A Late-Time View of the Progenitors of Five Type IIP Supernovae
MMon. Not. R. Astron. Soc. , 000–000 (0000) Printed 20 September 2018 (MN L A TEX style file v2.2)
A Late-Time View of the Progenitors of Five Type IIP Supernovae
Justyn R. Maund , , (cid:63) , Emma Reilly and Seppo Mattila Department of Physics and Astronomy, Queen’s University, Belfast BT7 1NN, Northern Ireland, UK Dark Cosmology Centre, Niels Bohr Institute, University of Copenhagen, Juliane Maries Vej 30, 2100 Copenhagen, DK. Royal Society Research Fellow Finnish Centre for Astronomy with ESO (FINCA), University of Turku, V¨ais¨al¨antie 20, FI-21500 Piikki¨o, Finland
20 September 2018
ABSTRACT
The acquisition of late-time imaging is an important step in the analysis of pre-explosionobservations of the progenitors of supernovae. We present late-time HST ACS WFC ob-servations of the sites of five Type IIP SNe: 1999ev, 2003gd, 2004A, 2005cs and 2006my.Observations were conducted using the F W , F W and F W filters. We confirm theprogenitor identifications for SNe 2003gd, 2004A and 2005cs, through their disappearance.We find that a source previously excluded as being the progenitor of SN 2006my has nowdisappeared. The late-time observations of the site of SN 1999ev cast significant doubt overthe nature of the source previously identified as the progenitor in pre-explosion WFPC2 im-ages. The use of image subtraction techniques yields improved precision over photometryconducted on just the pre-explosion images alone. In particular, we note the increased depthof detection limits derived on pre-explosion frames in conjunction with late-time images. Weuse SED fitting techniques to explore the e ff ect of di ff erent reddening components towards theprogenitors. For SNe 2003gd and 2005cs, the pre-explosion observations are su ffi ciently con-straining that only limited amounts of dust (either interstellar or circumstellar) are permitted.Assuming only a Galactic reddening law, we determine the initial masses for the progenitorsof SNe 2003gd, 2004A, 2005cs and 2006my of 8 . ± .
0, 12 . ± .
1, 9 . + . − . and 9 . ± . M (cid:12) ,respectively. Key words: stars : evolution – supernovae : general – supernovae : individual : 1999ev –supernovae : individual : 2003gd – supernovae : individual : 2004A – supernovae : individual: 2005cs – supernovae : individual : 2006my
All stars with initial masses > M (cid:12) are expected to end their livesas core-collapse supernovae (CCSNe). In the last decade, the di-rect observation of the progenitors of CCSNe in fortuitous pre-explosion imaging, has become an integral step in the study of allnearby events (for a review see Smartt 2009).The majority of the success in the actual detection of progeni-tors has been for the Red Supergiant (RSG) precursors to hydrogen-rich Type II Plateau (IIP) SNe. The ensemble of progenitor de-tections and detection limits for these RSGs, however, presented aconflict with the theoretical expectation from stellar evolution mod-els (Smartt et al. 2009). The “Red Supergiant Problem” refers tothe apparent absence of progenitors of Type IIP SNe with mass > M (cid:12) , whereas the theoretical expectation is that stars withmasses up to 25 − M (cid:12) should end their lives as RSGs. RecentlySmith et al. (2011) and Walmswell & Eldridge (2012) consideredthe role of circumstellar dust, which is otherwise not probed by theresulting SN or the surrounding stellar population, as a possible so- (cid:63) Email: [email protected] lution to the RSG problem. In particular, Walmswell & Eldridgeestablished that the amount of circumstellar dust, and hence red-dening, is larger for higher luminosity RSGs, that arise at the upperend of the mass range for stars to explode as Type IIP SNe. In thecase of SN 2012aw, Fraser et al. (2012) and Van Dyk et al. (2012)observed a significant di ff erence between the reddening determinedtowards the progenitor and the reddening inferred towards the sub-sequent SN; suggesting a significant amount of dust was associatedwith the progenitor and that it was destroyed in the SN explosion,such that it could not be measured post-explosion. Kochanek et al.(2012) and Van Dyk et al. (2012) also considered the additionalpossibility that the nature of the reddening, due to a circumstellardust component, was di ff erent to ordinary reddening laws appro-priate for the interstellar medium (e.g. Cardelli et al. 1989).A further issue with the RSG problem is that the maximummass is poorly constrained, due to the paucity of progenitors withhigh inferred masses. This may reflect either a real deficit of highmass progenitors (i.e. that the upper mass limit for stars to explodeas Type IIP SNe is low) or may be due to low numbers of high massprogenitors according to the Initial Mass Function (IMF). The exactarticulation and quantification of the RSG problem is compounded c (cid:13) a r X i v : . [ a s t r o - ph . S R ] F e b Maund et al. by the reliance on, generally, poor fortuitous pre-explosion images,limited detections in pre-explosion images in multiple filters, andambiguity as to whether the object observed at the SN position isindeed the progenitor object.Maund & Smartt (2009) demonstrated that it is possible, us-ing deep late-time images that are tailored (in a way that the pre-explosion images, by their fortuitous nature, cannot be) to observethe site of a given SN and overcome the severe limitations of anal-ysis on just the pre-explosion images alone. Using image subtrac-tion techniques, such as isis (Alard & Lupton 1998; Alard 2000),the late-time images can be used as templates, to both demonstratethe disappearance of the progenitor candidate (confirm the originalidentification) and conduct precise photometry (without contami-nation from nearby and underlying objects).Here we present a new analysis of five previously identi-fied progenitor candidates using newly acquired late-time HSTAdvance Camera for Surveys (ACS) Wide Field Channel (WFC)imaging. These progenitors are for SNe 1999ev, 2003gd, 2004A,2005cs and 2006my, and details of these SNe are presented in Ta-ble 1.
The analysis of the pre-explosion HST observations of the sitesof the five target SN progenitors has been previously presented byMaund & Smartt (2005) (1999ev); Van Dyk et al. (2003b), Smarttet al. (2004) and (Maund & Smartt 2009) (2003gd); Hendry et al.(2006) (2004A); Maund et al. (2005) and Li et al. (2006a) (2005cs);and Li et al. (2007), Leonard et al. (2008) and Crockett et al. (2011)(2006my). The pre- and post-explosion and late-time observationsof the sites of the five target Type IIP SN progenitors are presentedin Table 2.
Due to the availability of more recent and appropriate calibrations,new versions of the pre-explosion WFPC2 and ACS data were re-trieved from the HST archive . The pre-explosion observations ac-quired with WFPC2 were drizzled together following the standardprocedure . These images were produced for the purposes of imagesubtraction (see Section 2.6). In parallel, a separate “reduction” andphotometric analysis was conducted on the same data using HST-phot (Dolphin 2000b). For SN 2005cs, pre-explosion images wereacquired with the Advanced Camera for Surveys (ACS) Wide-FieldChannel (WFC). These images were acquired using a four-pointbox dither pattern. Alignment between the images was checked us-ing the P y RAF task tweakshifts , and the images were combinedusing multidrizzle . Due to the half-integer pixel shifts of the ditherpattern it was possible to enhance the spatial sampling of the finalimage, providing a final pixel scale of 0 . (cid:48)(cid:48) . This is larger thanexact half-sampling (0 . (cid:48)(cid:48) px − ), but is used to match the sam-pling achieved for the late-time ACS WFC images of SN2005cs(see Section 2.3). http: // archive.stsci.edu http: // / hst / wfpc2 / analysis / WFPC2 drizzle.html
For the purposes of this study, the principal interest in the immedi-ate post-explosion images (acquired up to 3 years post-explosion)was to provide a position for the SN relative to the surroundingstars, such that the SN position could be identified on the pre-explosion and late-time frames through di ff erential astrometry. The“reduction” procedure for these images was the same as outlinedfor the pre-explosion observations (see Section 2.1). Late-time observations of the sites of four of the target Type IIPSNe were acquired using the HST ACS WFC 1 for the programGO-11675 (PI Maund). The WFC chip was windowed to an arrayof 1k ×
1k pixels to reduce the readout time and mitigate the role ofCharge Transfer Ine ffi ciency (CTI). Observations were conductedin three filters F W , F W and F W . Importantly, for twoSNe (2003gd and 2004A) the use of the F W filter in theselate-time observations does not match the pre-explosion images ac-quired with the wider F W filter. In the pre-explosion frames, themajority of the flux from the progenitor at these wavelengths is rep-resentative of the continuum (requiring a small F W − F W colour correction). At late-times, however, the F W filter en-compasses the wavelength of H α , which is a characteristic emis-sion feature of late-time Type IIP SN spectra. A more appropriatecomparison between before and after continuum fluxes is, there-fore, achieved with the F W filter although, as discussed below,corrections for the slightly di ff erent filter transmission functionsalso need to be considered. Each of these late-time observations iscomposed of four separate sub-exposures acquired in a 4-point boxdither pattern. This arrangement was used to permit the acquisitionof better spatial sampling of the point-spread function (PSF) andfor removal of fixed hot-pixel features. The P y RAF task multidriz-zle was used to drizzle each of the sub-exposures (for a given filter)together. The task tweakshifts was used to fine-tune the alignmentbetween each of the sub-exposures prior to drizzling. Although the4-point box dither pattern employs half-integer pixel shifts, poten-tially giving an improvement of spatial sampling by a factor of 2,the presence of aliasing (alternating bands across the frames) pro-hibited reaching this final pixel-scale. This phenomenon was rela-tively insensitive to the choice of the drizzling kernel. Instead thefinal pixel scale is greater than one half of the original 0 . (cid:48)(cid:48) pixelscales of ACS / WFC (0 . (cid:48)(cid:48) px − ).For SN 2006my, observed for program GO-12282 (P.I. D.Leonard), the late-time images were only acquired in two bands( F W and F W ). For each filter, two exposures were ac-quired, for the rejection of cosmic rays, but at the same pointingsuch that no spatial resampling was possible. These images for eachfilter were combined using multidrizzle , but with the output imageshaving the original ACS / WFC pixel scale (0 . (cid:48)(cid:48) px − ). A series of transformations were calculated for the completedatasets to determine the positions for objects in a common ref-erence frame. Geometric transformations were calculated betweenthe pre-, post-explosion and late-time F W images (or, if un-available, F W images) using the iraf task geomap , assumingonly simple o ff sets, rotations and scalings. For the data at a givenepoch, shifts between images with other filters and the correspond-ing reference F W image were calculated by cross-correlating c (cid:13) , 000–000 ate-time view of progenitors of Type IIP supernovae Table 1.
