The distance to NGC1316 (Fornax A): yet another curious case
Michele Cantiello, Aniello Grado, John P. Blakeslee, Gabriella Raimondo, Gianluca Di Rico, Luca Limatola, Enzo Brocato, Massimo Della Valle, Roberto Gilmozzi
aa r X i v : . [ a s t r o - ph . C O ] M a r Astronomy & Astrophysics manuscript no. ms˙printer c (cid:13)
ESO 2018September 8, 2018
The distance to NGC 1316 (Fornax A): yet another curious case ⋆ Michele Cantiello , Aniello Grado , John P. Blakeslee , Gabriella Raimondo , Gianluca Di Rico , LucaLimatola , Enzo Brocato , , Massimo Della Valle , , and Roberto Gilmozzi INAF Osservatorio Astronomico di Teramo, via M. Maggini snc, I-64100, Teramo, Italye-mail: [email protected] INAF Osservatorio Astronomico di Capodimonte, salita Moiariello I-80131 Napoli, Italy Dominion Astrophysical Observatory, Herzberg Institute of Astrophysics, National Research Council of Canada,Victoria, Canada INAF Osservatorio Astronomico di Roma, Via Frascati 33, I-00040, Monte Porzio Catone, Roma, Italy International Centre for Relativistic Astrophysics, Piazzale della Repubblica 2, I-65122, Pescara, Italy European Southern Observatory, Karl-Schwarzschild-Str. 2, 85748 Garching bei M¨unchen, GermanyReceived –; accepted –
ABSTRACT
Aims.
The distance of NGC 1316, the brightest galaxy in the Fornax cluster, provides an interesting test for the cosmo-logical distance scale. First, because Fornax is the second largest cluster of galaxies within ∼ <
25 Mpc after Virgo and,in contrast to Virgo, has a small line-of-sight depth; and second, because NGC 1316 is the single galaxy with the largestnumber of detected Type Ia supernovae (SNe Ia), giving the opportunity to test the consistency of SNe Ia distancesboth internally and against other distance indicators.
Methods.
We measure surface brightness fluctuations (SBF) in NGC 1316 from ground- and space-based imaging data.The sample provides a homogeneous set of measurements over a wide wavelength interval. The SBF magnitudes, cou-pled with empirical and theoretical absolute SBF calibrations, are used to estimate the distance to the galaxy. We alsopresent the first B -band SBF measurements of NGC 1316 and use them together with the optical and near-IR SBFdata to analyze the properties of field stars in the galaxy. Results.
We obtain ( m − M ) = 31 . ± . . ) ± . . ) mag, or d = 20 . ± . . ) ± . . ) Mpc. Whenplaced in a consistent Cepheid distance scale, our result agrees with the distances from other indicators. On the otherhand, our distance is ∼
17% larger than the most recent estimate based on SNe Ia. Possible explanations for thisdisagreement are the uncertain level of internal extinction, and/or calibration issues. Concerning the stellar populationanalysis, we confirm the results from other spectro-photometric indicators: the field stars in NGC 1316 are dominatedby a component with roughly solar metallicity and intermediate age. A non-negligible mismatch exists between B -bandSBF models and data. We confirm that such behavior can be accounted for by an enhanced percentage of hot horizontalbranch stars. Conclusions.
Our study of the SBF distance to NGC 1316, and the comparison with distances from other indicators,raises some concern about the homogeneity between the calibrations of different indicators. If not properly placed inthe same reference scale, significant differences can occur, with dramatic impact on the cosmological distance ladder.Our results on the stellar populations properties show that SBF data over a broad wavelength interval are an efficientmeans of studying the properties of unresolved systems in peculiar cases like NGC 1316.
Key words. galaxies: elliptical and lenticular, cD – galaxies: distances and redshift – galaxies: clusters: individual:NGC 1316 – galaxies: photometry – galaxies: stellar content – galaxies: peculiar
1. Introduction
The past three decades have seen remarkable progressin the study of the distance scale of the Universe(Jacoby et al. 1992; Ferrarese et al. 2000; Freedman et al.2001; Freedman & Madore 2010), resulting in a gen-eral convergence of distances based on different indica-tors. Nevertheless, to resolve the lingering discrepancies(Tammann et al. 2008; Freedman & Madore 2010), there isan urgent need to ( i ) lower the statistical (intrinsic) and sys-tematic (external) errors for single distance indicators; ( ii )obtain distance measurements from indicators with a large range of applicability in terms of distances and of usefultargets in order to minimize the error propagation over thecosmological distance scale; and ( iii ) improve/analyze thematching between independent indicators. A particularlypromising distance indicator for addressing the three listeditems is the surface brightness fluctuations technique (SBF,Tonry & Schneider 1988; Tonry et al. 1990; Blakeslee et al.2009).After the first two “rungs” of the cosmological distancescale, represented by geometric methods and by primaryindicators (variable stars, main-sequence fitting, etc.), theSBF method is one of the most accurate indicators, with a median 0.14 mag accuracy on distance moduli, or ∼ ∼
100 Mpc (Mei et al. 2003; Biscardi et al.2008). However, the accuracy gets considerably smaller, toa mean ∼ .
08 mag ( ∼
4% in distance) from space-basedoptical data (Table 1)By definition, the SBF signal corresponds to the ratioof the second to the first moments of the luminosity func-tion of stars in a galaxy. As opposed to the surface bright-ness that does not scale with the distance of the galaxy,the SBF signal scales inversely with the distance squared.Observationally, the SBF method relies on the measure-ment of the intrinsic flux variance in a galaxy, generated bythe Poissonian fluctuations in the surface brightness dueto the statistical variation of the stellar counts in adjacentresolution elements. The variance, normalized to the localmean surface brightness, is converted to an apparent mag-nitude, ¯ m , from which the distance modulus, ¯ m − ¯ M , followsonce the absolute ¯ M is known.Given its definition, ¯ M in a given bandpass is dependenton the properties of the underlying stellar populations. Theanalysis of large samples of early-type systems, includinggalaxy in groups, has made it possible to characterize thedependence of ¯ M on stellar population properties using lin-ear relations with respect to some broad-band optical colour(Tonry et al. 2001; Mei et al. 2007; Cantiello et al. 2007a).In Table 1 we report the median errors on SBF mea-surements, δ ( ¯ m ), and on the associated distance moduli, δ ( ¯ m − ¯ M ), derived from different samples. For optical SBF,the typical accuracy for ground-based measurements hasbeen ∼ < ∼ .
20 mag in the V band, to 0 .
09 mag in the Spitzer [3.6] µm and [4.5] µm bands (e.g., Ngeow et al. 2009). The mean uncertaintiesfor the near-IR SBF sample quoted in Table 1 are gener-ally larger than the optical ones, ∼ < ∼
10% ondistances) even in the case of space-based data. This isdue both to observational/technical issues (e.g. dark cur-rent patterns, “wormy” background, see Jensen et al. 1998,2001), and to the scatter of the calibration (mostly re-lated to the sensitivity of the SBF signal to the propertiesof AGB and TP-AGB stars in these bands, see Mei et al.2001; Liu et al. 2002; Jensen et al. 2003; Raimondo et al.2005; Gonz´alez-L´opezlira et al. 2010). The availability ofWide Field Camera 3 (WFC3, on board of the
HubbleSpace Telescope , HST) is expected to significantly improvethe situation in this wavelength regime. Finally, the zeropoint of the calibration is typically tied to the Cepheid dis-tance scale to an accuracy ∼ < i ) in the list above). Thetechnique has been used to estimate distances for LocalGroup galaxies (even closer than that if one takes into ac-count the work on Galactic globular clusters by Ajhar et al.1994), out to galaxies at ∼ >
100 Mpc (Jensen et al. 2001;Biscardi et al. 2008). With the highly improved near-IRimaging capabilities of the WFC3/IR and similar instru-ments, and thanks to the much brighter SBF signal in thenear-IR, the upper limit on SBF distances is expected toincrease significantly. Thus, SBF can encompass more than two orders of magnitude in distance, bridging local to cos-mological distances with the use of a single indicator (item( ii ) above).In this paper we present measurements of SBFmagnitudes for the intermediate-age merger remnantNGC 1316, also known as Fornax A (e.g., Schweizer 1980;Terlevich & Forbes 2002). This galaxy is peculiar in manyways. Although by far the brightest member of the Fornaxcluster, it is not near the cluster center, being ∼ . ◦ away from the central giant elliptical NGC 1399 (projectedseparation of ∼ . Hα filaments, loops, and tidal tails originally analyzed bySchweizer (1980). Moreover, it is a powerful radio-galaxyand, to date NGC 1316 is the single galaxy with the largestnumber of discovered Type Ia supernovae (SNe Ia hereafter;four events recorded). The latter property makes NGC 1316a remarkable place to test the extragalactic distance scale,because of the role of SNe Ia for cosmological distances,and because NGC 1316 is one of the nearest massive post-merger galaxies.We have collected data covering a large wavelength in-terval (from B to H band), with the specific purpose ofperforming a self-consistent analysis of SBF data for thisgalaxy in order to carry out a comprehensive study of theSBF in NGC 1316, and consequently, of the galaxy distance(item ( iii ) above).Furthermore, given their dependence on the square ofthe stellar luminosity, SBF magnitudes are especially sen-sitive to the brightest stars at a particular wavelength andat a given evolutionary phase of a stellar population. Asshown by various authors SBF magnitudes and, in particu-lar, SBF colours, can be used to investigate the properties ofa specific stellar component in the host stellar population,depending on the observing wavelength (Worthey 1993a;Blakeslee et al. 2001; Cantiello et al. 2003; Jensen et al.2003; Raimondo et al. 2005). As an example, SBF coloursinvolving bluer bands, like B , have been used to studythe hot stellar component in unresolved systems (e.g.,Cantiello et al. 2007b). Likewise, specific phenomena likethe mass-loss rates in the AGB phase have been analyzedby taking advantage of near-IR SBF data (Raimondo 2009;Gonz´alez-L´opezlira et al. 2010). As a consequence of thequoted relation between the SBF signal and stellar pop-ulation properties, we take advantage of the broad pass-band coverage to characterize the properties of field starsin the galaxy, in order also to provide new constraints onthe formation history and evolution of the peculiar galaxyNGC 1316.The organization of the paper is as follows. In § § § § §
6, and summarize our conclusions in §
7. InAppendix A a detailed comparison of SBF and PNLF dis-tances is presented. Finally, Appendix B presents some de-tails on the SBF versus SNe Ia comparison.
Table 1.
Median ¯ m and ( ¯ m − ¯ M ) errors from the literature Sample δ ( ¯ m ) δ ( ¯ m − ¯ M ) Filter Number of sources(mag) (mag) (space/ground obs.)Optical bandsTonry et al. (2001) 0.18 0.20 I
280 (ground)Cantiello et al. (2007a) 0.02 0.09 F W ( ∼ I ) 13 (space)Blakeslee et al. (2009) 0.04 0.08 F LP ( ∼ g SDSS ) 134 (space)Blakeslee et al. (2010) 0.02 0.07 F W ( ∼ I ) 9 (space)Near-IR bandsJensen et al. (1998) 0.14 0.19 K ′
16 (ground)Liu et al. (2002) 0.07 0.20 K S
19 (ground)Jensen et al. (2003) 0.08 0.17 F W ( ∼ H ) 79 (space) Table 2.
Main properties of NGC 1316.
