1: \documentclass[12pt,preprint]{aastex}
2: \usepackage{psfig,rotate}
3:
4: \shortauthors{Zeh, Klose, \& Hartmann}
5: \shorttitle{GRB supernovae}
6:
7: \def\gr{\hbox{ \raisebox{-1.0mm}{$\stackrel{>}{\sim}$} }}
8: \def\kr{\hbox{ \raisebox{-1.0mm}{$\stackrel{<}{\sim}$} }}
9:
10: \epsscale{0.6}
11:
12: %---------------------------------------------------------------------------
13: \begin{document}
14:
15: \title{A Systematic Analysis of Supernova Light in Gamma-Ray Burst Afterglows}
16:
17: \author{A. Zeh, S. Klose}
18: \affil{Th\"uringer Landessternwarte Tautenburg, 07778 Tautenburg, Germany}
19: \author{D. H. Hartmann}
20: \affil{Department of Physics and Astronomy, Clemson University, Clemson, SC
21: 29634-0978}
22:
23: \date{Revised \today / Accepted}
24:
25: %---------------------------------------------------------------
26: \begin{abstract}
27: We systematically reanalyzed all Gamma-Ray Burst (GRB) afterglow data
28: published through the end of 2002, in an attempt to detect the predicted
29: supernova light component and to gain statistical insight on its
30: phenomenological properties. We fit the observed photometric light curves as
31: the sum of an afterglow, an underlying host galaxy, and a supernova
32: component. The latter is modeled using published multi-color light curves of
33: SN 1998bw as a template. The total sample of afterglows with established
34: redshifts contains 21 bursts (GRB 970228 - GRB 021211). For nine of these GRBs
35: a weak supernova excess (scaled to SN 1998bw) was found, what makes this to
36: one of the first samples of high-$z$ core collapse supernovae. Among this
37: sample are all bursts with redshifts less than $\sim$0.7. These results
38: strongly support the notion that in fact all afterglows of long-duration GRBs
39: contain light from an associated supernova. A statistics of the physical
40: parameters of these GRB-supernovae shows that SN 1998bw was at the bright end
41: of its class, while it was not special with respect to its light curve
42: shape. Finally, we have searched for a potential correlation of the supernova
43: luminosities with the properties of the corresponding bursts and optical
44: afterglows, but we have not found such a relation.
45: \end{abstract}
46:
47: \keywords{gamma-rays: bursts -- supernovae: general}
48:
49: %-----------------------------------------------------------------------%
50: \section{Introduction \label{Intro}}
51:
52: Observational and theoretical evidence suggest that the majority of GRB
53: progenitors are stars at the endpoint in stellar evolution (e.g., Fryer,
54: Woosley, \& Hartmann 1999; Heger et al. 2003). Since the discovery of a
55: near-by Type Ic supernova (SN 1998bw) in the error circle of the X-ray
56: afterglow for GRB 980425 (Galama et al. 1998; Kulkarni et al. 1998), evidence
57: is accumulating that core collapse supernovae are physically related to
58: long-duration GRBs. The supernova picture is further supported by the
59: observation that all GRB hosts are star-forming, and in some cases even
60: star-bursting galaxies (e.g., Frail et al. 2002; Sokolov et
61: al. 2001). Evidence for host extinction by cosmic dust in GRB afterglows
62: (e.g., Castro-Tirado et al. 1999; Klose et al. 2000), and the discovery of an
63: ensemble of optically 'dark bursts' (for a recent discussion, see Fynbo et
64: al. 2001; Klose et al. 2003; Lazzati, Covino, \& Ghisellini 2002) also is
65: consistent with the picture that GRB progenitors are young, massive stars
66: (Groot et al. 1998; Paczy\'nski 1998). Furthermore, for several GRB afterglows
67: X-ray data suggest a period of nucleosynthesis preceding or accompanying the
68: burst (e.g., Antonelli et al. 2000; Lazzati, Campana, \& Ghisellini 1999;
69: M\'esz\'aros \& Rees 2001). The angular distribution of the afterglows with
70: respect to their hosts also favors a physical relation of young, massive stars
71: to GRBs (Bloom, Kulkarni, \& Djorgovski 2002).
72:
73: As a natural consequence of a physical relation between the explosion of
74: massive stars and GRBs supernova light should contribute to the afterglow
75: flux, and even dominate under favorable conditions. The most convincing
76: example is GRB 030329 (Peterson \& Price 2003) at $z$=0.1685 (Greiner et
77: al. 2003a) with spectral confirmation of supernova light in its afterglow
78: (Hjorth et al. 2003; Kawabata et al. 2003; Matheson et al. 2003; Stanek et
79: al. 2003). In contrast to this unique spectroscopic evidence, several cases of
80: photometric evidence for extra light in GRB afterglows have been reported,
81: starting with GRB 980326 (Bloom et al. 1999). Various groups successfully fit
82: SN 1998bw templates to explain these late-time bumps, the most convincing case
83: being that of GRB 011121 (Bloom et al. 2002; Garnavich et al. 2003; Greiner et
84: al. 2003b).
85:
86: The goal of the present paper is to analyze the supernova bumps in GRB
87: afterglow light curves using a systematic approach. While this was done also
88: for several bursts by Dado, Dar, \& de R\'ujula (2002a; and references
89: therein) with respect to a verification of their cannonball model (Dar \& de
90: R\'ujula 2003), we tackle this issue in an independent and different
91: way. First, from the numerical site, we have developed our own computational
92: approach. This includes a numerical procedure to redshift SN 1998bw light
93: curves (see \S~\ref{SNfits}) and to fit afterglow data within the context of
94: the fireball model. Second, from the observational site, when necessary and
95: possible we have performed late-time observations of some GRB host galaxies
96: (\S~2). A considerable part of the data we have included in our study is
97: based on observing runs where we have been involved. Additional data have been
98: collected from the literature and checked for photometric consistency. Nearly
99: two dozen afterglows had sufficient data quality, and a known redshift, to
100: search for a late-time bump in the light curve (\S~3). Third, we concentrate
101: our attention on a statistical analysis of the phenomenological parameters for
102: this class of GRB-SNe (\S~4). In this respect our investigation goes beyond
103: the approaches undertaken by others to explain late-time bumps in individual
104: afterglow light curves.
105:
106: %-----------------------------------------------------------------------%
107: \section{Observations and data processing}
108:
109: Some of the GRB afterglows we analyzed had poorly sampled late-time data,
110: which made it difficult to find or determine the parameters of a SN bump (GRBs
111: 000418, 991208, and 010921). In order to perform late-time photometry of these
112: GRB hosts, we carried out two observing runs at the Calar Alto 3.5-m telescope
113: on March 13--14 and May 23--25, 2003. Observations were performed using the
114: multi-purpose camera \it MOSCA, \rm which uses a SITe 2k$\,\times4$k CCD with
115: a plate scale of 0.32 arcsec per pixel. The field of view is $11\,\times\,11$
116: arcmin$^2$. During the first observing run the seeing varied between 1.4 and
117: 1.6 arcsec; in May the seeing was better than 0.8 arcsec. Data reduction
118: followed standard procedures.