Details of the five target Type II SNe, with progenitor candidates identified in pre-explosion HST images.Supernova Type Host Galaxy Distance E ( B − V ) gal [ O / H ] (Mpc) (dex)1999ev II(P) NGC 4274 15 . ± . . ± . . ± . . ± . . ± . After Smartt et al. (2009). After Schlafly & Finkbeiner (2011), as quoted by NED. Smartt et al. (2009). these images with the F W image as the reference frame. Thecross-correlation was facilitated using the P y RAF task crosscor ,with the corresponding shifts calculated using shiftfind . For WFPC2 pre-explosion images the HSTphot package (Dolphin2000b) was utilised to conduct PSF-fitting photometry of the inputimages. HSTphot provides the appropriate corrections for aperturesize and charge transfer ine ffi ciency, as well as tools for conductingartificial star tests. We note that HSTphot only provides aperturecorrections to a final aperture size of 0 . (cid:48)(cid:48) . Using the corrections ofHoltzman et al. (1995), we apply a term to correct to the photometryto an infinite aperture.For data acquired for program GO-11675 (PI Maund), princi-pal photometry was conducted using iraf DAOphot on the final out-put drizzled images with the enhanced spatial sampling. We usedthe latest zeropoints appropriate for ACS WFC . Aperture correc-tions were calculated for each image to an aperture of 0 . (cid:48)(cid:48) , with afurther correction to infinity adopted from Sirianni et al. (2005).A key concern for the fidelity of the derived photometry is theine ffi ciency of charge transfer (CTI) for charged coupled detectorson HST. As photometry was conducted on the drizzled, subsampledimages, evaluating the magnitude of the CTI on the final photom-etry is non-trivial and is based on the position of a given star onthe original undrizzled, distorted FLT images. Geometric transfor-mations were calculated between the final drizzled images and the
FLT images, using 3rd order polynomials in x and y , using ge-omap . This approach was used, over using a simple pre-computeddistortion table, as non-negligible shifts were found between theexpected pointings in the dither pattern. The positions of stars onthe output drizzled images were transformed to the correspondinglocations on each of the four constituent FLT images. FollowingAnnibali et al. (2008), we conducted small aperture (3px) photom-etry on the individual
FLT images and used the measured flux andsky background values to calculate the magnitude loss due to CTIfollowing the analytic prescription of Chiaberge et al. (2009). Priorto conducting aperture photometry, the
FLT images were scaledwith the corresponding Pixel Area Map . For a given star it mightbe only possible to calculate the magnitude loss for 1 or 2 of the in-put FLT images, because of the relative positions of bad pixels orcosmic rays. An average CTI loss (as a magnitude) was determinedover the four constituent
FLT images and applied to the photometryderived from the drizzled images. http: // / hst / acs / analysis / zeropoints / http: // / hst / acs / analysis / PAMS
In addition, photometry of the ACS images was also con-ducted using the DOLPHOT package . We utilised two implemen-tations of DOLPHOT for photometry of the ACS data: DOL phot with the ACS module on the distorted CRJ and
FLT frames (whichwe refer to as DOLPHOT / ACS ) and DOLPHOT as a generic pho-tometry package for the distortion-corrected drizzled frames. Wefind excellent agreement between our DAOphot photometry andthe photometry derived using HSTphot and DOLPHOT, within thelimits of the photometric uncertainties. Similarly to HSTphot, theACS photometry was corrected from a 0 . (cid:48)(cid:48) to an infinite aperture,using the corrections tabulated by Sirianni et al. (2005)The data for SN 2006my were analysed separately using onlythe HSTphot and DOLPHOT packages for the WFPC2 and ACSdata, respectively. Image subtraction techniques were used to conduct template sub-traction of the late-time images from the pre-explosion images to:1) confirm the identities of the progenitors through disappearance;and 2) conduct optimal di ff erential photometry, independently ofthe background, of the now absent progenitors. We utilised the isis v / referenceimages (as well as refining the alignment between the two imagesand scaling the flux levels). In addition, isis also provides automaticobject detection and photometry on the di ff erence images. We alsotested our image subtractions using the HOTPANTS image subtrac-tion package , and examined the resulting di ff erence images usingDAOphot. We found no systematic di ff erence between the photom-etry of di ff erence images calculated using the two packages, andfor this study use photometry derived from di ff erence images con-structed using isis For each SN the late-time images were, generally, used as thereference images. The late-time images were transformed to matchthe pre-explosion images using iraf task geotran . This ensured thatwe avoided resampling the lower quality pre-explosion images tomatch the superior late-time images.The evaluation of the systematic uncertainties was conductedby varying the key parameters in ISIS that principally a ff ected theoutput photometry: the number and size of the stamps used for cal-culating the kernel, the degree of the kernel variation across thefield and the degree of the background fitting function. A major http: // americano.dolphinsim.com / dolphot / http: // / users / becker / hotpants.htmlc (cid:13) , 000–000 Maund et al.
Table 2.
HST observations of the sites of the five target Type II SNe.Dataset Date Instrument Filter Exposure Final Pixel ProgramTime (s) Size ( (cid:48)(cid:48) ) SN 1999ev
Pre-explosion U2JF0101 / /
03T 1995 Feb 5 WFPC2 / WF2 F555W 280 0.1 5741 Post-explosion J8DT03010 2001 Dec 31 ACS / WFC1 F555W 450 0.05 9353 J8DT03020 2001 Dec 31 ACS / WFC1 F814W 450 0.05 9353J8DT03030 2001 Dec 31 ACS / WFC1 F435W 400 0.05 9353Late-time JB4T01010 2010 Nov 14 ACS / WFC1 F555W 1368 0.035 11675 JB4T01020 2010 Nov 14 ACS / WFC1 F814W 1408 0.035 11675JB4T01030 2010 Nov 14 ACS / WFC1 F435W 1608 0.035 11675
SN 2003gd
Pre-explosion U8IXCA01M /
02M 2002 Aug 25 WFPC2 / WF2 F606W 1000 0.1 9676 U8IXCY01M / /
03M 2002 Aug 28 WFPC2 / WF2 F606W 2100 0.1 9676Post-explosion J8NV01020 2003 Aug 1 ACS / HRC F435W 2200 0.025 9733 J8NV01040 2003 Aug 1 ACS / HRC F555W 1000 0.025 9733J8NV01050 2003 Aug 1 ACS / HRC F814W 1350 0.025 9733Late-time JB4T02010 2010 Nov 14 ACS / WFC F555W 1364 0.035 11675 JB4T02020 2010 Nov 14 ACS / WFC F814W 1398 0.035 11675JB4T02030 2010 Nov 14 ACS / WFC F435W 1600 0.035 11675
SN 2004A
Pre-explosion U6EAD001R /
02R 2001 Jul 2 WFPC2 / WF3 F814W 460 0.1 9042 U6EAD003R /
04R 2001 Jul 2 WFPC2 / WF3 F606W 460 0.1 9042Post-explosion J8NV03010 2004 Sep 23 ACS / WFC1 F435W 1400 0.05 9733 J8NV03020 2004 Sep 23 ACS / WFC1 F555W 1509 0.05 9733J8NV03030 2004 Sep 23 ACS / WFC1 F814W 1360 0.05 9733Late-time JB4T03010 2010 Sep 09 ACS / WFC F555W 1400 0.035 11675 JB4T03020 2010 Sep 09 ACS / WFC F814W 1434 0.035 11675JB4T03030 2010 Sep 09 ACS / WFC F435W 1636 0.035 11675
SN 2005cs
Pre-explosion J97C5 2005 Jan 20-21 ACS / WFC F435W 2720 0.035 10452 J97C5 2005 Jan 20-21 ACS / WFC F555W 1360 0.035 10452J97C5 2005 Jan 20-21 ACS / WFC F658N 2720 0.035 10452J97C5 2005 Jan 20-21 ACS / WFC F814W 1360 0.035 10452Post-explosion J9AR01011-31 2005 Jul 24 ACS / HRC F555W 1944 0.025 11675Late-time JB4T04010 2010 Jul 30 ACS / WFC1 F555W 1460 0.035 11675JB4T04020 2010 Jul 30 ACS / WFC1 F814W 1494 0.035 11675JB4T04030 2010 Jul 30 ACS / WFC1 F435W 1696 0.035 11675
SN 2006my
Pre-explosion U2DT0901T / /
03T 1994 May 20 WFPC2 / WF2 F555W 660 0.1 5375 U2DT0904T / /
06T 1994 May 20 WFPC2 / WF2 F814W 660 0.1 5375Post-explosion U9OX0301M / / /
04M 26 Apr 2007 WFPC2 / PC F555W 1200 0.05 10803 U9OX0305M /
06M 26 Apr 2007 WFPC2 / PC F814W 1200 0.05 10803U9OX0307M /
08M 26 Apr 2007 WFPC2 / PC F450W 1400 0.05 10803Late-time JBKS01010 21 Nov 2010 ACS / WFC F555W 1090 0.05 12282 JBKS01020 21 Nov 2010 ACS / WFC F814W 1090 0.05 12282 P.I. J. Westphal P.I. S.J. Smartt P.I. J.R. Maund P.I. S.J. Smartt P.I. S.J. Smartt P.I. J. Rhoads P.I. S. Beckwith P.I. Filippenko P.I. M. Meixner P.I. V. Rubin P.I. S.J. Smartt P.I. D. Leonard c (cid:13)000
08M 26 Apr 2007 WFPC2 / PC F450W 1400 0.05 10803Late-time JBKS01010 21 Nov 2010 ACS / WFC F555W 1090 0.05 12282 JBKS01020 21 Nov 2010 ACS / WFC F814W 1090 0.05 12282 P.I. J. Westphal P.I. S.J. Smartt P.I. J.R. Maund P.I. S.J. Smartt P.I. S.J. Smartt P.I. J. Rhoads P.I. S. Beckwith P.I. Filippenko P.I. M. Meixner P.I. V. Rubin P.I. S.J. Smartt P.I. D. Leonard c (cid:13)000 , 000–000 ate-time view of progenitors of Type IIP supernovae concern for conducting image subtraction analysis using HST im-ages is the e ff ect of the degree by which the PSF is subsampled.Subsampling of the PSF means that the majority of the flux in WFPC2 images will fall in a single pixel, which may induce sys-tematic errors in the construction of the convolution kernel. In orderto assess the systematic uncertainty associated with the degree ofsubsampling, iterations of the image subtraction routine were con-ducted with di ff erent degrees of Gaussian smoothing applied to theinput and / or reference images. We conservatively estimate that thetotal systematic uncertainties associated with the image subtractionprocess are ∼ −
10% of the flux observed in the di ff erence image(as discussed in Section 3; with the larger systematic uncertaintyassociated with the fainter progenitors, in addition to the commen-surate increase in the relative Poisson noise). The systematic uncer-tainties dominate over the Poisson noise, and we assume dominateover other noise sources, such as read noise, that are propagated tothe di ff erence images.As we are concerned with pre-explosion and late-time im-ages acquired with HST, the image subtraction process will involveCTI in both the input and reference images. There are two keys is-sues with the evaluating the e ff ect of CTI on photometry that hasbeen derived using di ff erence imaging: 1) the CTI that a ff ects theprogenitor flux is in the flux system of the pre-explosion image,whereas the photometry of the progenitor from the di ff erence im-age is found in the flux system of the late-time reference image; and2) the evaluation of CTI for a source using its photometry from thepre-explosion image explicitly undermines any increase in the pre-cision of photometry that might be derived using image subtractiontechniques.For both WFPC2 and ACS / WFC the CTI is principally depen-dent on the flux of the object, the level of the nearby background,the position of the object on the chip (the number of charge trans-fers to be made to readout the electrons) and the date at which theobservations were made. Rather than attempting to explicitly calcu-late the counts associated with the progenitor on the pre-explosionimage, we instead use photometry of nearby or artificial stars, inthe vicinity of the progenitor, to serve as proxies for the calculationof the CTI for the progenitor. As these stars are at approximatelythe same position on the chip, on similar backgrounds as the pro-genitor and observed at the same epoch, we can reduce the problemof determining the CTI to just the dependence on the brightness (ormagnitude) of the progenitor. The nearby stars (real or fake) sam-ple a range of brightnesses and simple expressions can be derivedrelating the magnitude of an object in the di ff erence image to theCTI (in magnitudes) directly.We can consider the flux measured on the images, uncorrectedfor CTI, as f (cid:48) . The flux corrected for CTI is then simply: f = f (cid:48) − . CTI (1)Under the assumption that the CTI is a relatively small e ff ect, weuse the initial approximation that m ≈ m (cid:48) to derive the CTI in thepre-explosion frame using photometry, derived from di ff erence im-ages, in the photometric system of the pre-explosion images.On the ACS images, we model the CTI as being e ff ectively de-pendent on only two parameters: the brightness of the object and its y -position on the FLT images. We established, for the images con-sidered here, that the e ff ect of nearby background inhomogeneities,around the progenitor positions, are well within the stated uncer-tainties of the CTI expressions. We consider the CTI for a givenrange of pixels to be approximately given by a power law depen-dent only on the magnitude of the object: CT I ( m ) = β m α (2)We evaluated the coe ffi cients of Equation 2 for each pre-explosionACS / WFC image (as the coe ffi cients are dependent on the date onwhich the observations were made and the specific background atthe progenitor position), using real and artificial stars for whichthe CTI had been evaluated using the equations of Chiaberge et al.(2009). Due to significant uncertainty in the expression used to de-termine the CTI, and its slowing varying nature with pixel position,it is possible to consider the CTI to be approximately fixed over ∼
50 pixel ranges in y .We make a similar approximation for the WFPC2 observa-tions, using the CTI formulation presented by Dolphin (2000a) ,derived using artificial stars generated using HSTphot. We find theaverage dependence of the CTI (over a 50 pixel range in y -position)to correspond to a second order polynomial: CT I ( m ) = α + β m + γ m (3)For both WFPC
ACS observations, the dependence on m is relatively weak; the di ff erence in its evaluation using m or m (cid:48) is negligible. The importance of this approach is that it avoidsspecifically determining fluxes and sky background values from thepre-explosion images (which defeats both the purpose and preci-sion a ff orded by using image subtraction techniques to derive thephotometry of the pre-explosion source). The zeropoint in the pho-tometric scale of the reference image Z R can be derived from pho-tometry m R of reference stars, identified by isis . Using a packagesuch as isis , f (cid:48) R of the reference objects in the reference image canbe measured directly using aperture photometry and can be com-pared directly with the photometry of the same stars derived us-ing DAOphot, DOLPHOT or HSTphot; such that Z R contains notonly the absolute zeropoint, but also all the relevant aperture cor-rections. The magnitude m d of the progenitor candidate can then befound directly. As the image subtraction procedure determines thedi ff erence in observed fluxes f (cid:48) , these fluxes must be further cor-rected for CTI derived on the input image (containing the progen-itor) using the scheme outlined above. Given the di ff erence mea-sured from reference and input images, the final magnitude of thepre-explosion source is given as: m d = − . (cid:0) f (cid:48) d (cid:1) + Z R + CT I i (4)An additional source of systematic uncertainty is the di ff er-ences in the filter transmission functions between images usedfor image subtraction analysis. We note that, although some fil-ters are nominally identical, there may also be di ff erences betweenthe same filters used on di ff erent instruments. We used syntheticphotometry of ATLAS9 (Castelli & Kurucz 2004) and MARCS(Gustafsson et al. 2008) model SEDs, using the total transmission(filter and instrument) functions, to determine the relative colourdi ff erences between the filter sets used here. For most combinationsof filters, in particular between nominally identical HST filters, thecolour di ff erence as a function of temperature is < . In previous studies (e.g. Maund & Smartt 2005; Crockett et al.2011), the derivation of the detection thresholds has been con-ducted using analytical expressions for the background and source with updates from http : // purcell . as . arizona . edu / wfpc2 calib / c (cid:13) , 000–000 Maund et al.