Alternative names Fornax A, FCC 21, Arp 154, ESO 357-G 022RA(J2000) − S0Morphological Type − ± B -band magnitude − cz (km/s, Heliocentric) 1788 ± E ( B − V ) V − I ) ± Hyperleda (http://leda.univ-lyon1.fr); Schlegel et al. (1998); Tonry et al. (2001)
2. Observations and reductions
This work is based on data of NGC 1316 from the
Very Large Telescope (VLT) and HST archives. We used i ) B , V and I -band observations obtained with theFORS1 Imager at ESO’s VLT in Paranal (Program 64.H-0624(A), P.I. M. Della Valle), and from the HST ii ) ACS F W and F LP -band data from the ACSFCS sur-vey (Jord´an et al. 2007), iii ) WFC3/IR observations inthe F W and F W filters, plus WFC3/UVIS F W (HST Program ID 11691, P.I. P. Goudfrooij).It is useful to note, for the forthcoming discussion, thatthe VLT observations were part of a project aimed at dis-covering, monitoring and characterizing the properties ofthe nova population in NGC 1316, with the specific pur-pose of deriving the distance to the galaxy using novae(Della Valle & Gilmozzi 2002). Some relevant properties ofthe target are listed in Table 2.In the remainder of this section we describe the re-duction and calibration procedures adopted for the datafrom both telescopes. In all cases we used SExtractor(Bertin & Arnouts 1996) for the source photometry, andthe IRAF/STSDAS task ELLIPSE (based on the methoddescribed by Jedrzejewski 1987) to fit the galaxy isophotes. The data reduction was carried out with the VST-Tubeimaging pipeline (Grado et al. 2004, 2012), specifically de-veloped by one of the co-authors of this work (A.G.) fordata from the VLT Survey Telescope (Capaccioli et al.2005; Capaccioli & Schipani 2011, VST). VST-Tube is a very versatile software for astronomical data analysis,tested against imaging data taken with different tele-scopes/detectors, adaptable to existing or future multi-CCD cameras (more details will be given in a dedicatedforthcoming publication, A. Grado et al., in prep.). Further,VST-Tube offers the great advantage of fully controllingeach step of the data processing.In Table 3 we report the total exposure times availablefor each VLT filter. Unfortunately, technical problems withthe camera made a fraction of the total observing time un-usable (55%, 60% and 44% of the total exposure time in B , V , and I band respectively). The VLT images downloadedfrom the archive and used for this work showed a strongdegradation during the thirteen observing runs spanningnearly two months, from December 1999 to February 2000(Figure 1). This is partly due to moon illumination, and alsoto the problem of instrument contamination. A decontami-nation process of FORS1 was performed in November 1999,and one more in January 2000 (Cavadore et al. 1999). Theeffect of the last intervention is clearly visible in a ∼ > th and, in case of the I filter, we also includedthe last two exposures of 2000 February 4 th (column “DataUsed” in Table 7). The total exposure times used after theselection for good frames are also reported in Table 3.The images were reduced as usual removing the instru-mental signatures (overscan correction, bias subtraction,flat field correction and, in the case of the I band, fringepattern removal). The resulting co-added mosaic are ap-proximately 6 . × . arcmin , a colour combined image ofthe three mosaics is shown in Figure 2 (upper panel). Fig. 1.
Left panel: Mosaic single exposure in B on 1999 December 27 th (600s total, the first observing run of the proposal).Right panel: As left, but for an exposure taken during the night of 2000 January 20 th . Table 3.
Total and used exposure times for each band. B (s) V (s) I (s)Total available 13200 9000 10800Used 6000 3600 6000 Fig. 2.
Left panel: false colour combined VLT
BV I frames. Middle panel: as upper panel, but for residual frames. Thelocation of the ACS frames is outlined in black. Right panel: ACS and WFC3/UVIS combined residual images. Colourcoding is chosen to enhance the presence of dust. The sites of three over four of the SNe Ia host by the galaxy are shown.
The VLT data were calibrated adopting theGoudfrooij et al. (2001b, G01 hereafter) photometryas reference, then solving the photometric equations ac-cordingly. G01 obtained optical photometry of the sourcesin the field of NGC 1316 using archival NTT/EMMI datain B , V , and I filters. Furthermore, the authors alsoderived JHK S photometry for eight candidate globular clusters (GCs) in the galaxy, using the IRAC2 cameramounted on the ESO/MPI 2.2m telescope.The panels in Figure 3 show the B , V and I VLTphotometry versus the G01 calibrating data. Note thatG01 adopted Galactic extinction E ( B − V ) = 0 . E ( B − V ) =0 .
021 from Schlegel et al. (1998) . The new measurements of dust reddening fromSchlafly & Finkbeiner (2011) provide ∼ .
002 mag smaller4antiello et al.: The distance of NGC 1316
Fig. 3.
Comparison between the VLT (this work, t.w.)
BV I photometry and NTT data from G01 used for calibratingFORS1 photometry.The comparisons in Figure 3 are obtained in the sameobservational conditions, prior to corrections for Galacticextinction.
HST instruments, thanks to the high resolution andthe sharp PSF, provide ideal observational datasets forSBF analysis (e.g., Ajhar et al. 1997; Jensen et al. 2001;Biscardi et al. 2008; Blakeslee et al. 2010). We analyze theoptical ACS data and provide the first near-IR WFC3/IRSBF analysis, with the specific purpose of deriving a con-sistent set of ground and space-based SBF measurementsto secure a reliable distance to NGC 1316.
We analyzed the ACS F W ( ∼ SDSS g band) and F LP ( ∼ SDSS z ) observations of NGC 1316 obtainedfor the ACSFCS survey (see Cˆot´e et al. 2004; Jord´an et al.2007, and references therein for details on the observationsand the data reduction procedures for the ACSFCS and itstwin ACSVCS survey) .Other ACS observations in the F W , F W , and F W ( ∼ BV and I , respectively) bands are also availablefrom the HST archive. The B -band data are not suitablefor SBF analysis, while the V and I band have already beenanalyzed by us (Cantiello et al. 2007a), providing results ingood agreement with the present study (see below). Suchobservations will be used to improve the mapping of thedust in NGC 1316. The image processing, including cosmic- Galactic extinction, being E ( B − V ) = 0 . ∼ .
02, for B magnitudes, and ∼ .
002 magon the colour indices used throughout the present work. Fig. 4.
Comparison between our photometry and theACSFCS results. Black dots and labels refer to the wholesample of ∼
620 matched sources, gray circles and labels tothe ∼
40 selected sources. The median difference and theassociated rms are also reported.ray rejection, alignment, and final image combination, isperformed with the APSIS ACS data reduction software(Blakeslee et al. 2003), also used for ACSFCS image align-ment.For sake of homogeneity with the other datasets used inthis work, our photometric analysis of the ACS frames wasperformed independently from the ACSFCS one. Hence, itis a noteworthy result that our g F W and z F LP surfacebrightness profiles, derived as described in the followingsection, match within < .
05 mag the ACSFCS ones fromCˆot´e et al. (2007).The situation is a bit more complex for the photom-etry of point-like sources and/or “slightly-resolved” ones.
For this comparison we used the preliminary catalog of GCcandidates from the ACSFCS team, derived as describedin Jord´an et al. (2004, 2009), and our photometric catalogobtained as described in Section 3. The results of the com-parison are shown in Figure 4. If the full sample of ∼ rms scatter. The large scatter between the catalogs isdue to the independent analysis procedures, especially inthe way slightly-resolved sources are treated. At the dis-tance of the Fornax cluster with the resolution of ACS,the GCs hosted by NGC 1316 appear slightly resolved. TheACSFCS analysis is optimized to generate accurate pho-tometry of GCs with different radii. The aperture correctionfor such sources needs to be evaluated using more refinedanalysis methods (Jord´an et al. 2004, 2009) than the onesadopted here (see next section). The gray circles in Figure4 show a selection of GC candidates with i ) ∆ m ≤ . m = 24 mag in both ACS bands, ii ) galactocentric radius ≥ ′′ , to reduce the number ofobjects highly contaminated by dust, and iii ) ACSFCSestimated radius ≤ . ′′ , to select only the most com-pact sources, for which the issue of a different treatmentof the aperture correction should not complicate the com-parison. For the selected sample of sources the matchingis significantly improved (gray symbols in the figure). The g F W data are statistically consistent in the two cata-logs, while a small 0.01 mag offset is seen in the z F LP band. However, for the purposes of the present work, suchan offset only affects the estimate of the contribution to thefluctuation amplitude due to external sources (Tonry et al.1990; Sodemann & Thomsen 1995; Blakeslee et al. 1999).Given the level of completeness of the GC catalog, and theamplitude of the offset, the impact on ¯ z F LP is negligible.As will be shown in section 3, in fact, our SBF measure-ments are in very good agreement with the ACSFCS ones(Blakeslee et al. 2009) . The F W ( ∼ J ) and F W ( ∼ H ) band images ofNGC 1316 taken with the WFC3/IR were downloaded fromthe Hubble Legacy Archive together with the WFC3/UVIS F W ( ∼ U -band) image. The near-IR images were down-loaded for the specific purpose of deriving SBF magnitudes,while the U F W has been used to improve the detectionand masking of dust over the entire set of available images.We calibrated the WFC3/IR photometry using theVEGA zero points given by Kalirai et al. (2009). To checkthe WFC3/IR photometric data, we made two differentcomparisons, both shown in Figure 5. Figure 5 (left panel)plots the WFC3/IR colour J F W − H F W versus theACS g F W − z F LP for the combined ACS-WFC3 dataset. The crosses give the full sample of sources. In orderto select the best GC candidates, the black filled circlesshow the sources with z F LP ≤ . z F LP ∼ < . ≥ ≥ ′′ . The We adopt as reference the photometric VEGA zero points,while ACSFCS results use the AB ones. All ACSFCS data, in-cluding the calibration of SBF magnitudes, are transformed tothe VEGA magnitude system using Sirianni et al. (2005) zeropoint transformations. simple stellar population (SSP) models from the TeramoStellar Population Tools (SPoT, see below) group for theage range from 3 to 14 Gyr, and [Fe/H] from − ∼ < . J − H ) G − ( J F W − H F W ) ∼ .
3. SBF measurements
To derive the photometry of sources in each of the selectedVLT and HST frames, and measure the fluctuation ampli-tudes, we used the procedures described in our previousworks (see Cantiello et al. 2011a, and references therein).The procedure is basically the same for VLT/FORS1,HST/ACS, and HST/WFC3/IR data with minor differ-ences outlined below.The main steps of SBF measurement involve: sky back-ground determination and subtraction; galaxy model andlarge scale residual subtraction; photometry and masking ofpoint-like and extended sources, including dust; power spec-trum analysis of the residual frame. We determined the skybackground by fitting the surface brightness profile of thegalaxy with a Sersic law (Sersic 1968) plus a constant term.After sky determination, a first model of the galaxy wasobtained and subtracted from the sky-subtracted frame,and a mask of the bright sources was obtained. The largescale residuals, still present in the frame after subtractingthe galaxy model, were removed using the background mapobtained with SExtractor adopting a mesh size ∼
10 timesthe FWHM (Tonry et al. 1990; Cantiello et al. 2005). Inthe following we refer to the sky, galaxy-model and largescale residuals subtracted image as the residual frame.The procedure of i ) surface brightness analysis andsky determination, ii ) model fitting and subtraction, iii )sources/dust masking, and iv ) large scale residual subtrac-tion was iteratively repeated until the residual frame ap-peared “flat” in the regions of interest for SBF measure-ments, i.e., until the residual did not show any (local) arti-fact due to the subtracted galaxy model. The middle panelof Figure 2 show the false colour combined BV I image ofthe residual frames. The right panel of the figure, instead,shows a combination of the ACS and WFC3/UVIS frames,used to map the dust around the center of the galaxy. Thepositions of three of the four SNe Ia host in the galaxy isalso indicated in the figure (the region of SN 1980D is notcovered by either the HST or VLT frames; we adopted therevised SNe Ia coordinates from Stritzinger et al. 2010).The surface brightness profiles for all bands analyzed,as well as the difference between modeled and observedprofiles, are shown in Figure 6.The photometry of fore/background sources and of GCswas derived running SExtractor on the residual frames. Fig. 5.
