The Lick AGN Monitoring Project 2011: Reverberation Mapping of Markarian 50
A. J. Barth, A. Pancoast, S. J. Thorman, V. N. Bennert, D. J. Sand, W. Li, G. Canalizo, A. V. Filippenko, E. L. Gates, J. E. Greene, M. A. Malkan, D. Stern, T. Treu, J.-H. Woo, R. J. Assef, H.-J. Bae, B. J. Brewer, T. Buehler, S. B. Cenko, K. I. Clubb, M. C. Cooper, A. M. Diamond-Stanic, K. D. Hiner, S. F. Hoenig, M. D. Joner, M. T. Kandrashoff, C. D. Laney, M. S. Lazarova, A. M. Nierenberg, D. Park, J. M. Silverman, D. Son, A. Sonnenfeld, E. J. Tollerud, J. L. Walsh, R. Walters, R. L. da Silva, M. Fumagalli, M. D. Gregg, C. E. Harris, E. Y. Hsiao, J. Lee, L. Lopez, J. Rex, N. Suzuki, J. R. Trump, D. Tytler, G. Worseck, H. M. Yesuf
aa r X i v : . [ a s t r o - ph . C O ] O c t A CCEPTED FOR PUBLICATION IN A P J L
ETTERS
Preprint typeset using L A TEX style emulateapj v. 11/10/09
THE LICK AGN MONITORING PROJECT 2011: REVERBERATION MAPPING OF MARKARIAN 50 A ARON
J. B
ARTH , A NNA P ANCOAST , S HAWN
J. T
HORMAN , V ARDHA
N. B
ENNERT , D
AVID
J. S
AND , W
EIDONG L I , G ABRIELA C ANALIZO , A LEXEI
V. F
ILIPPENKO , E LINOR
L. G
ATES , J ENNY
E. G
REENE , M ATTHEW
A. M
ALKAN , D ANIEL S TERN ,T OMMASO T REU , J ONG -H AK W OO , R OBERTO
J. A
SSEF , H
YUN -J IN B AE , B RENDON
J. B
REWER , T ABITHA B UEHLER , S.B RADLEY C ENKO , K ELSEY
I. C
LUBB , M ICHAEL
C. C
OOPER , A
LEKSANDAR
M. D
IAMOND -S TANIC , K
YLE
D. H
INER ,S EBASTIAN
F. H
ÖNIG , M ICHAEL
D. J
ONER , M ICHAEL
T. K
ANDRASHOFF , C. D AVID L ANEY , M ARIANA
S. L
AZAROVA , A. M.N IERENBERG , D AWOO P ARK , J EFFREY
M. S
ILVERMAN , D
ONGHOON S ON , A LESSANDRO S ONNENFELD , E RIK
J. T
OLLERUD ,J ONELLE
L. W
ALSH , R
ICHARD W ALTERS , R OBERT L. DA S ILVA , M ICHELE F UMAGALLI , M ICHAEL
D. G
REGG , C HELSEA
E.H
ARRIS , E RIC
Y. H
SIAO , J EFFREY L EE , L ILIANA L OPEZ , J ACOB R EX , N AO S UZUKI , J ONATHAN
R. T
RUMP , D AVID T YTLER , G ÁBOR W ORSECK , AND H ASSEN
M. Y
ESUF Draft version October 8, 2018
ABSTRACTThe Lick AGN Monitoring Project 2011 observing campaign was carried out over the course of 11 weeks inSpring 2011. Here we present the first results from this program, a measurement of the broad-line reverberationlag in the Seyfert 1 galaxy Mrk 50. Combining our data with supplemental observations obtained prior to thestart of the main observing campaign, our dataset covers a total duration of 4.5 months. During this time, Mrk50 was highly variable, exhibiting a maximum variability amplitude of a factor of ∼ U -band continuumand a factor of ∼ β line. Using standard cross-correlation techniques, we find that H β and H γ lagthe V -band continuum by τ cen = 10 . + . - . and 8 . + . - . days, respectively, while the lag of He II λ β line exhibits a symmetric velocity-resolved reverberation signature with shorter lags in thehigh-velocity wings than in the line core, consistent with an origin in a broad-line region dominated by orbitalmotion rather than infall or outflow. Assuming a virial normalization factor of f = 5 .