Figure 1.
Colour terms, between filter sets, as a function of temperaturefor ATLAS9 (heavy lines; Castelli & Kurucz 2004) and MARCS (grey line;Gustafsson et al. 2008) model SEDs (appropriate for supergiants with E ( B − V ) =
0; see Section 4). noise for an “ideal” observation. This approach, however, does notaccurately reflect the way in which stars are actually detected in thephotometry process, using iraf tasks such as DAO find , and the ef-fect of crowding. We consider the insertion and attempted recoveryof artificial stars to derive the detection threshold.Artificial stars were generated using the PSFs derived from thedata themselves, with a randomly selected magnitude from a uni-form distribution and with a position uniformly distributed within ± . x and y of the SN location. The first approach involvedrepeating the original detection and aperture and PSF photometryroutine on the pre-explosion images, with artificial stars inserted,and considering a detection to be any recovery of a star within 1pixel and 0.5 magnitudes of the input star’s parameters. The sec-ond approach utilised isis to conduct image subtraction betweenthe late-time images and pre-explosion images, in which artificialstars had been inserted in the latter. For this latter approach, we setthe coordinates at which isis was to conduct aperture photometryand classified a detection to be any instance in which the recov-ered flux was 3 times that of the corresponding noise (including thesystematic uncertainty; see section 2.6).We consider the detection threshold to be the magnitude atwhich we recover 50% of input artificial stars, using a 3 σ detectionthreshold with DAOphot. As noted by Maund (2013, in prep.), thecompleteness function can be considered in terms of the comple-mentary cumulative Gaussian distribution. We therefore quote thecorresponding width of the completeness function as an e ff ectiveuncertainty on the derived detection threshold. SN 1999ev was discovered by T. Boles (Hurst et al. 1999) on 1999Nov 7.225 in the galaxy NGC 4274. Garnavich et al. (1999) sub- sequently classified the SN as being of Type II, although no fur-ther sub-classification of the SN has been reported. Van Dyk et al.(2003a) attempted to identify the progenitor object in pre-explosionWFPC2 F W images from 1995 Feb 1, although were not ableto conclusively identify a single object as the progenitor. In an in-dependent analysis, using a di ff erential astrometric solution derivedusing post-explosion ACS WFC images containing the SN, Maund& Smartt (2005) were able to identify a star in the pre-explosion im-ages coincident with the SN position (with an uncertainty of 0 . (cid:48)(cid:48) ).Late-time ACS WFC F W , F W and F W images(with pixel scale 0 . (cid:48)(cid:48) px − ) of the site of SN 1999ev were ac-quired on 2010 Nov 14 (11 years post-discovery). A late-time im-age of the site of SN 1999ev is shown as Fig. 2. A geometric trans-formation was calculated between the post-explosion and late-time F W images using 24 commons stars, with an uncertainty onthe transformation of ∆ r = . (cid:48)(cid:48) . A source is recovered in thelate-time images at the transformed position of the SN as identi-fied in the post-explosion images by Maund & Smartt (2005), asshown in Fig. 3. The source is detected in all three filters: m F W = . ± . m F W = . ± .
06 and m F W = . ± . m F W = . ± . m F W = . ± .
14 and m F W = . ± .
10. We note thatthe new measurement of the post-explosion F W photometry re-ported here is slightly fainter than measured previously, althoughthe F W and F W magnitudes are approximately similar tothose of Maund & Smartt (2005). The 8 .
98 years between the post-explosion and late-time images reveals significant evolution in thelight echo discovered by Maund & Smartt (2005), which has ex-panded to a radius of 0 . (cid:48)(cid:48) from 0 . (cid:48)(cid:48) (as shown on Figs. 3 and4). A transformation was calculated between the post-explosionand pre-explosion F W images using 24 stars (with a transfor-mation uncertainty of ∆ r = . (cid:48)(cid:48) ). In the pre-explosion images,we identify the same source that Maund & Smartt (2005) identi-fied as the progenitor source with m F W = . ± .
17. In ad-dition, we also find a nearby source with m F W = . ± . . (cid:48)(cid:48) (2 WF pixels) from the progenitor. We note that, forthe period in which the pre-explosion observations were conducted , the nearest logged warm pixel is 5 pixels away from the SN po-sition and not coincident with either the progenitor candidate orthe nearby object. In the late-time image, however, we do not re-cover any source at the corresponding transformed position. Theposition of the pre-explosion source at the SN position was esti-mated using the three centring algorithms available to DAOphot(centroid, Gaussian and optimal filter) and the position determinedusing HSTphot PSF fitting. The positions are shown, with respectto the transformed position of the SN on the pre-explosion image,in Fig. 5. Although there is an apparent discrepancy in the posi-tions for the source and the transformed SN position, the discrep-ancy is not significant. It does, however, raise concerns about howpositions are determined on subsampled images such as this pre-explosion WFPC2 WF2 F W image. The position determinedusing the optimal filter centring algorithm is noticeably di ff erentfrom the other three positions derived from the pre-explosion im-age and is o ff set in the direction of the nearby apparent neighbour-ing star. The standard deviation of the four measurements made onthe pre-explosion image is 0 . (cid:48)(cid:48) . Given the apparent brightness http: // / ftp / instrument news / WFPC2 / Wfpc2 hotpix / / vary 950113 950211 2.dat.Z c (cid:13)000
17. In ad-dition, we also find a nearby source with m F W = . ± . . (cid:48)(cid:48) (2 WF pixels) from the progenitor. We note that, forthe period in which the pre-explosion observations were conducted , the nearest logged warm pixel is 5 pixels away from the SN po-sition and not coincident with either the progenitor candidate orthe nearby object. In the late-time image, however, we do not re-cover any source at the corresponding transformed position. Theposition of the pre-explosion source at the SN position was esti-mated using the three centring algorithms available to DAOphot(centroid, Gaussian and optimal filter) and the position determinedusing HSTphot PSF fitting. The positions are shown, with respectto the transformed position of the SN on the pre-explosion image,in Fig. 5. Although there is an apparent discrepancy in the posi-tions for the source and the transformed SN position, the discrep-ancy is not significant. It does, however, raise concerns about howpositions are determined on subsampled images such as this pre-explosion WFPC2 WF2 F W image. The position determinedusing the optimal filter centring algorithm is noticeably di ff erentfrom the other three positions derived from the pre-explosion im-age and is o ff set in the direction of the nearby apparent neighbour-ing star. The standard deviation of the four measurements made onthe pre-explosion image is 0 . (cid:48)(cid:48) . Given the apparent brightness http: // / ftp / instrument news / WFPC2 / Wfpc2 hotpix / / vary 950113 950211 2.dat.Z c (cid:13)000 , 000–000 ate-time view of progenitors of Type IIP supernovae Figure 2.
Late-time colour image (composed of F W , F W and F W images) of the area of NGC 4274 containing SN 1999ev. The imagehas dimensions 15 (cid:48)(cid:48) × (cid:48)(cid:48) , and is oriented such that North is up, and Eastis left. The position of SN 1999ev is located at the centre of the image. Figure 4. Di ff erence image between the post-explosion and late-time F W observations of the site of SN 1999ev. The evolution of the ex-panding light echo is evident; the white echo appears in the post-explosionimage, while the outer dark echo occurs in the late-time frame. of the source, however, the astrometric uncertainty for the positionderived using HSTphot alone may be as large as 0 . (cid:48)(cid:48) or 0.4 WFpixels (Dolphin 2000b). SN 2003gd was discovered by R. Evans on 2003 Jun 12.82, inthe galaxy M74 (Evans & McNaught 2003). Kotak et al. (2003)
Figure 5.