Left panel: colour-colour diagram with the ACS and WFC3/IR magnitudes. Gray crosses mark the full sampleof common sources, black circles show selected GC candidates (see text). Right panel: Comparison between WFC3/IRphotometry and the IRAC2 data from Goudfrooij et al. (2001a). In both panels the predictions from SPoT simple stellarpopulation models with different metallicity are shown with different colour/line styles ([Fe/H]= − − − − [See electronic version of the Journal for a colour version of the figure.] As described in our previous work, we modified the in-put weighting image of SExtractor by adding the galaxymodel (times a factor between 0.5 and 10, dependingon the expected amplitude of the SBF signal; for detailssee Jord´an et al. 2004; Cantiello et al. 2005) so that theSBFs were not detected as real objects. The aperture cor-rection was obtained from a number of isolated point-source candidates in the frames and by making a curveof growth analysis out to large radii (Cantiello et al. 2009,2011b). The outer radius used for FORS1 data was 6 . ′′ . ′′ . ′′
6) and then added anextra aperture correction term to infinite radius by us-ing the instrument encircled energy tables (for ACS weused Sirianni et al. 2005, while for WFC3 we adopted theInstrument Handbook, version 4)Once the catalog of sources was derived, the next stepwas to fit the luminosity function of the sources, to beused to estimate the already mentioned background fluctu-ation term due to unmasked faint sources. We obtained thefit to the GC and background galaxy luminosity functionsfrom the photometric catalog of sources, after removing thebrightest/saturated point-like sources and the brightest andmost extended objects. The best fit to the sum of the twoluminosity functions, and the background fluctuation cor-rection term, P r , were derived as in Cantiello et al. (2005).To measure SBF magnitudes we estimated the az-imuthal average of the residual frame power spectrum, P ( k ), then matched it to the power spectrum of a templatePSF convolved with the mask image, E ( k ). The total fluctu-ation amplitude P was obtained via a robust minimizationmethod (Press et al. 1992) as the multiplicative factor in the power spectrum representation P ( k ) = P × E ( k ) + P ,where P is the white noise constant term. We used one tofive different isolated bright point sources in each residualframe for the template PSFs. Each PSF, after normaliza-tion, was singularly adopted to estimate the SBF signal ofthe galaxy. Finally, the SBF amplitude, P f = P − P r , wasestimated within circular annuli. The results of the powerspectrum analysis are summarized in Figure 7, with oneexample for each band.The results of the SBF and colour measurements for allbands considered are reported in Table 4. For each filter,in addition to the statistical error, we report the system-atic uncertainty due to PSF fitting. For H F W , since wecould only find one good candidate PSF in the frame, weassumed a conservative PSF scatter of 0.2 mag . In thecase of J F W no good PSF was found over the WFC3/IRframe, thus we used a PSF star taken from different ob-servations associated with the same HST proposal. As for H F W we assumed 0.2 mag PSF uncertainty.
4. The SBF distance to NGC 1316
The estimate of distances in the SBF method relies onknowledge of the absolute SBF magnitudes. Using the mea-surements reported in Table 4, together with empirical andtheoretical calibrations given in Table 5, we obtained thedistance moduli reported in the ( m − M ) column of thetable. In particular we obtained a mean distance modulus Previous studies indicate SBF variations < . V-6
Fig. 6.
Upper left panel: U F w /WFC3/UVIS, BV I /FORS1, g F W z F LP /ACS, and J F W H F W /WFC3/IRsurface brightness profiles for NGC 1316. Each band is shown with a different line style/colour. A vertical shift is appliedfor sake of clarity as labeled. Lower left panel: As upper left panel, but in R / scale. The Sersic profile fitting procedureused leaves the n -index free to vary. Here we use the median n = 6 value, which is also consistent with Cˆot´e et al. (2007).The full dots are extrapolated from Carter et al. (1983), accordingly shifted for each band (see colour image). Rightpanels: the difference between observed and fitted surface brightness profiles. [See electronic version of the Journal for acolour version of the figure.] of 31.59 ± ± The empirical calibrations of SBF magnitudes in opticalbands, in particular in the I band, are the most thor- oughly analyzed (Tonry et al. 1990, 2001; Mei et al. 2007;Blakeslee et al. 2010). The two aforementioned HST sur-veys of the Virgo and Fornax clusters provided an extremelyaccurate calibration of ¯ z F LP , including some degree ofnon-linearity in the calibration. Some debate still exists onnear-IR bands calibrations (see Gonz´alez-L´opezlira et al.2010, and references therein), although, as already men-tioned, relevant progress will be done thanks to the instal-lation of the WFC3/IR (Blakeslee 2012; French et al. 2012). Fig. 7.
Power spectrum analysis of the
BV Iz F LP J F W H F W frames. Each panel shows a different band, as labeled.For all bands, the upper panels show the logarithm of the power spectrum of the residual frame (gray dots) and the bestfit P ( k ) curve. In the middle panels the difference between observed and fitted power spectra is shown. The flat regionof log P ( k fit > k ) between vertical dashed lines (lower panels) is used to evaluate the best fit parameters P , and P . Table 4.
Surface Brightness Fluctuation and colour measurements corrected for galactic extinction.
VLT data h r i B − V V − I ¯ B ¯ V ¯ I (arcsec) (mag) (mag) (mag) (mag) (mag)170 0.84 ± ± ± ± ± h r i g − z ¯ z (arcsec) (mag) (mag)80 1.954 ± ± h r i V − I a J − H ¯ J ¯ H (arcsec) (mag) (mag) (mag)80 1.102 ± ± ± ± a ] Colour data obtained from VLT data using the same masks adopted for WFC3/IR measurements. In the upper part of Table 5 we report the distancemoduli obtained using the empirical calibrations taken fromliterature, together with the adopted calibrations.As a first general comment on the empirical equations, we must emphasize that the numbers reported in the ta-ble are all tied to the same common zero point, i.e., tothe Cepheid distances with metallicity correction to the PLrelation from Freedman et al. (2001).
The resulting zeropoints of the ¯ V , ¯ I and ¯ H F W versus V − I calibrationsare shifted of +0 .
06, +0 .
06 and − .
10 mag with respect tothe calibrations in the original papers (see appendix A inBlakeslee et al. 2010).For the ¯ I versus B − I calibration, taken fromCantiello et al. (2005), we do not make any revision since the zero point is already based on the chosen set ofCepheids. Similarly, the z F LP -band calibration does notneed any change (Blakeslee et al. 2009).For the H F W -band distance estimate, we used twoindependent calibrations. The first derived by Jensen et al.(2003) from HST/NICMOS data. We adopted the SPoTSSP models to evaluate the changes in the calibrationdue to the difference between the NICMOS2 and theWFC3/IR H F W passbands. The result is that theWFC3/IR ¯ H F W zero point is 0.2 mag fainter than theNICMOS2 H F W one. This is partly expected becauseof the cut at larger/redder wavelengths of the WFC3/IRfilter (1400-1700 nm passband, versus 1400-1800 nm forNICMOS2). Hence, we add a 0.2 mag to the zero point of Table 5.
Distances from SBF measurements.
Passband ¯ M calibration eq. ( m − M ) ReferenceEmpirical calibrations V (0 . ± .
12) + (5 . ± . V − I ) − .
15] 31.53 ± I ( − . ± .
08) + (4 . ± . V − I ) − .
15] 31.65 ± I ( − . ± .
1) + (3 . ± . B − I ) − .
0] 31.55 ± z − x +2.60 x +3.72 x , x ≡ ( g − z ) − .
94 31.66 ± J F W ... ... ... H F W ( − . ± .
1) + (5 . ± . V − I ) − .
16] 31.3 ± H F W − .
17 + 0 . x + 2 . x , x ≡ ( g − z ) − .
94 31.4 ± ± V .
89 + 4 . V − I ) − .
15] (0.3) 31.4 ± I − .
63 + 5 . V − I ) − .
15] (0.3) 31.6 ± I − .
63 + 2 . B − I ) − .
0] (0.3) 31.6 ± z − .
77 + 2 . g − z ) − .
94] (0.2) 31.8 ± J F W − .
75 + 3 . V − I ) − .
16] (0.3) 31.3 ± H F W − .
86 + 3 . V − I ) − .
16] (0.3) 31.4 ± ± (1) Blakeslee et al. (2001) (2)
Tonry et al. (2001) with revised Cepeheid distances (3)
Cantiello et al. (2005) (4)
Calibration using Blakeslee et al. (2009), uncertainties evaluated from eq. (1) in Mei et al. (2007) (5)
Calibration from Jensen et al. (2003) with metallicity correction on Cepheids and NICMOS to WFC3 H F W zeropointcorrection (see text). (6) Calibration from Cho, H. et al. (2013). the Jensen et al. (2003) empirical relation, assuming a de-fault 0.1 mag uncertainty because of the model-dependentcorrection term. The second calibration is a preliminaryresult obtained by Cho, H. et al. (2013), based on the ob-servations of 16 early-type galaxies in Virgo and Fornaxspecifically obtained to empirically calibrate the SBF forthe WFC3/IR passband. For the near-IR data, in contrastwith the optical measurements, the distance modulus in-cludes the systematic PSF uncertainty since it is dominantwith respect to the statistical errors of the SBF measure-ment and of the calibration zero point.All distances based on the empirical calibrations re-ported in Table 5 agree with each other within the quoteduncertainties. The weighted mean of distance moduli is alsogiven in the table.
Various authors have analyzed the possibility to calibrateabsolute SBF magnitudes using stellar populations synthe-sis models, thereby making it a primary distance indica-tor, not linked to the Cepheids zero point (Buzzoni 1993;Worthey 1993b; Blakeslee et al. 2001; Biscardi et al. 2008).In this work we have taken as reference the SBF versuscolour equations derived using the simple stellar popula-tion models from the Teramo SPoT group. For a detailedreview of the SPoT models we refer to Raimondo et al.(2005) and Raimondo (2009), and references therein. Thesemodels have been shown to be very effective in match-ing the empirical SBF calibration in different bands, aswell as in reproducing the resolved (colour magnitude di-agrams) and unresolved (colours, magnitudes) propertiesof stellar populations (Brocato et al. 2000; Cantiello et al. 2007a; Cantiello 2012). We used the updated version of theSPoT models, which for the photometric bands and chem-ical composition used in this section confirms the resultsobtained from the previous Raimondo et al. (2005) models(G. Raimondo, private communication). The grid of mod-els used has [Fe/H]= − § M versus colour re-lation is also tabulated. The uncertainties on distance mod-uli are derived by summing in quadrature the scatter of thetheoretical calibration and the uncertainty on ¯ m .The distance moduli derived with the theoretical cali-brations from the SPoT models are given in the Table 5. Allreported distances agree to within the quoted uncertainties. The SBF distance of NGC 1316 presented in § § V to H F W , i.e.,a wavelength interval that, in terms of SBF and colours,samples very different stellar population (sub)components. Fig. 8. V , I and H F W -band SBF versus ( V − I ) pre-dictions from five independent stellar population synthe-sis codes: solid line (black), dotted line (red), dashedline (blue), long-dashed line (green), dot-dashed (cyan)line show the updated SPoT, Blakeslee et al. (2001),Liu et al. (2002), Mar´ın-Franch & Aparicio (2006), andGonz´alez-L´opezlira et al. (2010) best fit lines, respectively.Empty (magenta) circles show the corresponding empiricalrelations, also given in Table 5. [See electronic version ofthe Journal for a colour version of the figure.] It is noteworthy that the empirical and theoretical evalu-ations agree very well with each other. Taking into accountthat the two methods are based on independent calibrationprocedures, subject to different types of systematic and sta-tistical uncertainties, this result suggests that both types oferrors are reasonably well constrained.For what concerns the systematic errors in our measure-ments, summing up all expected sources of uncertainty foreach one of the three instruments considered – filter zeropoint, data reduction, calibration zero point, PSF normal-ization – the expected systematic uncertainty is ∼ < ∼ < . .Estimating the systematic uncertainty in theoreticalSBF calibrations is not a simple task. One way to getsuch an estimate would be to change the ingredients inthe SSP code (stellar tracks, initial mass function, atmo- A further component to the systematic error comes frombandpass mismatch with various telescopes. Blakeslee et al.(2001) presented a discussion of this issue (see their Sect. 5.5,Fig. 15 in particular), showing that at the colour of the SBF,( ¯ V − ¯ I ) ∼ . I and the HST I can vary by ± .
