Spectral optical monitoring of 3C 390.3 in 1995-2007: II. Variability of the spectral line parameters

1.2 MB·English

by  L. C. Popovic

Astronomy&Astrophysicsmanuscriptno.Paper˙2˙astro˙ph (cid:13)c ESO2011
January26,2011
Spectral optical monitoring of 3C 390.3 in 1995-2007: II. Variability
of the spectral line parameters
L.Cˇ.Popovic´1,2,A.I.Shapovalova3,D.Ilic´2,4,A.Kovacˇevic´2,4,W.Kollatschny5,A.N.Burenkov3,V.H.Chavushyan6,
N.G.Bochkarev7,andJ.Leo´n-Tavares8
1 AstronomicalObservatory,Volgina7,11160Belgrade74,Serbia
2 IsaacNewtonInstituteofChile,YugoslaviaBranch
3 SpecialAstrophysicalObservatoryoftheRussianAS,NizhnijArkhyz,Karachaevo-Cherkesia369167,Russia
1 4 DepartmentofAstronomy,FacultyofMathematics,UniversityofBelgrade,Studentskitrg16,11000Belgrade,Serbia
1 5 Institutfu¨rAstrophysik,Friedrich-Hund-Platz1,Go¨ttingen,Germany
0 6 InstitutoNacionaldeAstrof´ısica,O´pticayElectro´nica,ApartadoPostal51,CP72000,Puebla,Pue.Me´xico
2 7 SternbergAstronomicalInstitute,Moscow,Russia
8 Metsa¨hoviRadioObservatory,HelsinkiUniversityofTechnologyTKK,Metsa¨hovintie114,FIN-02540Kylma¨la¨,Finland
n
a
Received/Accepted
J

5 ABSTRACT
2
Context.A study of the variability of the broad emission-line parameters of 3C390.3, an active galaxy with the double-peaked
] emission-lineprofiles,ispresented. Herewegive adetail analysisof variationinthebroad HαandHβ emission-lineprofiles,the
O
ratios,andtheBalmerdecrementofdifferentlinesegments.
C Aims.Withinvestigationofthevariabilityofthebroadlineprofilesweexplorethediskstructure,thatisassumedtoemitthebroad
. double-peakedHβandHαemissionlinesinthespectrumof3C390.3.
h Methods.Wedividedtheobservedspectraintwoperiods(beforeandaftertheoutburstin2002)andanalyzedseparatelythevariation
p inthesetwoperiods.Firstweanalyzedthespectralemission-lineprofilesoftheHαandHβlines,measuringthepeakpositions.Then,
- wedividedlinesintoseveralsegments,andwemeasuredtheline-segmentfluxes.TheBalmerdecrementvariationfortotalHαand
o
Hβfluxes,aswellasforthelinesegmentshasbeeninvestigatedanddiscussed.Additionally,wemodeledthelineparametersvariation
r
t usinganaccretiondiskmodelandcompareourmodeledlineparametervariationswithobservedones.
s
Results.Wecomparedthevariabilityintheobservedlineparameterswiththediskmodelpredictionsandwefoundthatthevariation
a
inlineprofilesandinthelinesegmentscorrespondstotheemissionofadisk-likeBLR.But,alsothereisprobablyoneadditional
[
emissioncomponent thatcontributestotheHαandHβlinecenter.Wefoundthatthevariationinthelineprofilesiscausedbythe

1 variation in the parameters of the disk-like BLR, first of all in the inner (outer) radius which can well explain the line parameter
v variationsinthePeriodI.TheBalmerdecrementacrossthelineprofilehasabell-likeshape,anditisaffectednotonlybyphysical
7 processesinthedisk,butalsobydifferentemittingdiskdimensionoftheHαandHβline.
6 Conclusions.ThegeometryoftheBLRof3C390.3seemstobeverycomplex,andinflows/outflowsmightbepresent,butitisevident
8 thatthebroadlineregionwithdisk-likegeometryhasdominantemission.
4
Keywords.galaxies:active–galaxies:quasar:individual(3C390.3)–line:profiles
.
1
0
1
1. Introduction accretion disk emission. On the other hand, the presence of an
1
accretion-disk emission in the BLR is expected, and double-
:
v The broad emission lines (BELs) are often observed in opti- peaked line profiles of some AGN indicate this (Perezetal.,
Xi calandultravioletspectraofacitvegalacticnuclei(AGN).The 1988;Eracleous&Halpern,1994,2004;Eracleousetal.,2009).
studyoftheprofilesandintensitiesofBELscangiveusrelevant
One of the well known AGN with broad double-peaked emis-
r informationaboutthegeometryandphysicsofthebroadlinere-
a sion lines in its spectrum is the radio-loud active galaxy 3C
gion(BLR).ThephysicsandgeometryoftheBLRareuncertain
390.3. Although, the double-peaked line profiles can be ex-
and investigation of BEL shape variability in a long period is
plained by different hypothesis (see e.g. Veilleux&Zheng,
veryusefulfordeterminationoftheBLRnature.Theprofilesof
1991),ase.g.super-massivebinaryblackholes(Gaskell,1996),
theBELsinAGNcanindicatethegeometryofemittingplasma
outflowing bi-conical gas streams (Zheng, 1996)), it seems
in the BLR (see e.g. Sulenticetal., 2000; Popovic´etal., 2004;
that in this case a disk emission is present in the BLR
Gaskell,2009;Zamfiretal.,2010,etc.).
(Shapovalovaetal., 2010, hereinafter Paper I). There is a pos-
Particularly,veryinterestingobjectsareAGNswithunusual sibilitythata jetemissioncanaffecttheopticalemissionin 3C
broad emission-line profiles, where broad Balmer lines show
390.3 (Arshakianetal., 2010), and some perturbations in disk
double peaks or double ”shoulders”, so called double-peaked couldbepresent(Jovanovic´etal.,2010)andtheycanalsoaffect
emitters. The doublepeaked line profilesmay be caused by an thedouble-peakedlineprofiles.
Sendoffprintrequeststo:L.Cˇ.Popovic´, Long-term variability in the line/continuumflux as well as
e-mail:[email protected] in the line profiles is observed in objects with broad double
peaked lines (see e.g. Dietrichetal., 1988; Shapovalovaetal.,
1
L.Cˇ.Popovic´etal.:Spectralmonitoringof3C390.3
2001; Sergeevetal., 2002, 2010; Shapovalovaetal., 2010; VanGroningen&Wanders (1992) (see also Shapovalovaetal.,
Lewisetal., 2010). Long-term variability of broad line pro- 2004)usingthespectraltemplatefornarrowcomponentsasref-
files is intriguing because it is usually unrelated to more rapid erencespectrum.Moredetailsaboutobservationscanbefound
changesinthecontinuumflux,butprobablyisrelatedtophysi- inPaperI.
calchangesintheaccretiondisk,ase.g.brightnessofsomepart
ofthedisk,orchangesinthedisksizeanddistancetothecentral
blackhole.Thedouble-peakedbroadlineprofilevariabilitycan 3. Analysisofthebroadlineprofiles
beexploitedtotestvariousmodelsfortheaccretiondisk(ase.g.