Close-up of the pre-explosion site of SN 1999ev. In red ( × ) is thetransformed position of the SN, derived from the post-explosion images,and the corresponding r.m.s. error ellipse. Also shown (in green) are the fourseparate measures of the position of the pre-explosion source derived usingthe centroid ( (cid:3) ), Gaussian ( (cid:52) ) and optimal filter ( ◦ ) centring algorithms inDAOphot and the position determined using HSTphot ( ∗ ). Also shown arethe average ( × ) and the corresponding standard deviation ellipse of the fourmeasurements. spectroscopically classified 2003gd as being a Type II SN, approx-imately 2 days post-explosion. Subsequent photometric and spec-troscopic observations of SN 2003gd, however, showed it to be aType IIP SN discovered at the end of the plateau phase (Hendryet al. 2005). Smartt et al. (2003) made a preliminary identificationof the progenitor in pre-explosion HS T
WFPC2 F W and Gem-ini GMOS-N i (cid:48) images. As such, SN 2003gd was the third SN, afterSNe 1987A and 1993J, to have a progenitor identified in fortuitouspre-explosion images. Independent analyses by Smartt et al. (2004)and Van Dyk et al. (2003b) showed the candidate progenitor to bea RSG, corresponding to a star with initial mass of ∼ − M (cid:12) . Theconfirmation of this star as the progenitor was finally provided in2009, when the star was observed to no longer be present in late-time Gemini GMOS-N i (cid:48) images (Maund & Smartt 2009); makingit the first conclusively confirmed RSG progenitor for a Type IIPSN. Late-time ACS WFC observations of the site of SN 2003gdwere acquired on 2010 Nov 14, and are presented on Figure 6. TheSN position in the late-time images was determined with respectto the ACS HRC post-explosion images, with an uncertainty onthe transformation of 0 . (cid:48)(cid:48) . The position of the SN on the pre-explosion HST and Gemini images, as presented by Smartt et al.(2004) and Maund & Smartt (2009), were recalculated to within0 . (cid:48)(cid:48) and 0 . (cid:48)(cid:48) respectively. The pre-explosion, post-explosionand late-time data F W images are shown on Figure 7 and, forcompleteness, we also show the corresponding i (cid:48) data presented byMaund & Smartt (2009) as Figure 8.In the late-time images we observe a source, termed Source A (cid:48) , that is clearly recovered in all late-time HST images at the trans-formed SN position, with magnitudes 25 . ± .
04, 25 . ± .
05 and24 . ± .
04 in the F W , F W and F W filters respectively.This source was detected in late-time Gemini GMOS-N g (cid:48) and r (cid:48) c (cid:13) , 000–000 Maund et al.
Figure 3.
HST imaging of the site of SN 1999ev. From left to right: Pre-explosion WFPC2 WF2 F W image; Post-explosion ACS / WFC F W image;and late-time ACS / WFC F W image. images, but not recovered significantly in the corresponding i (cid:48) im-age, acquired on 2008 Sep 06. Maund & Smartt (2009) measured g (cid:48) = . ± .
04 and r (cid:48) = . ± .
05 (in Vega magnitudes) for thesource at the SN position, and placed a detection limit of i (cid:48) > . HS T
WFPC2 on 2007 Aug 11(for program GO − m F W = . ± . m F W = . ± .
26. We note that this photometry is ap-proximately 0 . ff erence between thesource in the 2007 WFPC2 F W image and our late-time ACS F W image is consistent with the photometry conducted on theimages directly.We recalculated the photometry of the source at the SN posi-tion in the pre-explosion WFPC2 F W image, labeled Source A by Smartt et al. (2004), using HSTphot finding m F W = . ± .
06. Maund & Smartt (2009) derived the i (cid:48) magnitude of theprogenitor, using image subtraction techniques (see Figure 8), of23 . ± .
04 (with a possible 0.15 magnitude systematic uncer-tainty on underlying residual flux in the late-time Gemini image).Unlike the obviously red Source A observed in the pre-explosion HS T
WFPC2 and Gemini GMOS images, it is apparent that Source A (cid:48) in the late-time images is a blue-yellow object (see Figure 6).Even taking into account a colour correction between the late-time F W and Gemini GMOS i (cid:48) photometry (see Fig. 1), a signifi-cant increase in brightness is evident between the two observationsseparated by two years. SN 2004A was discovered by K. Itagaki (Nakano et al. 2004) on2004 Jan 9.4 in the galaxy NGC 6207. Kawakita et al. (2004) spec-troscopically classified the SN as a being a young Type II SN.Hendry et al. (2006) presented photometric and spectroscopic ob-servations of SN 2004A and showed it to be consistent with othernormal Type IIP SNe, such as SN 1999em. Hendry et al. also pre-sented an analysis of the pre-explosion
HS T
WFPC2 observationsof the site of SN 2004A from 2001 Jul 02, in conjunction withpost-explosion ACS WFC observations of the SN acquired on 2004Sep 23. A source was barely recovered at 4 . σ at the SN positionin the pre-explosion F W image. There was no correspondingsource in the pre-explosion F W image, consistent with a star Figure 6.
Late-time colour image (composed of F W , F W and F W images) of the area of M74 around the site of SN 2003gd. Theimage has dimensions 8 (cid:48)(cid:48) × (cid:48)(cid:48) , and is oriented such that North is up, andEast is left. The position of SN 2003gd is located at the centre of the image. with F W − F W > .
05. Hendry et al. concluded that if thiswas the progenitor star an RSG with initial mass 9 + − M (cid:12) ; althoughgiven concerns about the significance of the detection of the sourceat the SN position, Hendry et al. placed a conservative limit on theinitial mass of an undetected progenitor of < M (cid:12) .Late-time observations of the site of SN 2004A were acquiredon 2010 Sep 09, approximately 6.7 years post-discovery. Thelate-time F W observation and the corresponding pre-explosion F W observation are presented on Figure 9. The pre-explosionand late-time observations of the site of SN 2004A are presentedin Figure 9. Using the post-explosion F W image, acquired on2004 Sep 23 with ACS WFC, the position of the SN on the pre-explosion and late-time frames was determined to within 0 . (cid:48)(cid:48) and 0 . (cid:48)(cid:48) , respectively. The SN is not detected significantly in anyof the late-time ACS WFC images. The 3 σ detection limits at theSN position, in the late-time images, were evaluated with artificial c (cid:13)000
05. Hendry et al. concluded that if thiswas the progenitor star an RSG with initial mass 9 + − M (cid:12) ; althoughgiven concerns about the significance of the detection of the sourceat the SN position, Hendry et al. placed a conservative limit on theinitial mass of an undetected progenitor of < M (cid:12) .Late-time observations of the site of SN 2004A were acquiredon 2010 Sep 09, approximately 6.7 years post-discovery. Thelate-time F W observation and the corresponding pre-explosion F W observation are presented on Figure 9. The pre-explosionand late-time observations of the site of SN 2004A are presentedin Figure 9. Using the post-explosion F W image, acquired on2004 Sep 23 with ACS WFC, the position of the SN on the pre-explosion and late-time frames was determined to within 0 . (cid:48)(cid:48) and 0 . (cid:48)(cid:48) , respectively. The SN is not detected significantly in anyof the late-time ACS WFC images. The 3 σ detection limits at theSN position, in the late-time images, were evaluated with artificial c (cid:13)000 , 000–000 ate-time view of progenitors of Type IIP supernovae Figure 7.
HST imaging of the site of SN 2003gd. From left to right: Pre-explosion WFPC2 WF2 F606W image; Post-explosion ACS / HRC F555W image;and late-time ACS / WFC F555W image.
Figure 8.
Pre-explosion and late-time Gemini i (cid:48) observations of the site of SN 2003gd (for further details see Maund & Smartt 2009). star tests to be m F W = . m F W = . m F W = . .In the pre-explosion F W image, HSTphot finds a sourcewithin 0 . (cid:48)(cid:48) of the transformed SN position with m F W = . ± .
19 mags, detected with a signal-to-noise ratio of 5.6 (asshown on Figure 9). The positional uncertainty is slightly largerthan the formal 1 σ uncertainty of the geometric transformationalone. The ability of HSTphot to determine the position of objectsof such brightness, however, is limited, such that the expected un-certainty on the position of the object on the pre-explosion F W image is ∆ r (cid:62) . (cid:48)(cid:48) (Dolphin 2000b). We note that this is thesame object identified by Hendry et al. (2006) as the possible can-didate progenitor, although they measured the source to be ∼ . .
05 magnitudeslarger). We also find the two sources A and B identified by Hendryet al. (2006) are hot pixels . The pixel immediately adjacent tothe pixel hosting the majority of the progenitor candidate’s fluxis a warm pixel; however this pixel has a low dark current (with http: // etc.stsci.edu / etc / input / acs / imaging / http: // / hst / wfpc2 / analysis / wfpc2 hotpix.html low variability) and was corrected by the OTFR pipeline. Wedetermined the 50% completeness level for 3 σ detections for thepre-explosion images using artificial star tests conducting the HST-phot. Artificial stars were placed in a 5px radius around the trans-formed SN position. We find the corresponding detection limits tobe m F W = . ± .
40 and m F W = . ± .
45 mags.The late-time F W image was subtracted from the pre-explosion image, and the di ff erence images is presented on Fig.9. We find residuals at the location of the progenitor source andthe two hotpixels. The disappearance of the pre-explosion progen-itor candidate in these late-time images confirms the authentic-ity of the object as the progenitor. We derived photometry of theresidual at the transformed SN position of m F W = . ± . ∼ − . ± . m F W = . ± .
12 mags. This is similar to the photometry de-rived by Hendry et al. (2006), although for very di ff erent reasonsand improved precision.A similar di ff erence image was determined for the pre-explosion F W image and the late-time F W image. No sig-nificant residual was found in the di ff erence image as expected, http: // / instruments / wfpc2 / Wfpc2 hotpix / / vary 010617 010711 3.dat.Zc (cid:13) , 000–000 Maund et al.
Figure 9.
The site of SN 2004A in NGC 6207. Left) Pre-explosion WFPC2 F W image from 2001 Jul 02 with scale 0 . (cid:48)(cid:48) px − . The centred cross hairsmark the transformed position of SN 2004A and the nearby sources A and B are known hot pixels on the WFPC2 WF2 chip. Centre) Late-time ACS / WFC F W image. Right) Di ff erence image between the late and pre-explosion F W images. The observed residual sources correspond to the two hot pixels, A and B , and the progenitor. given the absence of a source at the transformed SN position in thepre-explosion image. Due to di ff erences between the pre-explosion F W and late-time F W filter transmission functions, we didnot use the di ff erence image to derive detection limits for the pro-genitor. SN 2005cs was discovered by Kloehr et al. (2005) on 2005 Jun27.933 in the galaxy M51. Modjaz et al. (2005) spectroscopicallyclassified SN 2005cs as being a young Type II SN. Richmond &Modjaz (2005) provisionally identified a blue supergiant in the fieldas a possible candidate for the progenitor, although later analysis (inconjunction with high resolution post-explosion HST ACS HRCimages) by Maund et al. (2005) and Li et al. (2006a) found theprogenitor star to be a RSG with initial mass M ZAMS ∼ M (cid:12) .Late-time observations of the site of SN 2005cs were acquiredon 2010 Jul 30 with the ACS / WFC, 5.1 years post-discovery. Acomparison of the pre-explosion and late-time observations of thesite of SN 2005cs is shown as Figs. 10 and 11. In this case, the late-time observations exactly match the pre-explosion observations, us-ing the same filters and detectors. The pre-explosion and late-timeobservations were drizzled to a final common pixel scale of 0 . (cid:48)(cid:48) .Utilising post-explosion ACS HRC observations of SN 2005cs, theSN position was located on the pre-explosion and late-time imagesto within 0 . (cid:48)(cid:48) and 0 . (cid:48)(cid:48) , respectively.Direct photometry of the source detected at the SN position inthe pre-explosion F W yielded m F W = . ± . . ∼ . m F W = . ± . m F W = . ± .
45 and m F W = . ± .
15 mags. These limits areparticularly high, relative to the expected depth for ACS / WFC im-ages of these durations, due to extended emission from the nearbycluster overlapping SN position.As noted by Maund et al. (2005) and Li et al. (2006b), this un-derlying emission can complicate the determination of the photom-etry of the progenitor from the pre-explosion imaging alone. Thishighlights the importance of using image subtraction techniques to accurately derive the progenitor photometry (by subtracting thebackground emission that is constant at both epochs). The di ff er-ence image between the pre-explosion and late-time F W ob-servations is presented on Fig. 11. We measure the brightness ofthe progenitor to be m F W = . ± .