02 mag. sphere models, etc.), and then analyze the effects on SBF.A rough (indirect) estimation of the uncertainty on the-oretical calibrations is to compare the results from inde-pendent models, obtained from different SSP codes relyingon independent physical input and algorithms. A first at-tempt along these lines was carried out by Cantiello et al.(2003). To estimate this uncertainty for the present work,we compare SSP model predictions from Blakeslee et al.(2001), Liu et al. (2002), Mar´ın-Franch & Aparicio (2006),Gonz´alez-L´opezlira et al. (2010) and the SPoT models. Forall these models, we obtain linear fits to the V , I and H F W SBF amplitudes versus ( V − I ) colour (Figure 8).The empirical calibration for each band is also shown inthe figure. Note the different range of colour used for eachequation, depending on the range of validity of the em-pirical equation (Tonry et al. 2001; Blakeslee et al. 2001;Jensen et al. 2003; Cantiello et al. 2007a). We find that ascatter of ∼ m − ¯ M ) empirical = 31 . ± . . ) ± . . )mag and ( ¯ m − ¯ M ) theoretical = 31 . ± . . ) ± . . )mag, we conclude that our best estimate of the distance ofNGC 1316 is ( ¯ m − ¯ M ) = 31 . ± . . ) ± . . )mag, or d = 20 . ± . . ) ± . . ) Mpc.
5. Comparison with distances from the literature
As already mentioned, the VLT observations used here wereobtained to detect and study the population of novae inNGC 1316, and to use them to derive the galaxy distance.Della Valle & Gilmozzi (2002), using the data of four de-tected novae (the first detected beyond the Virgo clusterat the time), and the Buscombe-de Vaucouleurs relation(Buscombe & de Vaucouleurs 1955; Capaccioli et al. 1989,1990) set an upper limit on the distance of the galaxy equalto 22 . Ajhar et al. (2001), and later Cantiello et al. (2011a), havepresented a comparison of SBF and Type Ia SNe distancesfor a total of 15 different SNe Ia in 14 galaxies. Both stud-ies found excellent overall agreement between the two dis-tance indicators, including the case of NGC 1316, providedthat a consistent set of Cepheid-based distances is used.Similarly, Freedman et al. (2001) and Freedman & Madore(2010) derive essentially identical values of H from thesetwo methods when calibrated consistently via Cepheids.NGC 1316 has been a prolific producer of Type Ia su-pernovae, with four recorded events: SN 1980N, SN 1981D,SN 2006dd, and the fast declining SN 2006mr. All SN Ialight-curves have been used to determine the distance ofthe host galaxy using several methods. Furthermore, beingone of the nearest bright galaxies with well sampled SNe Ialight-curves, NGC 1316 is frequently used in calibratingsamples for high-redshift SNe Ia (e.g., Jha et al. 2007;Burns et al. 2011). Based on the MLCS SN Ia distancemethod (Riess et al. 1998), Goudfrooij et al. (2001a) re-ported ( m − M ) = 31 . ± .
05 mag for NGC 1316 and con-cluded that it was therefore ∼ .
25 mag more distant thanthe rest of the Fornax cluster, for which Ferrarese et al.(2000) gave a mean distance modulus of 31.54 mag fromCepheids and other indicators. However, no details on thecalibration, etc., were given by Goudfrooij et al. (2001a).Ajhar et al. (2001) reported both MLCS and ∆ m dis-tances (Phillips 1993; Hamuy et al. 1996) for SN 1980Nin NGC 1316 under different calibrations. Rescaling theirresults to H = 72 km s − M pc − gives ( m − M ) =31 . ± .
15 and ( m − M ) = 31 . ± .
10 from the MLCSand ∆ m methods, respectively. Interestingly, the ∆ m value coincides exactly with the mean SBF distance for theFornax cluster by Blakeslee et al. (2009).On the other hand, Stritzinger et al. (2010, Str10 here-after) have recently reanalyzed the SN Ia distance toNGC 1316, and obtained ( m − M ) = 31 . ± .
03 (stat.) ± .
04 (sys.) mag, based on the analysis of the three “nor-mal” SNe Ia (adopting H = 72 km s − M pc − , see Table6). The authors also obtain values of ( m − M ) ∼ . − . ∼ H = 72 km s − Mpc − . In the table wedo not report the systematic errors, which are not givenby all authors. Taking the weighted mean of all measure-ments in Table 6, except those of Str10, adopting 0.4 magdefault uncertainty where no error is reported, we obtain( m − M ) = 31 . ± .
05 mag. Although such value agreeswith our estimates better than the Str10 result, we cautionthe reader against these “general” averages, reported hereonly to emphasize the relative distribution of distances withrespect to a reference point. Some of the values reportedin Table 6, in fact, are obtained using the same meth-ods/objects but under different assumptions (e.g. MLCSand MLCS2k2), and/or with different calibrators, so thatthe reported average does not necessarily have a correctphysical meaning.Str10 derives the distance to NGC 1316 using three dif-ferent methods: the EBV, the Tripp method, and the near-IR light-curves of Type Ia SNe. Each method has its owncalibration, and is quantitatively independent from the oth-ers. Hence, there can be various possible causes for the dif-ference between our and Str10 distances. First, for the EBVmethod, the authors adopt negligible internal extinction forall four SNe Ia, but also comment that the spectroscopicanalysis provides results “totally inconsistent with the lowhost galaxy reddening” (Str10, Section 4). However, the op-tical and optical/near-IR colours of these objects are con-sistent with minimal to unreddened supernova (Figure 13of Str10). Second, for the near-IR method, which is intrinsi-cally much less affected by the host internal extinction, wefind that the calibration used, from Krisciunas et al. (2009),is based in part on a compilation of Cepheid and/or SBFdistances that is not internally consistent (see the discus-sion in Appendix B).Interestingly, Str10 obtain a SN Ia distance modulusfor the Fornax cluster member NGC 1380 of ( m − M ) =31 . ± .
008 mag, which is consistent with the SBF resultfor the same galaxy of 31 . ± .
075 mag (Blakeslee et al.2009), and very similar to the SBF distance for NGC 1316.In fact, there is little significant variation in the SBF dis-tances among the magnitude-limited sample of 43 early-type Fornax galaxies studied by Blakeslee et al. (2009), andmost are consistent with the mean Fornax SBF modulus of31 . ± .
03 mag .Finally, we must note that the Tripp, the maximumnear-IR magnitudes, and the MLCS2k2 methods appliedto the fast declining SN 2006mr give distance estimates inagreement with the ones presented here. In spite of this,it should also be emphasized that the debate on whetherfast declining SNe Ia can be used to determine precise dis-tances is still open (so that different authors include or notthese objects in their final samples; e.g. Jha et al. 2007;Folatelli et al. 2010; Burns et al. 2011, Str10). The recent Type Ia supernova SN 2012fr (Childress et al.2012) occurred in NGC 1365, a giant barred spiral galaxyin the direction of the Fornax cluster with a measuredCepheid distance (Silbermann et al. 1999; Freedman et al.2001). However, from a comparison of Cepheid and SNe Ia dis-tances, Suntzeff et al. (1999) suggested that NGC 1365 is ac-tually ∼ . Table 6.
Type Ia Supernova distances to NGC 1316 available in the literature.
SN ( m − M ) Method References/NotesMeasurements from Str1080N+81D+06dd 31.180 ± ± ± m − M ) max Near-IR Author’s best estimateSN 2006mr 31.83 ± ± m − M ) max Near-IR Flagged as doubtfulOther Measurements a SN 1980N 31.45 M max Hamuy et al. (1991)SN 1981D 31.35 M max Hamuy et al. (1991)SN 1980N 31.44 ± ± ± ± m Ajhar et al. (2001)SN 1980N 31.35 M max Reindl et al. (2005)SN 1981D 30.98 M max Reindl et al. (2005)SN 1980N 31.38 ± ± ± ( a ) All distances are converted to a scale with H = 72 km s − Mpc − . Distances to NGC 1316 derived from the properties of theglobular cluster system have generally been flagged as un-reliable by the authors due to the peculiar properties of thegalaxy and its GC system.G´omez et al. (2001) made one of the first attemptsto constrain the distance to NGC 1316 with the GlobularCluster Luminosity Function (GCLF, Harris 2001), usingESO/EFOSC2 data. The
BV I weighted average distancemodulus they provided is ( m − M ) ∼ . ? had noted that overall the GCLF ofNGC 1316 was not well fitted by a Gaussian (reduced χ n >
3, as compared to χ n ≈ “NGC 1316 is [...] not included in the fits becausethe observed GC system in this galaxy is highly influencedby its interaction and proximity with its satellite galaxies,and therefore our GCLF fit is not reliable.” Masters et al. (2010) use the GC-radii method(Jord´an et al. 2005) to analyze the distances of FornaxCluster galaxies. However, the authors warn that, whilethe method is effective for typical GC systems, it cannot beapplied to NGC 1316 due to the large number of extendedGCs.In conclusion, the GC–based distances reported, for ex-ample in the NED archive, ( m − M ) GCLF,g = 33 .
55 mag,( m − M ) GCLF,z = 33 .
68 mag, and ( m − M ) GC − radius =30 . ± .
11 mag, are excluded from the comparison.
Feldmeier et al. (2007) reported a PNLF distance mod-ulus to NGC 1316 of ( m − M ) = 31 . +0 . − . mag. Thisresult relies on the PNLF calibration by Ciardullo et al.(2002), whose zero point is tied to the Freedman et al.(2001) sample of Cepheid’s distance moduli with no de-pendence on metallicity of the PL relations. As discussedabove ( § M ∗ = − . ± .
05 to M ∗ = − . ± .
04 mag (exter-nal scatter σ = 0 .
16 mag). Feldmeier et al. (2007) actuallyused M ∗ = − .
47; thus their PNLF distance modulus forNGC 1316 becomes ( m − M ) = 31 . +0 . − . mag .Both PNLF distances given above agree within quoteduncertainties with the Str10 distance. However, the SBFand the updated PNLF distances are consistent withinthe given statistical and systematic errors, notwithstandingthat the difference between them remains non-negligible. Inother words, within the given uncertainties, the SBF dis-tance is consistent with the PNLF, and the latter with theSNe Ia, but SBF and SNe Ia are not consistent with eachother. Moreover, with the revised zero point that correctsfor the metallicity dependence of the Cepheids, the otherPNLF distances published in the past decade for Fornaxgalaxies are ( m − M ) = 31 . +0 . − . mag for NGC 1380(Feldmeier et al. 2007) and ( m − M ) = 31 . ± .