circular or elliptical). Moreover, the double-peaked broad line In Paper I we explained the continuum and narrow line sub-
studies can provide important information about the accretion traction in order to obtain only the broad Hα and Hβ line pro-
disk,ase.g.inclination,dimensionandemissivityofthediskas files, and here that procedure will not be repeated. Using only
well as probes of dynamical phenomenathat may occur in the broad line component we analyze the line parameters: month-
disk (see in more details Eracleousetal., 2009, and reference andyear-averagedlineprofiles(seeFigs.1–3,Figs.2–3-avail-
therein). ableelectronicallyonly),theposition(andseparation)ofpromi-
nentpeaks,thefluxesofthelinesegmentsandtheHα/Hβratios
In Paper I we have presented the results of the long-term
(or Balmer decrement – BD). In this section we first give the
(1995–2007) spectral monitoring of 3C 390.3. We have ana-
analysisoftheobservedbroadlineparameters,andafterthatwe
lyzed the light curves of the broad Hα and Hβ line fluxes and
analyzethecorrespondingmodeledlineparameter,obtainedus-
the continuum flux in the 13-year period. We also found that
ing an accretiondisk model,in orderto learn aboutchangesin
quasi-periodicaloscillations (QPO) may be presentin the con-
thediskstructurewhichmaycausethelineparametervariability.
tinuumandHβlightcurves.We studiedaveragedprofileofthe
Hα and Hβ line in two periods (Period I from 1995 to 2002, AsreportedinPaperIthereisadifferencesintheHαandHβ
and Period II from 2003 to 2007) and their characteristics (as lineprofilesbeforeandafterthebeginningoftheactivityphase
e.g.peaksseparationandtheirintensityratio,orFWHM).From in2002.Therefore,weconsiderseparatelylineprofilesobtained
thecross-correlations(ICCFandZCCF)betweenthecontinuum intheperiodbeforeMarch05,2002(PeriodI,JD2452339.01
fluxandHβandHαlineswefoundthelagof∼95daysforHβ according the minimum in Hβ) and after that date (Period II,
and ∼120 days for Hα (see Paper I for details). We concluded seePaperI).Also,wefoundthatobservationsin2001and2002
thatthebroademissionregionhasdisk-likestructure,butthere have sometimes closer characteristics as the data in Period II,
couldprobablyexistsanadditionalcomponent,non-diskoralso thereforeinsomeplotswemarkedseparatelytheseobservations
disk-like, with different parameters that contributes to the line as2001–2002.
emission.WefoundadifferencesintheHαandHβlineprofiles
beforeandafterthebeginningoftheactivityphasein2002,con-
3.1.TheaveragedprofilesoftheHαandHβbroademission
sequentlywedividedourspectraintotwoperiods(beforeMarch
lines
05,2002–PeriodIandafterthat–PeriodII,seePaperI).
In this paper we study in more details the Hα and Hβ line It was also reported in Paper I that the line profiles of Hα and
profiles and ratios, taking into account the changes during the Hβ were changingduring the monitoringperiod. In Fig. 1, we
monitoring period. The aim of this paper is to investigate the show the month- and in Figs. 2 and 3 (available electronically
changesinthetheBLRstructureof3C390.3thatcausetheline only)year-averagedprofilesoftheHαandHβ.Asitcanbeseen
profilevariations.Toperformthisinvestigationweanalyzedthe inFig.1thelineprofilesofHα(left)andHβ(right)vary,show-
peakseparationvariations,variationsinlinesegmentsandvaria- ing clearly two, and sometimes three peaks. Similar variations
tioninBalmerdecrement.Usingarelativelysimplediskmodel, can be seen from the year-averagedprofiles (see Fig. 2 for Hα
wetrytoexplainqualitativelythechangesindiskstructurethat andFig.3forHβ).Noteherethatinbothperiodsthebluepeak
cancausethelineparametervariations. is higher as it is expected in the case of a relativistic accretion
The paper is organizedas follow: in §2 we describe of our disk.Themostinterestingisthecentralpeakatthezeroveloc-
observations; in §3 we present the analysis of the Hα and Hβ ity (clearlyseen in 1995and 1996)thatmay indicateperturba-
lineprofilesvariability,thepeak-velocityvariabilityandBalmer tionin thedisk oranadditionalemittingregion.We foundthat
decrement;in§4westudytheline-segmentvariations;in§5the thiscentralpeak(sometimesweaker)existsalmostalwaysdur-
Balmer decrement variation is analyzed; in §6 we discuss ob- ingbothperiods,thereforewemeasuredthepositionoftheblue,
tainedresults,andfinallyin§7weoutlineourconclusions. central and red peak. Our measurements are given in Table 1.
Sometimestheredpeakcouldnotbeproperlymeasured(since
it was too weak oreven absent),thusin thatcase the measure-
2. Observations mentsarenotgiveninTable1.Themeasurementsshowthatthe
position of the peaks are changing, and as can be seen in Fig.
In1995-2007spectraof3C390.3weretakenwiththe6mand 4changesarehigherintheHβthaninHαline.Itisinteresting
1mtelescopesoftheSAO RASandwithINAOE’s2.1mtele- thatthepositionofthebluepeakvariesaroundseveralhundreds
scopeofthe”GuillermoHaroObservatory”(GHO)atCananea, kms−1(e.g.inHαitis-3550±250kms−1,inHβ-3500±400km
Sonora, Me´xico and with a long slit spectrographs, equipped s−1),whilethepositionoftheredpeakhashighervariation(for
withCCDdetectorarrays.Thetypicalwavelengthintervalcov- Hα ∼+5000±1000kms−1,andforHβ ∼5250±1700kms−1).
ered was from 4000Å to 7500Å, the spectral resolution var- Thismayindicatethattheemissioniscomingrelativelycloseto
ied between (4.5-15) Å. Spectra were scaled using the [Oiii] thecentralblackhole.
λλ4959+5007 (for blue spectra) and the [OI]6300A (for red There is a difference between the Hα and Hβ red and blue
spectra) integrated line flux under the assumption that fluxes peakvelocities(seeFig.4),thatcanbeexpectedduetothestrat-
of this lines did not change during the time interval covered ificationofthedisk,i.e.thedifferentdimensionsofthediskre-
by our observations (1995–2007). The narrow components of gionthatemitsHα orHβ. Itis interestingthatthe centralpeak
Hα andHβwere removedbyapplyingthe modifiedmethodof seems to stay at similar position in both lines, i.e. as it can be
2
L.Cˇ.Popovic´etal.:Spectralmonitoringof3C390.3
seeninFig.4(paneltop)thepointsfollowwellthelinearbijec- radius)andfromsomedistance,duetotheprobabilityoftransi-
tion function (represented as dashed line). This could indicate tions,theremaybesignificantemissionintheHαlineandvery
that the central peak is connected with some kind of perturba- weak(orabsent)Hβemission.
tioninadiskoranadditionalregionthatinthesamewayaffects Also, we tested whatwill be the consequencewhenchang-
theHβandHαlineprofiles.Noteherethattheso-calledcentral ing disk emissivity in the modelsand foundthat it has smaller
peak was slightly shifted to the blue side (or close to the zero influenceonthedoublepeakedlineprofiles(seealsoBonetal.,
shift)in1996–1997(seeFig.1),butafterthatthispeakhasbeen 2009).
redshifted(between1500and2500kms−1). Toprobedifferenttypeofvariability,firstweconsiderthatin
In order to compare the variability of the blue peak posi- differenttime, differentpartof disk can contributeto the emis-
tion during a longer period, in Fig. 5 we presented our mea- sion of the broad Hα and Hβ lines, i.e. we keep constant the
surementsforthebluepeakpositionandmeasurementsobtained relativesize of the disk regionsresponsibleforthe emission of
byVeilleux&Zheng(1991),Gaskell(1996)andEracleousetal. Hα and Hβ lines but we change the locations of these regions
(1997).Noteherethattheestimatesofthebluepeakpositionin withrespecttothecentralblackhole.Thentheemittingregion
theHαlineismoreuncertainthaninHβ,duetothesubstraction canbecloser,i.e.innerandouterradiusclosertotheblackhole.
