gisella de rosa the ohio state university on behalf of the agn-storm collaboration probing agn...

GISELLA DE ROSATHE OHIO STATE UNIVERSITY

ON BEHALF OF THE AGN-STORM COLLABORATION

Probing AGN Structure on Microarcsecond Scales:

The Space Telescope and Optical Reverberation Mapping Program

Hubble 25 – April, 21st, 2015

AGN STORMTROOPERS (incomplete...) 2

P. Arevalo, A.J. Barth, V.N. Bennert, M.C. Bentz, A. Bigley, M. Bottorff, T.A. Boroson, W.N. Brandt, A.A. Breeveld, M. Brotherton, B.J. Brewer,

E. Cackett, M. Carini, L. Carrasco, S. Chul, K. Clubb, J. Comerford, E.M. Corsini, D.M. Crenshaw, S. Croft, E. Dalla Bontà, A. Deason, A. De Lorenzo-Caceres,

K. D. Denney, G. De Rosa, M. Dietrich, T. Dwelly, R. Edelson, S. Eftekharzadeh, J. Ely, M. Eracleous, P.A. Evans, M.M. Fausnaugh, A.V. Filippenko, O. Fox, N. Gehrels,

J.M. Gelbord, M.R. Goad, C.J. V. Gorjian, M. Graham, J.E. Greene, C.J. Grier, D. Grupe, J. Gutierrez,P.B. Hall, E. Holmbeck, K. Horne, M. Im, C. Johnson, M. Joner, J.

Kaastra, S. Kaspi, B.C. Kelly, P. Kelly, J.A. Kennea, M. Kim, C.S. Kochanek, K.T. Korista, G.A. Kriss, D. Laney, M.W. Lau, J.C. Lee, D. Leonard, P. Lira,

M. Lundquist, S. Mathur, J. Mauderhan, J. McCormac, R. McGurk, I.M. McHardy, N. Milgram, L. Morelli, F. Muller Sanchez, H. Netzer,M. Nguyen, J. Nousek, P. Ochner,

A. Pancoast, I. Papadakis, L. Pei, B.M. Peterson, A. Pizzella, R.W. Pogge, J.-U. Pott, F. Pozo Nunez, X. Prochaska, G. Rude, D. Sand, J.S. Schimoia, K. Schnulle,

S.G. Sergeev, B. Shappee, I. Shivvers, M. Siegel, A. Siviero, D. Starkey, K. Steenbrugge, M. Strauss, A,-L. Sun, H.I. Sung, M. Tejos, T. Treu, B. Tucker, P. Uttley, S. Vaughan,

M. Vestergaard, L. Vican, C. Villforth, W.F. Welsh, J.-H. Woo, H. Yan, H. Yuk, S. Young, N. Zakamska, W. Zheng, Y. Zu

Hubble 25 – April, 21st, 2015

Outline3

Studying AGN using reverberation mapping

AGN STORM: program details

AGN STORM: early results

Reverberation mapping with HST, the future

Hubble 25 – April, 21st, 2015

AGN structure4

Hubble 25 – April, 21st, 2015

Urry & Padovani, 1995

NGC 5548

Rest wavelength (Å)

f l (1

0-1

3 e

rg c

m-2 s

-1 Å

-1)

BLR

NLR

DISK

Problem: microarcsecond resolution needed to study the details of the inner structure (e.g. size, dynamics)

Figure courtesy of L. Pei

Reverberation mapping: time vs spatial resolution

The continuum luminosity varies stochastically in time

The gas responsible for the broad line emission is ionized by the AGN continuum and “responds” to variations in the continuum luminosity on a time scale

t = RBLR / c

We can estimate the distance of the gas from the timescale for response.