18 magfor NGC 1344 (Teodorescu et al. 2005). Thus, although thePNLF method finds systematically lower distances to boththe Virgo and Fornax clusters, it is consistent with SBFin placing NGC 1316 well within the distance range of theother Fornax cluster galaxies. Further details on the SBF-PNLF comparison are given in Appendix A. We have not taken into account a further − m − ¯ M ) ∼ .
55 mag. 13antiello et al.: The distance of NGC 1316
6. SBF and integrated colours to constrain stellarpopulation properties
The comparison of model predictions with observed SBFmagnitudes and SBF colours to understand the prop-erties of the host galaxy has already been successfullyused by different authors (Tonry et al. 1990; Buzzoni 1993;Jensen et al. 2003; Raimondo et al. 2005; Cantiello et al.2007b; Buzzoni & Gonz´alez-L´opezlira 2008).The SBF colour versus integrated B − V for NGC 1316,plus two more Virgo cluster galaxies, NGC 4374 andNGC 4621 (data from Cantiello et al. 2011a), are shown inFigures 9 and 10. The data in the figures are compared toSPoT SSP models computed with standard assumptions fordifferent metallicity and age (panels a − c ), while the pan-els from d to n show the predictions for solar-metallicitystandard and non-standard models: specifically, i ) in pan-els d − f SSPs with an enhanced hot horizontal branchcomponent (HHB, having ∼
50% stars in the canonical HBand ∼
50% HHB stars) are considered; ii ) in panels g − i results obtained by adding a fraction of very young stars of30 or 100 Myr to the old, solar [Fe/H] component (in pro-portion 1:1000) are plotted; and iii ) in panels from l to n predictions for models with a mix of an old solar metallic-ity component, and a further one with [Fe/H]= − − − − c in Figure 9 clearly shows the well-knownage-metallicity degeneracy that affects classical integratedcolours (e.g., Worthey 1994). The position of galaxies in thispanel overlaps nicely with SSP models, though nothing canbe said about the stellar content. Because of the overlapbetween models, the field stellar component in NGC 1316could be either older and more metal poor or younger andmore metal rich than the other two galaxies. In contrast,the SBF colour versus B − V models shown in panels a − b are much less affected by the degeneracy, especially for the¯ V − ¯ H F W colour. The data for the three galaxies lie nearthe region of solar metallicity models. A possible interpre-tation of the relative positions of the three galaxies in the¯ V − ¯ H F W and ¯ V − ¯ I versus ( B − V ) colour planes, is thatNGC 1316 hosts a field component that is as metal rich as inNGC 4621, with [Fe/H] ∼ ≤ t ( Gyr ) ≤ l − n we find that, while the colour-colour panel n shows the expected age-metallicity degener-acy, the situation changes for SBF colours (panels l − m ).The Virgo cluster galaxies, in fact, overlap with the regionof models obtained with the mixing of old t=14 Gyr SSPswith different [Fe/H]s. In both the l and m panels NGC 1316lies above the line of mixed old SSPs, suggesting that ayounger SSP is necessary to obtain a good match withthe models. Again, this is not surprising since NGC 1316 isa known example of an intermediate-age merger remnant,and also has Mg ∼ .
25 mag, while the other two targetshave Mg > .
28 mag.The data to models comparison is less straightforwardwhen B -band SBF magnitudes are considered. As discussedby other authors (e.g., Worthey 1993a; Cantiello et al.2007b), B -band SBF cannot be used to get reliable galaxydistances, both because the amplitude of the signal is 2-3 orders of magnitude fainter than in optical/near-IR bands,and because of the strong sensitivity to stellar populationproperties. This is depicted in Figure 10 (panels a − c ),where we plot SBF colours obtained with ¯ B versus the B − V for the three galaxies, and the standard SPoT mod-els. One major difference with the results in Figure 9 isthe substantial mismatch between data and models seen inpanel b and, especially, in panel c . The ¯ B − ¯ V versus B − V models are clearly affected by the age-metallicity degener-acy, but in this case the data are offset by ∼ . B − ¯ H F W colour, due to the much larger baseline of thiscolour.If non-standard SPoT SSP models are taken into ac-count (panels from d to n in Figure 10) we notice that: – models with an enhanced number of hot HB stars pro-vide a good match to the data; – the mean metallicity of the dominant stellar componentin NGC 1316 is apparently lower than that of the twoVirgo galaxies. The models shown in panels d − f , infact, are obtained assuming HHB “enhancement” forthree different ages and solar metallicity. Hence, takinginto account SSP models with lower [Fe/H], i.e., bluercolours, the matching with the position of NGC 1316will improve; – the presence of a diffuse very young stellar component,with an age of 30 Myr, seems to reduce the mismatchwith data (panels g − i ). However, even in the case of themerger remnant NGC 1316, the presence of such a youngdiffuse stellar component is unlikely even if the pres-ence of a relatively young stellar population is expectedin strong radio-emitter galaxies (Della Valle & Panagia2003); – the improved data and models matching in panel l seemsto support the possibility of an old [Fe/H] mixed stellarcomponent. However, this is ruled out by the compar-ison in panel n , showing a poor match to the ¯ B − ¯ V predictions for the same data set.At this stage it is useful to recall that the HHB scenariois supported by other independent observations. First, wemention the results by Brown et al. (2000) and Brown et al.(2008) who found a significant fraction of HHB stars inM 32 using HST U V data. Second, the puzzling pres-ence of a strong
U V emission in some regular early-typegalaxies, discovered several decades ago (Code et al. 1972;Bertola et al. 1980), is now widely interpreted as the pres-ence of an old hot stellar component. Although the mech-anisms regulating this component are not well understood(Park & Lee 1997; Kaviraj et al. 2007; Han 2008), some ofthese old hot stellar sources may have effects on ¯ B , as in-dependently predicted by various SSP models (Worthey1993a; Cantiello et al. 2003), or based on empirical evi-dence (Shopbell et al. 1993; Sodemann & Thomsen 1996;Cantiello et al. 2007b).We emphasize that, when computing the SBF calibra-tion equations using the models with HHB, we find that ¯ B brightens by ∼ ∼ < . z ). Such be-havior, again, highlights the uselessness of ¯ B for distancedeterminations, and confirms the interest on SBF in bluebands for analyzing the properties of unresolved blue hotstellar components. Fig. 9.
Panels a − c : SPoT standard SSP models compared with SBF measurements of NGC 1316 (full black circle). Thedata of NGC 4374 (filled square), and NGC 4621 (filled triangle) from Cantiello et al. (2011a) are also shown. Differentline styles mark different [Fe/H] contents: dot-dashed (magenta), long-dashed (green), solid (blue) and dashed (drown)refer to [Fe/H]= − . , − . , . . a indicates the direction of increasing age. Panels d − n : SPoT models obtained with non-canonical assumptions.In all cases the initial (reference) population has solar metallicity, t ∼
14 Gyr, and is shown with a blue star, while thefinal composite population is shown with a different colour and connected with a dotted line to the reference model.The symbols for observational data are the same as in panels a − c . Panels d − f : A fraction equal to 50% of the totalHB-stars is simulated being HHB. For these models three ages are considered (t=10, 12 and 14 Gyr, increasing ages aremarked with larger symbols). Panels g − i : A young population of t=30 Myr (red), or t=100 Myr (magenta) is added tothe old solar one. The fraction in mass of old to young stars is reported in the lower panel. Panels l − n : old SSPs withvarious [Fe/H] are mixed to the solar one, as labeled. The mass fraction metal-poor to standard SSPs is shown in thelower panel. [See electronic version of the Journal for a colour version of the figure.] In conclusion, the present analysis of the stellar popula-tion properties for NGC 1316 seems to confirm the resultsof NGC 4374 and NGC 4621, i.e., that a diffuse componentof hot old stars contributes to the SBF signal in the B -band. Furthermore, the relative comparison of SBF andcolour data of NGC 1316 with those of the two galaxies inVirgo seems to indicate that the dominant stellar compo-nent in NGC 1316 is younger and slightly less metal rich,as expected for an intermediate-age merger remnant.
7. Summary
We have measured SBF magnitudes in NGC 1316, thebrightest galaxy in the Fornax cluster, using ground-basedVLT/FORS1 data in
BV I -bands, and space-based g F W and z F LP from ACS/WFC, plus J F W and H F W WFC3/IR observations.The distance of NGC 1316 is particularly interesting inthe context of the cosmological distance scale. The Fornaxcluster, in fact, is the second largest cluster of galaxieswithin ∼ <
25 Mpc after the Virgo cluster. However, in con-trast to Virgo, the line-of-sight depth of Fornax is small,enabling accurate calibration of distances without the ad-
Fig. 10.
Same as in Figure 9, but ¯ B is considered instead of ¯ V . [See electronic version of the Journal for a colour versionof the figure.] ditional scatter intrinsic to the spatial extent of the clus-ter. Furthermore, NGC 1316 is among the galaxies with thelargest number of detected Type Ia supernovae (SN 1980N,SN 1981D, SN 2006dd and SN 2006mr). For this reason, itis a unique place to test the consistency of SNe Ia distances,both internally and against other distance indicators.Using our SBF measurements in V Iz F LP H F W and available empirical calibration of the absolute ¯ M , weobtained a weighted mean distance modulus to NGC 1316( ¯ m − ¯ M ) empirical = 31 . ± . . ) ± . . ) mag.Additionally, we obtained SBF distances from V Iz F LP J F W H F W data based on theoreti-cal calibrations derived from the SPoT SSP models.The resulting weighted mean distance modulus is( ¯ m − ¯ M ) theoretical = 31 . ± . . ) ± . . ) mag.The good agreement between SBF distances obtainedfrom empirical and theoretical calibrations is a notable re-sult. The two classes of calibration are, in fact, completelyindependent of each other, one relying on the first two rungsof the cosmic distance scale, the other on the present knowl-edge of the various ingredients that go into stellar popula-tion synthesis (stellar evolution theory, stellar atmospheres,etc.). Furthermore, since SBF magnitudes over such a wide range of wavelengths depend on the properties of stars indifferent evolutionary stages, even the agreement betweendistance moduli obtained in different bands should be re-garded as a remarkable result particularly in terms of thehigh degree of reliability reached by SSP models.By combining the distance moduli from the two typesof calibrations, we obtain ( ¯ m − ¯ M ) = 31 . ± . . ) ± . . ) mag, or d = 20 . ± . . ) ± . . ) Mpc.This distance modulus agrees generally well with es-timates obtained from other indicators, and with SNe Ialight-curves analysis obtained before 2010. A non-negligibledifference exists with the most recent analysis based onSNe Ia by Stritzinger et al. (2010), who obtained a bestestimate for the distance ∼
17% smaller than ours. Thepossible sources of the disagreement may be related to thecomplex issue of the internal extinction, and to zero pointcalibration issues of both distance indicators.The comparison to PNLF is also subject to the linger-ing problem of zero point calibration. When placed in aCepheid distance scale consistent with ours, the PNLF dis-tance to NGC 1316 is d = 18 . ± . . ) ± . . ) Mpcand agrees within the quoted errors with our SBF distance.However, the difference between SBF and (the updated) PNLF distance remains non-negligible, and lowering theuncertainties (especially systematic; see Appendix A)would be a desirable result for both indicators.In order to analyze the properties of the dominant stel-lar component in the galaxy, we compared SBF colours andintegrated colours to SSP model predictions. We found thatthe stellar light of the galaxy seems to be dominated by a[Fe/H] ∼ < . B -bandSBF data are included in the model comparison. As inthe cases of NGC 4374 and NGC 4612, which showed asimilar mismatch to models, we used the SPoT stellarpopulation synthesis code to generate SSP models withnon-canonical properties. In particular, we considered thefollowing three cases: starting from an old t ∼
14 Gyrpopulation with solar metallicity we have 1) enhanced thecontent of hot HB stars, 2) added a very young diffusesecondary component, and 3) added a more metal poorSSP. As in the previous case (Cantiello et al. 2011a), thesimulations seem to favor the HHB component scenario.Assuming a contribution to ¯ B from such hot HB starsremoves the discrepancy between the data and models inthis band, yet has negligible effect on SBF in other bands,i.e., it does not affect the theoretical calibrations used toobtain distances.Our results on the distance and stellar population prop-erties of NGC 1316 based on SBF analysis have shown that,despite the great progress in recent years, many issues re-main open on both topics. Concerning distances, the cali-bration of distance indicators, and the treatment of errorpropagation in the distance scale, still need to be accuratelyand consistently analyzed. Concerning stellar populations,SBF colours, as a new and independent stellar populationanalysis technique, seem to provide useful constraints tothe properties of field stars, hidden to many classical pho-tometric indicators. Acknowledgements.