07 mags, which is fainterthan the brightness determined from direct photometry of the pre-explosion source (see above). This magnitude is also significantlyfainter than the photometry of Maund et al. (2005), and slightlybrighter than the photometry of Li et al. (2006a) (who attempted toaccount for pre-explosion flux at the SN position due to the nearbycluster). We note that we find no significant source at the SN posi-tion in the corresponding F W and F W di ff erence images.Artificial star tests, in conjunction with image subtractiontechniques, were used to derive alternative detection limits (seeSection 2.7) for the pre-explosion F W and F W images. Inthe absence of a corresponding late-time F N ACS / WFC im-age, the detection limit on the pre-explosion F N image couldonly be derived using direct recovery of artificial stars on the pre-explosion frame. The photometric completeness functions for thepre-explosion observations in which the progenitor was not de-tected is shown on Fig. 12 and presented in Table 3. The detectionof a residual in isis di ff erence images requires only a significant de-gree of residual flux at the SN position and is less dependent onthe amount of background flux than the direct recovery of artificialstars. There are di ff erences between the detection limits derived onthe pre-explosion images here and the limits presented by Maundet al. (2005) and Li et al. (2006b). These studies used combinationsof the analytical noise expression and artificial star tests, and treatedthe e ff ect of flux from the nearby cluster di ff erently. We note thatour detection limits derived using image subtraction techniques aresignificantly deeper, highlighting the importance of late-time im-ages even in cases where detections of the progenitor are dubiousor unavailable.We also find that there are a number of other sources that areclearly variable between the pre-explosion and late-time images inthe vicinity of SN 2005cs (as shown on Fig. 11). Inspection of thepre-explosion and late-time images shows that the apparent resid-uals in the di ff erence images are associated with stars which areclearly brighter or fainter in the late-time images compared withthe pre-explosion images. Given the density of stars in this field,compared with the sites of the other SNe considered here, it is to c (cid:13) , 000–000 ate-time view of progenitors of Type IIP supernovae Figure 10.
Colour images of the site of SN 2005cs using ACS / WFC beforeexplosion and at late-times. The position of the SN is indicated by the cross-hairs. In the pre-explosion image a red source is clearly visible at the SNposition, and is found to be absent in the late-time image. Each image hasdimension 6 (cid:48)(cid:48) × (cid:48)(cid:48) , and is oriented such that North is up, East is left. be expected that there would be other variable sources in the fieldaround SN 2005cs. The other residuals in the di ff erence imagesare, therefore, consistent with other real variables, but the progeni-tor object is the only star to be absent in one of the two sets of theimages. SN 2006my was discovered on 2006 Nov 8.82UT by K. Itagaki(Nakano & Itagaki 2006) in the galaxy NGC 4651. Stanishev &Nielsen (2006) spectroscopically classified the SN as being a TypeII SN similar to SN 1999em. Li et al. (2007), Leonard et al. (2008)and Crockett et al. (2011) analysed the pre-explosion WFPC2 im-
Figure 12.
Detection completeness functions for pre-explosion
ACS WFC observations of the site of SN 2005cs for the F W , F W and F N filters (in which the progenitor was not detected). The dashed curves in-dicate the completeness functions derived from attempted direct recoveryof artificial stars on the original pre-explosion images, while solid curvesare for detection limits derived in conjunction with image subtraction tech-niques. Table 3.
Detection limits for pre-explosion ACS / WFC F W , F W and F N images of the site of SN 2005cs measured using artificial star testsand direct recovery and image subtraction techniques. Previously deriveddetections limits from Maund et al. (2005) and Li et al. (2006b) are shownfor comparisonRecovery ISIS Maund et al. Li et al.F435W 24 . ± .
20 25 . ± . · · · F555W 24 . ± .
30 26 . ± . . ± . · · · · · · ages of the site of SN 2006my. All three studies commented on thesignificant o ff set between the transformed SN position and a nearbysource recovered in the pre-explosion F W image. Leonard et al.and Crockett et al. concluded that the F W source was unrelatedto the SN, and that the progenitor was not detected in either thepre-explosion F W or F W images.The pre-, post-explosion and late-time F W and F W imaging of the site of SN 2006my is shown on Fig. 13. Followingthe analyses presented by Li et al. (2007), Leonard et al. (2008) andCrockett et al. (2011), we analysed the pre-explosion WFPC2 WF2 F W and F W covering the position of SN 2006my. The SNposition, derived from post-explosion WFPC2 PC1 images, was de-termined on the pre-explosion images using 19 common stars witha resulting uncertainty of 0 . (cid:48)(cid:48) ; larger than achieved by Crockettet al. (2011). The transformed position is found to be in the prox-imity of a cluster of bright pixels and, as previously found by Liet al. (2007), Leonard et al. (2008) and Crockett et al. (2011), thetransformed position is not consistent with the position of the near-est source found by HSTphot in the pre-explosion F W image(see Fig. 13). The source detected by HSTphot is located 0 .
97 pix-els from the transformed SN position, an o ff set significantly largerthan the transformation uncertainty. The transformed SN positionis close to the position of a source in the pre-explosion F W im-age but, as noted by the previous studies, based on the sharpness c (cid:13) , 000–000 Maund et al.
Figure 11.
ACS WFC F W observations of the site of SN 2005cs. Left) Pre-explosion F W image (0.44 years prior to discovery). The cross hairs indicatethe transformed position of the SN and the identified progenitor candidate. Centre) Late-time F W image (5.09 years post-discovery) with the progenitorcandidate absent. Right) Di ff erence image, between the pre-explosion and late-time F W observations, clearly showing a residual at the SN position. value derived by HSTphot, it is not consistent with a point sourceand is located 1 .
19 pixels from the source detected in the F W image. The shift between the pre-explosion F W and F W image was found to be very small: ∆ x = − . ∆ y = . F W source was mea-sured to have brightness 24 . ± .
18, which is approximately0 . ff erent HSTphot settings used here). Sec-tions of the late-time F W and F W images were trans-formed to match the pre-explosion images; resampling the ACSWFC 0 . (cid:48)(cid:48) pixels to 0 . (cid:48)(cid:48) . These images were processed using isis ,and the resulting di ff erence images are shown in Fig. 13. isis sig-nificantly detects a residual, at a distance of only 0 .
11 pixels fromthe transformed position of the SN. Using 5 reference stars, andthe DOLPHOT photometry derived for these stars in the late-time F W image, we derive a magnitude of 24 . ± .
13 mags forthe residual in the di ff erence image. Artificial star tests on the pre-explosion images were used to derive the CTI correction, for a starat the position of the observed residual on the pre-explosion frame,of 0 . ± . F W di ff erence imagecorresponds, therefore, to an object with m F W = . ± . F W image. The fainter F W magnitude derivedusing ISIS, compared to our own HSTphot photometry of the pre-explosion images and the previously reported values, and the appar-ent discrepancy in the position of the pre-explosion source and theSN, most likely reflects that the source in the pre-explosion F W image is a blend of the progenitor with a source due East of the SNposition (which also skews the apparent position of the progenitorsource in that direction). In the analysis of the pre-explosion andlate-time F W images no residual was found in the di ff erenceimage. The possible nature of the pre-explosion F W source isrevealed in the late-time images as a complicated, extended back-ground feature.We used artificial star tests to probe the detection limit of thepre-explosion F W and F W images within a 10 pixel ra-dius of the SN position; deriving 50% completeness limits at 3 σ of m F W = . ± .
65 and m F W = . ± .
75 mags.
In previous studies (e.g. Maund & Smartt 2005; Smartt et al. 2004;Hendry et al. 2006; Li et al. 2006a; Maund et al. 2005), the photo-metric properties of the progenitor and surrounding stars were de-rived through comparison with the ideal supergiant colour sequencepresented by Drilling & Landolt (2000). As noted by Maund (2013,in prep.) there are significant deficiencies with this approach; suchas the requirement for colour transformation equations to transformthe observed photometry to the photometric system of Drilling &Landolt. More recent studies (e.g. Van Dyk et al. 2012; Maund et al.2011; Van Dyk et al. 2011; Fraser et al. 2012; Maund et al. 2013)have shown the benefit in fitting directly to SEDs constructed fromsynthetic spectra with known parameters using the same filters asthe observations.In considering the observed photometry of objects identifiedat the position of the target SN in the pre-explosion images weutilised the BIX nested sampling and BASIE Markov Chain MonteCarlo SED fitting packages described by Maund (2013, in prep.).These two packages allow us to comprehensively probe the e ff ectsof di ff erent stellar parameters on the interpretation of the observedphotometry on a densely sampled grid of model stellar photometry,in the native filter system of the observations. By design, both ofthese SED fitting packages can handle detections and upper limitssimultaneously, although for limited data (i.e. the number of de-tections is less than the number of free parameters) only the BASIEcode can be used to explore the allowed parameter space rather thanlocate a unique solution. Crucially, we can explore the degenera-cies between the parameters (such as temperature and reddening),and implicitly account for correlations between the temperature andbolometric luminosity (through the bolometric correction).Here we use two families of stellar SED models: the ATLAS9(Castelli & Kurucz 2004) and MARCS (Gustafsson et al. 2008)models. Synthetic photometry of these models was conducted usingour own codes.As we expect the progenitors to be cool RSGs ( < K ),we interpret the observed photometry (and upper limits) with re-spect to the 5 M (cid:12) spherical MARCS SEDs, which have been suc-cessfully compared with observations of RSGs in a number of pre-vious studies (e.g. see Levesque et al. 2005; Davies et al. 2013).We assume that RSGs are well described by models with surfacegravity log g = .
0, and fit for the e ff ective temperature ( T ef f ) and c (cid:13) , 000–000 ate-time view of progenitors of Type IIP supernovae Figure 13.
Pre-explosion and late-time HST images of the site of SN 2006my in the F W (top row) and F W (bottom row) bands. The transformedposition of the SN and the positions of the sources detected in the pre-explosion F W and F W (labeled “source”) images are indicated by the circles. the reddening. We consider the e ff ects of foreground and host red-dening to be due to Galactic-like dust (parameterised by E ( B − V );Cardelli et al. 1989). To constrain the e ff ect of reddening due todust expected to be found around RSGs, we consider the redden-ing laws for graphite or silicate dust (parameterised by the opticaldepth τ ν ), contained in spherical shells with ratios for the inner andouter radii of R out / R in = T ef f and E ( B − V ) are de-rived with respect to the ATLAS9 models. We selected those mod-els from the ATLAS9 grid that are consistent with supergiant sur-face gravities (Laidler et al. 2008) . Total reddenings are derivedwith respect to a Cardelli et al. (1989) R V = . http: // / hst / HST overview / documents / synphot / AppA Catalogs4.html the SEDs of compact clusters we adopt the model spectra producedusing the starburst
99 code (Leitherer et al. 1999).For stellar progenitors, we derive masses using our various lu-minosity estimates (depending on the type of dust and star) follow-ing the technique of Smartt et al. (2009). Smartt et al. use predictedluminosities for the end-phases of STARS stellar evolution mod-els (Eldridge & Tout 2004) to derive initial masses for progenitorswith luminosities constraints derived from the observations. For agiven luminosity, the progenitor is considered to lie in the massrange bounded at one end by the most massive star to end core Heburning at that luminosity, and at the other by the least massive starto proceed to model termination (the onset of core Ne burning) atthat luminosity (see Smartt et al. 2009, and their Fig. 1). We useSTARS models calculated at integer initial masses, with the appro-priate metallicities, and interpolate to determine the luminosities atthe end of core He burning and the beginning of core Ne burning.This scheme characterises possible RSG progenitors. We also notethat, at lower masses, some stars will undergo second dredge-up,causing them to become Asymptotic Giant Branch (AGB) stars thatare cooler but more luminous than the similar mass stars that die asRSGs. Based on the observed temperature range of RSGs, derivedusing MARCS spectra (Levesque et al. 2005), we use a tempera-ture threshold of 3400 K , above and below which we consider starsto be RSGs or AGB stars, respectively.In deriving posterior probability density functions (pdfs) forthe initial masses for the progenitors, we also consider the e ff ectof prior information from the IMF. We apply a weighting factor tothe posterior pdfs ∝ M − . for a Salpeter (1955) IMF, to follow the c (cid:13) , 000–000 Maund et al. weighting scheme applied by Smartt et al. (2009) for their analysisof the Type IIP SN progenitor population.