oftheatmosphereB-band(PaperI),whileHβismoreuncertain For obtaining such models, we have been shifting the emitting
intheredpartduetosubtractionofthe[OIII]lines(see§4.1).As region closer to the black hole for a step of ∆R = 50 Rg. This
itcanbeseeninFig.5themeasurementfollowwelltheprevious was performed by varying the inner and outer radius, keeping
observations,andthere is an increaseof the bluepeak position allotherparametersasconstant(exceptoftherandomvelocity
velocityfrom1970to1990(1995),whileafter1995,thetheblue thatwas changedaccordingly,butthis hasalso small influence
peakpositionvelocityisslightlydecreasing. ontheline profiles).InFig. 6we presentedsimulatedlinepro-
files. The range of the inner radius was R = 250− 600 R ,
inn g
where the broadestline in Fig. 6 correspondsto the R =250
inn
3.2.Simulationofthelineparametersvariations R . Consequently, the outer radius was in the range of R =
g out
1200−1550R forHαandR = 1000−1350R forHβ(see
g out g
Onecanfitthebroaddoublepeakedlineprofilestoextractsome
Fig.6).
disk parameters (see e.g. Eracleous&Halpern, 1994, 2004;
Fromthemodelswehavemeasuredthevelocitiesoftheblue
Flohic&Eracleous, 2008; Lewisetal., 2010; Jovanovic´etal.,
andredpeak(V andV ).Noteherethatwecouldnotobtain
blue red
2010, etc.),buthere,since wehavea largesetofobservational
the central peak since, as it was mentioned above, it probably
data,wewilluseasimplemodeltosimulatevariationsindiffer- hasadifferentorigin,eitheritoriginatesoutsideofthedisk,orit
entdiskparametersandgivesomequalitativeconclusions.
isduetosomediskperturbations.
Toqualitativelyexplainthechangesinthediskstructurethat Tocomparethemodeledvaluesandobservedones,weplot
can cause the line parameter variations, we simulate the disk in Fig. 7 the differencebetweenthe blueand read peakveloci-
emission, using a disc model given by Chenetal. (1989) and ties V −V as a functionof the blue peak velocity for Hα
red blue
Chen&Halpern(1989).Themodelassumesarelativistic,geo- (top) and Hβ line (bottom). The modeled values are presented
metricallythinandcirculardisk(seeinmoredetailsChenetal., asfullsquares.Fig.7showsthatthemodeledvaluesareingen-
1989;Chen&Halpern,1989).Noteherethatinthismodelrela- eral agreementwith the observedones, which howevershow a
tivisticeffectsareapproximativelyincludedandthelimitofthe considerablescatteringaroundthemodeledtrend.Itseemsthat
inner radiusis Rinn > 100 Rg (Rg = GM/c2 – gravitationalra- theemissionofthedifferentpartsofthediskindifferentepochs
dius),buttheestimatedinnerradiusinthecaseof3C390.3issig- can explain the peak position variations. In such scenario, the
nificantly larger (see Flohic&Eracleous, 2008), consequently, line profile variability can be explainedas following:when we
the model given by Chenetal. (1989) and Chen&Halpern have the inner radius closer to the central black hole, the sep-
(1989) can be properly used, i.e. it is not necessary to include aration between peaks stays larger and the blue peak is more
afullrelativisticcalculation(ase.g.inJovanovic´etal.,2010). blue-shifted,but when the inner radius is farther from the cen-
To obtain the double-peakedline profiles we generated the tral black hole, the distance between peaks is smaller and the
setofdifferentdisk-likelineprofiles.Forthestartingparameters blue peak is shifted to the center of the line. Note here that
oftheHαlineweusedtheresultsofFlohic&Eracleous(2008). Shapovalovaetal.(2001)foundtheanticorrelationbetweenthe
Theyobtainedthefollowingdiskparametersfromfittingtheob- continuumfluxandV −V intheperiod1995-1999.
red blue
servedHαlineof3C390.3:theinnerradiusR = 450R ,the
inn g
outerradiusR = 1400R ,thediskinclinationi = 27deg,the
out g 4. TheHαandHβline-segmentsvariation
random velocity in the disk σ = 1300 kms−1, with the emis-
sivity r−q, where q = 3 is assumed. To model the parameters To see if there are any changes in the structure of the disk or
for the Hβ line, we took into account the results from Paper I disk like-region,we investigatedthe light curvesfor the differ-
thatthedimensionoftheHαemittingregionis∼120lightdays, entsegmentsof the Hα andHβ broademission lines. First, we
and of Hβ is ∼ 95 light days, consequently we proportionally dividedthelineprofilesalongthevelocityscaleinto9segments.
(95/120) rescaled the Hβ disk parameters for outer radius (i.e. The size of intervals are defined as 2000 kms−1 for segments
R = 1200R ) keepingotherparametersas in the case of the from0to±3and3000kms−1forsegment±4(thesegmentsand
out g
Hα disk. Note here that we assumed that the inner radius for correspondingintervalsaregiveninTable2).
both, Hα and Hβ is the same. It is an approximation, but one The observational uncertainties were determined for each
mayexpectthattheplasmaconditions(temperatureanddensity) segmentoftheHαandHβlightcurves.Inevaluatingtheuncer-
areprobablychangingfastintheinnerpartofdisk,thereforeat tainties,weaccountforerrorsduetotheeffectofthesubtraction
some distance fromthe black hole,the plasma hassuch condi- of the template spectrum (or the narrow components and con-
tionsthatcanemitBalmerlines(i.e.recombinationstayseffec- tinuum).We comparefluxesof pairs of spectra obtainedin the
tive).Ontheotherside,atlargerdistancesfromtheblackhole, timeintervalfrom0to2days.InTable3(availableelectronically
the conditionsin plasma are slowly changing(as a function of only),wepresenttheyear-averageduncertainties(inpercent)for
3
L.Cˇ.Popovic´etal.:Spectralmonitoringof3C390.3
Table1.Measurementsofthepeakpositionvelocities(blue,central,red)frommonth-averagedHαandHβprofiles.
Hα Hβ
Year Month blue central red V −V Month blue central red V −V
red blue red blue
km/s km/s km/s km/s km/s km/s km/s km/s
1996 Mar -3520±39 96±115 3840±35 7360±74 Feb-Mar -3269±136 -216±131 3773±16 7042±152
1996 Jul -3411±8 -457±100 4056±35 7467±43 Jun-Aug -3123±177 -595±7 3422±346 6546±523
1998 Feb -3390±100 2639±71 - - May-Jun -2977±112 2575±718 - -
1998 Jul-Sep -3378±7 3104±35 - - Jul -3226±74 3057±202 - -
1998 Dec -3433±22 2726±71 - - Oct-Dec -3591±177 2283±387 6359±162 9950±340
1999 Sep -3465±116 1914±56 5896±180 9361±296 Aug-Oct -3708±94 1830±201 6783±24 10491±118
2000 Jul -3747±23 2649±117 5063±133 8810±156 Jul-Nov -3854±30 1713±35 5599±87 9453±116
2001 Jan -3660±24 1665±21 4803±11 8463±35 Jan-Mar -3942±12 1874±56 5453±128 9395±140
2001 May-Jun -3714±53 2682±112 5182±88 8896±141 May-Jun -3693±8 2429±98 5673±107 9366±115
2001 Oct -3693±22 2390±66 4803±72 8496±94 Oct-Nov -3357±12 2116±151 - -
2002 Feb-Mar -3389±100 1881±41 4750±26 8139±126 Feb-Apr -2977±29 2298±36 - -
2002 Jun-Jul -3313±84 2098±21 4662±96 7975±181 Jun -3138±198 2137±67 - -
2002 Oct-Nov -3443±99 1892±36 4619±35 8062±134 Oct-Dec -3372±33 2020±15 5176±17 8548±50
2003 May -3649±23 1899±95 4695±19 8344±42 May-Jun -3810±8 2049±139 5001±100 8811±108
2003 Sep-Oct -3649±23 1773±71 4652±19 8301±42 Sep-Oct -3591±94 1727±97 4358±16 7948±111
2004 Mar-Aug -3584±8 2487±82 4500±11 8085±19 Mar-Jun -3489±50 2575±25 4357±99 7846±149
2004 Dec -3433±39 2833±70 4740±109 8173±147 Dec -3007±12 2678±5 4767±182 7773±194
2005 Apr -3563±22 2801±35 4533±35 8095±57 Apr-Jun -3170±12 2502±6 - -
2007 Jan -3682±7 1643±113 - - Jan-Feb -3591±12 1961±68 - -
2007 Feb -3703±24 1329±37 - - May-Jun -3708±12 1800±6 - -
2007 Aug-Nov -3800±8 1470±297 - - Aug-Nov -3781±32 1713±130 - -
eachsegmentofHαandHβandthe correspondingmeanyear- red wing with significantcorrelation(r = 0.84, P =0.35E-15).