119:
120: Most of the light curves we investigated have been followed in more than one
121: photometric band. For each of the GRBs we chose the best-sampled light curve
122: as a reference light curve for the fit in the other photometric bands. In the
123: most cases this was the $R$ band light curve. In brief, our approach was as
124: follows. \it First, \rm we assumed that the afterglow slopes and break time
125: are the same in all filters (Eq.~\ref{AG}), in reasonable agreement with
126: observational data. Consequently, once we fit the reference light curve of an
127: optical transient and deduced the corresponding afterglow parameters, we
128: treated them as fixed parameters when fitting the light curves of the optical
129: transient in other photometric bands. Thereby the fit procedure was based on a
130: $\chi^2$-minimization with a Levenberg - Marquardt iteration. \it Second, \rm
131: for the representation of the supernova component we used published $UBVRI$
132: data of the light curve of SN 1998bw (Galama et al. 1998) as a template, and
133: redshifted them to the corresponding cosmological distance of the burster
134: (\S~\ref{SNfits}). These light curves are different from band to band. In
135: addition, we allowed a variation of the SN luminosity with respect to SN
136: 1998bw and a stretch in time (Eq.~\ref{ot} in \S~\ref{numerics}). \it Third,
137: \rm the host magnitude, which represents a constant component in the integral
138: light of the optical transient, was usually considered as a free
139: parameter. Only for GRB 011121 and 020405 we used host-subtracted magnitudes
140: to fit the light curves.
141:
142: Before performing a numerical fit, the observational data was corrected for
143: Galactic extinction along the line of sight using the maps of Schlegel,
144: Finkbeiner, \& Davis (1998). This also holds for SN 1998bw, where we assumed
145: $E(B-V)$= 0.06 mag. We calculated the Galactic visual extinction according to
146: $A_V^{\rm Gal}$ = 3.1 $E(B-V)$, whereas the extinction in $U$ and $B$ were
147: obtained via Rieke \& Lebofsky (1985), and in $R_c$ and $I_c$ by means of the
148: numerical functions compiled by Reichart (2001).
149:
150: %-----------------------------------------------------------------------%
151: \section{Results \label{results}}
152:
153: Among the 36$\pm1$ GRBs with detected optical afterglows up to the end of
154: 2002\footnote{http://www.mpe.mpg.de/$^{\sim}$jcg/grbgen.html}, 21 had
155: sufficient data quality, and a known redshift, for a meaningful search for an
156: underlying supernova component (Table~1). Among them in nine cases evidence
157: for a late-time bump was found. The results are summarized in Table~2. A
158: general inspection of this table makes clear that the burst ensemble with
159: detected late-time bumps in their afterglows separates into a group with
160: statistically significant evidence for a bump (990712, 991208, 011121, and
161: 020405), mostly in at least two photometric bands, and a group with less
162: significant bumps (970228, 980703, 000911, 010921, and 021211). Given the fact
163: that evidence for these bumps has also been reported by other groups (with the
164: only exception being GRB 010921), we feel confident that the results
165: presented in Table~2 can be used for a first statistical approach to
166: understand this type of GRB-SNe.
167:
168: The most interesting result is that our numerical procedure found evidence for
169: a late-time bump in \it all \rm GRB afterglows with a measured redshift
170: $z\kr$0.7. We believe that the interpretation of these bumps as an underlying
171: supernova component is the most natural and observationally most founded
172: explanation. Among the higher redshifted bursts, we confirm the finding by
173: Holland et al. (2001) of a possible bump in the afterglow of GRB 980703, the
174: discovery by Lazzati et al. (2001) of a bump in the afterglow of GRB 000911
175: ($z$=1.06; Price et al. 2002b), and a bump in the afterglow of GRB 021211,
176: which was discovered by Della Valle et al. (2003) and is also discussed by
177: Dado, Dar, \& de R\'ujula (2003b).
178:
179: For five afterglows (GRB 970508, 991216, 000418, 010222, 020813) with $0.7 <
180: z < 1.5$, we can only place upper limits on the luminosity of any underlying
181: supernova component. These five upper limits have typical uncertainties of a
182: factor of two. The only exception is GRB 020813 where the uncertainty is
183: much larger, so that no firm conclusions can be drawn here.
184: The remaining seven bursts in our sample (GRB 971214, 990123,
185: 990510, 000301, 000926, 011211, 021004) have redshifts $z>1.5$ and, therefore,
186: have not been investigated here, since this would have required a substantial
187: extrapolation of the SN 1998bw template into the UV domain. Finally, the
188: late-time bump clearly seen in the afterglow of GRB 980326 (Bloom et
189: al. 1999) is not included in our study, because the redshift of the burster is
190: not yet known.
191:
192: As we have outlined in the previous section, we fit the SN component using the
193: light curves of SN 1998bw as a template. Thereby, we allowed the luminosity
194: and the light curve shape to be a free parameter. The former means a scaling
195: of the luminosity of SN 1998bw by a factor $k$ (Eq.~\ref{ot}), whereby $k$
196: always refers to the corresponding wavelength region in the redshifted SN
197: frame. Differences in the light curve shapes were modeled by means of a
198: stretch factor $s$, which allows the supernova light curve to develop slower
199: ($s>1$) or faster ($s<1$) than the one of SN 1998bw (Eq.~\ref{ot}). In this
200: respect we follow Hjorth et al. (2003) to describe the light curve of GRB
201: 030329/SN 2003dh. The advantage of such an approach is that we can use
202: these two parameters to explore the entire ensemble of GRB-SNe in a
203: statistical sense. Table~2 shows that for the bursts with the photometrically
204: best sampled late-time bumps in their optical light curves (GRB 011121,
205: 020405) the deduced parameters $k$ and $s$ are consistent with each other in
206: different wavelength regions. A comparison of these parameters of the
207: nine afterglows with late-time bumps, which are at different redshifts, seems
208: to be a reasonable first approach in order to constrain the width of the
209: luminosity distribution of GRB-SNe.
210:
211: \begin{figure*}
212: \plotone{f1.eps}
213: \caption{The deduced luminosities of GRB-SNe in units of
214: the luminosity of SN 1998bw in the same spectral region (parameter $k$,
215: Eq.~\ref{ot}). All data are based on observations in the $R$ band. The broken
216: line corresponds to SN1998bw; it is
217: $\Delta m = -2.5$ log $k$, which measures the
218: magnitude difference at maximum light between the GRB-SN and SN 1998bw in the
219: corresponding wavelength regime. The lower panel (b) is for a stretch
220: parameter fixed at $s$=1, while in the upper panel (a) $s$ is not fixed. In
221: the former case we can set upper limits on $k$ for further four
222: bursts (GRB 970508, 991216, 000418, 010222).
223: Moreover, the numerical procedure can fit the afterglow light curve of
224: GRB 980703. Note that the data are not corrected for a possible extinction in
225: the GRB host galaxies. \label{luminosity1}}
226: \end{figure*}
227:
228: In Fig.~\ref{luminosity1} we display the deduced parameter $k$ (luminosity)
229: for every individual GRB-SNe. We plot luminosity versus redshift just to look
230: for a potential evolutionary effect (which is not apparent) and to separate
231: the individual SNe from each other. While in Fig.~\ref{luminosity1}a we
232: allowed the stretch factor $s$ to be a free parameter during the numerical
233: fit, Fig.~\ref{luminosity1}b shows the results obtained when we fixed
234: $s=1$. The reason for the latter was twofold. First, if $s=1$ we can constrain
235: the luminosity of an underlying SN for those GRBs, where we do not detect a
236: bump in the late-time light curve. This is not possible in a reasonable way,
237: if we allow $s$ to be a free parameter. Second, sometimes the data base is too
238: small in order to include also the stretch parameter in the fitting procedure,
239: so that we have to fix $s$. Note that in Fig.~\ref{luminosity1} there are
240: three small sets of bursts at redshifts 0.4, 0.7 and 1.0. Between them is no
241: difference apparent neither in the luminosities of the GRB-SNe nor in the
242: width of the luminosity distribution. What is apparent from a comparison of
243: Fig.~\ref{luminosity1}a,b is that introducing the stretch factor reduces the
244: width of the luminosity distribution of the GRB-SNe and brings the luminosity
245: of all SNe a little closer to those of SN 1998bw ($k=1$). The distribution of
246: the deduced stretch factor itself is shown in Fig.~\ref{s}. Although $s$
247: varies by a factor of two in both directions around $s=1$, within their
248: individual 1$\sigma$ error bars most data are close to $s=1$. Finally,
249: no correlation was found between the deduced SN luminosity (parameter $k$)
250: and the stretch factor $s$ (Fig.~\ref{sk}).
