1: \documentclass[12pt,preprint]{aastex}

2: \begin{document}

3: 

4: \title{Re-analysis of VLT Data for M83 with Image Subtraction ---

5: Nine-fold Increase in Number of Cepheids}

6: 

7: \author{A. Z. Bonanos, K. Z. Stanek}

8: \affil{Harvard-Smithsonian Center for Astrophysics, 60 Garden St.,

9: Cambridge, MA~02138}

10: \affil{\tt e-mail: abonanos@cfa.harvard.edu, kstanek@cfa.harvard.edu}

11: 

12: \begin{abstract}

13: 

14: We apply the image subtraction method to re-analyze the ESO Very Large

15: Telescope data on M83 (NGC 5236), obtained and analyzed by Thim et

16: al. Whereas Thim et al. found 12 Cepheids with periods between 12-55

17: days, we find 112 Cepheids with periods ranging from 7-91 days, as

18: well as $\sim60$ other variables. These include 2 candidate eclipsing

19: binaries, which, if confirmed, would be the first optically discovered

20: outside the Local Group. We thus demonstrate that the image

21: subtraction method is much more powerful for detecting variability,

22: especially in crowded fields. However, {\em HST}\/ observations are

23: necessary to obtain a Cepheid period-luminosity distance not dominated

24: by blending and crowding.  We propose a ``hybrid'' approach, where

25: numerous Cepheids are discovered and characterized using large

26: ground-based telescopes and then followed-up with the {\em HST}\/ to

27: obtain precise distances.

28: 

29: \end{abstract}

30: \keywords{Cepheids --- distance scale --- galaxies: individual (M83,

31: NGC 5236)}

32: \section{Introduction}

33: 

34: Cepheid variables are one of the most important primary distance

35: indicators due to the period-luminosity (PL) relation they obey.

36: Precise Cepheid distances require a large number of Cepheids with

37: accurate photometry and periods. However, the more distant the galaxy,

38: the more important become the effects of crowding and blending with

39: nearby luminous stars (Mochejska et al. 2000; Stanek \& Udalski 1999;

40: Mochejska et al. 2001), which introduce systematic, one-sided errors

41: in the distance measurements and skew the distance towards

42: artificially lower values. These effects can only be disentangled

43: satisfactorily with the {\em Hubble Space Telescope}\/, which has the

44: required spatial resolution, especially now with the new

45: instrumentation (ACS).

46: 

47: The spiral galaxy M83, the principal member of the nearby M83 group,

48: is a good target to obtain a Cepheid distance to. Observations of M83

49: with the Very Large Telescope (VLT) were analyzed by \citet{Thi03}

50: with point-spread function (PSF) photometry, discovering 12 Cepheids,

51: which were used to derive the Cepheid distance to the galaxy. However,

52: the ``traditional'' method of doing PSF fitting photometry on all the

53: stars on an image and looking for variations in the light from night

54: to night is not very efficient or effective in finding Cepheids in

55: crowded fields. In this paper we re-analyze the VLT data for M83 using

56: image subtraction to demonstrate this fact.

57: 

58: Observations of M83, at a distance of $\sim4.5$ Mpc \citep{Thi03},

59: with the 8.2 meter VLT and $0.76\arcsec$ median seeing are roughly

60: equivalent to the DIRECT project observations of M31/M33 with the FLWO

61: 1.2 meter telescope, at a distance of $\sim780$ kpc (Stanek \&

62: Garnavich 1998) with $\sim4\arcsec$ seeing. This fact partly motivated

63: this paper -- we would consider such poor seeing data unacceptable to

64: run PSF photometry on. The three most recent DIRECT papers

65: \citep{Moc01a,Moc01b, Bon03} have used Alard's image subtraction

66: package ISIS \citep{Ala98,Ala00} to discover variables, in particular

67: detached eclipsing binaries and Cepheids, in M31 and M33. This method

68: has become a method of choice for variability searches in crowded

69: fields.