0
flux. We adopted the years for each line where frequency and Inothersegments,onecanseethatthechangeinthebluewing
numberofobservationswereenoughtoestimateerror-bars.The inPeriodIwassmallerthanintheredone(Fig.9).Itisinterest-
meanvaluesofuncertaintiesforallsegmentsarealsogiven.As ing that in the Period I, the flux of the far blue wing (-4)stays
onecanseefromTable3,forthefarwings(segments±4)theer- nearlyconstant,whilethefluxoftheredpart(+4)isincreasing.
rorbarsaregreater(∼30%-50%)inHβthaninHα(∼20%).But EvenifweconsiderthePeriodIwithouttheperiod2001-2002,
when comparingthe errorbarsin the far red and blue wings of thecorrelationisweakandinsignificant(r=0.03,P =0.84)and
0
eachlines,wefindthattheerrorbarsaresimilar.Largererrorbars infavorofthepreviousstatement.
canalsobeseeninthecentralpartoftheHαduetothenarrow To see whether there is a similar response of the line seg-
linesubtraction. mentsto the continuumfluxwe constructplotsin Figs. 10 and
Ourmeasurementsofline-segmentfluxesforHαandHβare 11 (Fig. 11 available electronicallyonly), where the flux of all
given in Table 4 and 5 (available electronically only), respec- segmentsagainstthecontinuumfluxarepresented.Asitcanbe
tively.ThelightcurvesforeachsegmentoftheHαandHβlines seeninFigs.10and11thereisarelativelygoodresponseofthe
arepresentedinFig.8(availableelectronicallyonly),wherewe linesegmentstothecontinuumflux(seeplotsforthevaluesof
showbluepart(crosses),redpart(pluses),core(fullcircles),and the Pearsoncorrelationcoefficientr and the nullhypothesisP
0
continuum(solidline)inarbitraryunitsforcomparison.Ascan forbothperiods).However,thelinesegments(±4)(especiallyin
beseeninFig.8,thefluxesfrom0to±4segmentsofbothlines thecaseoftheHβline(Fig.10,top))inthePeriodIhaveaweak
have similar behavior during the monitoring period. As a rule, response (or even absent response, in the case segment +4) to
thebluesegmentsarebrighterorequaltotheredones.Onlyin thecontinuumflux.Forexample,thefluxinthefarredsegment
±4intheperiodof1997–1999theredsegmentsarebrighterthan (Hβ+4)increasesaround4-5timeswithouttheincreaseofthe
theblueones.ItisinterestingthattheredandbluewingsofHβ continuumflux.
in 1995–1997where with small difference in brightness, while NoteherethatintheHβsegment+4onecanexpecthigher
in2006thebluepeakwasaround30%brighterthantheredone.
errorsduetothesubtractionofthebright[OIII]lines.Asacon-
ThevariationinlinesegmentsoftheHαandHβaresimilar, sistency check, we plot in Fig. 10 (bottom) Hα (±4) segments
andtherearemoreobservationsofHβthanofHα,furtherinthe fluxesvs.continuumflux.Asitcanbeseen,thetrendissimilar
analysiswe will consideronlythe segmentvariationof the Hβ as in Hβ, and also in Period I Hα segment (±4) fluxes do not
line,exceptinthecaseofsegments±4,whereweconsiderboth respondtothecontinuumflux.Thismayindicatethatthereare
HαandHβ. periodswhensomeprocessesinthediskthatarenotconnected
withthecentralcontinuumsource.
As mentioned above, there is a central peak that is proba-
4.1.Linesegmentvs.linesegmentandcontinuumvariations
blycomingfromanadditionalemissionregionoriscreateddue
We study the response of symmetrical line segments (-4 vs. 4, to some perturbation in the disk. Therefore we plot in Fig. 12
etc.), and as it shown in Fig. 9, there is only a weak response theline-segmentfluxesvs.thefluxofthecentralline-segment0
fromthefarbluetothefarredsegmentinPeriodI.Weplotthe (measuredbetween-1000and+1000kms−1).Asitcanbeseen
bestlinearfitsforPeriodIandPeriodII,andfindthatthecorre- thereisagoodcorrelationbetweenthecentralline-segmentflux
lationbetweenthefarwings(segment-4vs.segment+4)hasa andfluxesofothersegments,onlyonecanconcludethatthere-
negativetendencywithlowsignificance(thePearsoncorrelation sponses was better in the Period II. Thus, it seems that in this
coefficientr = −0.29,the nullhypothesisvalue P =0.13E-01), period both the disk-line and additional component effectively
0
butin the Period II there is a linear response of the blue to the respond to the continuum variation. Also, one can see that the
4
L.Cˇ.Popovic´etal.:Spectralmonitoringof3C390.3
3.5
II Ip peerrioiodd rI=-0.29 (P0=0.13E-01) 4.5 II Ip peerrioiodd
3 Fit I period rII=0.84 (P0=0.35E-15) 4 Fit I period
Fit II period Fit II period
2001-2002
3.5
2.5
4 3 3
β - 2 β -
H H 2.5
of of
ux 1.5 ux 2
Fl Fl
1.5
1
1
0.5
0.5 rI=0.83 (P0=0.20E-18)
rII=0.91 (P0=0.94E-21)
0 0
0 0.5 1 1.5 2 2.5 3 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
Flux of Hβ 4 Flux of Hβ 3
8 8
I period I period
II period II period
7 Fit I period 7 Fit I period
Fit II period Fit II period
6 6
2 1
β - 5 β - 5
H H
of of
ux 4 ux 4
Fl Fl
3 3
2 rI=0.87 (P0=0.15E-22) 2 rI=0.90 (P0=0.84E-26)
rII=0.97 (P0=0.88E-33) rII=0.97 (P0=0.11E-33)
1 1
1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 1 2 3 4 5 6
Flux of Hβ 2 Flux of Hβ 1
Fig.9.Thelinesegmentvs.linesegmentresponse.TheobservationsinPeriodIaredenotedwithfullcirclesandinPeriodIIwith
opensquares.Theline-segmentfluxisinunits10−14ergm−2s−1.ThePearsoncorrelationcoefficientrandthenullhypothesisvalue
P aregivenoneachplotforbothperiods.