The presence of a source in the late-time images precludes the useof image subtraction techniques to further analyse the nature of thepre-explosion source. We note that the pre-explosion F W mag-nitude of the source at the SN position is of similar magnitude tothe source in the late-time F W image. The di ff erence betweenthe pre-explosion and late-time photometry is not significant witha p-value of 0.71 (using a simple z -test). This lends support to thehypothesis that the source observed in the pre-explosion images atthe SN position has been recovered in the late-time images, and thatthe original identification of the progenitor presented by Maund &Smartt (2005) is, at least partially, incorrect. We suggest three pos-sible scenarios for the nature of the source at the SN position:(i) The source at the SN position in the pre-explosion and late-time images is a host cluster that contained the now absent progen-itor.(ii) The source observed in the pre-explosion and late-time im-age is an unrelated star that is coincident with the line-of-sight tothe SN. While the late-time observations do suggest a large, youngstellar population hidden by the large dust sheet, the determinationof the likelihood of a chance alignment is non-trivial. Given theastrometric coincidence < . (cid:48)(cid:48) , it is likely to be very low.(iii) The source observed in the late-time images is an unre-solved light echo, and the pre-explosion source has now disap-peared. Given the observation of evolving light echoes around theposition of the SN, and the obvious amount of dust in the vicinityof the SN, the apparent late-time brightness may be due to a lightecho from dust immediately behind the SN. We find this scenariounlikely, as it requires the progenitor and light echo to have coinci-dentally similar brightness.Given the nature of the late-time three-colour imaging it is not pos-sible to unambiguously distinguish between the di ff erent scenarios.The late-time images were used to examine the consequences of theprogenitor residing in a host cluster that was observed in the pre-explosion and late-time images. The shape of the source in the late-time images was measured using the ishape package (Larsen 1999).In each filter band, ishape returned a significantly better fit with aMo ff at function over a delta function ( χ (Mo ff at) /χ (delta) < . ff ective radius R ef f > . × the Full Width at Half Maxi-mum (Larsen 1999). The e ff ective radius of the source was mea-sured to be 0 . + . − . , 1 . + . − . and 1 . + . − . pixels in the F W , F W and F W , respectively (with a pixel scale correspondingto 2 . − at the distance of NGC 4274). The large error barsare symptomatic of the complexity of the region hosting the SN,in particular with the proximity of light echoes and the apparentfaintness of the source. The ishape analysis was also conducted onsix nearby objects that were all found to be consistent with point-like, stellar sources - suggesting that ishape does have the capa-bility, under the conditions of the late-time images, to di ff erenti-ate extended sources from point-like sources. To further explorethe implications of a host cluster for the progenitor, the late-time ACS photometry was compared with S tarburst
99 models (Lei-therer et al. 1999); and the results of this fit is shown as Figure14. Given the three colour photometry, there are two allowed so-lutions: a moderately reddened older solution (40 −
100 Myr) im-plying M ZAMS < M (cid:12) ; and a heavily reddened younger solution ( <
10 Myr) implying M ZAMS > M (cid:12) . In addition, given the cri-terion presented by Bastian et al. (2005), that point-like objectswith M V < − . E ( B − V ) (cid:62) .
76 (assuming an R V = . E ( B − V ) = . ± .
32 was measured. The large errorbar is consistent with both the poor photometric errors for eachof the six nearby sources and the large scatter in reddenings inthe sample. The reddenings towards these objects are significantlygreater than expected for just pure foreground Galactic reddening( E ( B − V ) = .
2) towards NGC 4274. This may reflect a com-plex dust distribution where some of the sources are in front of thedust sheet and others may be embedded. This suggests that theremight be significant reddening towards the source at the SN posi-tion, however this is not conclusive.If the late-time source is, in fact, a light echo, then the late-time images do not provide any further insight into the properties ofthe source in the pre-explosion images. Our inability to the confirmthe disappearance of the progenitor also means that the late-timeimages cannot be used to rule out the possibility that the object inthe pre-explosion and late-time images is an unrelated object in theline-of-sight. High-resolution near-infrared observations, with theHST, could be used to probe the nature of the stellar populationbehind the dust sheet (to examine the density of objects along theline-of-sight) as well as provide further constraints on the nature ofthe source in late-time images and nature of the obscuring dust.The ambiguity of the nature of the object at the SN positionin the pre-explosion and late-time images means that, althoughSN 1999ev may have had an identified progenitor of some kind,the previously derived progenitor properties are unreliable; even inthe interpretation that the source is a cluster, the reliance on three-colour photometry leads to degeneracies in the reddening-age solu-tions that prohibit a precise initial mass estimate for the progenitor.
The photometry of 30 stars within 4 (cid:48)(cid:48) ( ∼ E ( B − V ) = . ± .
04 (see Figure 15). The amountof reddening is consistent with the reddening previously estimatedfrom three colour photometry of the surrounding stars using earlypost-explosion
ACS HRC images (Smartt et al. 2004) and from thecolour evolution of the SN itself (Hendry et al. 2005).Given the detection of the progenitor in pre-explosion obser-vations in two filters, we consider the roles of three di ff erent typesof reddening: 1) unconstrained reddening, with an interstellar red-dening law; 2) an interstellar reddening component consistent withthe observed reddening to the surrounding stars and an uncon-strained degree of reddening arising from Graphite dust around theprogenitor; and 3) the same as 2, except with Silicate dust. Thecorresponding regions of the parameter space and the Hertzsprung-Russell (HR) diagram, allowed by the pre-explosion observationsin conjunction with half-solar metallicity MARCS SEDs, are pre-sented on Fig. 16. Given the observed colour of the progenitor, re-gardless of the amount of reddening, we find T ef f > K and the We note that the starburst
99 code uses the Padova stellar evolution (Gi-rardi et al. 2002) models, and the masses we report for the derived ages arefrom these models. c (cid:13) , 000–000 ate-time view of progenitors of Type IIP supernovae Figure 15.
Reddenings for stars surrounding the sites of SNe 2003gd, 2004A, 2005cs and 2006my (as a function of distance from the SN positions).
Figure 14.
The posterior probability distribution for SED fits of the late-time photometry of the source at the position of SN 1999ev to cluster SEDmodels produced using S tarburst
99. The contours contain the 68% and95% of the probability, while the most likely solution is indicated with thecross. The vertical dashed line corresponds to E ( B − V ) = .
76; for E ( B − V ) (cid:62) .
76, the absolute magnitude of the source is M F W < − .
6. Thehorizontal lines indicate the lifetimes of stars with a given initial mass aspredicted by the Padova stellar evolution code (Girardi et al. 2002). radius of the progenitor constrained to be 200 < R < R (cid:12) . Theflatter nature of the graphite and silicate reddening laws, comparedto the Cardelli et al. reddening law, leads to tighter luminosity con-straints and a weaker dependence on the temperature / colour of theprogenitor. The masses inferred for the progenitor are relatively in-sensitive to the choice of reddening law, with M init ∼ − ± . M (cid:12) .Given the reddening derived towards SN 2003gd, we find thatthe SED of the object recovered at the SN position in the late-timeimages ( A (cid:48) ) is consistent with a black body with temperature T ∼ The three-colour photometry of 12 stars within 5.25 (cid:48)(cid:48) ( ≈ E ( B − V ) = . ± . E ( B − V ) = . ± .
03, derived using three colour photometryfrom the post-explosion
ACS / WFC images.Due to the single F W detection of the progenitor, andthe upper F W limit, an SED fit with > F W upper limit constrains the temperatureof the progenitor to be lower than 3700K; although this is sensitiveto the reddening prior, such that higher reddenings may permit hot-ter progenitors. The corresponding region on the HR diagram isalso shown on Figure 17. As the reddening is e ff ectively fixed, theslope of the contours reflects the increasing bolometric correctionfor cooler stars. If the reddening inferred from the surrounding starsis, instead, a lower limit on the reddening towards the progenitor(due to an additional component of reddening due to circumstellerdust), then we can only derive a lower limit on the luminosity ofthe progenitor. The contours also include those stars in the initialmass range 5 − M (cid:12) that are expected, at this metallicity, to undergosecond dredge up and become AGB stars. The photometry, from the pre-explosion images, of 20 stars within2 (cid:48)(cid:48) ( ∼ E ( B − V ) = . ± .
05 (see Fig. 15).In order to constrain the properties of the progenitors weutilised the F W magnitude presented here, measured using im-age subtraction, and the revised detection thresholds for the pre-explosion ACS F W , F W and F N images. Furthermore,we adopted the infrared detection limits for the progenitor reportedfor pre-explosion Gemini NIRI JHK images (Maund et al. 2005)and
NICMOS F W , F W F M images (Li et al. 2006a).As the NICMOS images are deeper, they form the principal con-straint on the SED of the progenitor in the IR, however we includethe Gemini
NIRI limits in our calculation for completeness. Weexplored the same parameter space, for the same combinations ofreddening components, as for the progenitor of SN 2003gd (seeSection 4.2) and the results are presented on Fig. 18.Similarly to the progenitor of SN 2003gd, the graphite and c (cid:13) , 000–000 Maund et al.
Figure 16.
The parameters of the progenitor of SN 2003gd for di ff erent reddening components. In each panel the contours contain 68% and 95% of thetotal probability. Top Row),Left)
The temperature and reddening of the progenitor assuming Galactic-like dust;
Centre) the progenitor’s location on the HRdiagram (also shown are the locations of stellar evolution models of given initial mass for the end of core He burning ( (cid:78) ), the onset of Ne burning ( (cid:4) ) and theendpoints for those models that undergo second dredge-up ( • ). Dotted grey lines indicate lines of constant progenitor radius.; and Right) the inferred initialmass probability density function with no weighting (solid line) and weighting according to the initial mass function (dotted line).
Middle Row)
The same asthe top row, but for a Cardelli et al. (1989) reddening component of E ( B − V ) = . ± .
04, derived from the surrounding stars and a unconstrained reddeningcomponent for graphite dust around the progenitor (solid contours are for R out / R in = R out / R in = R out / R in = Bottom Row)
The same as the middle row, but for silicate dust. silicate reddening laws yield flatter contours of the HR diagram,making the dependence of the luminosity on the temperature lessextreme. For each reddening type, there are two islands of preferredsolutions: a cool, low reddened solution and a hotter, reddened so-lution; reflecting the severe constraints from non-detections in theinfrared and in blue, respectively. The bimodal probability distribu-tion in T ef f and E ( B − V ) leads to a skewed mass probability den-sity function for a Cardelli et al. reddening law extending to highermasses, although the peak of the distribution occurs at ∼ M (cid:12) .For the graphite and silicate reddening laws, the unweighted massprobability density function is more symmetric and peaks around ∼ M (cid:12) . The strict infrared limits exclude the possibility of the progenitor being a massive AGB stars for all the reddening types(Eldridge et al. 2007). Previously Li et al. (2007) suggested that, apart from a Galacticreddening component, there was no evidence for a host redden-ing component in spectra of SN 2006my; this value was similarlyused by progenitor studies conducted by Crockett et al. (2011) andLeonard et al. (2008). We utilised the HSTphot photometry of thepost-explosion WFPC2 images of SN 2006my to study the redden-ing associated with the surrounding stellar population. We selected46 good stars ( χ < . | sharp | < .
3) with complete three-colour c (cid:13) , 000–000 ate-time view of progenitors of Type IIP supernovae Figure 17.