251:
252: \begin{figure}
253: \plotone{f2.eps}
254: \caption{The distribution of the
255: parameter $s$ (Eq.~\ref{ot}) describing a stretching of the SN light curve
256: relative to those of SN 1998bw (for which by definition $s=1$, broken line).
257: The mean value of $s$ is close to 1.0. \label{s}}
258: \end{figure}
259:
260: \begin{figure}
261: \plotone{f3.eps}
262: \caption{Drawn here is the luminosity ratio $k$ versus
263: the stretch factor $s$ for the eight GRB-SNe of Figs.~\ref{luminosity1}a and
264: \ref{s}. No correlation between $k$ and $s$ is apparent in the
265: data. \label{sk}}
266: \end{figure}
267:
268: %-----------------------------------------------------------------------%
269: \section{Discussion \label{statistics}}
270:
271: \subsection{The luminosity distribution of the GRB-SNe}
272:
273: When plotting the parameter $k$ deduced for the $R$ band in the observer
274: frame, the width of the luminosity distribution of the class of GRB-SNe
275: (Fig.~\ref{histogram1}) is similar to what is observed for other classes of
276: core collapse SNe (Richardson et al. 2002). The mean of $k$ in the $R$ band is
277: 0.7 independent of whether or not we fix the stretch parameter at $s=1$, while
278: our template SN 1998bw is at the bright end of the GRB-SN luminosity
279: distribution. The latter conclusion is supported when we plot $k$ for the same
280: wavelength region in the SN host frame, which is a better indicator of the
281: spread in SN luminosities. Most of the GRB-SNe we explored have a data point
282: for the photometric band centered around 395$\pm$10 nm in the corresponding
283: host frame (Table~\ref{res}). The distribution of $k$ now indicates again that
284: SN 1998bw is among the most luminous GRB-SNe. It also indicates that in fact
285: there is no peak around $k$=0.8 (Fig.~\ref{histogram1}), but we may so far
286: have only sampled the bright part of the GRB-SN luminosity function. Some
287: caution is of course required, given the partly large error bars of the $k$
288: factors, which are not shown in Figs.~\ref{histogram1} and
289: ~\ref{histogram2}. While the conclusion that SN 1998bw was among the most
290: luminous members of its class seems to be robust, the shape of the GRB-SNe
291: luminosity function is still less well-determined. Extinction by interstellar
292: dust in the host galaxies could in principle also affect these results,
293: although only for GRB 010921 was a significant host extinction (\gr 1 mag)
294: reported (Price et al. 2003).
295:
296: As we have outlined before, for redshifts $z$\kr0.7 all GRB
297: afterglows show evidence for an underlying late-time bump. Within
298: our context this means that we trace a complete set of GRB-SNe,
299: i.e., not only the brightest members of this class. The width of
300: this GRB-SNe luminosity distribution in the photometric band
301: centered around 395$\pm$10 nm in the SN frame is $\sim$1 to 1.5
302: magnitudes. This wavelength region is roughly placed in the $B$
303: band, so that we can compare the corresponding luminosities with
304: those of other Type Ib/c supernovae (Richardson et al. 2002),
305: i.e., those class of SNe, which is believed to include the
306: progenitors of GRBs. It turns out that the GRB-SNe do fit into a
307: region between approximately $M_B=-19.5$ and $M_B=-18$ in figure 6
308: of Richardson et al., where no data on Type Ib/c SNe are known. If
309: all GRB-SNe are indeed of type Ib/c this would favor the
310: conclusion that the luminosity function of Type Ib/c SNe is rather
311: described by a broad Gaussian than by a bimodal distribution
312: (figures 6 and 7 in Richardson et al.).
313:
314: \begin{figure}
315: \plotone{f4.eps}
316: \caption{The distribution
317: of the luminosity parameter $k$ (Eq.~\ref{ot}) as measured in the $R$ band in
318: the observer frame (Table~\ref{res}, with $s$ being a free parameter). GRB
319: 980703 is not included here because the fitting procedure did not find a
320: solution in this case. Note that the histogram does not include the
321: 1$\sigma$ error bars of the individual $k$ factors, which are relatively
322: large. \label{histogram1}}
323: \end{figure}
324:
325: The non-detection of a supernova bump in more than half of the investigated
326: GRB afterglow light curves may be accounted for by several reasons, like a
327: relatively bright host, or a faint supernova. In particular, finding a SN bump
328: for high-$z$ bursts is an observational challenge. For $z\gr0.7$ and $k=1$,
329: this peak magnitude exceeds $R_c=24$, which poses a major challenge for 3-m
330: class telescopes, given the usually very limited amount of target of
331: opportunity time for such observations. It is thus no surprise that only for
332: three of the GRBs above $z=0.7$ a supernova component was found (GRB 980703,
333: GRB 000911, GRB 021211), even though we can not rule out that the SN
334: 'drop-out' is due to some evolutionary effect of the underlying burster
335: population and their environment.
336:
337: \begin{figure}
338: \plotone{f5.eps}
339: \caption{The same as
340: Fig.~\ref{histogram1}, but for the photometric band centered around
341: 395$\pm$10 nm in the SN host frame (Table~\ref{res}, with $s$ being a free
342: parameter). Not included here are GRB 980703, 010921, and 021211 since there
343: is no data point in this wavelength range. \label{histogram2}}
344: \end{figure}
345:
346: %-----------------------------------------------------------------------%
347: \subsection{The SupraNova model}
348:
349: Vietri \& Stella (2000) argued that GRBs are the result of delayed black hole
350: formation, which implies that the core collapse and its subsequent supernova
351: may significantly precede the burst. The delay could be of order months to
352: years (Vietri \& Stella 2000), or perhaps as short as hours (Woosley, Zhang, \&
353: Heger 2002). While constraining the latter possibility can not be accomplished
354: with the data at hand, the longer time scales are easily constrained. For only
355: two of the SN light curves the fit indeed improved if we allowed for a shift
356: in time between the onset of the burst and the onset of the SN (GRB 990712,
357: 011121). The offsets never exceeded 5 days, and were both negative and
358: positive. However, the uncertainties in this parameter are large, due to the
359: poorly sampled shape of the underlying supernova (e.g., Garnavich et
360: al. 2003). The average shift is basically consistent with
361: zero. Presumably, these shifts are due to an underlying correlation between
362: luminosity and light curve shape, as observed in other types of supernovae
363: (e.g., Candia et al. 2003; Stritzinger et al. 2002). This is just what the
364: parameter $s$ takes into account. On the other hand, its clear that we have
365: no information about this issue in those cases where we have not found
366: evidence for a SN bump at late times. While this still leads open the
367: possibility of two populations of bursters (collapsars \& SupraNovae), we
368: emphasize again that we find a late time bump in all afterglows with
369: redshift $z<0.7$.