25, the virial estimate ofthe black hole mass is (3 . ± . × M ⊙ . These observations demonstrate that Mrk 50 is among the mostpromising nearby active galaxies for detailed investigations of broad-line region structure and dynamics. Subject headings: galaxies: active — galaxies: individual (Mrk 50) — galaxies: nuclei Department of Physics and Astronomy, 4129 Frederick Reines Hall,University of California, Irvine, CA, 92697-4575, USA; [email protected] Department of Physics, University of California, Santa Barbara, CA93106, USA Physics Department, California Polytechnic State University, San LuisObispo, CA 93407, USA Las Cumbres Observatory Global Telescope Network, 6740 CortonaDrive, Suite 102, Santa Barbara, CA 93117, USA Department of Astronomy, University of California, Berkeley, CA94720-3411, USA Department of Physics and Astronomy, University of California,Riverside, CA 92521, USA Lick Observatory, P. O. Box 85, Mount Hamilton, CA 95140, USA Department of Astrophysical Sciences, Princeton University, Prince-ton, NJ 08544, USA Department of Physics and Astronomy, University of California, LosAngeles, CA 90095-1547, USA Jet Propulsion Laboratory, California Institute of Technology, 4800Oak Grove Boulevard, Pasadena, CA 91109, USA Astronomy Program, Department of Physics and Astronomy, SeoulNational University, Seoul 151-742, Republic of Korea NASA Postdoctoral Program Fellow Department of Astronomy and Center for Galaxy Evolution Research,Yonsei University, Seoul 120-749, Republic of Korea Department of Physics and Astronomy, N283 ESC, Brigham YoungUniversity, Provo, UT 84602-4360, USA Hubble Fellow Southern California Center for Galaxy Evolution Fellow Center for Astrophysics and Space Sciences, University of California,San Diego, CA 92093-0424, USA Physics Division, Lawrence Berkeley National Laboratory, 1 Cy-clotron Road, Berkeley, CA 94720, USA Department of Astronomy, The University of Texas at Austin, Austin,TX 78712, USA Caltech Optical Observatories, California Institute of Technology,Pasadena, CA 91125, USA Department of Astronomy and Astrophysics, UCO/Lick Observatory, University of California, 1156 High Street, Santa Cruz, CA 95064, USA Department of Physics, University of California Davis, Davis, CA95616, USA; IGPP, Lawrence Livermore National Laboratory, Livermore,CA 94550, USA
BARTH ET AL. INTRODUCTION
Observations of broad-line variability in nearby Seyfertgalaxies play a central role in the interpretation of the demo-graphics and cosmological evolution of supermassive blackholes in active galactic nuclei (AGNs). By measuring thetime delay between AGN continuum variations and the subse-quent response of the broad-line region (BLR) gas, the light-travel time across the BLR, and hence the mean BLR ra-dius ( r BLR ), can be directly measured. These reverberation-mapping measurements have been carried out for a few dozenlow-redshift AGNs (e.g., Kaspi et al. 2000; Peterson et al.2004; Bentz et al. 2009a), and the observed BLR sizes mea-sured via H β reverberation range from typically a few light-days up to several light-months. The BLR size is observed toscale with AGN continuum luminosity roughly as r BLR ∝ L . (Bentz et al. 2009b), and this relationship makes it possible toestimate BLR radii from a single observation of an AGN.With a direct measurement or estimate of r BLR , and as-suming virial motion of BLR clouds, it becomes possible toestimate the mass of the black hole in an AGN as M BH = f r BLR ( ∆ V ) / G , where ∆ V is the width of the broad line,and f is a dimensionless scaling factor (e.g., Ulrich et al.1984; Kaspi et al. 2000; Onken et al. 2004). This methodhas been used to estimate black hole masses in large sam-ples of AGNs out to the highest observed redshifts (for areview, see Vestergaard 2011). Currently, nearly all obser-vational constraints on the cosmological growth history ofsupermassive black holes depend on masses derived fromthis virial equation. While the assumption of virial mo-tion in the BLR has gained support from a variety of con-sistency checks (e.g., Peterson & Wandel 2000; Onken et al.2004; Nelson et al. 2004), it remains extremely difficult to ob-tain direct constraints on the structure and dynamical state ofthe BLR in any individual AGN, and in the absence of suchconstraints, the inferred black