5

t2= t1+ t = t1+RBLR/c

RBLR

Accretion disk

Broad line gas

Hubble 25 – April, 21st, 2015

Observer

Reverberation mapping: estimating t6

t t t

Continuum

CROSS CORRELATION

Lya line

Hubble 25 – April, 21st, 2015

De Rosa+ 2015

Reverberation mapping: MBH estimate

MBH= f RBLR Dv2 G-1

Virial Product

RBLR from RM delay estimate

Dv from width of the varying part of the emission line

Calibration of <f> through MBH-s*

7

Hubble 25 – April, 21st, 2015

Quiescent galaxyAGN, new dataAGN

Grier+2013

Reverberation mapping: R-L relation8

Radius of the emitting region increases with a power of the ionizing continuum luminosity

Single Epoch Mass Estimates: MBH ∝ Dv2 RBLR

line width2 Llg

All relations are anchored to the RM MBH

RM is not only vital to estimate MBH in the local universe, but also in distant sources

Hubble 25 – April, 21st, 2015

9

Observer

v1

v2

v3 t1 < t2 < t3

Reverberation mapping: resolving in t and vLOS

Hubble 25 – April, 21st, 2015

Required time sampling, duration, and S/N makes velocity-delay map recovery very difficult.

Reverberation mapping: status before AGN STORM

10

Reverberation lags measured for ~50 AGNs, mostly Hβ

AGNs with lags for multiple lines: highest ionization emission lines respond most rapidly ionization stratification

Handful of optical datasets good enough to probe BLR structure/kinematics. Balmer lines: flattened geometries, combination of virialized motion and infalling gas

Little known about high-ionization lines (IUE, HST-FOS). Estimated lags but limited information about detailed response (few orbits, low S/N)

Hubble 25 – April, 21st, 2015

11

AGN STORM: program

AGN STORM program

12

Multiwavelength reverberation mapping monitoring program to study the Seyfert 1 galaxy NGC 5548,

2014 January - August

NIR-MIR:photometry

ground based, Spitzer

Optical: ground based

spectroscopy &

photometrydaily

X-Ray:Chandra, SWIFT

Optical-NUV:

photometrySWIFT

3 times per day

FUV: HST

once per day171 orbits

AGN STORM

Hubble 25 – April, 21st, 2015

AGN STORM: scientific goals

13

Structure and kinematics of the high-ionization BLR with HST COS Ly α λ1215, C IV λ1549, Si IV λ1400, NV λ1240, He II λ1640

Simultaneous ground-based optical spectroscopy programBalmer lines, He II λ4686

Accretion disk structure with HST, Swift, and ground-based data (X-rays to mid-IR)

Physical structure of emitting gas Dynamical processes at microarcsec scales

Uncertainties in MBH based on RM measurementsHubble 25 – April, 21st, 2015

AGN STORM: HST program

14

Daily observations with Cosmic Origins Spectrograph:

Each visit: G130M and G160M exposures in single orbit (~1100-1750 Å )

179 visits (2014 February 1 to July 27) 8 unsuccessful visits (e.g., safing events)

Customized data reduction pipeline developed: Precision of flux calibration better than 1.5% locally Wavelength calibration precise to < 6 km s−1

Hubble 25 – April, 21st, 2015

15

AGN STORM: early results

AGN STORM HST program

Mean lags relative to 1367 Å continuum

16

Hubble 25 – April, 21st, 2015 De Rosa+ 2015

Preliminary analysisCIV emission line

Strong velocity dependence of lags

Higher velocity gas responds most rapidly

First half: larger variations, shorter lags

Second half: smaller variations, longer lags

17

Hubble 25 – April, 21st, 2015 De Rosa+ 2015

Preliminary analysisLya emission line

Strong velocity dependence of lags

Situation complicated by strong absorption systems

First half: larger variations, shorter lags

Second half: smaller variations, longer lags

18

Hubble 25 – April, 21st, 2015 De Rosa+ 2015

AGN STORMHST program

challenges

Broad UV absorption gradually disappearing during the campaign

Analysis of absorption features ongoing ( Kriss+ in prep )

19

Hubble 25 – April, 21st, 2015

Figure courtesy of G. Kriss

AGN STORMHST program

challenges

Line variations are not simple continuum echoes

Superb data, challenge to model in detail

Velocity delay map analysis ongoing

(Horne+ in prep; Pancoast+ in prep)

20

Hubble 25 – April, 21st, 2015

Figure courtesy of K. Horne

Continuum

CIV

AGN STORM HST & optical

spectroscopy

Mean lags relative to UV and optical continua

21

Pei+ in prep

Hubble 25 – April, 21st, 2015

AGN STORM: HST & optical spectroscopy

22

topt >> tUV RBLR measured from optical data alone is systematically lower than true RBLR