370:
371: %-----------------------------------------------------------------------%
372: \subsection{X-ray lines and supernova bumps}
373:
374: The identification of late-time bumps in afterglow light curves with SN light
375: would benefit from observations in the X-ray band (for the cannonball model,
376: e.g., Dado, Dar, \& de R\'ujula 2003a). If the X-ray lines seen in some
377: afterglows (e.g., Reeves et al. 2002) have their origin in the circumburster
378: medium (e.g., Lazzati et al. 2001) and not in the exploding star (e.g.,
379: M\'esz\'aros \& Rees 2001) this would be difficult to reconcile with the
380: interpretation of a late-time bump with an underlying SN
381: component. Unfortunately, the majority of bursts with high-resolution \it
382: XMM-Newton \rm or \it Chandra \rm spectroscopic X-ray follow-up observations
383: have no well-observed optical light curves. Among the afterglows with a
384: detected optical late-time bump listed in Table~2 only for GRB 020405 such
385: observations exist (Mirabal, Paerels, \& Halpern 2003); no evidence for X-ray
386: lines has been found there. \it BeppoSAX \rm observed the afterglow of GRB
387: 970228 (Frontera et al. 1998) with comparable low energy resolution, and no
388: X-ray lines have been reported. Among those low-$z$ bursts with well-defined
389: optical light curves and no evidence for a late-time bump in the data, only
390: for GRB 970508 \it BeppoSAX \rm X-ray follow-up observations have been
391: published (Piro et al. 1999). Evidence for an iron line was found. Although
392: one might add to the list of bursts with well-observed late-time light curves
393: and additional spectral information in the X-ray band GRBs 990123, 000926,
394: 010222, 011211, and 021004, the redshift of these bursts was \gr 1.5, making
395: it more or less hopeless to find an underlying SN component in the available
396: data base (an upper limit for GRB 010222 is reported in
397: Fig.~\ref{luminosity1}b; Henden et al. 2004, in preparation). While it is
398: very interesting that neither for GRB 020405 (Mirabal et al. 2003) nor for GRB
399: 030329 (Tiengo et al. 2003) lines have been found in X-ray spectra of their
400: afterglows, at present we cannot confirm a possible anti-correlation between
401: the occurrence of X-ray lines and the appearance of SN light in GRB
402: afterglows. This important issue remains to be investigated in the \it Swift
403: \rm era.
404:
405: \begin{figure}
406: \plotone{f6.eps}
407: \caption{The correlation
408: between the afterglow parameters $\alpha_1$ and $\alpha_2$ (Eq.~\ref{AG}) for
409: all afterglows with a break in their light curves. The dotted line is the
410: theoretical prediction in the simplest model ($\Theta_{\rm jet}$=const;
411: M\'esz\'aros \& Rees 1999). Open diamonds indicate the four
412: afterglows with a break and a detected underlying late-time bump, i.e., a SN
413: component (GRB 980703, 011121, 020405, 021211).
414: Note that several afterglows listed in Table~\ref{allgrbs}
415: showed no evidence for a break in their light curves, so
416: that they are not included in this figure. \label{a2a1e}}
417: \end{figure}
418:
419: %-----------------------------------------------------------------------%
420: \subsection{SN properties vs. afterglow parameters}
421:
422: Of particular interest is whether the burst and afterglow properties are to
423: some degree related to the existence of an underlying SN component. For this
424: reason we have investigated if the deduced SN luminosity is correlated with
425: the corresponding energy release in the gamma-ray band (as given in Bloom,
426: Frail, \& Kulkarni 2003). No such correlation was found. We have also checked
427: whether the afterglow parameters $\alpha_1$ (pre-break decay slope),
428: $\alpha_2$ (post-break decay slope), and $t_b$ (break time; Eq.~\ref{AG}) from
429: those GRBs with detected SN component are different from those without such a
430: component. Again, no correlation was apparent, even though one should keep in
431: mind that the data base is still very small. For illustration, in
432: Fig.~\ref{a2a1e} we display the relation between the afterglow parameters
433: $\alpha_1$ and $\alpha_2$ for all GRB afterglows we have investigated. While
434: there is a 'forbidden region' with $\alpha_2 \kr \alpha_1 + 3/4$ apparent in
435: the data, with the border line representing the simplest jet model with no
436: sideways expansion of the jet (M\'esz\'aros \& Rees 1999), no bimodality in
437: this distribution is visible. A tendency in our data that GRB afterglows with
438: detected underlying SN component seem to prefer pre-break slopes $\alpha_1>1$
439: should not be overinterpreted, since the $\alpha_1<1$ sample includes several
440: bursts with redshifts $z>1.2$, where the photometric detection of an
441: underlying SN component is difficult. On the other hand, at least one
442: selection effect does occur here. If a bright SN component is apparent in the
443: data then the parameter $\alpha_2$ of the afterglow light curve is usually
444: much more difficult to quantify, because the late-time evolution of the
445: genuine afterglow is less well-defined. This problem is well seen in the
446: afterglow light curve of GRB 011121 (e.g., Greiner et al. 2003) and GRB 030329
447: (e.g., Lipkin et al. 2004).
448:
449: %--------------------------------------------------------------------------%
450: \section{Summary}
451:
452: In an attempt to study the GRB-SN association, we have re-analyzed in a
453: systematic way all GRB afterglow data published by the end of 2002. We have
454: found that in nine cases evidence for extra light at late times is apparent in
455: the optical afterglow light curves. In most cases this is seen in more than
456: one photometric band. This extra light can be modeled well as supernova
457: light peaking $(1+z) 15...20$ days after a burst. Our main finding is that all
458: GRB afterglows with redshift $z\kr0.7$ showed evidence for extra light at
459: later times. For larger redshifts the data base is usually not of sufficient
460: quality, or the SN is simply too faint, in order to search for such a feature
461: in the late-time afterglow light curve.
462:
463: The cut off date of our sample (end of 2002) was chosen to ensure that all
464: GRBs had published follow-up observations. Since that date, five new
465: afterglows with redshifts have been established. All but one were at redshifts
466: above 0.7, and again for the only nearby event (GRB 030329) a supernova
467: component was established. This is consistent with the statistical inferences
468: from the sample of earlier long-duration GRBs and leads us to conclude that
469: the current world sample of GRB afterglow measurements provides strong
470: statistical support for the link between (long-duration) GRBs and the final
471: stages of massive star evolution. This conclusion basically agrees with
472: earlier reports by Dado, Dar, \& de R\'ujula (2002a; and references therein),
473: and is essentially independent of the underlying GRB model. While so far only
474: one event (GRB 030329) allowed a direct spectroscopic confirmation of this
475: link, the larger photometric sample discussed here supports this idea by
476: statistical means.
477:
478: Based on our sample of nine GRB-supernovae we have performed a first
479: statistical approach to get insight on the characteristic luminosities of this
480: type of supernovae. We have found strong evidence that SN 1998bw is at the
481: bright end of the GRB-SN luminosity distribution, with the latter matching
482: well into what is known so far about the luminosities of the brightest members
483: of other types of core collapse supernovae (Richardson et al. 2002). While
484: GRB-SNe are not standard candles, their peak luminosities are
485: comparable to those of Type Ia. In fact, within the context of the SN
486: interpretation of the late-time bumps in afterglows, our results demonstrate
487: once more that the first years in GRB research have already provided a first
488: sample of high-$z$ core collapse SNe up to a redshift of 1. This sample might
489: grow rapidly in the near future if indeed all long-duration GRBs tell us when
490: and where in the universe a massive star explodes.
491:
492: Some caution is of course required. First, there is some bias
493: in the sample of bursts with detected optical afterglows. None of the bursts
494: with a detected SN bump was classified as an X-ray rich burst or an X-ray
495: flash, and for none of them were X-ray lines reported in the literature. In
496: other words, it is still possible that bursts with SN bumps do not belong to
497: these classes of events (but see Fynbo et al. 2004). Second, for most of the
498: bursts discussed here evidence for a SN bump is based on a very small number
499: of data points around the SN peak time (say, between 10 to 40 days after the
500: burst), with the most critical cases being GRB 991208 and 010921. However,
501: we see no reason why we should disregard these events.