0
Table2.Thebeginningandendingradialvelocities,V andV ,fordifferentsegmentsinthelineprofilesofHαandHβ.
beg end
P
P
Psegment
PP -4 -3 -2 -1 0(C) +1 +2 +3 +4
Vr P
V -10000 -7000 -5000 -3000 -1000 1000 3000 5000 7000
beg
V -7000 -5000 -3000 -1000 1000 3000 5000 7000 10000
end
fluxoftheredpart(segment+4)inPeriodIhasaweakresponse present.ThemodeledF vs.F relationship,marked
bluewing redwing
to the central line-segment 0 (r = 0.58, P =0.76E-07), that is with full triangles, appears to follow the trend of the observed
0
similarasinthecaseofthesegmenttothecontinuumfluxcor- pointsin Period I but is shifted below the observedcorrelation
relation(Fig.10).Thismayhavethesamephysicalreason,that pattern.Tofindthecausesforsuchdisagreement,wemeasured
somekindofperturbationindiskwerepresentinPeriodI. the shift of the Hβ averaged profiles for two periods (given in
Paper I) with respect to the averaged profile constructed from
the modeledprofiles(fordifferentpositionsof the Hβ emitting
4.2.ModelingofthelinesegmentvariationoftheHβline disk).WemeasuredtheshiftatthecenterofthewidthoftheHβ
line at half maximum and at 20% of the maximum. We found
Weusedthesamemodelsasdescribedin§3.2,butnowwehave that the shift of the modeledaverage profile is +900 km s−1 at
measuredtheline-segmentsfluxesforthemodeledlines,defined 20%ofthemaximumand+600kms−1athalfmaximum,while
inthesamewayasfortheobservedlines.HereweplotinFig.13 themeasurementsforPeriodIare∼+700kms−1 at20%ofthe
thesumof-4and-3segmentfluxesasafunctionofthesumof maximumand∼+110kms−1 athalfmaximum.InPeriodIIthe
+4and+3segmentfluxesnormalizedonthecentral0segment, observedaveragedHβlinewasmoreblue-shiftedwithrespectto
i.e. Fbluewing = [F(−4)+F(−3)]/F(0)vs. Fredwing = [F(+4)+ the modeledone.We found∼+310kms−1 at20%ofthe max-
F(+3)]/F(0). imumand∼ -330kms−1 athalfmaximum.Thereforewe sim-
ObservationsinPeriodIaredenotedwithfullcirclesandin ulated the segment variation (Fig. 13) taking into account that
PeriodIIwith opensquares(see Fig.13). Also we see thatob- the whole disk line is blue-shiftedby 300km s−1 (opencircles
servations in 2001 and 2002 are closer to the Period II, so we in Fig. 13), and 850 km s−1 (diamonds in Fig. 13). As can be
denoted these with open triangles. It is interesting that in the seeninFig.13themodeledvalueswellfittheobservationsfrom
firstperiodagoodcorrelationbetweenF andF is Period I (line blue-shifted for 300 km s−1) and Period II (line
bluewing redwing
5
L.Cˇ.Popovic´etal.:Spectralmonitoringof3C390.3
I period I period
3 II period rI=0.19 (P0=0.11E+00) 3 II period rI=0.37 (P0=0.15E-02)
-2-2m s] 2.5 2001-2002 rII=0.77 (P0=0.38E-11) -2-2m s] 2.5 2001-2002 rII=0.79 (P0=0.89E-12)
-14βH -4 [10 erg c 1 .125 -14βH 4 [10 erg c 1 .125
Flux of 0.5 Flux of 0.5
0 0
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5
F [10-15 erg cm-2 s-2 A-1] F [10-15 erg cm-2 s-2 A-1]
cnt cnt
10 12
I period I period
r=0.48 (P=0.12E-01)
9 II period II period I 0
-2-2m s] 78 2001-2002 -2-2m s] 10 2001-2002 rII=0.80 (P0=0.94E-05)
-14αFlux of H -4 [10 erg c 23456 rI=-0.09 (P0=0.66E+00) -14αFlux of H 4 [10 erg c 2468
1 rII=0.76 (P0=0.43E-04)
0 0
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5
F [10-15 erg cm-2 s-2 A-1] F [10-15 erg cm-2 s-2 A-1]
cnt cnt
Fig.10. The far blue (left) and far red (right) wings flux against the continuum-fluxvariation for Hβ (top) and Hα (down). The
notationisthesameasinFig.9.ThePearsoncorrelationcoefficientr andthenullhypothesisvalue P aregivenoneachplotfor
0
bothperiods.
3 5 8 6.5
I period I period I period I period
2.5 II period 4.5 II period 7 II period 5. 56 II period
4
6 5
βFlux of H -4 1 .125 βFlux of H -3 23 ..2355 βFlux of H -2 45 βFlux of H -1 34.. 5534
3 2.5
1.5
0 .05 rrII=I=-00..7081 ((PP00==00..8985EE-+1020)) 0 .15 rrII=I=00..4794 ((PP00==00..1120EE--0049)) 12 rrII=I=00..8746 ((PP00==00..2160EE--1190)) 1. 512 rrII=I=00..9841 ((PP00==00..4362EE--3143))
1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5
Flux of Hβ 0 Flux of Hβ 0 Flux of Hβ 0 Flux of Hβ 0
6.5 5.5 4 3
I period I period I period I period
5 .65 II period 5 II period 3.5 II period 2.5 II period
4.5
5 3
βFlux of H 1 34 ..3455 βFlux of H 2 23 ..3455 βFlux of H 3 2. 52 βFlux of H 4 1. 512
2.5 1.5
2
1 .25 rrII=I=00..9850 ((PP00==00..5878EE--3153)) 1.5 rrII=I=00..9835 ((PP00==00..4903EE--3126)) 1 rrII=I=00..7845 ((PP00==00..6265EE--1135)) 0.5 rrII=I=00..5880 ((PP00==00..7261EE--0172))
1 1 0.5 0
1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5
Flux of Hβ 0 Flux of Hβ 0 Flux of Hβ 0 Flux of Hβ 0
Fig.12. The response of the line segments to the central segment 0 (from -1000 to 1000 km s−1). The observations in Period I
are denotedwith fullcircles and in Period II with opensquares. The line-segmentflux is in units 10−14ergm−2s−1. The Pearson
correlationcoefficientrandthenullhypothesisvalueP aregivenoneachplotforbothperiods.
0
blue-shifted for 800 km s−1). It is interesting that in Period I in thefirst period(excluding2001–2002)thevariabilitycanbe
therearechangesinthelocationofdiskregionsresponsiblefor well explainedby variationof the disk positionwith respectto
thelinesemission(fromR =250R toR =550R ),while theblackhole,whileinthePeriodIIand2001–2002(whenout-
inn g inn g
inPeriodII,theinnerdiskradiusappearstohavebeenchanged burstisstarting),probablythediskpositiondoesnotchangeso
byasmalleramount(from350R to450R ).Thisindicatesthat
g g
6
L.Cˇ.Popovic´etal.:Spectralmonitoringof3C390.3
much.Itseemsthatinthiscasewehavethatthepartofthedisk andredwings(around±9000kms−1)aswellastwomaximain
emittingthebroadlineshasonlybecomebrighter. redandbluepart(around±6000kms−1).Thereisalsoacentral
minimum(around0kms−1)thatin2001and2002ismovingto
+2000kms−1, whilein 2005and2006ismovingto-2000km
4.3.2-DCCFoftheHβlineprofiles s−1.Itisinterestingthatfrom2007totheendofthemonitoring
Wenowinvestigateindetailthecrosscorrelationfunction(CCF) periodonlyapeakinthebluepart(between-6000to-4000km
of the line-profile variations of Hβ. We proceed in the same s−1) is dominant,while BD from-2000to +6000km s−1 tends
way as it was studied the line-profile variations in Mrk 110 tobeconstanthavingvaluesbetween3.5and4.5.
(Kollatschny&Bischoff,2002;Kollatschny,2003).