hole masses remain subject tosubstantial systematic uncertainty (Krolik 2001). The mostpromising method to examine the kinematics of BLR gas isvelocity-resolved reverberation mapping, in which the time-delay response of emission-line variability relative to contin-uum fluctuations can be measured as a function of line-of-sight velocity. Velocity-resolved reverberation data can en-code a wealth of information about BLR structure on spatialscales that are orders of magnitude too small to be resolved byany other technique (e.g., Welsh & Horne 1991; Horne et al.2004).Recently, high-cadence observing campaigns have pro-duced major improvements in the quality of velocity-resolvedreverberation data for the broad Balmer lines (Bentz et al.2009a; Denney et al. 2010). For the most highly variableobjects, it is possible to examine the shape of the two-dimensional transfer function, which describes the distribu-tion of broad-line lag response time as a function of veloc-ity (Bentz et al. 2010), and to apply new modeling techniquesthat can directly constrain the BLR geometry and potentiallytest the critical assumption of virial motion (Pancoast et al.2011a; Brewer et al. 2011). In order to increase the numberof objects with data suitable for such analysis, we carried outa new reverberation-mapping campaign in Spring 2011.In this Letter , we present our first results for Mrk 50,a Seyfert 1 galaxy at redshift z = 0 . β emission line. OBSERVATIONS AND REDUCTIONS
Photometry
From 2011 January 21 until June 13 (all dates are UT), weobtained queue-scheduled V -band images using the 0.76 mKatzman Automatic Imaging Telescope at Lick Observa-tory (Filippenko et al. 2001), the 0.9 m telescope at theBrigham Young West Mountain Observatory (WMO), theSuper-LOTIS 0.6 m telescope at Kitt Peak, the Faulkes Tele-scope South at Siding Spring Observatory, and the Palomar1.5 m telescope (Cenko et al. 2006). Exposure times weretypically 180–300 s. We attempted to observe Mrk 50 on anightly basis, but poor weather and telescope scheduling is-sues left some gaps in temporal coverage.All images were bias-subtracted and flattened, and cosmic-ray hits were removed using the LA-COSMIC routine(van Dokkum 2001). In order to remove the host galaxyand obtain a light curve of the variable AGN flux, we em-ployed image subtraction using the HOTPANTS package byA. Becker , which is based on the algorithm described byAlard (2000). For each telescope, a high-quality templateimage was chosen, and the template was then aligned witheach night’s image and convolved with a spatially varyingkernel to match the point-spread function of that image. Af-ter subtracting the scaled template image, the variable AGNflux is left as a point source in the subtracted image, allow-ing for aperture photometry using the IRAF DAOPHOT pack-age. The photometric aperture radius used for each telescopewas set to match the average point-source full width at half-maximum intensity (FWHM) for images from that telescope.Light curves were constructed separately for each telescope,and each was then normalized to the WMO light curve by de-termining an average flux scaling factor based on nights whenMrk 50 was observed at both locations. We find that the im-age subtraction yields a better-quality light curve than simpleaperture photometry, and provides closer agreement betweenthe light curves obtained with different telescopes.The overall flux scale was calibrated using observations ofLandolt (1992) standard stars observed during a few photo-metric nights at WMO. Observations taken within 6 hr of oneanother were combined by taking a weighted average. Thefinal V light curve is illustrated in Figure 1. Spectroscopy