70: 

71: At the distance of M83, the issue of blending must be taken into

72: account in deriving the Cepheid distance. The median seeing of the VLT

73: data is $0.76\arcsec$, which corresponds to $17\;$pc in M83. As first

74: discussed by Mochejska et al. (2000), blending is the close

75: association of a Cepheid with one or more intrinsically luminous

76: stars, which is the result of the higher value of the star-star

77: correlation function for massive stars, such as Cepheids, compared to

78: random field stars.  This effect cannot be detected within the

79: observed PSF by usual analysis. \citet{Moc00,Moc01c} address the

80: effect of blending on the Cepheid distances to M31 and M33. They

81: compare high resolution {\em HST}\/ images to the ground-based DIRECT

82: data and find that blending can affect the flux of a Cepheid typically

83: by $\sim20-30\%$, which leads to an underestimation of the true

84: distance. This phenomenon is different from crowding, which is the

85: random background luminosity fluctuation in each resolution

86: element. In M83, a large fraction of the flux of a blended Cepheid

87: could come from its companions, assuming the effect scales to that in

88: M31 and M33. Based on a simulation by Stanek \& Udalski (1999), this

89: would result in a significant distance bias.  Whereas the discovery of

90: Cepheids in nearby galaxies can be done adequately from the ground

91: given good signal-to-noise photometry, deriving the Cepheid PL

92: distance requires high spatial resolution {\em HST}\/ imaging.

93: 

94: In this paper, we apply the image subtraction method to the M83

95: dataset. In Section 2 we describe the observations, in Section 3 the

96: image subtraction method and finally the results, which are discussed

97: in Section 4.

98: 

99: \section{Observations}

100: 

101: We have retrieved the M83 data obtained by FORS1 on the ESO Very Large

102: Telescope, from the ESO/ST-ECF Science Archive Facility. The field of

103: view is $6.8\times6.8$ arcminutes, with a pixel scale of $0.2\arcsec$

104: per pixel. The $2048\times2048$ Tektronix CCD has 24 $\mu$m

105: pixels. There are 34 epochs in $V$-band spanning a period of 1.5

106: years, from January 2000 to July 2001. Each epoch consists of 3-4

107: subexposures of 400 or 500 sec. More details are given by

108: \citet{Thi03}.

109: \vspace{-0.5cm}

110: \section{The Method of Image Subtraction and Results}

111: The images were overscan corrected and flat fielded with standard

112: IRAF\footnote{IRAF is distributed by the National Optical Astronomy

113: Observatories, which are operated by the Association of Universities

114: for Research in Astronomy, Inc., under cooperative agreement with the

115: NSF.} routines. The transformation from rectangular to equatorial

116: coordinates was derived using 84 transformation stars from the

117: USNO-B1.0 \citep{Mon03} catalog. The average difference between the

118: catalog and the computed coordinates for the transformation stars was

119: less than $0.\arcsec3$ in RA and $0.\arcsec3$ in Dec. Next, we ran the

120: image subtraction package ISIS \citep{Ala98,Ala00} on the 104 $V$-band

121: images of M83.

122: 

123: The ISIS reduction procedure consists of several steps. Initially, all

124: the frames are transformed to a common coordinate grid. The best

125: seeing frame, FORS.2000-04-05T06:16:02.924 in our case, is chosen to

126: be the reference image. Next, a composite reference image is created

127: by stacking several best seeing frames. For each frame, the composite

128: reference image is convolved with a kernel to match its PSF and then

129: subtracted. On the subtracted images, the constant stars cancel out,

130: and only the signal from variable stars remains. A median image is

131: constructed of all the subtracted images, and the variable stars are

132: identified visually as significant peaks. Finally, profile photometry

133: is extracted from the subtracted images. \citet{Moc01a} describes this

134: procedure in more detail.