Fig.14showsthemaximumresponseofthecorrelationfunc-
5.2.ModelingoftheBalmerdecrement
tions of the Hβ line segment light curves with the continuum
light curve and Fig. 15 shows the delays of the individual Hβ It is known that the BD depends on physical processes. But
line segmentsin velocity space. Fig. 15 givesthe delaysof the BD asfunctionofthevelocitiesalongthe lineprofilemayalso
individuallinesegmentsofHβseparatelyforthetwoobserving depend on the geometry of the Hα and Hβ emitting region, as
periods: from 1995–2002(filled squares) and from 2003–2007 e.g. if the line emission regionof Hα and Hβ are notthe same
(filledtriangles). onemayexpectthattheHαandHβfluxratiowillbeafunction
The outer blue and red line wings show the shortest de- ofthegasvelocity.Therefore,wemodeledtheshapeoftheBD
lay clearly indicatingthat the line emitting regionis connected vs.velocity.
withanaccretiondisk(e.g.Welsh&Horne,1991;Horneetal., Inorderto test the influenceof the geometryto the BD vs.
2004). A carefulinspection of the delaysof the individualline velocityalongtheline,weusedmodeleddiskprofiles(see§3.2),
segments in Fig. 15 indicates that a second trend is superim- butnowweconsidertwocases:a)thediskregionsemittingHα
posed:Anearlierresponseoftheredlinewingcomparedtothe and Hβ have different sizes but the same inner radii that vary
blue line wing is seen in Fig. 15 for the second half of the ob- accordingly(asdiscussedin§3.2);b)theinnerradiusoftheHα
serving period. This behavior is consistent with the prediction emittingregiondifferfromthefixedradiusoftheHβregionby
fromdisk-windmodels(Ko¨nigl&Kartje,1994). ±50R ,±100R .
g g
Itisintriguingthatthevery-broadlineAGN3C390.3shows OursimulationsoftheBDvs.velocityaregiveninFig.18,
the same pattern in the velocity-delay maps as the narrow-line whereitcanbeseenthatthediskmodel(assumingdifferentdi-
AGNMrk110(seeKollatschny&Bischoff,2002;Kollatschny, mensions of the emission disk that emits Hα and Hβ) can re-
2003).Theouterredandbluewingsrespondmuchfastertocon- produceverysimilarBDprofilesalongvelocityfield,especially
tinuumvariationsthanthecentralregions. as they are observedin Period I. Thereis a differencein shape
thatisprobablycausedbythephysicalconditionsinthediskas
well as due to the centralcomponentemission. We foundbell-
5. BalmerDecrement(BD)variation like profiles of BD, and they also have two peaks (in red and
bluepart),aswellasasmallminimuminthecenter.Theshape
The ratio of Hα and Hβ depends on the physical processes in
of the BD vs. velocity observed in Period II, where a peak in
the BLR and variations in the Balmer decrement (BD) during
thebluepartisprominent(seeFig.17),andinsomecaseshasa
monitoringtime can indicate changes in physical propertiesof
deeperredminimumthanblueone,cannotbeobtainedinmod-
the BLR. Using the month-averaged profiles of the broad Hα
eled BD vs. velocity. This may be due to the influence of the
andHβlines,wedeterminedtheirintegratedfluxesintherange
centralcomponent,butalso,duetotheoutburstinPeriodII,or
between -10000 and +10000 kms−1 in radial velocity, i.e. the
duetodifferentphysicalprocessesinthedisk.
integratedfluxratioF(Hα)/F(Hβ)orintegratedBD.InFig.16
AsitcanbeseeninFig.18(bottompanel,opensquares),a
we plot the time variation of the continuum flux, Hα and Hβ
big difference between modeled BD vs. velocity and observed
line fluxes and the BD. The BD reached its maximum in 2002
oneisinthecase wheretheHαemittingregionisclosertothe
when the fluxes of lines and continuumwere in the minimum.
blackholefor100R .Thisindicatesthatthediskpartemitting
AsonecanseefromFig.16thereisnocorrelationbetweenBD g
Hαcannottobesignificantlyclosertothecentralblackholethan
andline/continuumvariation.Only,asitcanbeseeninthevery
theonethatemitstheHβline.
bottompanelinFig.16,whichshowsseparateBDsfortwoperi-
ods(fullcirclesrepresentPeriodI,andopensquaresPeriodII),
thereissometendencythattheBDishigherforweakercontin-
6. Discussion
uumflux.WhenBDisplottedversusthecontinuumflux(Fig.16
bottom panel) a negativetrend seems to be present below 1.75 In this paper we performed detailed study of the line parame-
×10−15erg cm−2s−1Å−1. For higher values of the flux BD stays ter variations (peak separation, variation of the line segments,
roughlyconstant.NotethatasimilarBDbehaviorwasobserved Balmer decrement)during a 13-year long period. As we noted
bySergeevetal.(2010). above,thelineprofilesofthe3C390.3inthemonitoringperiod
have a disk-like shape, i.e. there are mainly two peaks, where
theblueoneismoreenhanced.Insomeperiods,thereisacentral
5.1.BDasfunctionofthevelocityalonglineprofile
peak,thatprobablyiscomingfromanadditionalemissionregion
Furthermore,weinvestigateBDsforeachlinesegment(asitwas or extra emission caused by some perturbationin the disk (see
describedin§4),andplotsofBDagainstvelocityarepresentin e.g. Jovanovic´etal., 2010). In order to confirm the disk emis-
Fig.17(noteherethatapartofthisFigureisavailableelectron- sionhypothesiswemodeledthepeakpositionvariationusinga
icallyonly).Fig.17showssomecommonfeaturesinthetrends semi-relativisticmodel,andfoundthatpeakpositionvariations
ofthedifferentlinesegmentsBDplottedvs.velocityoveranex- canbeexplainedwiththediskmodel.
tendedperiodoftimerangingfrom1995to2007.Forexample, We also investigated line-segment flux variation and found
for almost all observations there is a minimum in the far blue thattheratiosofthelinesegmentsarewellfittedbyadiskmodel.
7
L.Cˇ.Popovic´etal.:Spectralmonitoringof3C390.3
8 7
7 6
a/Hb) 56 Ha/Hb) 5
D(H 4 BD( 4
B
3 Year=1996.25 3 Year=1997.58
MJD=50143 MJD=50265
2 2
-10000-8000-6000-4000-2000 0 20004000 6000800010000 -10000-8000-6000-4000-2000 0 20004000 6000800010000
velocity(km/s) velocity(km/s)
6 5
5 4
b) b)
H 4 H 3
a/ a/
H H
D( 3 D( 2
B B
2 Year=1997.99 1 Year=1998.17
MJD=50783 MJD=50845
1 0
-10000-8000-6000-4000-2000 0 20004000 6000800010000 -10000-8000-6000-4000-2000 0 20004000 6000800010000
velocity (km/s) velocity(km/s)
8 7
7 6
b) 6 b)
H H 5
Ha/ 5 Ha/
D( D( 4
B 4 B
3 Year=1998.58 3 Year=1998.71
MJD=50995 MJD=51056
2 2
-10000-8000-6000-4000-2000 0 20004000 6000800010000 -10000-8000-6000-4000-2000 0 20004000 6000800010000
velocity(km/s) velocity(km/s)
8 7
7
6
b) 6 b)
H H 5
a/ 5 a/
H H
D( 4 D( 4
B B
3 Year=2007.17 3 Year=2007.42
2 MJD=54132 MJD=54221
2
-10000-8000-6000-4000-2000 0 2000 4000 6000 800010000 -10000-8000-6000-4000-2000 0 2000 4000 6000 800010000
velocity(km/s) velocity(km/s)
6 6
5
b) 5 Hb)
D(Ha/H 4 BD(Ha/ 34
B
3
Year=2007.67 2 Year=2007.75
MJD=54313 MJD=54344
2 1
-10000-8000-6000-4000-2000 0 2000 4000 6000 800010000 -10000-8000-6000-4000-2000 0 2000 4000 6000 800010000
velocity(km/s) velocity(km/s)
5 6
4 5
b) b)
H H
Ha/ 3 Ha/ 4
D( D(
B B
2 3
Year=2007.83 Year=2007.92
MJD=54374 MJD=54405
1 2
-10000-8000-6000-4000-2000 0 2000 4000 6000 800010000 -10000-8000-6000-4000-2000 0 2000 4000 6000 800010000
velocity(km/s) velocity(km/s)
Fig.17. Variations in the Balmer decrement(BD=Hα/Hβ) as a function of the radial velocity for month-averagedspectra of 3C
390.3.UpperpanelrepresentsBDfromtheperiodof1995–1998,andbottompanelfrom2007.Thesetwopanelsarechosensince
theyrepresentthecharacteristicbehaviorofBDvs.velocity.Theabscissa(OX)indicatestheradialvelocitiesrelativetothenarrow
components.Thepositionoftheblue,centralandredpeaksaremarkedwithblue,blackandpinkline,respectivelyforHα(solid
line)andHβ(dashedline).Therestofthepanelsareavailableelectronicallyonly.