Part of this work was supported by PRIN-INAF2010 (P.I.: G. Clementini), PRIN-INAF 2011 (P.I. A. Grado) andPRIN-INAF 2011 (P.I.: G. Marconi), and the FIRB-MIUR 2008 (P.I.G. Imbriani). We are grateful to M. Capaccioli, E. Di Carlo, and I.Biscardi for useful discussions related to this work.This research has made use of the NASA/IPAC ExtragalacticData-base (NED) which is operated by the Jet Propulsion Laboratory,California Institute of Technology, under contract with the NationalAeronautics and Space Administration. This research has also madeuse of the SIMBAD database, operated at CDS, Strasbourg, France,and of the HyperLeda database (http://leda.univ-lyon1.fr).
Appendix A: Some musings on PNLF and SBFdistances
One of the most intriguing issues in the extragalactic dis-tance scale is the ∼ +0 . H obtainedfrom SNe Ia was 10% higher when the SN Ia distances werecalibrated via PNLF distances, as compared to the H ob-tained by calibrating the SN Ia distances by either SBF ordirectly from Cepheid distances.As discussed in detail in Section § §
5, to reli-ably compare two or more distance indicators, it is ofparamount importance to verify the consistency of the cal-ibrations (or calibrators) used. Both PNLF and SBF cali-brations are tied to the same primary indicator, the period-luminosity relation of Cepheids, and to the same dataset,i.e., the Cepheids from Freedman et al. (2001). However,the zero points of present SBF calibrations are tied to theCepheid distances obtained with metallicity-dependent PLrelations ( D P LZ hereafter; Mei et al. 2007; Blakeslee et al.2009, 2010, this paper), while the standard PNLF calibra-tion relies on Cepheid distances with no dependence onmetallicity ( D P L ; Ciardullo et al. 2002; Feldmeier et al.2007; Ciardullo 2012).In this appendix, we take the detailed discussion pre-sented in Ciardullo et al. (2002) – who found the ∼ . D P LZ . For PNLF, instead, Ciardullo (2012) re-ported a +0.07 mag correction to the PNLF distance mod-uli (0.07 mag brighter zeropoint, M ∗ ) when the D P LZ dis-tances are used instead of D P L . It is useful to note thatthe author also finds that the best-fit value to M ∗ cali-brated against the RGB-Tip distances, i.e., independentlyfrom Cepheid distances, is again +0.07 mag brighter thanthe PNLF calibration obtained from D P L .The two corrections: a ) − b ) +0.07 mag for PNLF, both deriving from the adop-tion of the D P LZ , can justify ∼ .
13 mag of the PNLF-SBFoffset.Figure A.1 shows the histogram of the differences be-tween PNLF and SBF distance moduli based on variousSBF and PNLF distance and/or calibration. Panel a ) inthe figure shows the PNLF to SBF ( m − M ) difference, Table 7.
Quality control statistics for the exposures.
Name Background Background seeing Used Data- (counts) ( mag/arcsec ) (”) - B bandFORS.1999-12-28T02:42:27.813.fits 888.0 22.25 0.86 yesFORS.1999-12-28T02:54:22.346.fits 907.4 22.22 0.81 yesFORS.2000-01-09T02:41:46.659.fits 983.8 22.13 1.07 yesFORS.2000-01-09T02:52:43.240.fits 998.9 22.12 1.01 yesFORS.2000-01-12T01:11:59.202.fits 1556.3 21.64 1.06 yesFORS.2000-01-12T01:22:57.678.fits 1527.3 21.66 1.12 yesFORS.2000-01-13T00:56:41.525.fits 2319.5 21.20 1.20 yesFORS.2000-01-13T01:07:40.539.fits 2236.8 21.24 1.14 yesFORS.2000-01-14T03:50:58.405.fits 1958.7 21.39 1.15 yesFORS.2000-01-14T04:01:58.759.fits 1580.1 21.62 1.19 yesFORS.2000-01-15T03:10:31.420.fits 7298.6 19.96 0.77 noFORS.2000-01-15T03:21:27.996.fits 7231.2 19.97 0.86 noFORS.2000-01-17T03:28:21.613.fits 11877.6 19.43 1.36 noFORS.2000-01-17T03:39:19.186.fits 12162.4 19.40 1.30 noFORS.2000-01-18T01:20:21.447.fits 12832.7 19.35 0.87 noFORS.2000-01-18T01:31:20.684.fits 12983.9 19.33 0.77 noFORS.2000-01-19T01:30:49.601.fits 20047.5 18.86 1.38 noFORS.2000-01-19T01:41:43.505.fits 21026.0 18.81 1.31 noFORS.2000-01-20T02:36:07.201.fits 32829.0 18.33 1.15 noFORS.2000-01-20T02:47:02.044.fits 37640.4 18.18 0.70 noFORS.2000-01-21T01:58:13.531.fits 35260.4 18.25 0.70 noFORS.2000-01-21T02:09:11.551.fits 35681.1 18.24 0.70 no V bandFORS.2000-01-09T03:28:34.384.fits 3261.5 21.13 0.91 yesFORS.2000-01-09T03:39:32.352.fits 3333.3 21.10 1.10 yesFORS.2000-01-13T01:20:04.617.fits 10012.2 19.91 0.98 yesFORS.2000-01-13T01:31:04.242.fits 69.2 19.93 0.84 yesFORS.2000-01-13T01:39:32.083.fits 5141.9 20.63 1.03 yesFORS.2000-01-13T01:50:31.241.fits 5130.5 20.64 1.09 yesFORS.2000-01-17T04:16:19.860.fits 24234.6 18.95 1.51 noFORS.2000-01-17T04:27:17.584.fits 25234.8 18.91 1.59 noFORS.2000-01-18T02:06:59.548.fits 25642.0 18.89 0.76 noFORS.2000-01-18T02:17:56.431.fits 23947.3 18.96 0.93 noFORS.2000-01-19T01:07:22.117.fits 38819.6 18.44 1.21 noFORS.2000-01-19T01:18:19.487.fits 37909.1 18.46 1.28 noFORS.2000-01-20T01:52:46.180.fits 22531.1 18.28 1.25 noFORS.2000-01-20T01:58:44.418.fits 29233.8 17.99 1.36 noFORS.2000-01-20T02:05:45.069.fits 39325.9 17.67 0.95 noFORS.2000-01-20T02:11:42.821.fits 38475.5 17.70 0.76 noFORS.2000-01-21T00:54:09.577.fits 6202.4 17.93 1.14 noFORS.2000-01-21T01:06:08.681.fits 5118.1 18.14 1.06 noFORS.2000-01-21T01:30:59.576.fits 16978.9 18.14 1.09 noFORS.2000-01-21T01:35:17.720.fits 17047.7 18.14 1.11 noFORS.2000-01-21T01:39:35.548.fits 15737.2 18.23 0.98 noFORS.2000-01-21T01:43:50.433.fits 15366.4 18.25 1.03 noFORS.2000-01-21T01:48:07.248.fits 15653.2 18.23 1.06 noFORS.2000-01-21T01:52:24.597.fits 16583.5 18.17 1.08 no I bandFORS.1999-12-27T03:46:00.987.fits 9675.7 19.07 0.69 yesFORS.1999-12-27T03:57:55.778.fits 10469.2 18.99 0.78 yesFORS.2000-01-09T03:05:09.924.fits 8138.7 19.26 0.87 yesFORS.2000-01-09T03:16:07.432.fits 8755.6 19.18 0.82 yesFORS.2000-01-13T02:03:14.285.fits 9061.0 19.14 0.93 yesFORS.2000-01-13T02:14:12.265.fits 9847.6 19.05 0.84 yesFORS.2000-01-15T03:34:22.320.fits 19878.0 18.29 0.79 yesFORS.2000-01-15T03:45:21.020.fits 20169.8 18.27 1.00 yesFORS.2000-01-17T03:52:07.180.fits 18776.9 18.35 1.22 noFORS.2000-01-17T04:03:05.404.fits 19359.7 18.32 1.36 noFORS.2000-01-18T01:43:30.890.fits 19114.6 18.33 0.65 noFORS.2000-01-18T01:54:29.531.fits 19684.5 18.30 0.74 noFORS.2000-01-19T00:44:09.869.fits 27488.0 17.94 1.13 noFORS.2000-01-19T00:55:07.408.fits 28624.7 17.89 1.09 noFORS.2000-01-21T02:22:08.776.fits 16986.9 17.71 1.01 noFORS.2000-01-21T02:28:02.578.fits 17131.9 17.70 1.05 noFORS.2000-01-21T02:33:59.741.fits 16976.3 17.71 0.99 noFORS.2000-01-21T02:39:57.532.fits 16960.7 17.71 1.08 noFORS.2000-02-04T02:06:34.213.fits 9268.9 19.12 1.03 yesFORS.2000-02-04T02:17:28.971.fits 8959.7 19.16 0.85 yes ∆ P NLF − SBF , using the original sample of 28 galaxies byCiardullo et al. (2002), with updated zero points for bothdistance indicators. The mean is ∆
P NLF − SBF = − . P NLF − SBF = − .
36 mag beforezero-point correction.After the Ciardullo et al. (2002) paper, few PNLF dis-tances have been obtained for galaxies with SBF measure-ments. Panel b ) in Figure A.1 shows the PNLF to SBFdifference for a total of 33 galaxies, including the dis-tances obtained after 2002. The comparison shown in panel c ) is obtained using the recent SBF distances from theACSVCS and ACSFCS surveys (when available) in placeof the Tonry et al. (2001) distances. Finally, panel d ) usesthe same SBF and PNLF distances of panel c ) except thatfor the galaxies with old Tonry et al. (2001) corrected dis-tances we include the further “Q-correction” term using eq.A1 from Blakeslee et al. (2010). Table A.1 presents the av-erage and median differences for all assumptions shown inFigure A.1.It is worth mentioning that, according toCiardullo et al. (2002), to properly compare PNLFand SBF, the latter distance moduli should be furtherreduced by 0.04 mag. If one includes this correction term,all differences reported in Table A.1 becomes smaller,with the best sample (i.e., d in the table) providing∆ P NLF − SBF = − . ± .