135: 

136: After locating the positions of the bright peaks, we obtained light

137: curves for $\sim650$ candidate variables. We ran them through the

138: DIRECT pipeline \citep{Kal98,Sta98} which fits model Cepheid light

139: curves and eclipsing binary light curves and classifies the stars as

140: Cepheids, eclipsing binaries (EBs) or other variables. After checking

141: these light curves, we found 112 Cepheids, with periods ranging from 7

142: to 91 days, 2 candidate EBs and $\sim60$ other variable

143: stars. Figure~\ref{CephLC} presents 10 sample Cepheid flux light

144: curves. Table~\ref{tab:ceph} lists the coordinates (RA, Dec) and (X,

145: Y) on the reference image for each Cepheid and the periods. We recover

146: all but one of the Cepheids found by \citet{Thi03}, the Cepheid

147: C4. The periods we derive agree well with those of \citet{Thi03}, the

148: median difference being 0.04 days and the largest difference being

149: 0.22 days for Cepheid C6. Figure~\ref{CephDist} plots the location of

150: the Cepheids as circles on the reference image, their size being

151: proportional to the period. A 29 day Cepheid is labeled for scale. On

152: the same plot, the stars show the position of the 12 Cepheids from

153: \citet{Thi03}. The image subtraction method detects many Cepheids in

154: the more crowded part of the field, which is much harder to do with

155: the ``traditional'' PSF photometry method.

156: 

157: \begin{figure}   

158: \plotone{f1.eps}

159: \caption{Flux light curves of 10 sample Cepheids found in M83. For

160: each Cepheid, the coordinates (RA, Dec) are given in parentheses,

161: along with the period. The median point for each epoch is plotted.}

162: \label{CephLC}

163: \end{figure}

164: 

165: \begin{figure}   

166: \plotone{f2.eps}

167: \caption{Location of the 112 Cepheids we found with image subtraction

168: on the reference image. Circles are centered at the position of the

169: Cepheids and their size is proportional to the period of the Cepheid

170: (see labeled 29 day Cepheid for reference). Stars indicate the

171: positions of the 12 Cepheids found by \citet{Thi03}. All but one were

172: recovered by our analysis. [{\it See the electronic edition of the

173: Journal for a color version of this figure.}]}

174: \label{CephDist}

175: \end{figure}   

176: 

177: We do not claim completeness in finding Cepheids, however we do show

178: that Cepheids in the highly crowded and blended central regions of the

179: galaxy can be detected from the ground with high quality data, such as

180: these VLT observations. High resolution observations are necessary to

181: disentangle blending for Cepheids both in the outer and inner parts of

182: the galaxy.

183: \vspace{-0.5cm}

184: \section{Discussion}

185: 

186: This paper re-analyzes the excellent VLT data of M83 obtained by

187: \citet{Thi03} using the image subtraction method. The resulting

188: nine-fold increase in the number of Cepheids detected indicates that

189: image subtraction should be used in crowded fields. We also present

190: parameters and sample light curves for the 112 Cepheids we have

191: found. These additional Cepheids are valuable for determining the PL

192: distance to M83 accurately. However, {\em HST}\/ observations are

193: necessary to resolve blending effects. After we started working on

194: this project, we became aware of the Cycle 12 program with the title

195: ``M83: Calibrating the Cepheid PL Relation'' (PI: B. Madore). We note

196: that some of these 112 Cepheids might be Population II variables,

197: however, a PL diagram would be necessary to distinguish them.

198: 

199: In Figure~\ref{Ampl}, we plot the flux amplitude versus period

200: relation for the 112 Cepheids. We have taken the full flux amplitude

201: from the light curve and plotted the logarithm of this quantity versus

202: logarithm of the period $P$. There is a definite correlation, however,

203: it is not very tight. \citet{Pac00} demonstrate that the period-flux

204: amplitude relation is not universal and needs to be calibrated before

205: being used to measure distances accurately.

206: 

207: In addition to these Cepheids, we also find $\sim60$ other variable

208: stars, including 2 candidate eclipsing binaries (EBs). If these are

209: confirmed from their position on a CMD to be located in M83, they

210: would be the first optically discovered EBs outside the Local

211: Group. Figure~\ref{Misc} presents some of the more interesting light

212: curves of other periodic or non periodic variables that we find,

213: including the candidate EBs. There are two long period variables,

214: which are possibly Cepheids with periods of 132 and 195 days. Their

215: positions on the PL diagram would verify this.