8
L.Cˇ.Popovic´etal.:Spectralmonitoringof3C390.3
Itisinterestingthattheresponseoftheline-segmentfluxtoline- ellipticaldisk(Flohic&Eracleous,2008)orperturbationsinthe
segmentfluxisbetterinPeriodIIthaninPeriodI.Thesameisin disk(Jovanovic´etal.,2010).
thecaseoftheresponseofthesegmentfluxestothecontinuum Takingintoaccountallmentionedabove,onecanspeculate
flux. It seems that in Period I, the variability in the line profile aboutthe BLR structure and the nature of these variations.Let
was not caused only by an outburstin the ionizing continuum, usconsiderthat,inprinciple,thereisanaccretingmaterialthat
but also by some perturbations in the accretion disk (see e.g. is optically thick, and the lightfrom a centralsource is able to
Jovanovic´etal., 2010). While in Period II, it could be that the photoionize only a small thin region above (below) the accret-
disk structure is not changing much and that the variability in ingmaterial(orthickdisk,seeFig.19),letuscallthisregiona
the line parameters is primarily caused by the outburst in the “disk-likebroadline region”(disk-likeBLR1). Emittinggasin
continuum. thisregionhasdisk-likemotion,sinceitislocatedinthediskpe-
TheBalmerdecrementasafunctionofthecontinuumhasa ripheryandfollowsthediskkinematics.Thebrightnessofbroad
decreasingtrendinPeriodI(withtheincreaseofthecontinuum) lines which are coming from this region will be caused by the
and is constant in Period II. Also, there is a different shape of centralionizingcontinuumsource,but the line parameterswill
theBDvs.velocityindifferentperiods.Thecharacteristicshape dependon the dimensionsof the regionas well as the location
of the BD vs. velocity has two peaks (blue and red) which do of the BLR1 with respect to the central black hole. Of course,
notcorrespondto the line peaks(they are in the far wings, see other parameters as e.g. emissivity or local velocity dispersion
Fig.17),thereareminimainthefarblueandredwing,aswell canslightlyaffectthelineprofiles,butitseemsthatthechange
as a minimum in the center. The shape of the BD vs. velocity intheinner/outerradiusisthemostimportant.
is similar to what we expectfrom an accretion disk, where the
The shape of the broad lines confirm the disk-like geome-
dimensions of the disk for Hα and Hβ are different. In the pe-
try,andcorrelationbetweenthebroad-lineandcontinuumfluxes
riod(PeriodII)ofthehighestbrightenersinthecontinuumand
confirmstheinfluenceofthecentralsourcetothelineintensity
inlines,theBDvs.velocityshapesshowabluepeakandmore-
(seePaperI).TheopticalcontinuumandtheHβfluxvariations
less flat red part, that is probably caused by some other emis-
areprobablyrelatedtochangesintheX-rayemissionmodulated
sion, additional to the disk one, and also by different physical
byvariableaccretionrate,i.e.thechangeofthesurfacetempera-
conditionsinthesetwosub-regions.Ourmodeledshapesofthe
tureofthediskasaresultofthevariableX-rayirradiationUlrich
BD vs. velocity show that the case of the disk emission where
(2000). But question is, what can cause the changes in the di-
the dimensions of the Hα and Hβ emitting disks are different,
mension(aswellinposition)ofthebroadlineemittingregion?
can explain bell-like shapes of the BD vs. velocity. Note here
It is obvious that it is not the continuum flux, since the varia-
thatwedidnotconsiderphysicalprocessesintheemittingdisk
tioninthecontinuumdoesnotcorrelatewiththelineparameter
whichcanaffecttheshapeoftheBDvs.velocity.Comparingthe
variation.
dimensionsofthedisk,wecametothesameconclusionasfrom
The variation in the line parameters may be related to the
the CCF (see Paper I), that the Hα emitting disk is larger than
perturbationsin the accreting material that is feeding the disk-
theHβone,butalsothattheinnerradiusofbothdisksseemsto
like BLR with scattered gas. The density and optical depth of
besimilar.
suchgasfromtheaccretingmaterialmaycausethatindifferent
epochsthedisk-likeBLRhasdifferentdimensionsandpositions
6.1.VariabilityandtheBLRstructureof3C390.3 withrespecttothecentralblackholes.As,e.g.ifscatteredmate-
rial(gas)fromaccretingmaterialisabsentinsomeparts,itwill
Beforewe discuss theBLR structure,letusrecallsome results
affect the line parameters similar as change in the dimensions
thatweobtainedinPaperIandinthispaper:i)theCCFanalysis
(position)ofthedisk-likeBLR.
shows that the dimension of the BLR that emits Hβ is smaller
than one that emits Hα; ii) there is a possibility that there are On the other side, the observed broad double-peaked lines
quasi-periodicoscillations,whichmaybeconnectedwiththein- haveablue-shiftwithrespecttothemodeledone.Thismayindi-
stabilityinthediskordisk-likeemittingregion;iii)concerning catethatthereissomekindofwindinthedisk,probablycaused
thelineprofilevariations,thevariabilityobservedin13-yearpe- by the radiation of the central source. Recently Tombesietal.
riod can be divided into two periods, before and after the out- (2010) performed an uniform and systematic search for the
burstin2002;iv)therearethreepeaks,twowhichareexpected blueshifted Fe Kα absorption lines in the X-ray spectra of 3C
inthediskemission,andone-central,thatisprobablycoming 390.3observedwithSuzakuanddetectedabsorptionlinesaten-
fromdiskperturbationsorfromanadditionalemittingregion;v) ergiesgreaterthan7 keV. Thatimpliesthe originof the Fe Kα
the shapesof Hβand Hαobservedin the same periodare sim- is in the highly ionized gas outflowing with mildly relativistic
ilar,especiallyinPeriodI(noteherethatinPeriodII,thereare velocities(inthe velocityrangefrom0.04to 0.15lightspeed).
differences in the center and red wings); vi) our simulation of Taking into account that the optical line emission is probably
the variability, taking into account only different disk position originatingfartherawaythanthe X-ray,onecan expectsignifi-
with respectto the centralblack hole, can qualitativelyexplain cantlysmalleroutflowvelocitiesintheopticallines.
theobservedbroadlineparametervariations;vii)comparingthe One can expect that perturbations in the thick disk (or ac-
shiftsofthemodeledandobservedHβline,wefoundthatthere cretingmaterial)canaffecttheemittingdisk-likeregion.Itmay
isablue-shiftintheobservedHβcomparingtothemodeledone, cause some perturbation or spiral shocks in the disk-like BLR
thatisalsodifferentintwoperiods:around300kms−1inPeriod (see e.g. Chakrabarti&Wiita, 1994; Jovanovic´etal., 2010).