Not a problem for statistical MBH calibrated via M-s relationship as long as UV/opt continuum lag does not scale strongly with L or M (systematics simply subsumed into f)

More simultaneous optical/UV monitoring programs needed to determine continuum lag to determine possible dependence on L or M

Hubble 25 – April, 21st, 2015

Swift UVOT - Feb to June 2014

12 hour cadence ~280 epochs

23

AGN STORM: HST & NUV-optical photometry

Ground based – 16 observatories

> 600 epochs in V

Edelson+ 2015

Hubble 25 – April, 21st, 2015 Fausnaugh+ in prep

HST & NUV-optical photometry

Some tension with irradiated thin disk model

Disk structure larger than expected for a standard thin disk (Edelson+2015, Fausnaugh+ in prep)

Optical lags are comparable to, or greater than, He II lags. Emission regions are of similar physical size

24

Hubble 25 – April, 21st, 2015

Fausnaugh+ in prep

Edelson+ 2015

AGN STORM: summary

25

Successfully acquired 171 UV spectra of NGC 5548 with HST COS.

Emission-line lags range from ~2.5 days for He II λ1640 t0 ~6 days for Lyα λ1215.

Both Lyα and C IV lags show velocity-dependence: the data set has information on BLR kinematics.

Lag of a few days between UV and optical continuum light curves: implications for RBLR and virial factor.

Accretion disk: tension with thin disk model. Similar physical size than emission regions.

Hubble 25 – April, 21st, 2015

AGN STORM: publication plan 26

I: HST-COS dataset – accepted

II: Swift-HST RM of the accretion disk - accepted

III: Continuum interband lags HST-SWIFT-Ground based ~ submission IV: Optical emission line variations ~ submission

V: Velocity delay maps - ongoing

VI: Modeling of accretion disk structure - ongoing

VII: Direct modeling of spectra - ongoing

VIII: Absorption line variations - ongoing

IX: Dynamical modeling - ongoing

X: Spectral energy distribution analysisXI: Photoionization modelingXII: NIR observationsXIII: Wrap up

Hubble 25 – April, 21st, 2015

AGN STORM: a look into the (near) future

27

Hubble 25 – April, 21st, 2015

The AGN-STORM program shows that:

• RM with HST-COS is feasible (schedule/instrument) and successful

• UV coverage is needed when absolute size of BLR matters (e.g. dynamical modeling)

• Unexpected results – need to test uniqueness

We have ~5 years to observe more sources:AGN ID tCIV lt-days

NGC 3783 ~4, 2

Fairall 9 ~14

3C390.3 ~30, 6

NGC 7469 ~3, 5

NGC 4151 ~3.5, 3

NGC 4593 ~0.1, 1.5

28

Backup

STScI – August, 14th, 2014

Ground base observability vs z29

Figure courtesy of Sarah Gallagher

Reverberation mapping: geometry and kinematics

30

Observer

1

2

3

Toy model, stationary clouds

τ=r/c All points on an “isodelay surface” have the same extra light-travel time to the observer, relative to photons from the continuum source.

t1 < t2 < t3Shell: emission-line gas

Hubble 25 – April, 21st, 2015

AGN STORM: continuum variations31

Typical precision < 1.5%

Hubble 25 – April, 21st, 2015

Reverberation Mapping: assumptions32

Continuum originates in a single central sourceMBH=107-108 Msun , accretion disk size=1013-14 cm – BLR size ~1016 cm

Light travel is the most important time scaleEmission line gas responds instantaneously (trec ~ 40)

RM experiments carried out on time scales shorter than BLR dynamical time (3-5 years)

Simple (though not necessarily linear) relation between observed continuum (optical/UV) and the ionizing continuum (FUV)

Hubble 25 – April, 21st, 2015

Reverberation Mapping: simplified approach

33

Under these assumptions, velocity independent transfer equation:

However, most of the past data sets were inadequate for transfer function solution, so simpler analyses were used: e.g. Cross Correlation Function Analysis

Line light curve

Continuum light curve

TransferFunction

Lag estimates34

Mean response time (“lag”) obtained from cross-correlation function:

Figure courtesy of B. Peterson

Emission line width

Measure line widths from the rms residual spectrum:

Captures the velocity dispersion of the gas that is “reverberating”.