Our campaign at the Lick Shane 3 m telescope consistedof 69 nights allocated between 2011 March 27 and June 13,during which time we observed Mrk 50 on 41 nights usingthe Kast dual spectrograph (Miller & Stone 1993). In this pa-per, we discuss only measurements from the blue arm of thespectrograph, where we used a 600 lines mm - grism over ∼ - . All observationswere done with a 4 ′′ -wide slit oriented at PA = 180 ◦ . Standardcalibration frames including arc lamps and dome flats wereobserved each afternoon, and flux standards were observedduring twilight. The exposure time for Mrk 50 was normally2 × EVERBERATION MAPPING OF MRK 50 3 F IG . 1.— Mrk 50 light curves for the V band, the U s -band continuum mea-sured from the spectra over 3550–3800 Å, and the H β , H γ , and He II emis-sion lines. The V -band panel shows a difference imaging light curve illus-trating changes in flux relative to the mean. Units for the V and U s bands are10 - erg cm - s - Å - , and units for broad-line fluxes are 10 - erg cm - s - . The arrow in the H β light curve marks the start of the dedicated Lickobserving campaign. and 4 ′′ slit, but the wavelength coverage was slightly differenteach time. The exposure for these observations was typically900 s.Spectroscopic reductions and calibrations followed stan-dard methods implemented in IRAF and IDL. A large extrac-tion width of 10 . ′′ SPECTROSCOPIC DATA ANALYSIS
The reduced spectra were first normalized to auniform flux scale by employing the procedure ofvan Groningen & Wanders (1992). This method appliesa flux scaling factor, a linear wavelength shift, and a Gaussianconvolution to each spectrum in order to minimize theresiduals between the data and a reference spectrum con-structed from several of the best-quality nights. The scalingis determined using a wavelength range containing the [O
III ] λ II tem- F IG . 2.— Mean and rms spectra. In each panel, the upper spectrum isconstructed from the set of scaled spectra of Mrk 50, and the lower spectrumis constructed from the continuum-subtracted, scaled spectra. plate from Boroson & Green (1992) broadened by convolu-tion with a Gaussian kernel, several emission-line compo-nents represented by either Gaussians or Gauss-Hermite func-tions (van der Marel & Franx 1993), and a starlight modelconsisting of simple stellar population models at solar metal-licity from Bruzual & Charlot (2003) which were broadenedby convolution with a Gaussian kernel. Additionally, aforeground extinction was applied to the model spectrum.The model fit was optimized by χ minimization using aLevenberg-Marquardt technique (Markwardt 2009). Then,the best-fitting model components representing the starlightand nonstellar continuum were subtracted from the data, leav-ing a pure emission-line spectrum. For the starlight model,we obtained a good fit using an 11 Gyr-old population as thedominant component, and adding contributions from youngerpopulations did not significantly improve the fit. The medianstarlight fraction at λ rest = 5100 Å is 41%, and Fe II contributesjust ∼ II λ β , 4370–4510 Å forH γ (also including the narrow [O III ] λ II . Emission-line light curves are shown in Fig-ure 1, along with the continuum flux density measured fromthe scaled spectra over 3550–3800 Å (which we refer to as the U s band). The continuum and broad lines were highly variableduring the monitoring period: over a 30-day span the U s -band BARTH ET AL. F IG . 3.— Cross-correlation functions for H β , H γ , and He II against theAGN continuum flux, the autocorrelation function of the V -band continuum,and the cross-correlation of the continuum bands U s vs. V . continuum dropped by a factor of ∼
4, and the H β line re-sponded with a factor of ∼ β line in the rms spectrum is mostoften used as the measure of line width (Peterson et al. 2004).We find σ (H β rms ) = 1740 ±
101 km s - , where the measure-ment uncertainty is determined through a bootstrap resam-pling procedure (Bentz et al. 2009a), and the line width hasbeen corrected for instrumental broadening of σ inst ≈ - following Barth et al. (2011). REVERBERATION LAG MEASUREMENTS
In order to measure the cross-correlation function (CCF)for unevenly sampled time series, we employ the interpola-tion cross-correlation function methodology and Monte Carloerror analysis techniques described by Gaskell & Peterson(1987), White & Peterson (1994), and Peterson et al. (2004);these methods have been employed in the majority of re-cent reverberation-mapping work (e.g., Bentz et al. 2009a;Denney et al. 2010). We measured the cross-correlations ofH β , H γ , and He II against the V light curve, and also of H β against the U s light curve. The CCFs were computed from -
20 to +
40 days in increments of 0.25 days. Table 1 lists twomeasures of the lag: τ peak , which is the lag at the peak of theCCF, and τ cen , the centroid of the CCF for all points above80% of the peak value (Peterson et al. 2004). Figure 3 illus-trates the CCF measurements.The H β and H γ lines have similar lag times of τ cen = 10 . II we find that both τ cen and τ peak are consistent with zero lag, indicating a very com-pact size for the inner, high-ionization portion of the BLR.The faster response time for He II is apparent in the lightcurves, particularly in the steep drop beginning in mid-April(HJD ≈ TABLE 1C
ROSS -C ORRELATION L AG R ESULTS
Measurement τ cen (days) τ peak (days)H β vs. V . + . - . . + . - . H γ vs. V . + . - . . + . - . He II vs. V - . + . - . - . + . - . H β vs. U s . + . - . . + . - . U s vs. V . + . - . . + . - . N OTE . — All lags are given in the observed frame. of the H β CCF in comparison with the continuum autocorre-lation function (ACF) suggests that the Balmer-line emittingzone of the BLR covers a fairly large radial extent, and thisinterpretation is supported by the velocity-resolved measure-ments described below. We also measured the CCF betweenthe U s and V bands in order to search for evidence of rever-beration due to Balmer continuum emission from the BLR(Maoz et al. 1993; Korista & Goad 2001), but no significantlag was found.Our Mrk 50 dataset, which benefits from high-amplitudevariability and high-cadence sampling, presents an excellentcase study for velocity-resolved variability. Light curves weremeasured for seven separate velocity segments across thewidth of the H β line, and each segment light curve was cross-correlated against the V light curve. Figure 4 illustrates thelag ( τ cen ) as a function of velocity across the H β line, reveal-ing a roughly symmetric trend of shorter lags in the line wingsand longer lags in the core, with a ∼
10 day difference in re-sponse time between the core and wings. This pattern, whichresembles the symmetric H β lag response seen in some otherAGNs including Mrk 110 (Kollatschny & Bischoff 2002) andNGC 5548 (Denney et al. 2010), is qualitatively consistentwith predictions for BLR clouds in circular orbits in the Kep-lerian potential of the black hole (e.g., Welsh & Horne 1991),with the highest line-of-sight velocities originating from gaslocated close to the black hole. In contrast, a BLR dominatedby either radial infall or outflow would result in an asymmet-ric velocity-lag map, with inflow producing longer lags on theblueshifted side of the line, and outflow producing longer lagson the red side. Such signatures of radial motion have beenseen in a few objects (Bentz et al. 2009a; Denney et al. 2010),but the current sample of AGNs with velocity-resolved dataof this quality is still too small to examine the distribution ofdifferent BLR kinematic states. BLACK HOLE MASS ESTIMATE
Following Peterson et al. (2004), we compute the H β “virial product” [defined as VP = r BLR ( ∆ V ) / G ] by using ∆ V = σ (H β rms ) and r BLR = c τ cen , where τ cen has been cor-rected to the AGN rest frame. For Mrk 50, r BLR = 10 . + . - . lt-days, and VP = (6 . ± . × M ⊙ .While the virial product is a straightforward combinationof measured quantities, the relationship between virial prod-uct and black hole mass is more indirect and uncertain. Mostrecent reverberation work has adopted a normalization ofthe virial mass scale based on the assumption that AGNs asa whole fall on the same M BH - σ relation as nearby inac-tive galaxies (Onken et al. 2004; Woo et al. 2010). Using the M BH - σ relation derived by Gültekin et al. (2009) as the lo-cal reference, this implies a mean value of log f = 0 . + . - . ,and the reverberation masses determined in this way showan intrinsic scatter of 0.44 dex about the M BH - σ relationEVERBERATION MAPPING OF MRK 50 5 F IG . 4.— Velocity-resolved reverberation in the H β line. The upper panelshows the mean lag measured for each velocity segment of the broad H β line, with the horizontal error bar representing the width of the velocity seg-ment. The overall lag for H β and its uncertainty range are shown by solid anddashed lines. The lower panels show the mean and rms continuum-subtractedspectra, and the error bar indicates the FWHM due to instrumental broaden-ing. (Woo et al. 2010). Adopting the Woo et al. value of f = 5 . M BH = (3 . ± . × M ⊙ for Mrk 50. For con-sistency with recent work (Peterson et al. 2004; Bentz et al.2009a; Denney et al. 2010), the quoted uncertainty includesonly the statistical error on the virial product, but not the un-certainty resulting from the choice of a specific f value; thetrue uncertainty in M BH is dominated by the uncertainty in f . Masses derived in this way depend on the assumption ofvirial motion, the assumption that AGNs should fall on thesame M BH - σ relation as quiescent galaxies, and the adop-tion of a specific M BH - σ relation as the local baseline. Dif-fering assessments of the form of the local M BH - σ relation,particularly at low masses, can potentially affect the normal-ization of the AGN virial mass scale at the factor of ∼ f can currently be determined to ∼ M BH - σ relation (Woo et al. 2010; Graham et al.2011), the actual uncertainty in the overall AGN mass scale remains significantly larger and is difficult to quantify, andthe observed 0.44 dex scatter in the AGN M BH - σ relationmust reflect, at least in part, the dispersion of the true f values for individual AGNs. Resolving these issues will re-quire enlarging the number of targets having high-qualityreverberation-mapping data, and the application of new ap-proaches to extract dynamical information from the observa-tions (e.g., Pancoast et al. 2011a; Brewer et al. 2011).Mrk 50 has an early-type morphology, and Malkan et al.(1998) classify it as an S0 galaxy based on Hubble SpaceTelescope imaging. The only published measurement of thestellar velocity dispersion of Mrk 50 is σ ⋆ = 78 ±
15 km s - based on a Sloan Digital Sky Survey spectrum (Greene & Ho2006). However, a new measurement from Keck LRIS datagives 109 ±
14 km s - (Harris et al., in preparation), and weconsider this to be more reliable than the SDSS measurement.The scaling relations calibrated by Gültekin et al. (2009) thenimply an expected M BH ≈ (0 . - . × M ⊙ from thegeneral M BH - σ relation including ellipticals and spirals, or(0 . - . × M ⊙ based on the M BH - σ relation fitted to el-lipticals only. Our reverberation-based mass is slightly higherthan these values, but Mrk 50 still lies well within the ∼ . M BH - σ relation (Woo et al. 2010). CONCLUSIONS AND FUTURE WORK
From our Spring 2011 monitoring campaign, we have ob-tained very high-quality reverberation mapping of Mrk 50.This is one of just a few nearby AGNs in which strongvelocity-resolved lag signatures have been detected, and it isnow among the most promising targets for detailed studies ofBLR geometry and kinematics.Our long observing campaign produced a large and veryrich dataset, and this paper presents just one set of early re-sults from this program. Future papers will include detaileddescriptions of the data-analysis procedures and results forthe entire observed sample. A major goal of our project is toexploit the potential of velocity-resolved reverberation map-ping to elucidate the structure of the BLR and to derive blackhole masses directly, and an upcoming paper (Pancoast et al.2011b) will describe new modeling of our Mrk 50 data. Theblack hole mass determined from dynamical modeling is con-sistent with the simple virial estimate presented here, andMrk 50 is now the second object (after Arp 151; Brewer et al.2011) to show agreement between the two approaches.We are extremely grateful to the Lick Observatory stafffor their outstanding support during our observing run. TheLick AGN Monitoring Project 2011 is supported by NSFgrants AST-1107812, 1107865, 1108665, and 1108835. T.T.acknowledges a Packard Research Fellowship. The WestMountain Observatory receives support from NSF grant AST-0618209. We thank Brad Peterson for a helpful referee report.
Facilities:
Shane (Kast), KAIT, BYU:0.9m, PO:1.5m,LCOGT (FTS), Super-LOTIS
REFERENCESAlard, C. 2000, A&AS, 144, 363Barth, A. J., et al. 2011, ApJ, 732, 121Bentz, M. C., et al. 2009a, ApJ, 705, 199Bentz, M. C., Peterson, B. M., Netzer, H., Pogge, R. W., & Vestergaard, M.2009b, ApJ, 697, 160Bentz, M. C., et al. 2010, ApJ, 720, L46Boroson, T. A., & Green, R. F. 1992, ApJS, 80, 109Brewer, B. J., et al. 2011, ApJ, 733, L33 Bruzual, G., & Charlot, S. 2003, MNRAS, 344, 1000Cenko, S. B., et al. 2006, PASP, 118, 1396Denney, K. D., et al. 2010, ApJ, 721, 715Filippenko, A. V., Li, W. D., Treffers, R. R., & Modjaz, M. 2001, in SmallTelescope Astronomy on Global Scales, ed. W. P. Chen, et al. (SanFrancisco: ASP), 121Gaskell, C. M., & Peterson, B. M. 1987, ApJS, 65, 1