502:
503: In their discovery paper, Klebesadel, Strong, \& Olson (1973) noted that a
504: potential relation of GRBs to supernovae might still be an option to explain
505: this new phenomenon. While the model they refer to (Colgate 1968) does not
506: describe what is today believed to be the underlying GRB mechanism,
507: historically it is nevertheless remarkable that the first paper ever about
508: GRBs might have given the right hint on the underlying source population,
509: followed by many years of trial and error.
510:
511: %--------------------------------------------------------------------------%
512: \acknowledgements
513:
514: S.K. and A.Z. acknowledge financial support by DFG grant KL 766/12-1 and from
515: the German Academic Exchange Service (DAAD) under grant No. D/0103745. D.H.H.
516: acknowledges support for this project under NSF grant INT-0128882. A.Z.
517: acknowledges the receipt of a scholarship from the
518: Friedrich-Schiller-University Jena, Germany. We thank Nicola Masetti and
519: Eliana Palazzi, Bologna, for providing host-subtracted photometric data on the
520: afterglow of GRB 020405. This work has profited from the GCN data base
521: maintained by Scott Barthelmy at NASA and the \it GRB big table \rm maintained
522: by Jochen Greiner, Garching. We thank Kevin Lindsay (Clemson) for a careful
523: review of the manuscript. We thank the referee for critical comments, which
524: helped to improve the paper.
525:
526: %--------------------------------------------------------------------------%
527: \appendix
528:
529: \section{Numerical approach}
530:
531: \subsection{The light curve of the optical transient \label{numerics}}
532:
533: We model the light curve of the optical transient (OT) following a GRB as a
534: composite of afterglow (AG) light, supernova (SN) light, and constant light
535: from the underlying host galaxy. The flux density, $F_\nu$, at a frequency
536: $\nu$ is then given by
537: \begin{equation}
538: F_\nu^{\rm OT}(t) = F_\nu^{\rm AG}(t) + k \, F_\nu^{\rm SN}(t/s)
539: + F_\nu^{\rm host}\,.
540: \label{ot}
541: \end{equation}
542: Here, the parameter $k$ describes the observed brightness ratio (in the host
543: frame) between the GRB-supernova, and the SN template (SN 1998bw) in the
544: considered photometric band (in the observer frame). We allowed $k$ to be
545: different in every photometric band, but within a band independent of
546: frequency. The parameter $s$ is a stretch factor with respect to the used
547: template. We have also explored the consequences of a shift in time between
548: the onset of the burst and the onset of the supernova explosion, as implied by
549: certain theoretical models (Vietri \& Stella 2000). Then, in Eq.~(\ref{ot})
550: $F_\nu^{\rm SN}(t/s)$ was replaced by $F_\nu^{\rm SN}(t+\tau)$. Here,
551: $\tau=0$ refers to GRB 980425/SN 1998bw (Iwamoto et al. 1998). If $\tau<0$
552: the SN preceded the onset of the GRB.
553:
554: We describe the afterglow light curve by a broken power-law (Beuermann et al.
555: 1999; Rhoads \& Fruchter 2001),
556: \begin{equation}
557: F_\nu^{\rm AG}(t) = \mbox{const}\
558: [(t/t_b)^{\alpha_1\,n}+(t/t_b)^{\alpha_2\,n}]^{-1/n}\,,
559: \label{AG}
560: \end{equation}
561: with const=$2^{1/n} \, 10^{-0.4m(t_b)}.$ Here $t$ is the time after the burst
562: (in the observer frame), $\alpha_1$ is the pre-break decay slope of the
563: afterglow light curve, $\alpha_2$ is the post-break decay slope, and $t_b$
564: is the break time. The parameter $n$ characterizes the sharpness of the break;
565: a larger $n$ implies a sharper break.
566: In most cases the parameter $n$ (Eq.~\ref{AG}) had to be fixed, otherwise
567: the iteration did not converge. The reason was the usually too small number of
568: data points around the break time. In these cases we set $n=10$, producing a
569: relatively sharp break in the light curve. However, this procedure did not
570: strongly affect the deduced supernova parameters $k$ and $s$ (Eq.~\ref{ot}).
571:
572: In the observer frame the flux density of the time-dependent supernova light
573: is given by (cf. Dado et al. 2002a)
574: \begin{equation}
575: F_\nu^{\rm SN}(t) = \frac{1+z_{\rm SN}}{1+z_{\rm bw}}\ \frac{d_{L,{\rm
576: bw}}^2} {d_{L,{\rm SN}}^2}\, F_{\rm bw} \Big(\nu\ \frac{1+z_{\rm
577: SN}}{1+z_{\rm bw}}; t \ \frac{1+z_{\rm bw}} {1+z_{\rm SN}}
578: \Big)\,.
579: \label{Dado}
580: \end{equation}
581: Here 'SN' stands for the GRB supernova under consideration, and 'bw'
582: represents SN 1998bw ($z$=0.0085; Tinney, Stathakis, \& Cannon 1998). We
583: calculated the luminosity distance, $d_L$, assuming a flat universe with
584: matter density $\Omega_M =$ 0.3, cosmological constant $\Omega_\Lambda=$ 0.7,
585: and Hubble-constant $H_0=65$ km s$^{-1}$ Mpc$^{-1}$.
586:
587: We always fitted photometric magnitudes. After manipulating
588: Eqs.~(\ref{ot}, \ref{AG}), the apparent magnitude of the OT in a given
589: photometric band is given as
590: \begin{equation}
591: m_{\rm OT}(t) = -2.5\,\log\{10^{-0.4\,m_c}[(t/t_b)^{\alpha_1\,n} +
592: (t/t_b)^{\alpha_2\,n}]^{-1/n} +
593: k\,10^{-0.4\,m_{\rm SN}(t/s)} + 10^{-0.4\,m_{\rm host}}\}\,.
594: \label{mag1}
595: \end{equation}
596: Again, $t/s$ was replaced by $t+\tau$ when we allow for a delay between SN and
597: GRB. Equation~(\ref{mag1}) has eight free parameters: $\alpha_1, \alpha_2, n,
598: t_b, k, s, m_{\rm host}$, and the constant $m_c$, which absorbs the
599: constant of Eq.~(\ref{AG}) and corresponds to the magnitude of the fitted
600: light curve for the case $n=\infty$ at the break time $t_b$. If there is no
601: break in the light curve, then Eq.~(\ref{mag1}) reduces to
602: \begin{equation}
603: m_{\rm OT}(t) = -2.5\,\log\{10^{-0.4\,m_1}t^{\alpha} +
604: k\,10^{-0.4\,m_{\rm SN}(t/s)} + 10^{-0.4\,m_{\rm host}}\}\,,
605: \label{mag2}
606: \end{equation}
607: where $m_1$ is the brightness of the afterglow at $t=1$ day after the
608: burst (if $t$ is measured in days).
609:
610: %-----------------------------------------------------------------------%
611: \subsection{Redshifting the SN 1998bw light curves \label{SNfits}}
612:
613: Equation~(\ref{ot}) requires as an input the function $F_\nu^{\rm SN}(t)$ for
614: arbitrarily frequencies in the optical bands. Spectra from SN 1998bw are
615: available in the literature, but the time coverage of published broad-band
616: photometry is much better. Therefore, we constructed $F_\nu^{\rm SN}(t)$ based
617: on published $UBVRI$ light curves (Galama et al. 1998), assuming that we can
618: smoothly interpolate between adjacent photometric bands. Thereby, we have
619: taken into account that various broad-band features are inherent to the
620: spectral energy distribution of SN 1998bw that develop with time (e.g., Patat
621: et al. 2001; Stathakis et al. 2000). Therefore, for different photometric
622: bands SN 1998bw light curves peak at different times $t_{\nu}^{\rm max}$, at
623: different flux densities, $F_\nu^{\rm max} = F_\nu(t_{\nu}^{\rm max})$, and
624: have different shapes.\footnote{For reasons of clarity, in this section we
625: omit the index 'SN' at $F_\nu$; all flux densities refer to SN 1998bw.} In the
626: following, we demonstrate our numerical approach using the $U$ and $B$ bands
627: as an example (Fig.~\ref{98bw}).
628:
629: Let $\nu_U$ be the central frequency of the $U$ band, $\nu_B$ be the central
630: frequency of the $B$ band, and $\varepsilon$ be defined as $0 \le \varepsilon
631: \le 1$. Let $\nu'$ be the frequency for which the light curve $F_\nu(t)$ is
632: required, then the relation $\nu' = \nu_U + \varepsilon \, (\nu_B - \nu_U)$
633: defines the value of $\varepsilon$. Assuming a smooth behavior of $F_\nu (t)$
634: between the $U$ band and the $B$ band, for the frequency-dependent peak flux of
635: the light curve at the frequency $\nu'$, we assume
636: \begin{equation}
637: \log F_{\nu'}^{\rm max} = \log F_U^{\rm max} \nonumber \\ +
638: \varepsilon \, (\log F_B^{\rm max} - \log F_U^{\rm max})\,.
639: \end{equation}
640: Similarly, for the frequency-dependent peak time of the supernova light curve
641: at a fixed frequency, $\nu'$, we write
642: \begin{equation}
643: t_{\nu'}^{\rm max} = t_U^{\rm max} + \varepsilon \, (t_B^{\rm max} -
644: t_U^{\rm max})\,.
645: \end{equation}
646:
647: Finally, in order to model the frequency-dependent shape of the
648: SN 1998bw light curves, we normalize them to their peak flux and peak time.
649: Then, at a given frequency we have $F_\nu(t) = \eta_\nu \ F_{\nu}^{\rm max},$
650: where $\eta$ is a function of the ratio $t/t_\nu^{\rm max}$ and $0 \le \eta \le
651: 1$. Correspondingly, our ansatz for the shape function $\eta_{\nu'}$ for a
652: redshifted SN 1998bw is
653: \begin{equation}
654: \log \eta_{\nu'}(x) = \log \eta_U (x) + \varepsilon \,[\log \eta_B
655: (x) - \log \eta_U(x) ]\,,
656: \end{equation}
657: where $x= t_{\rm host} / t_{\nu'}^{\rm max}$ and
658: \begin{equation}
659: t_{\rm host} = t \ \frac{1+z_{\rm bw}}{1+z_{\rm SN}}
660: \end{equation}
661: is measured in the host frame (symbols follow Eq.~\ref{Dado}).
662:
663: \begin{figure}
664: \epsscale{1.0}
665: \plotone{f7.eps}
666: \caption{\it Left: \rm The
667: $UBVRI$ light curves of SN 1998bw according to the data provided by Galama et
668: al. (1998). The fits (drawn through lines) are based on a purely empirical
669: equation which fits a supernova light curve very well and is not physical. It
670: extrapolates also beyond 60 days. Note that the light curves differ in peak
671: flux, peak time, and in shape. In order to predict the light curves of a
672: redshifted SN 1998bw one has to construct light curves for any frequency in
673: between the characteristic frequencies of the $UBVRI$ bands. \it Right: \rm
674: The broad-band light curves of SN 1998bw normalized to
675: their peak maxima and peak times. Now they differ only in their shapes.
676: \label{98bw}}
677: \end{figure}
678:
679: Once an ensemble of functions $F_\nu (t)$ has been calculated, the apparent
680: magnitude of the redshifted SN 1998bw in a given photometric band
681: (Eqs.~\ref{mag1}, \ref{mag2}) is obtained by integrating over the flux density
682: per unit wavelength ($F_\lambda (t) = -\nu^2/c \, F_\nu (t)$), multiplied by
683: the corresponding filter response function $S_\lambda$. For $S_\lambda$ we
684: used the transmission curves for Bessel filters provided on the internet pages
685: of the \it European Southern Observatory \rm for VLT-FORS1 with reference to
686: Bessel (1979). For the transformation between photometric magnitudes and flux
687: densities we used the calibration constants provided by Fukugita, Shimasaku,
688: \& Ichikawa (1995; their table 9) and Zombeck (1990). The so calculated
689: broad-band light curves of a redshifted SN 1998bw were then used as an input
690: function for Eqs.~(\ref{mag1}, \ref{mag2}).
691:
692: The results of our numerical procedure were compared with corresponding
693: results published by Dado et al. (2002a) and Bloom et al. (2002), and we found
694: close agreement. We used our procedure to correctly \it predict \rm the color
695: evolution of GRB-SN 030329 (Zeh, Klose, \& Greiner 2003), and have performed a
696: very good numerical fit for the light curves of GRB-SN 011121 (Greiner et
697: al. 2003b). The limit of our procedure is given by the chosen photometric band
698: in combination with the redshift of the burster. Once we can no longer
699: interpolate in between the $UBVRI$ bands, but have to extrapolate into the UV
700: domain (cf. Bloom et al. 1999), results become less accurate.
701:
702: %--------------------------------------------------------------------------%
703: \section{Notes on individual bursts with detected supernova bump
704: \label{individual}}
705:
706: \it GRB 970228: \rm \ The light curves were constructed based on the data
707: compiled by Galama et al. (2000). The $R_c$ band light curve shows evidence
708: for extra light appearing $\sim$2 weeks after the burst, which can be
709: attributed to an underlying SN 1998bw component at the redshift of the burster
710: (Galama et al. 2000; Reichart 1999). There is no evidence for a break, but its
711: presence cannot be excluded due to the rather sparse data set. In the $I_c$
712: band the SN bump is most visible, but the light curve suffers from a
713: lack of early-time data. In the $V$ band the significance for extra light is
714: even lower, again due to the lack of observational data.
715:
716: \it GRB 980703: \rm \ The search of an SN component in the afterglow of GRB
717: 980703 is affected by the relatively bright host. Most data were taken
718: from Bloom et al. (1998), Castro-Tirado et al. (1999), Holland et al. (2001),
719: and Vreeswijk et al. (1999). Evidence for a late-time bump is rather weak.
720:
721: \it GRB 990712: \rm \ This burst had a relatively bright host galaxy
722: hampering the long-term study of its afterglow. We used the data presented by
723: Fruchter et al. (2000a), Hjorth et al. (2000), and Sahu et al. (2000) to
724: analyze the light curves. Although the $R_c$ band light curve is well-sampled,
725: the bright host may have hidden the detectability of a break at later
726: times. Our numerical procedure finds evidence for extra-light, confirming the
727: finding by Bj\"ornsson et al. (2001).
728:
729: \it GRB 991208: \rm \ We used the compilation of data by Castro-Tirado et
730: al. (2001), with additional data from Dodonov et al. (1999), Halpern et
731: al. (1999), Garnavich et al. (1999), and Fruchter et al. (2000b), including
732: our late-time observation of the host in early 2003 to analyze
733: the light curves. The $R_c$ band data can be fitted with or without the
734: inclusion of a break. In the former case the break is mainly due to a
735: single data point at $t\sim7$ days. Most likely, the afterglow was
736: discovered after a break had already occurred in the light curve. The
737: afterglow parameters were determined in the $R_c$ band. These parameters fit
738: well in the $V$ band. Unfortunately, in the $I_c$ band no data were obtained
739: during the time of the SN bump.
740:
741: \it GRB 000911: \rm \ This was a long-lasting burst with a duration of
742: $\sim$500
743: seconds (Hurley et al. 2000; Price et al. 2002b). The optical afterglow
744: was in detail observed by Price et al. (2002b) and Lazzati et al. (2001). We
745: confirm the finding by Lazzati et al. that the published data show evidence
746: for a bump in $VRI$ at later times.
747:
748: \it GRB 010921: \rm \ This burst occurred in a rather crowded stellar field
749: and had a relatively large error box (Hurley et al. 2001), which hampered the
750: early detection of its afterglow (Price et al. 2001). The light curve is
751: therefore not well-sampled. Using the data published by Park et al. (2002) and
752: Price et al. (2002a) combined with our late-time observations of the host,
753: our numerical procedure finds evidence for extra light with its peak time
754: $\sim2$ weeks after the burst. This result is obtained when we adopted a
755: single power-law decay, where the decay slope $\alpha$ was deduced from the
756: $r'$ band light curve. Our procedure finds a SN component with $k=0.68\pm0.48$
757: and $s=0.68\pm0.28$. Within the given uncertainties this is not in conflict
758: with the upper limit reported by Price et al. (2003).
759:
760: \it GRB 011121: \rm \ This was the nearest known burst at the time of its
761: discovery (excluding GRB 980425/SN 1998bw). It showed clear evidence for an
762: underlying SN component in several photometric bands (Bloom et
763: al. 2002; Dado, Dar, \& de R\'ujula 2002b; Garnavich et al. 2003; Greiner et
764: al. 2003b). In our fit we included late-time \it Hubble Space Telescope \rm
765: data (Bloom et al. 2002). A break is apparent in the light curve at
766: $t\sim1$ day (see Greiner et al. 2003b; note that in Greiner et al. we
767: assumed a Galactic extinction towards SN 1998bw of 0 mag).
768:
769: \it GRB 020405: \rm \ This is the second burst with known redshift and a
770: well-observed bump in its late-time afterglow. We used host-subtracted data
771: provided by N. Masetti (private communication) to analyze the light
772: curves. Extra light apparent in the late-time light curve can be attributed to
773: an underlying SN component, as already noted by Masetti et al. (2003). Our
774: procedure also detects a break in the light curves at $t\sim2$ days.
775:
776: \it GRB 021211: \rm \ For the fit we included data published by Della Valle et
777: al. (2003), Fox et al. (2003), Li et al. (2003), and Pandey et al. (2003). A
778: weak bump is apparent at late times, which is most likely due to an underlying
779: SN component given its (weak) spectral confirmation (Della Valle et
780: al. 2003).
781:
782: %--------------------------------------------------------------------------%
783: \begin{references}
784:
785: \reference{} Antonelli, L. A., et al. 2000, ApJ, 545, L39
786: \reference{} Bessel, M. S. 1979, PASP, 91, 589
787: \reference{} Beuermann, K., et al. 1999, A\&A, 352, L26
788: \reference{} Bj\"ornsson, G., et al. 2001, ApJ, 552, L124
789: \reference{} Bloom, J. S., et al. 1998, ApJ, 508, L21
790: \reference{} Bloom, J. S., et al. 1999, Nature, 401, 453
791: \reference{} Bloom, J. S., et al. 2002, ApJ, 572, L45
792: \reference{} Bloom, J. S., Kulkarni, S. R., \& Djorgovski, S. G. 2002,
793: AJ, 123, 1111
794: \reference{} Bloom, J. S., Frail, D. A., \& Kulkarni, S. R. 2003,
795: ApJ, 594, 674
796: \reference{} Candia, P., et al. 2003, PASP, 115, 277
797: \reference{} Castro-Tirado, A. J., et al. 1999, ApJ, 511, L85
798: \reference{} Castro-Tirado, A. J., et al. 2001, A\&A, 370, 398
799: \reference{} Colgate, S. A. 1968, Canadian J. Phys., 46, S476
800: \reference{} Dado, S., Dar, A., \& de R\'ujula, A. 2002a, A\&A, 388, 1079
801: \reference{} Dado, S., Dar, A., \& de R\'ujula, A. 2002b, ApJ, 572, L143
802: \reference{} Dado, S., Dar, A., \& de R\'ujula, A. 2003a, ApJ, 585, 890
803: \reference{} Dado, S., Dar, A., \& de R\'ujula, A. 2003b, ApJ, 593, 961
804: \reference{} Dar, A. \& de R\'ujula, A. 2003, astro-ph/0308248
805: \reference{} Della Valle, M. et al. 2003, A\&A, 406, L33
806: \reference{} Dodonov, S. N., et al. 1999, GCN Circ. 461
807: \reference{} Djorgovski, S. G., et al. 1997, GCN Circ. 289
808: \reference{} Fox, D. W. et al. 2003, ApJ, 586, L5
809: \reference{} Frail, D. A., et al. 2002, ApJ, 565, 829
810: \reference{} Frontera, E. et al. 1998, ApJ, 493, L67
811: \reference{} Fruchter, A., et al. 2000a, GCN Circ. 752
812: \reference{} Fruchter, A., et al. 2000b, GCN Circ. 872
813: \reference{} Fryer, C. L., Woosley, S. E., \& Hartmann, D. H. 1999,
814: ApJ, 526, 152
815: \reference{} Fynbo, J. U., et al. 2001, A\&A, 369, 373
816: \reference{} Fynbo, J. U., et al. 2004, Proc. Santa Fe 2003 conference
817: \reference{} Fukugita, M., Shimasaku, K., \& Ichikawa, T. 1995, PASP, 107,945
818: \reference{} Galama, T. J., et al. 1998, Nature, 395, 670
819: \reference{} Galama, T. J., et al. 2000, ApJ, 536, 185
820: \reference{} Garnavich, P. M., et al. 1999, GCN Circ. 456
821: \reference{} Garnavich, P. M., et al. 2003, ApJ, 582, 924
822: \reference{} Greiner, J., et al. 2003a, GCN Circ. 2020
823: \reference{} Greiner, J., et al. 2003b, ApJ, 599, 1223
824: \reference{} Groot, P. J., et al. 1998, ApJ, 493, L27
825: \reference{} Halpern, J. P., et al. 1999, GCN Circ. 458
826: \reference{} Heger, A., et al. 2003, ApJ, 591, 288
827: \reference{} Hjorth, J., et al. 2000, ApJ, 534, L147
828: \reference{} Hjorth, J., et al. 2003, Nature, 423, 847
829: \reference{} Holland, S. et al. 2001, A\&A, 371, 52
830: \reference{} Hurley, K., et al. 2000, GCN Circ. 791
831: \reference{} Hurley, K., et al. 2001, GCN Circ. 1097
832: \reference{} Iwamoto, K., et al. 1998, Nature, 395, 672
833: \reference{} Kawabata, K. S., et al. 2003, ApJ, 593, L19