06 mag, and a median of − . ∼ < . ∼
10% larger SBF distances. Again, this difference occursmainly beyond ∼
10 Mpc, and is similar to the 10% larger H obtained when calibrating SNe Ia via PNLF instead ofCepheids (Feldmeier et al. 2007).Even though the data presented in Table A.1 (with thepossible further +0.04 mag improvement cited above) sug-gest that the best average difference is statistically consis-tent with zero – especially taking into account the system-atic uncertainties, not considered in this comparison – thescatter we find is larger, or nearly equal to the squared sumof the estimated internal scatters of both indicators. Takenat face value, this result either means that the internal scat-ter of one or both indicators is underestimated, or that areal systematic offset exists between the two.As another test, to further check the latter concern,we analyzed the PNLF to SBF offset by considering late-type and early-type galaxies separately. Using sample d inTable A.1, we find the following differences: ∆ EarlyP NLF − SBF = − . ± .
05 mag (median − .
30 mag) based on the data of24 galaxies, and ∆
LateP NLF − SBF = 0 . ± .
17 mag (median − .
02 mag) for the remaining 9 galaxies. The result showsthat for the class of galaxies used to derive SBF and PNLFzero points, i.e., the late types hosting Cepheids, there isno statistically significant offset between the two indicators.On the other hand, for early-type galaxies the offset is large,and statistically inconsistent with zero. Related to this, onefundamental difference between the two indicators is thatabsolute SBF magnitudes are “corrected” for galaxy stel-lar content, i.e., the difference between early- and late-typegalaxies is taken into account with SBF, while the PNLF distances are based on the constancy of M ∗ for both typesof galaxy.The calibration of SBF magnitudes, and its depen-dence on galaxy type, has been analyzed in detail overa ground-based sample of ∼
300 galaxies by Tonry et al.(2001), and more recently from HST data of ∼
150 galaxiesby Blakeslee et al. (2009, 2010). The derived SBF calibra-tions, as is well-known, include a colour-dependent term,which, as also shown by SSP models (Worthey 1993b;Cantiello et al. 2003; Raimondo et al. 2005), is basically ametallicity correction term. In optical bands, this correc-tion term implies fainter SBF magnitudes for redder/moremetal-rich systems.In contrast, the M ∗ calibration to PNLF does not in-clude any metallicity dependent term for bright galaxies. Adependence of the PNLF M ∗ to metallicity has been foundby Ciardullo et al. (2002) and Ciardullo (2012), howeverthe authors conclude that such dependence is relevant onlyin small, metal-poor systems.Inspecting the open circles in Figure 5 of Ciardullo(2012, open circles mark the data obtained from D P LZ dis-tances), reported in Figure A.2, one can see that a ) thetrend of M ∗ with metallicity has the opposite sign withrespect to SBF, meaning that M ∗ gets brighter for moremetal-rich systems, and that b ) some residual correlationof M ∗ with metallicity also appears in the high metallicityregime.This strongly suggests that there may be some unac-counted for residual dependence of the PNLF calibration onthe metallicity, presently unquantified because of the rela-tively limited sample – though detectable even in presentdatasets (Fig. A.2). If so, one possible explanation for the ir-reducible ∼ − . Appendix B: Some notes on Type Ia SNe distancesfrom Stritzinger et al. (2010)
To understand the possible causes of the difference betweenour and Str10 distance we must recall that, as for SBF,in order to calibrate SNe Ia light curves one must rely onsources at known distance and/or with well-known intrinsicproperties, and then standardize the absolute magnitude ofthe SN Ia (e.g., Phillips 1993).Str10 derives the distance to NGC 1316 using three dif-ferent methods: the EBV, the Tripp method, and the near-IR light-curves. The authors also use the MLCS2k2 methodon SN 2006mr, obtaining a distance that is 50% further thanthe average they derived from the normal events.In the following we discuss each one of the three meth-ods used by Str10, trying to highlight the possible causesleading to the observed difference.
B.1. The Colour Excess Method
Str10 adopted the calibration from Burns et al. (2011, AJin press at the time Str10 was published) to obtain the dis-tance moduli with the “EBV” model in their fitting packageSNooPy. For this model they adopt the recommended cal-
Table A.1.
SBF to PNLF comparison.
Sample Number of galaxies ∆
PNLF − SBF ( rms ) Median difference(mag) (mag)Original [1] − − [2] − − [3] − − [4] − − [5] − − b except for NGC 4697, NGC 1344 and NGC 821 whose PNLF distance is independent from Ciardullo et al. calibration.[4 ] Updated SBF distances from ACSVCS (Cˆot´e et al. 2004) & ACSFCS (Jord´an et al. 2007) (when available). Old SBF andPNLF distances as in sample b .[5 ] As sample c except that SBF with Q–corrected SBF distances (see text) are used for the old Tonry et al. (2001) distancemoduli. Fig. A.1.
Histograms of the difference between PNLFand SBF distance moduli. Each panel shows differentcomparison/calibration samples. Panel a ): Original sam-ple of 28 galaxies with PNLF and SBF distances fromCiardullo et al. (2002) and Tonry et al. (2001) respectively,with revised Cepheids calibration. Dotted/dashed verticallines show the mean/median of the difference. Panel b ) :As in a ) but 5 more galaxies with PNLF measurementsmade after Ciardullo et al. (2002) are added to the sample.Panel c ): As b ), but more recent ACSFCS and ACSVCSSBF distances are used when available. Panel d ): As c ),but the correction from eq. A1 by Blakeslee et al. (2010) isincluded to Tonry et al. (2001) distances.ibration from Burns et al. (2011), which uses a sub-sampleof unreddened SNe Ia from Folatelli et al. (2010), exclud-ing fast declining objects, as SN 2006mr. Folatelli et al. Fig. A.2.
Figure 5 from Ciardullo (2012) showing the soledata with metallicity corrected Cepheids distances. [Seeelectronic version of the Journal for a colour version ofthe figure.] (2010), in turn, calibrate their dataset using 26 SNe Ia at z > .
01, whose distances are based on Hubble’s law assum-ing H = 72 km s − M pc − , and three SNe Ia at z < . m − M ) = 31 . ± .
08 mag from the SBF measurementsby Cantiello et al. (2007a).The SNooPy/EBV method relies (also) on an estimateof the internal reddening around the SN Ia. Str10 find thatall four SNe Ia in NGC 1316 have negligible internal ex-tinction. The authors warn about some complications inthe interpretation of the data. With respect to spectro-scopic analysis of Na I D absorption in the spectra ofSN 2006dd and SN 2006mr, these authors state that “thevery strong Na I D absorption observed in SNe Ia 2006ddand SN 2006mr is totally inconsistent with the low hostgalaxy reddening we derive from the light curve observa- tions” (Str10, Section 4), as the colour evolution of thetwo SNe Ia closely resembles that of unreddened SNe Ia.However, while there is a general observational agreementon higher colour excess corresponding to higher Na I Dequivalent width (EW), this correlation is tight in high-resolution spectra, but the scatter increases substantiallyat lower resolution (Blondin et al. 2009; Poznanski et al.2011, 2012), with considerable confusion due to the blend-ing of the Na I D doublet. For instance, using Figure 5 inBlondin et al. (2009), at EW ∼ . E ( B − V ) host ranges from ∼ . ∼ . E ( B − V ) host > .
45 and > .
15 mag forSN 2006dd and SN 2006mr, respectively (with mean values ∼ . ∼ . ′′ × ′′ zoom of the regions around the four SNe Iais shown in Figure B.1. In the upper part of the fig-ure we show the HST UVIS/ F W , ACS/ F W andACS/ F W residual frames of the three SNe Ia lo-cated within the frames analyzed in this work (SN 1981D,SN 2006dd, SN 2006mr). For sake of completeness, we haveobtained archival g , r , and i -band Gemini/GMOS-S data ofthe region around the SN 1980N, shown in the lower panelsof Figure B.1. As evidenced in the figure, there are undeni-able patterns of dust near the position of the 2006 SNe Ia,though one cannot decide whether the SNe are behind, infront of, or within such dust lanes.The comparison with previous estimates of internal ex-tinction from the literature show that for SN 1980N andSN 1981D, Jha et al. (2007) find E ( B − V ) host values largerthan Str10 (three times larger in the case of SN 1981D)but in agreement within uncertainties with the SNooPyfits. The agreement gets worse if the extinctions derivedby Str10 from near-IR data are taken into account. Note,however, that Str10 corrected the optical and near-IR pho-tometry for host galaxy contamination. Such a correction,although negligible for the case of SN 1980N, does not seemto be discussed by Jha et al.For both the two most recent SNe Ia, 2006dd and2006mr, Maoz & Mannucci (2008) estimate an internal ex-tinction of ∼ E ( B − V ) host = 0 . ± .
008 mag; while, as men-tioned above, SNooPy cannot be used for fitting the light-curves of the fast declining SN 2006mr.In conclusion, the three SNe Ia used by Str10 to get thebest estimate of ( m − M ) could be controversial in terms ofinternal extinction, affecting both the estimate of ( m − M )with the EBV method, and the associated uncertainties. B.2. The Tripp Method
The second method adopted by Str10 is based on the two-parameter model of Tripp (1998) which, differently from the EBV method, can also be applied to fast declining SNe Ia.The calibrating sample is again taken from Folatelli et al.(2010), and SN 2006mr is omitted in the re-computed cali-bration relations, to avoid circularity.The distance to the three normal SNe Ia with thismethod is consistent with the estimates based on theSNooPy/EBV method. In contrast, the ( m − M ) obtainedwith the data of SN 2006mr is ∼ +0 . ∼ rms reported in the cited figure. Certainly, changing thedistance modulus of one of the calibrating data-points inthe Folatelli et al. sample implies changing the linear cali-bration relation shown in the cited figure, and possibly re-duces the offset between data and fit. In any case, though,using ( m − M ) ∼ . B.3. The near-IR method
The last method used by Str10 is based on near-IR light-curves of SNe Ia, calibrated using Krisciunas et al. (2009)absolute near-IR peak magnitudes without NGC 1316 data.The Krisciunas et al. calibration of near-IR peak mag-nitudes adopts new observations of SNe Ia, and datapreviously published by the same team (Krisciunas et al.2004a,b). For the nearby galaxies, the authors adopted dis-tances based on either SBF or Cepheids. From a carefulreading of the cited papers, we find that Krisciunas et al.(2004a) obtained the
JHK calibration from 16 SNe Ia.For three nearby galaxies, NGC 1316, NGC 4526, andNGC 5128, the authors adopt the SBF distance fromAjhar et al. (2001, based on D P LZ distances), while forNGC 4536 and NGC 3368 the D P L distances from Cepheidsis used. Both SBF and Cepehids distances are based on thesame Freedman et al. (2001) calibrating sample.Krisciunas et al. (2004b) extended the sample of SNe Iawith well-sampled near-IR light-curves to about 20 objects.The authors added two more supernova-host galaxies withSBF distances – NGC 4374 and NGC 3190 – to the pre-vious list of nearby galaxies. However, in contrast withKrisciunas et al. (2004a), they adopted the Tonry et al.(2001) distance moduli, which are based on D P L ; thatis, they are 0.06 mag larger than the ( m − M ) reported byAjhar et al. (2001).Finally, the most recent calibration of near-IR light-curves of SNe Ia by Krisciunas et al. (2009), used the dis-tance moduli for nearby galaxies from the SBF survey byTonry et al. (2001) with the revised Jensen et al. (2003)zero points (+0.16 mag with respect to Tonry et al. 2001).More specifically, in the new list of ∼
25 supernova-hostgalaxies, the authors added three new objects, NGC 936,NGC 1201 and NGC 1371, with SBF-based distances. In ad-dition, Krisciunas et al. adopted a revised SBF distance toNGC 5128 from Jensen et al. (2003). However, the authorscorrected the ( m − M ) by +0.16 mag, which is the differencebetween the Tonry et al. (2001) and Jensen et al. (2003) Fig. B.1.