Non varying features disappear.

35

De Rosa et al., in prep

DvLOS

A virialized BLR36

For all the sources for which it is has been testable (multiple emission line):

Suggesting that gravity is indeed the principal dynamical force in the BLR

DV ∝ RBLR-1/2

Peterson & Wandel, 2002

MBH cross-check37

Comparison between direct methods(MBH in units of 106 Msun)

Galaxy NGC 4258 NGC 3227 NGC 4151

Megamasers 38.2±0.1 N/A N/A

Stellar dynamics 33±2 7-20 47+11-14

Gas dynamics 25-260 20+10-4 30+7.5

-22

Reverberation mapping

N/A 7±1 26.5±1.5

New

AGN12 AGN12

MBH estimate

MBH= f RBLR Dv2 G-1

Virial Product

Observable

f is our ignorance factor: everything we do not know about BLR geometry and inclination

Putting everything together Calibration of <f> through MBH-s*

38

Grier et al., 2013

Quiescent galaxyAGN, new dataAGN

Hb RBLR-L AGN Relationship

Simple ionization argument:

To first order, all AGN spectra look the

same (Davidson 1972):

After a carful subtraction of the host-galaxy light:

R ∝ LBLR0.5

39

Bentz et al., 2013

AGN-STORM: Why 180 HST orbits?

40

1. Assume geometry & kinematics: inclined keplerian disk with two-armed spiral wave

2. Simulate continuum variation assuming dumped random walk

3. Create velocity delay maps for different lines (photoionization grid)

4. Create realistic mock spectra5. Create recovered velocity delay maps6. Play with free parameters: length of the

campaign & time resolution

Velocity delay map: edge-on ring41

rϑ

Rotating clockwise, Vorb

Observer

Line-of-sight velocity V (km/s)

2 r/c

r/c

Vorb-Vorb

Time delay τ

V=Vorb sinϑ

τ=(1+cosϑ)r/c

Circular orbit projects to an ellipse in the (V, τ) plane.

Velocity delay map: thick geometries 42

Generalization to a disk (multiple rings) or thick shell (multiple inclination) is trivial.

Observer

LOS V (km/s)

2 rout/c

Vout-Vout

τ

rout/c

-Vin Vin

2 rin/c

rin/c

Observer

Reverberation Mapping: full approach43

Under same assumptions as before:

Velocity resolved line light curve

Continuum light curve

Velocity delay map

Well-defined inversion problem dependent only on the quality of the data.

First recovered velocity-delay map ARP 151: Bentz et al. 2010

Problems: S/N, duration and time sampling requirements make the recovery of velocity delay maps difficult

AGN-STORM: program elements

44Spectroscopy UV 2.5m Hubble Space Telescope + COSOptical MDM 1.3m Telescope, Kitt Peak ARC 3.5m Telescope, Apache Point Asiago 1.22m Telescope, Asiago, Italy CrAO 2.6m Shajn Telescope, Crimea Lick 3m Shane Telescope, Mt. Hamilton, Calif

ornia 2.5m Nordic Optical Telescope + Spectrograp

h, La Palma, Canaries WIRO 2.3m Telescope, Wyoming 2m Liverpool Telescope with Frodospec spect

rograph, La Palma, Canaries

PhotometryX-Ray Chandra ACIS-S/LETG Swift X-Ray TelescopeUV Swift 0.3m UVOT (UV/Optical) Telescope

Optical  2m Liverpool Telescope with Imager, La Palma, Canari

es Bohyunsan Optical Astronomy Observatory 1.8m Teles

cope, Mt. Bohyun, South Korea  

CrAO 0.7m AZT-8 Telescope, Crimea, Ukraine CrAO 1.3m AZT-11 Telescope, Crimea, Ukraine (operat

ed by WKU) Fountainwood Observatory, Georgetown, Texas (SU) LOAO 1-m Telescope, Mt. Lemmon Arizona (operated

by KASI) Hard Labor Creek Observatory 0.5m Telescope, Rutle

dge GA (GSU) 1.3m RCT Telescope, Kitt Peak Arizona (operated by

WKU) Wise Observatory C18 Telescope, Mitzpe Ramon, Israe

l Bell Observatory 0.6m Telescope, Western Kentucky U

niversity West Mountain Observatory 0.91m Telescope, West M

ountain Utah (BYU) ESO 2.2m Telescope,LaSilla, Chile LCOGT 1-meter Telescope Maidanak Observatory 1.5m Telescope, Uzbekistan Mount Laguna Observatory 1m Telescope, California Swift 0.3m UVOT (UV/Optical) Telescope ASASSN

Near Infrared ESO 2.2m Telescope,LaSilla, Chile Infrared Imaging System 0.8m Telescope, Cerro Arma

zones, Chile 2m Liverpool Telescope with Imager, La Palma, Canari

es Guillermo Haro Observatory, Sonora, Mexico (INOAE) Spitzer Space Telescope, IRAC imager

Velocity binned time lags45

Hubble 25 – April, 21st, 2015

46

Wavelength solution

Offsets in Standard Pipeline Products

• Wavelength offsets are seen between • different CENWAVES• COS and STIS datasets

• Offsets are consistent: data shown is 170 orbits combined for each setting.

47

Correcting with STIS48

• Identify IGM features present in COS and STIS spectra

• Cross-correlate lines from each observation with the STIS dataset

• Compute a linear offset to the dispersion and apply

Improved Alignment49

50

Lowered RMS

51

Fixed pattern noiseRaise your hand if you see any potential sources of fixed-pattern noise.

Improving S/N with P-flats52

170 orbits (21 hours of G130M and 57 hours of G160M)1291 and 1327 at 4 FPPOS1600 at 2 FPPOS1623 at 1 FPPOS

Wide range of parameter space:Broad settings with maximum FPPOS contributionNarrow settings with minimal FPPOS contribution

Improving S/N with P-flats53

Prominent Effect on narrow modes with few FPPOS

Narrow profile keeps fixed-pattern features sharp

Few FPPOS lets fixed-pattern noise patterns repeat

Improving S/N with P-flats54

Prominent Effect on narrow modes with few FPPOS

Narrow profile keeps fixed-pattern features sharp

Few FPPOS lets fixed-pattern noise patterns repeat

Improving S/N with P-flats55

Much smaller, though still evident effect on wider modes with many FPPOS

G130M modes intrinsically average over more detector area

More FPPOS diminishes the effect of fixed-pattern features

S/N Limits56

57

Static sensitivity functions

Residuals at lifetime position 2: WD 0308-56558

Residuals at lifetime position 2: WD 1057+71959

AGN-STORM: Static sensitivity functions

60

WD 0308-565 G130M, WD 1057+719 G160M

Model each CENWAVE/FPPOS setting individually

Low order Chebyshev polynomial FUVA

Spline function FUVB

Mask absorption lines and bad pixels

Residuals at lifetime position 2: G130M standard pipeline

61

62

Residuals at lifetime position 2: G130M AGN-STORM pipeline

63

Residuals at lifetime position 2: G160M standard pipeline

64

Residuals at lifetime position 2: G160M AGN-STORM pipeline

65

Time dependent sensitivity

66

WD 0308-565, time evolution of the residuals

AGN-STORM: Time dependent sensitivity

67

1. Redefined wavelength ranges

2. Added additional break point

3. Computed correction from individual CENWAVE settings

68

TDS residuals: Standard pipeline

3 1

2

69

TDS residuals: AGN-STORM pipeline

3 1

2

70

TDS residuals: Time evolution of the distribution

71

TDS residuals: RMS

Summary72

Improved wavelength calibration by cross-correlating onto the STIS wavelength scale.

Increased S/N by applying pixel-to-pixel flatfields.

Improved static sensitivity function by modeling individual CENWAVE/FPPOS settings.

Improved TDS correction with a) Revised wavelength ranges b) Additional breakpoint c) Compute correction from individual CENWAVEs

= Needs

validation = Needs code

dev/testing = More

complicated