The same as Fig. 16 but for pre-explosion observations of the progenitor of SN 2004A, assuming only a Cardelli et al. reddening law with E ( B − V ) = . ± .
03 derived from surrounding stars.
Figure 18.
The same as Fig. 16, but for the pre-explosion observations of the progenitor of SN 2005cs.c (cid:13) , 000–000 Maund et al. photometry within 5 (cid:48)(cid:48) ( ∼ E ( B − V ) = . ± .
25 (seeFig. 15). The large scatter reflects the poor quality of the
WFPC2 photometry, as well as likely di ff erences in the reddening betweenthe individual stars. The relative colour constraint provided by thepre-explosion F W limit and the F W detection of the pro-genitor is insu ffi cient to limit the temperature of the progenitor fortemperatures below < K . The allowed regions of the HR di-agram, for the two reddening estimates, is shown on Figure 4.5.As for SN 2004A (see Section 4.3), the shape of the contours isdictated by the larger bolometric correction at cooler temperatures.The lack of an infrared detection or detection limit for the progen-itor means that both the RSG and AGB solutions are allowed. Thehigher reddening inferred from the photometry of the surroundingstars implies higher luminosities for the progenitor, than for justGalactic foreground reddening, and leads to a higher initial mass(13 . ± . . ± . M (cid:12) ).Given the apparent o ff set between the transformed SN po-sition and the source recovered in the pre-explosion F W im-age, we explored the possible causes for the apparent discrepancy.We note that all three previous studies, and our own results, agreethat there is a significant o ff set between the transformed SN po-sition and the pre-explosion F W source. As noted by Leonardet al. (2008), in conjunction with Dolphin (2000b), the positionaluncertainty for an isolated source with the brightness of the pre-explosion F W source is ∼ . x = . ± . y = . ± . ff set from the SN position by 0.4 pixels. From the outcome of theMonte Carlo simulations we note three e ff ects:(i) The positional uncertainty is larger for faint sources and willbe dependent on pixel noise statistics in the main pixel containingsource flux and the surrounding pixels.(ii) Given the subsampled nature of the PSF in WFPC
WFPC2 images for isolated sources, it isalso important to consider the e ff ects of nearby sources (within afew pixels) that may skew / bias the centring algorithm away fromthe true source position. From the Monte Carlo simulations, giventhe environment and pixel noise at the SN position, we estimatethe uncertainty on the position of the pre-explosion F W sourcemay be ∼ . (cid:48)(cid:48) . With such large uncertainties, the apparent o ff setbetween the SN and the pre-explosion source would be only ∼ σ given the o ff sets calculated by Crockett et al. (2011) and Leonardet al. (2008); however, we caution that our lower quality geometrictransformation results in a 1 . σ o ff set. A summary of the masses derived for the progenitors consideredhere is presented on Table 4. The biggest di ff erences between ourfindings and those of Smartt et al. (2009) are for the progenitors ofSNe 1999ev and 2006my. Our late-imaging shows that the natureof the source found at the SN position in pre-explosion observa-tions of the site of SN 1999ev is, at best, uncertain; whilst, at worst,a misidentification of a host cluster or unrelated coincident star. Asthe nature of this source is not clarified in the late-time imaging, theprogenitor of SN 1999ev should no longer be considered in progen-itor population statistics. In the analysis presented by Smartt et al.(2009), the progenitor of SN 1999ev had the distinction of havingthe highest mass inferred for a detected Type IIP SN progenitor,and so set the maximum mass limit for stars to explode as RSGsand produce Type IIP SNe. With the removal of the SN 1999evprogenitor from the population statistics, the maximum mass limitwill actually drop and make the “Red Supergiant Problem” evenmore severe. The late-time imaging of the site of SN 2006my hasshown that the progenitor was detected in the pre-explosion obser-vations, and the upper mass limit quoted by Smartt et al. shouldnow be quoted as a detected progenitor with a corresponding massestimate.The mass estimates derived for the confirmed progenitors aregenerally higher (by ∼ M (cid:12) ) than those presented by Smartt et al.;in part, due to the slightly larger foreground and host reddenings weinferred towards the progenitors from the colours of the surround-ing stars. In addition, for SN 2003gd and 2005cs we consideredadditional reddening components to the reddening from interstel-lar dust, in keeping with the expectation that there is dust local toprogenitor that is not probed by the surrounding stars or the obser-vations of the SNe. We also note that our uncertainties are smallerthan those quoted by Smartt et al.: we considered the e ff ects ofuncertainties of the luminosity convolved with the flat probabil-ity distribution of the star having a mass in the range bounded bythe maximum mass star to end core He-burning at that luminosityand the minimum mass star to begin Ne burning at that luminos-ity. The initial mass probability density functions are, apart fromSN 2005cs, approximately symmetric and almost follow a normaldistribution. We believe this is a fairer presentation of the initialmasses for the progenitors and their uncertainties. The applicationof a weighting to the initial mass pdf, according to the IMF, hasa small e ff ect in shifting the pdf to slightly lower masses. UnlikeSmartt et al. (2009), who used this weighting scheme to “truncate”their large uncertainties, we find that the e ff ect of weighting on therelative width of the initial mass pdfs is minor.The slight increase in the inferred progenitor mass does nothelp rectify the apparent discrepancy between these “evolutionary”masses and the progenitor masses derived from hydrodynamicalmodels. Utrobin & Chugai (2008) found an initial mass for the pro-genitor of SN 2005cs of 18 . ± M (cid:12) , which is at odds with thenew masses derived here, regardless of the choice of reddening. AsUtrobin & Chugai (2008) suggest, additional dust in a circumstel-lar shell could lower the apparent luminosity of the progenitor anddecrease the mass inferred from pre-explosion observations. Thenature of the pre-explosion observations of SN 2005cs, in particu-lar the strict near-infrared upper limits, severely limits the amountof reddening that the progenitor might undergo. The constraints onthe radius for the progenitor of SN 2005cs are also below the ra-dius inferred by Utrobin & Chugai (2008) of 600 ± R (cid:12) , but notsignificantly discrepant.It is clear from the analysis presented in Section 4, that the c (cid:13) , 000–000 ate-time view of progenitors of Type IIP supernovae Figure 19.
The parameters of the progenitor of SN 2006my.
Left)
The progenitor of SN 2006my on the HR diagram, assuming E ( B − V ) = . ± . E ( B − V ) = .
027 (dotted contours).
Right)
The initial mass probability density functions for the progenitor of SN 2006my, given E ( B − V ) = . ± .
25 (heavy curves) and E ( B − V ) = .
027 (grey curves).
Figure 20.
The results of position determinations for the pre-explosion source, using DAOphot and the ofilter centring algorithm, for 1000 Monte Carlosimulations of the pre-explosion F W image of the site of SN 2006my. Left)
The positions of the recovered sources in the coordinate frame of the original
WFPC + , while the mean position derivedform the Monte Carlo simulations is indicated by the (cid:63) ; Centre and Right)
The position distributions in x and y coordinates for the recovered sources; the greyline indicates a Gaussian fit to the distributions. available pre-explosion observations of a given SN progenitor dic-tates the degree of analysis that may be conducted. Given the pres-ence of constraining optical and infrared upper limits, the possiblee ff ect of circumstellar reddening on the progenitor of SN 2005cswas evaluated despite having a detection in only one band. Thisdemonstrates the importance for having good pre-explosion ob-servations in the near-infrared for studying the cool progenitorsof Type IIP SNe, even if the progenitor is not detected at thosewavelengths. The two detections of the progenitor of SN 2003gd,at di ff erent wavelengths, enabled similar constraints for reddeningdue to circumstellar dust. Conversely, the analyses of SN 2004Aand 2006my were limited by them having only a single detectionand only loose constraint on the progenitor colour at a bluer wave-length.Due to the lack of constraints on reddening due to circum-stellar dust, the final derived masses for the progenitors of SNe2004A and 2006my may represent, in actuality, lower mass lim-its. For this study, we have assumed the circumstellar dust followsthe reddenings laws proposed by Kochanek et al. (2012), for thecase of the progenitor of SN 2012aw. Conversely, Van Dyk et al. (2012) suggested a steeper reddening law, still following Cardelliet al., but with R V ≈ .
35; leading to an overall higher extinction.The Kochanek et al. (2012) reddening law was determined usingmodels of specific dust compositions, of either graphite or silicatedust, expected to be found around RSGs, whereas Van Dyk et al.(2012) estimated the change in reddening law based on observa-tions of Galactic RSGs. For SN 2012aw (Fraser et al. 2012; VanDyk et al. 2012), the progenitor was detected in four bands, andthe degeneracies between reddening, reddening law and tempera-ture could not be broken; suggesting the full determination of theparameters, independent of the assumptions of reddening laws, re-quires detections at > ff ect all progenitors but not be apparent post-explosion.For their sample of RSGs, Davies et al. (2013) measure extinctionsarising from circumstellar dust in the range A V = . − .
0. A fur-ther issue, that we have not explored, was suggested by Walmswell c (cid:13) , 000–000 Maund et al. & Eldridge (2012, and references therein) that the amount of dustin the circumstellar medium is related to the mass loss rate of theRSG and, ultimately, its bolometric luminosity.A corresponding issue to the reddening problem is the tem-perature. We note that, for the progenitors with constraining pre-explosion observations, we find the allowed temperature range tobe generally hotter than the predicted endpoints for the stellar evo-lution models, but are consistent with the recent reappraisal of RSGtemperatures by Davies et al. (2013). The lower limit of the temper-ature scales for the progenitors of SNe 2003gd and 2005cs suggeststhat they have a spectral type no later than M0, which correspondsto the predicted positions for stars that have just finished core He-burning; the pre-explosion observations of SNe 2003gd and 2005cssuggest that the progenitors were not massive AGB stars (Eldridgeet al. 2007; Siess 2007). It is only in the poorly constrained cases,for the progenitors of SNe 2004A and 2006my, that we cannot ex-clude cooler temperatures that might be associated with massiveAGB stars. With limited observations in the optical (in particularthe B and V bands), it is di ffi cult to place limits on the maximumtemperature of the progenitor, as hotter temperatures can always beaccommodated with additional reddening. In the case of the pro-genitor of SN 2005cs, the requirement that the progenitor ended itslife as an RSG has serious implications for the interpretation of thesubsequent SN as a low luminosity “Electron-Capture” SN (Janka2012).Our late-time imaging campaign has shown that, for the caseof SN 2003gd, the possibilities of recovering precise photometryof the progenitor through template subtraction may be underminedby rebrightening of the SN at late-times. In the case of SN 2003gd,the previous analysis of Maund & Smartt (2009) was fortuitous inthat it managed to observe the SN before it rebrightened with avery strict brightness limit of the SN in the late-time i (cid:48) image. Suchrebrightening is not without precedent; Kotak et al. (2009) observedthe optical lightcurve of SN 2004et to rebrighten (by ∼ V )at optical and infrared wavelengths.As noted in Section 4.5, the apparent discrepancy between theposition of the progenitor of SN 2006my on the pre-explosion im-ages and the transformed SN position does raise questions abouthow positional uncertainties are handled. Maund & Smartt (2005)determined the positional precision using the standard deviation ofthe four di ff erent centring methods available to DAOphot (centroid,ofilter, gauss and psf). It must be noted, however, that these centringtechniques are not providing independent estimates of an object’slocation. In the future, it may be preferable to choose a single cen-tering technique, and consider the role of Poisson noise (both objectand background) and read out noise in each pixel on the determina-tion of an object’s position. Furthermore, tests using the centroidingroutines in both the DAOphot and SAO image DS9 packages hasshown that, in the case of isolated objects in subsampled images,centroiding can be a relatively blunt tool (providing default posi-tions located in the centre of pixels). In considering flux deficitsusing image subtraction techniques, rather than appealing to as-trometric coincidence, we have shown that confirmation of a staras being the actual progenitor requires observing it to have disap-peared. We have presented late-time imaging of the sites of five Type IIPSNe with pre-explosion HST images, in which progenitor candi-dates were detected. In three of the cases (2003gd, 2004A and 2005cs), our previous identifications have been confirmed and wefind initial masses for these stars in the range 6 − M (cid:12) . The pre-explosion observations of SNe 2003gd and 2005cs are su ffi cient toplace constraints on the progenitor mass that are relatively insensi-tive to the amount and type of dust around these progenitors. Giventhe similarities in brightness between the pre-explosion and late-time sources detected at the position of SN 1999ev, we concludethe progenitor identification for this SN is unsafe and suggest thepre-explosion source may be a reddened host cluster; although thethree-colour late-time imaging is insu ffi cient to place a tight con-straint on the age or reddening of a such cluster. The analysis of thepre-explosion and late-time observations of the site of SN 2006myhave revealed that the source previously thought to be significantlyo ff set from the SN position has disappeared. The astrometric coin-cidence of the residual in the di ff erence image with the transformedSN position suggests it was the progenitor object.Far from providing just simple confirmation of a progenitor’sidentity (through its disappearance), our analysis shows late-timeimaging is crucial for conducting a deeper and more precise anal-ysis of the properties of a progenitor than is a ff orded by fortuitouspre-explosion observations alone. The power of the application oflate-time imaging, for studying progenitors, is demonstrated by thesignificantly deeper detection limits that may be achieved by us-ing artificial star tests in conjunction with image subtraction tech-niques. ACKNOWLEDGEMENTS
Based on observations made with the NASA / ESA Hubble SpaceTelescope, which is operated by the Association of Universities forResearch in Astronomy, Inc., under NASA contract NAS 5-26555.These observations are associated with program GO-11675. Theresearch of JRM is funded through a Royal Society University Re-search Fellowship. We thank Stephen Smartt and Steen Hansen fortheir useful comments.