Zoom in 6 × arcsec of the residual frames centered on the positions of the four SNe Ia, as labeled. Blackcircles mark the position of the respective SNa. UVIS/ F W , ACS/ F W and ACS/ F W frames (left, middle andright panels, respectively) are shown for the SN 1981D, SN 2006dd and SN 2006dd. The panels showing SN 1980N are g , r, and i observations taken from the Gemini Telescope archive.calibrations, and not by +0.10 mag, i.e., the difference be-tween the Ajhar et al. (2001) and Jensen et al. (2003) cal-ibrations.Hence, the calibration of the absolute magnitudes innear-IR bands used by Str10 relies on a sample of ∼ a ) internally homogeneous, and b ) consistent with theCepheid distances used in this work. As shown in the table,the revised distance moduli are on average 0.06 mag largerthan the ones used for the original SNe Ia near-IR cali-bration. If one simply takes the average of these numbers,adding a sample of 15 objects more (25 total SNe Ia minusthe nine nearby objects and NGC 1316) where no shift hasto be applied, the correction to the absolute magnitudes inTable 9 of Str10 is ∼ − .
02 mag (a correction that shouldbe applied to the calibration by Krisciunas et al. 2009).Although a +0.02 mag shift in Str10 distance moduligoes in the direction of reducing the mismatch between ourand Str10 distances, the amplitude of the correction is negli- gible. However, it suggests that the uncertainties associatedwith the near-IR calibrations might be underestimated.In conclusion, the analysis presented in this Appendix high-lights two main issues: i ) the homogeneity of the calibra-tors used, and ii ) the estimate of internal extinctions forSNe Ia (although one should not forget that the qualityof the SN 1980N and SN 1981D data is lower than others).While the first issue listed works in the direction of reduc-ing the difference between our and Str10 distance moduli,both issues imply an increase of the present levels of statis-tical/systematic uncertainties on SNe Ia distances. Table B.1.
Nearby galaxy sample used for near–IR calibration: original and revised distances.
Galaxy ( m − M ) near − IR Reference Method ( m − M ) Revised ∆( Revised − Orig. )NGC 4526 31.08 Krisciunas et al. (2004a) SBF 31.08 +0.00NGC 4536 30.80 Krisciunas et al. (2004a) Cepheids 30.87 +0.07NGC 3368 29.97 Krisciunas et al. (2004a) Cepheids 30.11 +0.14NGC 4374 31.32 Krisciunas et al. (2004b) SBF 31.26 − − ± ± References
Ajhar, E. A., Blakeslee, J. P., & Tonry, J. L. 1994, AJ, 108, 2087Ajhar, E. A., Lauer, T. R., Tonry, J. L., et al. 1997, AJ, 114, 626Ajhar, E. A., Tonry, J. L., Blakeslee, J. P., Riess, A. G., & Schmidt,B. P. 2001, ApJ, 559, 584Bertin, E. & Arnouts, S. 1996, A&AS, 117, 393Bertola, F., Capaccioli, M., Holm, A. V., & Oke, J. B. 1980, ApJ, 237,L65Biscardi, I., Raimondo, G., Cantiello, M., & Brocato, E. 2008, ApJ,678, 168Blakeslee, J. P. 2012, Ap&SS, 341, 179Blakeslee, J. P., Ajhar, E. A., & Tonry, J. L. 1999, in ASSL Vol. 237:Post-Hipparcos cosmic candles, 181Blakeslee, J. P., Anderson, K. R., Meurer, G. R., Ben´ıtez, N., &Magee, D. 2003, in ASP Conf. Ser. 295: Astronomical Data AnalysisSoftware and Systems XIIBlakeslee, J. P., Cantiello, M., Mei, S., et al. 2010, ApJ, 724, 657Blakeslee, J. P., Jord´an, A., Mei, S., et al. 2009, ApJ, 694, 556Blakeslee, J. P., Vazdekis, A., & Ajhar, E. A. 2001, MNRAS, 320, 193Blondin, S., Prieto, J. L., Patat, F., et al. 2009, ApJ, 693, 207Brocato, E., Castellani, V., Poli, F. M., & Raimondo, G. 2000, A&AS,146, 91Brown, T. M., Bowers, C. W., Kimble, R. A., Sweigart, A. V., &Ferguson, H. C. 2000, ApJ, 532, 308Brown, T. M., Smith, E., Ferguson, H. C., et al. 2008, ApJ, 682, 319Burns, C. R., Stritzinger, M., Phillips, M. M., et al. 2011, AJ, 141, 19Burstein, D. & Heiles, C. 1984, ApJS, 54, 33Buscombe, W. & de Vaucouleurs, G. 1955, The Observatory, 75, 170Buzzoni, A. 1993, A&A, 275, 433Buzzoni, A. & Gonz´alez-L´opezlira, R. A. 2008, ApJ, 686, 1007Cantiello, M. 2012, Ap&SS, 341, 187Cantiello, M., Biscardi, I., Brocato, E., & Raimondo, G. 2011a, A&A,532, A154Cantiello, M., Blakeslee, J., Raimondo, G., Brocato, E., & Capaccioli,M. 2007a, ApJ, 668, 130Cantiello, M., Blakeslee, J. P., Raimondo, G., et al. 2005, ApJ, 634,239Cantiello, M., Brocato, E., & Blakeslee, J. P. 2009, A&A, 503, 87Cantiello, M., Brocato, E., & Capaccioli, M. 2011b, A&A, 534, A35Cantiello, M., Raimondo, G., Blakeslee, J. P., Brocato, E., &Capaccioli, M. 2007b, ApJ, 662, 940Cantiello, M., Raimondo, G., Brocato, E., & Capaccioli, M. 2003, AJ,125, 2783Capaccioli, M., Della Valle, M., D’Onofrio, M., & Rosino, L. 1990,ApJ, 360, 63Capaccioli, M., Della Valle, M., Rosino, L., & D’Onofrio, M. 1989,AJ, 97, 1622Capaccioli, M., Mancini, D., & Sedmak, G. 2005, The Messenger, 120,10Capaccioli, M. & Schipani, P. 2011, The Messenger, 146, 2Carter, D., Jorden, P. R., Thorne, D. J., Wall, J. V., & Straede, J. C.1983, MNRAS, 205, 377Cavadore, C., Beletic, J., & Monnet, G. 1999, De-Contamination re-port of FORS1 CCD, Tech. rep.Childress, M., Zhou, G., Tucker, B., et al. 2012, Central BureauElectronic Telegrams, 3275, 2Cho, H. et al. 2013, in IAU Symposium, Vol. 289, IAU Symposium,in pressCiardullo, R. 2012, Ap&SS, 149 Ciardullo, R., Feldmeier, J. J., Jacoby, G. H., et al. 2002, ApJ, 577,31Code, A. D., Welch, G. A., & Page, T. L. 1972, in NASA SpecialPublication, Vol. 310, Scientific results from the orbiting astronom-ical observatory (OAO-2), ed. A. D. Code, 559Cˆot´e, P., Blakeslee, J. P., Ferrarese, L., et al. 2004, ApJS, 153, 223Cˆot´e, P., Ferrarese, L., Jord´an, A., et al. 2007, ApJ, 671, 1456Della Valle, M. & Gilmozzi, R. 2002, Science, 296, 1275Della Valle, M. & Panagia, N. 2003, ApJ, 587, L71Feldmeier, J. J., Jacoby, G. H., & Phillips, M. M. 2007, ApJ, 657, 76Ferrarese, L., Mould, J. R., Kennicutt, R. C., et al. 2000, ApJ, 529,745Folatelli, G., Phillips, M. M., Burns, C. R., et al. 2010, AJ, 139, 120Freedman, W. L. & Madore, B. F. 2010, ARA&A, 48, 673Freedman, W. L., Madore, B. F., Gibson, B. K., et al. 2001, ApJ, 553,47French, B. S., Jensen, J. B., & Blakeslee, J. P. 2012, in AmericanAstronomical Society Meeting Abstracts, Vol. 220, AmericanAstronomical Society Meeting Abstracts n. 220, 332.02G´omez, M., Richtler, T., Infante, L., & Drenkhahn, G. 2001, A&A,371, 875Gonz´alez-L´opezlira, R. A., Bruzual-A., G., Charlot, S., Ballesteros-Paredes, J., & Loinard, L. 2010, MNRAS, 403, 1213Goudfrooij, P., Alonso, M. V., Maraston, C., & Minniti, D. 2001a,MNRAS, 328, 237Goudfrooij, P., Gilmore, D., Whitmore, B. C., & Schweizer, F. 2004,ApJ, 613, L121Goudfrooij, P., Mack, J., Kissler-Patig, M., Meylan, G., & Minniti,D. 2001b, MNRAS, 322, 643Grado, A., Capaccioli, M., Limatola, L., & Getman, F. 2012, Memoriedella Societa Astronomica Italiana Supplementi, 19, 362Grado, A., Capaccioli, M., Silvotti, R., et al. 2004, AstronomischeNachrichten, 325, 601Hamuy, M., Phillips, M. M., Maza, J., et al. 1991, AJ, 102, 208Hamuy, M., Phillips, M. M., Suntzeff, N. B., et al. 1996, AJ, 112, 2391Han, Z. 2008, A&A, 484, L31Harris, W. E. 2001, in Saas-Fee Advanced Course 28: Star ClustersHerrmann, K. A., Ciardullo, R., Feldmeier, J. J., & Vinciguerra, M.2008, ApJ, 683, 630Jacoby, G. H., Branch, D., Ciardullo, R., et al. 1992, PASP, 104, 599Jedrzejewski, R. I. 1987, MNRAS, 226, 747Jensen, J. B., Tonry, J. L., Barris, B. J., et al. 2003, ApJ, 583, 712Jensen, J. B., Tonry, J. L., & Luppino, G. A. 1998, ApJ, 505, 111Jensen, J. B., Tonry, J. L., Thompson, R. I., et al. 2001, ApJ, 550,503Jha, S., Riess, A. G., & Kirshner, R. P. 2007, ApJ, 659, 122Jord´an, A., Blakeslee, J. P., Peng, E. W., et al. 2004, ApJS, 154, 509Jord´an, A., Cˆot´e, P., Blakeslee, J. P., et al. 2005, ApJ, 634, 1002Jord´an, A., McLaughlin, D. E., Cˆot´e, P., et al. 2007, ApJS, 171, 101Jord´an, A., Peng, E. W., Blakeslee, J. P., et al. 2009, ApJS, 180, 54Kalirai, J. S., MacKenty, J., Bohlin, R., et al. 2009, WFC3 SMOVProposal 11451: The Photometric Performance and Calibration ofWFC3/IR, Tech. rep.Kaviraj, S., Sohn, S. T., O’Connell, R. W., et al. 2007, MNRAS, 377,987Kelson, D. D., Illingworth, G. D., Tonry, J. L., et al. 2000, ApJ, 529,768Konami, S., Matsushita, K., Nagino, R., et al. 2010, PASJ, 62, 1435Krisciunas, K., Hastings, N. C., Loomis, K., et al. 2000, ApJ, 539, 658