Iandaround850kms−1 inPeriodII;viii)theinspectionofthe Chakrabarti&Wiita (1994) found that the observed variability
delaysoftheindividuallinesegments(see§4.3)indicatespres- ofthedouble-peakedbroademissionlines(seeninsomeactive
enceofadiskandalsoawind,i.e.thedisk-windmodelmaybe galacticnuclei)canbeduetotheexistenceoftwo-armedspiral
presentintheBLRof3C390.3. shocksintheaccretiondisk.Usingthismodeltheysuccessfully
Additionally, note here that the disk-like structure of the fittedtheobservationsof3C390.3madeatdifferentepochswith
3C 390.3 BLR is favored in several recent papers (see e.g. self-similar spiral shock models which incorporate relativistic
Flohic&Eracleous, 2008; Jovanovic´etal., 2010) assuming an corrections.
9
L.Cˇ.Popovic´etal.:Spectralmonitoringof3C390.3
Alternatively,Jovanovic´etal.(2010)fitted theobservations tinuum influence on the line intensities (especially in the far
of 3C 390.3 from different epochs with a model that includes wings),therewastheadditionaleffectthatinfluencethelinepro-
moving perturbation across the disk. The perturbations/shocks files and intensities. Probably it was caused by some perturba-
in the disk model can explain different line profiles, and also tionsorshocksinthedisk-likeBLR(Chakrabarti&Wiita,1994;
the central peak (see e.g. Jovanovic´etal., 2010). But the cen- Jovanovic´etal.,2010)thatproducedchangesinthestructureof
tral peak might be emitted from a broad line region that does thedisk-likeBLR.
notfollowthediskgeometry,itmaybeverysimilartothetwo-
component model, i.e. that we have composite emission from
thedisk andanadditionalregion(seee.g.Popovic´etal., 2004;
7. Conclusions
Ilic´etal.,2006;Bonetal.,2009).Also,Arshakianetal.(2010)
considered that the BLR of 3C390.3 has a complex structure
In this paper, we present line profile variations of 3C 390.3 in
composedfromtwocomponents:1)BLR1–thetraditionalBLR
a long period. Due to the change of line profiles, we divided
(Accretion Disk), 2) BLR2 – a sub-region (outflow) that sur-
theobservationsintotwoperiods(beforeandaftertheminimum
rounds the compact radio jet. In this paper the existence of a
in 2002: Period I and II, respectively) and found difference in
jet-excited outflowing BLR is suggested, which may question
thelinesegmentsandBalmerdecrementvariationsinthesetwo
thegeneralassumptionofavirializedBLR.
periods. From our investigation, the main conclusions are the
Itisinterestingthatthecentralcomponentispresentduring
following:
the whole monitoringperiod,and that in 1995–1996it was lo-
i) the line profiles during the monitoringperiod are chang-
catedinthecenter,andafterthatshiftedtothetheredpartofline.
ing, always showing the disk-like profile, with the higher blue
Suchredshiftis notexpectedif thereis anoutflowin theBLR.
peak. There is also the central peak that may come from the
But let us recall the results obtained in the study of dynamics
emissionregionadditionaltothedisk,butasitwasmentionedin
oftheline-drivendisk-wind,wherekinematicsofthegasshows
Jovanovic´etal.(2010),itmayalsobecausedbytheperturbation
thatbothinfallandoutflowcanoccurindifferentregionsofthe
inthedisk.
windatthesametime(seeProgaetal.,1998,2000),whichde-
ii) thefar-wingsflux variationin the first period,wherethe
pendsonradiationforce.Also,recentlyGaskell(2009)reported
far-red wing flux does not respondwell to the continuumflux,
aboutthepossibilitythataninflowispresentintheBLRofAGN.
isprobablycausedbysomephysicalprocessesintheinnermost
Finally, one can consider a model as given in Fig. 19: the
part of the disk. The observed changes in Hα and Hβ may be
disk-likeBLR1followsthekinematicsoftheaccretingmaterial,
interpretedintheframeworkofadiskmodelwithchangesinthe
and from time to time perturbations can appear in this region
locationandsizeofthedisklineemittingregions.InPeriodII,
(similar as in the fully accretion disk). The emitting gas in the
the good correlation between the continuum and line-segment
BLR1 is coming (ejected) from the accretingmaterial and it is
flux suggests that the brightness of the disk is connected with
ionized by the centralsource. The radiation pressure may con-
the ionizing continuum,and that the structure of the disk does
tributeinsuchwaythatwehaveawind(anoutflow)inthisre-
notsignificantlychange.
gion. There may be also a region above the disk-like BLR1 (a
BLR2inFig.19),wheretheionizedgasmayhavemotionwith iii)Balmerdecrementisalsodifferentforthesetwoperiods.
respecttothecentralblackhole,anddependingontheradiation InthefirstperiodtheBDdecreaseswithcontinuumflux,while
pressurethevelocitiesaredifferentindifferentperiods. in the second period the BD stays more-less constant (around
4.5). The segment BD shows two maxima (around ±6000 km
s−1) which do not correspondto the red and blue peak, but in-
6.2.Differentnatureofvariationinbroadlinesof3C390.3 steadtheyarefartherintheblueandredwing(thanpeaksveloc-
It is interesting to discuss the difference between the variation ities). Also oneminimumaroundzero velocityis present. This
minimumchangedpositionbetween±2000kms−1 aroundzero
in PeriodI andPeriodII.Itis obviousthatthe disk-parameters
velocity.Thiscentralminimum(aswellasshiftedmaxima)may
variationcanexplaintheobservedline-parametervariations(as
becausedbytheadditional(totheaccretiondisk)emissionthat
e.g.inthepeakseparationandline-segmentfluxesvariations).In
perhapsispresentinthe3C390.3broadlines.Theseresultssug-
Period I, there were several quasi-periodicallow-intensity out-
gestthat,in additionto the physicalconditionsacrossdisk,the
bursts(seePaperI),anditseemsthattheline,andconsequently
sizeoftheHαandHβemittingregionsofthediskplaysanim-
thedisk-likegeometricalparametersarechanged.Itmaybethat
portantrole.Wemodeledbell-likeBDvs.velocityprofilesinthe
in this period the outbursts are caused by some perturbations
casewhentheHαdiskislargerthanHβone.
in the disk (in the inner part that emits the X-ray radiation as
wellasintheouterpartemittingthebroadlines).Thesepertur- iv)thevariationobservedinthelineparameterscanbewell
bationsaffectedthelineprofilesandprobablypartlylinefluxes modeledifoneassumeschangesinpositionoftheemittingdisk
(especially in the far wings). As it can be seen from modeled withrespecttothecentralblackhole.Theemissionofthedisk-
variationtheinner/outerradiusissignificantlychanged. like region is dominant, but there is the indication of the addi-
In the Period II, there is a strong outburst, that caused the tional emission. Therefore, to explain the complex line-profile
increaseofthebrightnessinthecontinuumandlines.Thelines variabilityoneshouldconsideracomplexmodelthatmayhave
stay brighter, but the structure (inner/outerradius) of the disk- adiskgeometrytogetherwithoutflows/inflows(seeFig.19).
likeregionisnotchangingmuch.Thereisadifferentresponseof An important conclusion of this work is that, even if the
thecontinuumtothelinesegmentsinPeriodIandPeriodII,i.e. disk-like geometryplaysa dominantrole,the variabilityof the
inPeriodIIthereisahighercorrelationsbetweenthecontinuum Hα and Hβ line profilesand intensities (and probablypartly in
andline-segmentfluxes. thecontinuumflux)hasdifferentnaturefordifferentperiods.It
Ontheotherhand,theBDinPeriodIhasadecreasingtrend seems that in Period I, the perturbation(s) in the disk caused
with the increase of the continuum flux, while in Period II, it (at least partly) the line and continuum amplification, while in
staysalmostconstant.Thismayindicatedifferentnatureofthe Period II the ionizing continuum caused the line amplification
variability. It seems that in Period I, beside the ionizing con- withoutbigchangesinthedisk-likestructure.
10