References
Alard C., 2000, A&AS, 144, 363Alard C., Lupton R. H., 1998, ApJ, 503, 325Annibali F., Aloisi A., Mack J., Tosi M., van der Marel R. P., An-geretti L., Leitherer C., Sirianni M., 2008, AJ, 135, 1900Bastian N., Gieles M., Efremov Y. N., Lamers H. J. G. L. M., 2005,A&A, 443, 79Cardelli J. A., Clayton G. C., Mathis J. S., 1989, ApJ, 345, 245Castelli F., Kurucz R. L., 2004, ArXiv Astrophysics e-printsChiaberge M., Lim P. L., Kozhurina-Platais V., Sirianni M., MackJ., 2009, Technical report, Updated CTE photometric correc-tion for WFC and HRC. STScICrockett R. M., Smartt S. J., Pastorello A., Eldridge J. J., StephensA. W., Maund J. R., Mattila S., 2011, MNRAS, 410, 2767Davies B., Kudritzki R.-P., Plez B., Trager S., Lancon A., Gazak Z.,Bergemann M., Evans C., Chiavassa A., 2013, ArXiv e-printsDolphin A. E., 2000a, PASP, 112, 1397Dolphin A. E., 2000b, PASP, 112, 1383Drilling J. S., Landolt A. U., 2000, in Cox A. N., ed., , Allen’sAstrophysical Quantities, 4 edn, AIP, New YorkEldridge J. J., Mattila S., Smartt S. J., 2007, MNRAS, 376, L52Eldridge J. J., Tout C. A., 2004, MNRAS, 353, 87Evans R., McNaught R. H., 2003, IAUC, 8150, 2 c (cid:13) , 000–000 ate-time view of progenitors of Type IIP supernovae Table 4.
Final results for the progenitors of SNe 1999ev, 2003gd, 2004A, 2005cs and 2006my, for di ff erent reddenings due to instellar dust (CCM99) andcircumstellar dust reddening laws (CSM Graphite and CSM Silicate).SN CCM89 CSM Graphite CSM Silicatew / IMF w / o IMF w / IMF w / o IMF w / IMF w / o IMF1999ev Likely cluster · · · · · · · · · · · · . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . · · · · · · · · · · · · . + . − . a . + . − . a . ± . . ± . . ± . . ± . b . ± . . ± . · · · · · · · · · · · · c . ± . . ± . · · · · · · · · · · · · a The values corresponding the mode and 68% probability intervals. b Assuming foreground reddening E ( B − V ) = . c Assuming E ( B − V ) = . ± . Fraser M., Maund J. R., Smartt S. J., Botticella M.-T., Dall’Ora M.,Inserra C., et al. 2012, ArXiv e-printsGarnavich P., Jha S., Kirshner R., Challis P., 1999, IAUC, 7306, 1Girardi L., Bertelli G., Bressan A., Chiosi C., GroenewegenM. A. T., Marigo P., Salasnich B., Weiss A., 2002, A&A, 391,195Gustafsson B., Edvardsson B., Eriksson K., Jørgensen U. G., Nord-lund Å., Plez B., 2008, A&A, 486, 951Hendry M. A., Smartt S. J., Crockett R. M., Maund J. R., Gal-YamA., Moon D., Cenko S. B., Fox D. W., Kudritzki R. P., BennC. R., Østensen R., 2006, MNRAS, 369, 1303Hendry M. A., Smartt S. J., Maund J. R., Pastorello A., ZampieriL., Benetti S., Turatto M., et al. 2005, MNRAS, 359, 906Holtzman J. A., Hester J. J., Casertano S., et al. 1995, PASP, 107,156Hurst G. M., Boles T., Armstrong M., Schwartz M., 1999, IAUC,7306, 1Janka H.-T., 2012, Annual Review of Nuclear and Particle Science,62, 407Kawakita H., Kinugasa K., Ayani K., Yamaoka H., 2004, IAUC,8266, 2Kloehr W., Muendlein R., Li W., Yamaoka H., Itagaki K., 2005,IAUC, 8553, 1Kochanek C. S., Khan R., Dai X., 2012, ApJ, 759, 20Kotak R., Meikle W. P. S., Farrah D., Gerardy C. L., Foley R. J.,Van Dyk S. D., Fransson C., Lundqvist P., Sollerman J., FesenR., Filippenko A. V., Mattila S., Silverman J. M., AndersenA. C., H¨oflich P. A., Pozzo M., Wheeler J. C., 2009, ApJ, 704,306Kotak R., Meikle W. P. S., Smartt S. J., Benn C., 2003, IAUC, 8152,1Laidler V., Bo ffi F., Barlow T., Brown T., Friedman S., JesterS., Maiz Apellaniz J., Pro ffi t C., 2008, Synphot Data UsersGuide. ”STScI”, ”Baltimore”Larsen S. S., 1999, A&AS, 139, 393Leitherer C., Schaerer D., Goldader J. D., Gonz´alez Delgado R. M.,Robert C., Kune D. F., de Mello D. F., Devost D., HeckmanT. M., 1999, ApJS, 123, 3Leonard D. C., Gal-Yam A., Fox D. B., Cameron P. B., JohanssonE. M., Kraus A. L., Mignant D. L., van Dam M. A., 2008,PASP, 120, 1259Levesque E. M., Massey P., Olsen K. A. G., Plez B., Josselin E.,Maeder A., Meynet G., 2005, ApJ, 628, 973Li W., Van Dyk S. D., Filippenko A. V., Cuillandre J.-C., Jha S., Bloom J. S., Riess A. G., Livio M., 2006a, ApJ, 641, 1060Li W., Van Dyk S. D., Filippenko A. V., Cuillandre J.-C., Jha S.,Bloom J. S., Riess A. G., Livio M., 2006b, ApJ, 641, 1060Li W., Wang X., Van Dyk S. D., Cuillandre J.-C., Foley R. J., Fil-ippenko A. V., 2007, ApJ, 661, 1013Mattila S., Smartt S. J., Eldridge J. J., Maund J. R., Crockett R. M.,Danziger I. J., 2008, ApJL, 688, L91Maund J. R., Fraser M., Ergon M., Pastorello A., Smartt S. J.,Sollerman J., Benetti S., Botticella M. ., Bufano F., DanzigerI. J., Kotak R., Magill L., Stephens A. W., Valenti S., 2011,ArXiv e-printsMaund J. R., Fraser M., Smartt S. J., Botticella M. T., Barbarino C.,Childress M., Gal-Yam A., Inserra C., Pignata G., Reichart D.,Schmidt B., Sollerman J., Taddia F., Tomasella L., Valenti S.,Yaron O., 2013, ArXiv e-printsMaund J. R., Smartt S. J., 2005, MNRAS, 360, 288Maund J. R., Smartt S. J., 2009, Science, 324, 486Maund J. R., Smartt S. J., Danziger I. J., 2005, MNRAS, 364, L33Modjaz M., Kirshner R., Challis P., Hutchins R., 2005, IAUC,8555, 1Nakano S., Itagaki K., 2006, Central Bureau Electronic Telegrams,727, 1Nakano S., Itagaki K., Kushida R., Kushida Y., 2004, IAUC, 8265,1Otsuka M., Meixner M., Panagia N., Fabbri J., Barlow M. J., Clay-ton G. C., Gallagher J. S., Sugerman B. E. K., Wesson R.,Andrews J. E., Ercolano B., Welch D., 2012, ApJ, 744, 26Richmond M. W., Modjaz M., 2005, IAUC, 8555, 2Salpeter E. E., 1955, ApJ, 121, 161Schlafly E. F., Finkbeiner D. P., 2011, ApJ, 737, 103Siess L., 2007, A&A, 476, 893Sirianni M., Jee M. J., Ben´ıtez N., Blakeslee J. P., Martel A. R.,Meurer G., Clampin M., De Marchi G., Ford H. C., GillilandR., Hartig G. F., Illingworth G. D., Mack J., McCann W. J.,2005, PASP, 117, 1049Smartt S. J., 2009, ARAA, 47, 63Smartt S. J., Eldridge J. J., Crockett R. M., Maund J. R., 2009,MNRAS, 395, 1409Smartt S. J., Maund J. R., Hendry M. A., Benn C. R., 2003, IAUC,8152, 4Smartt S. J., Maund J. R., Hendry M. A., Tout C. A., Gilmore G. F.,Mattila S., Benn C. R., 2004, Science, 303, 499Smith N., Li W., Filippenko A. V., Chornock R., 2011, MNRAS,412, 1522 c (cid:13) , 000–000 Maund et al.
Stanishev V., Nielsen T. B., 2006, Central Bureau Electronic Tele-grams, 737, 1Utrobin V. P., Chugai N. N., 2008, A&A, 491, 507Van Dyk S. D., Cenko S. B., Poznanski D., Arcavi I., Gal-Yam A.,Filippenko A. V., Silverio K., Stockton A., Cuillandre J.-C.,Marcy G. W., Howard A. W., Isaacson H., 2012, ApJ, 756,131Van Dyk S. D., Davidge T. J., Elias-Rosa N., Taubenberger S., LiW., Levesque E. M., Howerton S., Pignata G., Morrell N.,Hamuy M., Filippenko A. V., 2012, AJ, 143, 19Van Dyk S. D., Li W., Cenko S. B., Kasliwal M. M., Horesh A.,Ofek E. O., Kraus A. L., Silverman J. M., Arcavi I., Filip-penko A. V., Gal-Yam A., Quimby R. M., Kulkarni S. R.,Yaron O., Polishook D., 2011, ArXiv e-printsVan Dyk S. D., Li W., Filippenko A. V., 2003a, PASP, 115, 1Van Dyk S. D., Li W., Filippenko A. V., 2003b, PASP, 115, 1289Walmswell J. J., Eldridge J. J., 2012, MNRAS, 419, 2054 c (cid:13)000