216: 

217: \begin{figure}   

218: \plotone{f3.eps}

219: \caption{Flux amplitude versus period for 112 Cepheids in M83.}

220: \label{Ampl}

221: \end{figure}   

222: 

223: \begin{figure}   

224: \plotone{f4.eps}

225: \caption{Flux light curves of some miscellaneous variables in M83,

226: with coordinates (RA, Dec). Two possible long period Cepheids and two

227: candidate eclipsing binaries are shown.}

228: \label{Misc}

229: \end{figure}   

230: \vspace{-0.5cm}

231: \acknowledgments{We thank Danny Steeghs, David Bersier and Barbara

232: Mochejska for helpful discussions. We thank Grzegorz Pojma\'nski for

233: his most useful ``lc'' program. We also thank Bohdan Paczy\'nski and

234: Dimitar Sasselov for their comments on an earlier version of this

235: paper. This work is based on observations made with the European

236: Southern Observatory telescopes obtained from the ESO/ST-ECF Science

237: Archive Facility.}

238: 

239: \vspace{-0.5cm}

240: \begin{thebibliography}{99}

241: 

242: \bibitem[Alard \& Lupton(1998)]{Ala98} 

243: Alard, C., \& Lupton, R. 1998, \apj, 503, 325

244: 

245: \bibitem[Alard(2000)]{Ala00} 

246: Alard, C. 2000, A\&AS, 144, 363

247: 

248: \bibitem[Bonanos et al.(2003)]{Bon03}

249: Bonanos, A.~Z., Stanek, K.~Z., Sasselov, D.~D., Mochejska, B.~J., Macri,

250: L.~M., \& Kaluzny, J. 2003, \aj, in press (astro-ph/0303307)

251: 

252: \bibitem[Kaluzny et al.(1998)]{Kal98}

253: Kaluzny, J., Stanek, K.~Z., Krockenberger, M., Sasselov, D.~D., Tonry,

254: J.~L., \& Mateo, M. 1998, \aj, 115, 1016

255: 

256: \bibitem[Mochejska et al.(2000)]{Moc00} 

257: Mochejska, B.~J., Macri, L.~M., Sasselov, D.~D., \& Stanek, K.~Z. 2000,

258: \aj, 120, 810

259: 

260: \bibitem[Mochejska et al.(2001a)]{Moc01a} 

261: Mochejska, B.~J., Kaluzny, J., Stanek, K.~Z., Sasselov, D.~D., \&

262: Szentgyorgyi, A.~H. 2001a, \aj, 121, 2032

263: 

264: \bibitem[Mochejska et al.(2001b)]{Moc01b} 

265: Mochejska, B.~J., Kaluzny, J., Stanek, K.~Z., Sasselov, D.~D., \& 

266: Szentgyorgyi, A.~H. 2001b, \aj, 121, 2032

267: 

268: \bibitem[Mochejska et al.(2001c)]{Moc01c} 

269: Mochejska, B.~J., Macri, L.~M., Sasselov, D.~D., \&  Stanek, K.~Z. 

270: 2001c, \aj, submitted (astro-ph/0103440)

271: 

272: \bibitem[Monet et al.(2003)]{Mon03} 

273: Monet, D., et al. 2003, \aj, 125, 984

274: 

275: \bibitem[Paczy\'nski \& Pindor(2000)]{Pac00}

276: Paczy\'nski, B., \& Pindor, B. 2000, \apj, 533, L103

277: 

278: \bibitem[Stanek \& Garnavich (1998)]{StaGar} 

279: Stanek, K.~Z., \& Garnavich, P.~M. 1998, ApJ, 503, L131

280: 

281: \bibitem[Stanek et al.(1998)]{Sta98} 

282: Stanek, K.~Z., Kaluzny, J., Krockenberger, M., Sasselov, D.~D., Tonry,

283: J.~L., \& Mateo, M. 1998, \aj, 115, 1894

284: 

285: \bibitem[Stanek \& Udalski(1999)]{Sta99} 

286: Stanek, K.~Z., \& Udalski, A. 1999, preprint (astro-ph/9909346)

287: 

288: \bibitem[Thim et al.(2003)]{Thi03}

289: Thim, F., Tammann, G.~A., Saha, A., Dolphin, A., Sandage, A., Tolstoy,

290: E., \& Labhardt, L. 2003, \aj, in press (astro-ph/0303101)

291: 

292: \end{thebibliography}

293: \input{tab1.stub.tex}

294: \end{document}

295: