 NuSTAR 1011 1793 AÅ 1750 AÅ INTEGRAL 1011 1793 AÅ 1750 AÅ INTEGRAL 1735 AÅ 1480 AÅ 1735 AÅ 1480 AÅ 1367 AÅ 1160 AÅ 1367 AÅ 1160 AÅ 0.001 0.01 0.1 1 10 100 0.001 0.01 0.1 1 10 100 Observed Energy (keV) Observed Energy (keV) Fig.6.Averagemulti-wavelengthdataofNGC5548fromoursummer2013campaign.Thedataaredisplayedbefore(leftpanel) andafter(rightpanel)correctionsforGalacticreddening(Sect4.1),hostgalaxystellaremission(Sect4.2),emissionlinesfromthe BLRandNLR(Sect4.3),blendedFeiiemission(Sect.4.4),Balmercontinuum(Sect.4.5)andGalacticinterstellarX-rayabsorption (Sect4.6).Thespectrahavebeenbinnedforclarityofpresentation.ThedisplayedX-raydataincludetheeffectsofheavyabsorption bytheobscurerandthewarmabsorber.TheshapeoftheNIR/optical/UVdataafterthecorrections(rightpanel)isremarkableand exhibitsathermaldiskspectrum. 8 M.Mehdipouretal.:AnatomyoftheAGNinNGC5548.I. 0.33 to 2.67, σ from 20 to 210 kms−1 and outflow velocities v 1013 rangingfromabout250to1220kms−1.Forthe2000and2001 NGC 5548 XMM-Newton data, when the warm absorber would have been exposed to normal ionising flux, parameters of the warm ab- sorber were taken from a re-analysis of their RGS data, using the same methodas in Kaastra et al. (2014).Thisworkwill be 1012 reportedinaforthcomingpaperonourcampaign(Ebreroetal. Hz) 2000 (24 Dec) inprep),wherethelong-termvariabilityofthewarmabsorberis Jy 2001 (09 Jul) goingtobepresented. F (ν 2001 (12 Jul) ν Summer 2013 4.9.SoftX-rayemissionlines 1011 The2013RGSdataclearlyshowthepresenceofseveralnarrow emission lines and radiative recombinationcontinuafrom pho- toionisedgas(Kaastraetal.2014).Theseemissionfeaturesare 0.01 0.1 1 10 moreapparentin2013asthesoftX-raycontinuumissuppressed Observed Energy (keV) duetoobscuration,buttheyarealsopresentatearlierepochs.In orderto take into accounttheir flux contributionin our contin- Fig.7.Optical/UV(OM)andX-ray(RGSandEPIC-pn)dataof uummodellingweusedthesamemodelandparametersreported NGC5548fromunobscured(2000,2001)andobscured(2013) inKaastraetal.(2014).Thecontributionoftheselinestotheto- epochs.Theoptical/UVdatahavebeencorrectedfortheeffects tal0.3–2.0keVfluxis1.4%(in2000),0.8–1.3%(in2001)and described in Sects. 4.1–4.5. The X-ray spectra have been cor- 7.7%(in2013).Sowhilstintheunobscuredepochstheircontri- rectedforGalacticabsorption(Sect.4.6),butnotforabsorption butionisrathersmall,in obscuredepochstheycontributemore bytheobscurerandthewarmabsorber.ThedisplayedX-raydata tothesoftX-rayflux.Inanotherforthcomingpaperonourcam- between0.3–1.4are RGS andathigherenergiesEPIC-pn.The paign by Whewell et al. (in prep), a detailed study of the soft X-ray spectra have been binned for clarity of presentation. In X-ray emission features as detected by RGS in NGC 5548 is the2013data,thesoftX-rayspectrumisthelowestoneandthe reported. optical/UVdatathehighestone. 5. ThesoftX-rayexcessinNGC5548 vandturbulentσ velocities.Theionisationparameterξ(Tarter v The AGN ‘softX-rayexcess’ is an excesscontinuumemission etal.1969)isdefinedas above the intrinsic X-ray power-law at the soft X-ray energies L (below ∼ 2 keV). Since its discoveryby Singh et al. (1985)in ξ ≡ (1) n r2 HEAO-1 observations of Mrk 509 and also by Arnaud et al. H (1985) in EXOSAT observations of Mrk 841, it has been ob- where L is the luminosity of the ionising source over the 1– served in manySeyfert-1 AGN. For example,in the Catalogue 1000 Ryd (13.6 eV to 13.6 keV) band in erg s−1, nH the hy- of AGN In the XMM-NewtonArchive(CAIXA, Bianchiet al. drogen density in cm−3 and r the distance between the ionised 2009), this component was commonly found in about 80% of gas and the ionising source in cm. The first component of the the AGN. In NGC 5548 the soft excess has been previously obscurercoversabout86%ofthecentralX-rayemittingregion, detected(e.g.Kaastra&Barr1989;Magdziarzetal.1998)and withlogξ = −1.2and NH =1.2×1022cm−2.Thesecondcom- itspresenceintheunobscuredstateisapparent.Figure8shows ponent of the obscurer covers 30% of the X-ray source and is the EPIC-pnspectrum obtainedin 2000.The continuumabove almost neutral (logξ = −4.0) with NH =9.6×1022cm−2. The 3.0 keV has been fitted with a Galactic-absorbed power-law parametersoftheobscurerwerefixedtothoseofKaastra etal. component (Γ∼1.8), including a reflection component for (2014)inourbroadbandspectralmodellingofthestackedsum- modelling the Fe Kα line. The best-fit is then extrapolated mer2013data.Ofcoursefortheunobscuredepochs(2000and to lower energies, displaying the presence of the soft excess 2001 XMM-Newton data), the obscurer components were ex- belowabout2keV.However,thesoftexcessinNGC5548only cluded from our modelby setting Cf of both obscurer compo- stands out in data before the X-ray obscuration began (i.e. < nentstozero. 2011).Becauseofthenewlydiscoveredobscurer(Kaastraetal. 2014), the soft X-ray flux is heavily absorbed in 2013. This demonstratesthatthesoftexcessemissionisproducedinasmall 4.8.Absorptionbythewarmabsorberoutflows regioninteriortothe obscurer(e.g.theBLR) andisnotfroma Sincetheobscurerislocatedbetweenthecentralionisingsource large-scalescatterer. Yet, the softX-raybandshowssignificant and the warm absorber, it prevents some of the ionising radia- variability which can be attributed to both the obscurer and/or tionfromreachingthewarmabsorber.Thusthedifferentphases soft excess component. Therefore, for any variability study of ofthewarmabsorberbecomelessionised,resultinginmoreX- the obscurer, the variability of the soft excess component also ray absorption than when NGC 5548 was unobscured. In our needs to be established. This means an appropriate model for modellingoftheobscureddataweusedthede-ionisedwarmab- thesoftexcessisrequiredwhichwedescribebelow. sorber modelobtainedby Kaastra et al. (2014),which consists of six different phases of photoionisation.Each phase is repre- There are different interpretationsand models in the litera- sented by an xabs componentin SPEX with N , ξ, flow v and tureforthenatureoftheenigmaticsoftexcessemissioninAGN H turbulentσ velocitieskeptfrozenthroughoutourmodellingto (seeSect.8.1).ForthecaseofNGC5548,evidence(described v the values given in Kaastra et al. (2014). The warm absorber below) suggests that the soft X-ray excess is primarily part phases have N ranging from 2 to 57 ×1020 cm−2, logξ from of an optical/UV/soft X-ray continuum component, produced H 9 M.Mehdipouretal.:AnatomyoftheAGNinNGC5548.I. NGC 5548 0.8 NGC 5548 EPIC-pn XXMMMM ((22000000)) XXMMMM ((22000011)) o ati 0.6 SSwwiifftt ((22000077)) ux r XXMMMM ((22001133)) y fl a soft excess X−r d 0.4 ar H oft / S 0.2 0.5 1.0 1.5 2.0 UVW2 flux (10−14 erg cm−2 s−1 AÅ−1) Fig.8.EPIC-pnspectrumofNGC5548from2000,fittedabove −1s) NGC 5548 3.0 keV with a Galactic-absorbedpower-law(Γ∼1.8), includ- erg 8.0 XXMMMM ((22000000)) ingareflectioncomponentfortheFeKα.Thefitisextrapolated 420 XXMMMM ((22000011)) 1 to lower-energies displaying the presence of a soft excess be- y ( SSwwiifftt ((22000077)) lowabout2keV.Thefitresiduals,(observed−model)/model,are nosit 6.0 XXMMMM ((22001133)) mi displayedinthebottompanel. u V) l ke 4.0 2 − by ‘warm Comptonisation’: up-scattering of the seed disk 0.3 photons in a Comptonising corona, which is distinct from the s ( es 2.0 one responsible for the hard X-ray power-law (i.e. the hot xc e corona),andischaracterisedbyalowertemperatureandhigher oft S optical depth (e.g. Magdziarz et al. 1998). Firstly, without any modelling, the unobscured NGC 5548 data (Fig. 9, top panel) 0.5 1.0 1.5 2.0 UVW2 flux (10−14 erg cm−2 s−1 AÅ−1) show that as the UV flux increases, the ratio of soft to hard X-ray flux (‘softness’ ratio) also increases. This indicates the Fig.9. Top panel:Ratio of observedsoft(0.3–2.0keV) to hard UV and soft excess are related to each other. In Fig. 9 (top (2.0–10 keV) X-ray flux plotted versus the observed UVW2 panel)theX-raysoftnessratioisplottedversustheUVfluxfor (2030 Å) flux. Bottom panel: Intrinsic luminosity of the soft the unobscured data (XMM-Newton 2000 and 2001 and Swift excess componentbetween 0.3–2.0 keV plotted versus the ob- 2007), and also for the obscured XMM-Newton data from the servedUVW2(2030Å)flux,andfittedwiththepower-lawfunc- average of the summer 2013 observations. The softness ratio tiongiveninEq.2. corresponds to the observed flux, without any modelling and correctionsforabsorption.Remarkably,fortheobscuredepoch in2013,thesoftnessratioisatitslowestleveleventhoughthe InFig.9(bottompanel)weshowthesoftexcessluminosity UV flux is at its highest. The reason for this sharp drop in the inthe0.3–2.0keVbandversustheUVflux,showingtherelation softnessratioisthepresenceoftheheavysoftX-rayabsorption between the soft excess and UV. The soft excess luminosity is bytheobscurerin2013(seeFig.7). calculated using the model described in Sect. 6 (i.e. corrected forobscurer,warmabsorberandGalacticabsorption).Thedata The link between the UV and soft excess flux was also inFig.9(bottompanel)havebeenfitted withafunctionwhich seen in Mrk 509 (Mehdipour et al. 2011). From broadband isfoundtobeapower-law,givenby modelling of its continuum, Mehdipour et al. (2011) and Petrucci et al. (2013) argued warm Comptonisation to be the Lsoftexcess =1.261(FUVW2)2.869 (2) most reasonable explanation for the soft excess. Furthermore, where L is the intrinsic luminosity of the soft excess warm Comptonisation has already been suggested for the soft softexcess componentinthe0.3–2.0keVbandinunitsof1042 ergs−1 and excessinNGC5548,inoneofthefirstpapersinvestigatingthis F istheUVW2filterfluxinunitsofergcm−2s−1Å−1.This interpretation, by Magdziarz et al. (1998). From broad-band UVW2 fluxintheUVW2filterisatthewavelengthof2030Åforboth spectral analysis of NGC 5548, using data from IUE, Ginga, UVOTandOM andisthe observedflux withoutanymodifica- ROSAT, CGRO and OSSE, Magdziarz et al. (1998)found that tionssuchasde-reddening,buttakesintoaccounttheoptical/UV thesoftexcessrequiresaseparatecontinuumcomponentwhich spectrumshapeofNGC5548(showninFig.5)intheUVW2fil- canbefittedbyComptonisationofthermalphotonsfromacold disk (T ∼ 3.2 eV), in a warm (∼ 0.3 keV), optically thick terbandpass.WefurtherdiscussthesoftexcessinSect.8.1. seed (τ∼30)corona.Theauthorsalsofoundtheoptical/UVandsoft X-ray fluxes in NGC 5548 to be closely correlated. Therefore 6. BroadbandcontinuummodellingofNGC5548 consideringtheaboveevidenceweadoptwarmComptonisation astheappropriatemodelforthesoftexcess. Here we describe our continuum modelling of the NGC 5548 data. The archival XMM-Newton data from 2000 and 2001 10