834: \reference{} Klebesadel, R. W., Strong, I. B., \& Olson, R. A. 1973, ApJ, 182,
835: L85
836: \reference{} Klose, S., et al. 2000, ApJ, 545, 271
837: \reference{} Klose, S., et al. 2003, ApJ, 592, 1025
838: \reference{} Kulkarni, S. R. et al. 1998, Nature, 395, 663
839: \reference{} Lazzati, D., Campana, S., \& Ghisellini, G. 1999, MNRAS, 304, L31
840: \reference{} Lazzati, D., et al. 2001, ApJ, 556, 471
841: \reference{} Lazzati, D., Covino, S., \& Ghisellini, G. 2002, MNRAS, 330, 583
842: \reference{} Lazzati, D., et al. 2001, A\&A, 378, 996
843: \reference{} Li, W. et al. 2003, ApJ, 586, L9
844: \reference{} Lipkin, Y. M. et al. 2004, ApJ, in press (astro-ph/0312594)
845: \reference{} Masetti, N., et al. 2003, A\&A, 404, 465
846: \reference{} Matheson, T., et al. 2003, ApJ, 599, 394
847: \reference{} M\'esz\'aros, P., \& Rees, M. J. 1999, MNRAS 306, L39
848: \reference{} M\'esz\'aros, P., \& Rees, M. J. 2001, ApJ, 556, L37
849: \reference{} Mirabal, N., Paerels, F., \& Halpern, J. P. 2003, ApJ, 587, 128
850: \reference{} Paczy\'nski, B. 1998, ApJ, 494, L45
851: \reference{} Park, H. S., et al. 2002, ApJ, 571, L131
852: \reference{} Pandey, S. B. et al. 2003, A\&A, 408, L21
853: \reference{} Patat, F., et al. 2001, ApJ, 555, 900
854: \reference{} Peterson, B. A., \& Price, P. A. 2003, GCN Circ. 1985
855: \reference{} Piro, L. et al. 1999, ApJ, 514, L73
856: \reference{} Price, P., \& Peterson, B. A. 2003, GCN Circ. 1987
857: \reference{} Price, P. A., et al. 2001, GCN Circ. 1107
858: \reference{} Price, P. A., et al. 2002a, ApJ, 571, L121
859: \reference{} Price, P. A., et al. 2002b, ApJ, 573, 85
860: \reference{} Price, P. A., et al. 2003, ApJ, 584, 931
861: \reference{} Reichart, D. 1999, ApJ, 521, L111
862: \reference{} Reichart, D. 2001, ApJ, 553, 235
863: \reference{} Reeves, J. N. et al. 2002, Nature, 416, 512
864: \reference{} Rhoads J. E., \& Fruchter A. S. 2001, ApJ, 546, 117
865: \reference{} Richardson, D., et al. 2002, AJ, 123, 745
866: \reference{} Rieke, G. H., \& Lebofsky, M. J. 1985, ApJ, 288, 618
867: \reference{} Sahu, K. C., et al. 2000, ApJ, 540, 74
868: \reference{} Schlegel, D., Finkbeiner, D., \& Davis, M. 1998, ApJ, 500, 525
869: \reference{} Sokolov, V. V., et al. 2001, A\&A, 372, 438
870: \reference{} Stanek, K. Z., et al. 2003, ApJ, 591, L17
871: \reference{} Stathakis, R. A., et al. 2000, MNRAS, 314, 807
872: \reference{} Stritzinger, M., et al. 2002, AJ, 124, 2100
873: \reference{} Tiengo, A. et al. 2003, A\&A, 409, 983
874: \reference{} Tinney, C., Stathakis, R., \& Cannon, R. 1998, IAU Circ. 6896
875: \reference{} Vietri M., \& Stella M. 2000, ApJ, 527, L43
876: \reference{} Vreeswijk, P. M. et al. 1999, ApJ, 523, 171
877: \reference{} Woosley, S. E., Zhang, W., \& Heger, A. 2002, in:
878: From Twilight to Highlight: The Physics of Supernovae.
879: Proceedings of the ESO/MPA/MPE Workshop held in Garching,
880: Germany, 29-31 July 2002, p. 87; astro-ph/0211063
881: \reference{} Zeh, A., Klose, S., \& Greiner, J. 2003, GCN Circ. 2081
882: \reference{} Zombeck, M. V. 1990, Handbook of Space Astronomy \&
883: Astrophysics. Cambridge University Press, p. 100
884: \end{references}
885:
886:
887: %--------------------------------------------------------------------------
888: \clearpage
889:
890: \begin{table*}
891: \caption{The input sample of GRB afterglows\tablenotemark{a}\label{allgrbs}}
892: \vspace{0.3cm}
893: \begin{tabular}{|ll|ll|ll|}
894: \hline \noalign{\smallskip}
895: GRB & $z$ & GRB & $z$ & GRB & $z$ \\
896: \noalign{\smallskip} \hline
897: \noalign{\smallskip}
898: 970228 & 0.695 & 991208 & 0.706 & 010921 & 0.450 \\
899: 970508 & 0.835 & 991216 & 1.02 & 011121 & 0.362 \\
900: 971214 & 3.42 & 000301C& 2.04 & 011211 & 2.140 \\
901: 980703 & 0.966 & 000418 & 1.118 & 020405 & 0.69 \\
902: 990123 & 1.600 & 000911 & 1.058 & 020813 & 1.25 \\
903: 990510 & 1.619 & 000926 & 2.066 & 021004 & 2.3 \\
904: 990712 & 0.434 & 010222 & 1.477 & 021211 & 1.01 \\
905: \noalign{\smallskip} \hline
906: \end{tabular}
907: \tablenotetext{a}{Redshifts were taken from the literature.}
908: \end{table*}
909:
910: %--------------------------------------------------------------------------
911: \clearpage
912:
913: \begin{table*}
914: \caption{Best-fit parameters for the SN component found in GRB
915: afterglows\tablenotemark{a} \label{res}}
916: \renewcommand{\tabcolsep}{4pt}
917: \vspace{0.3cm}
918: \begin{tabular}{|lccc|ccc|cc|cc|}
919: \hline \noalign{\smallskip}
920: GRB & $z$ & band & $\lambda_{\rm host}$& $k$ & $s$ & $\chi^2_{\rm d.o.f.}$
921: & $k$ if $s$=1& $\chi^2_{\rm d.o.f.}$ & $\chi^2_{\rm noSN}$ & data \\
922: \noalign{\smallskip} \hline
923: \noalign{\smallskip}
924: & & $I_c$ & 476 & \nodata & \nodata &\nodata& 0.66$\pm$0.27 & 0.01 & 2.16 & 4 \\
925: 970228 & 0.695 & $R_c$ & 389 & 0.40$\pm$0.24 & 1.46$\pm$0.80 & 0.70 & 0.33$\pm$0.30 & 0.71 & 0.77 & 10\\
926: & & $V$ & 325 & \nodata & \nodata &\nodata& 0.25$\pm$0.50 & 0.06 & 0.15 & 4 \\[1mm]
927:
928: & & $I_c$ & 410 & \nodata & \nodata &\nodata& \nodata &\nodata& 1.59 & 14\\
929: 980703 & 0.966 & $R_c$ & 335 & \nodata & \nodata &\nodata& 1.66$\pm$1.22 & 0.78 & 0.79 & 19\\
930: & & $V$ & 280 & \nodata & \nodata &\nodata& \nodata &\nodata& 1.50 & 7 \\[1mm]
931:
932: & & $I_c$ & 562 & 1.00$\pm$0.38 & 0.56$\pm$0.10 & 0.55 & \nodata &\nodata& 1.67 & 6 \\
933: 990712 & 0.434 & $R_c$ & 459 & 0.48$\pm$0.10 & 0.89$\pm$0.10 & 1.00 & 0.43$\pm$0.08 & 1.01 & 2.25 & 23\\
934: & & $V$ & 384 & 0.37$\pm$0.44 & 0.71$\pm$0.36 & 2.30 & 0.29$\pm$0.18 & 1.62 & 1.99 & 16\\[1mm]
935:
936: & & $I_c$ & 472 & \nodata & \nodata &\nodata& \nodata &\nodata& 1.82 & 13\\
937: 991208 & 0.706 & $R_c$ & 386 & 0.90$\pm$0.35 & 1.12$\pm$0.28 & 1.64 & 1.02$\pm$0.32 & 1.56 & 2.52 & 20\\
938: & & $V$ & 323 & 1.16$\pm$0.19 & 1.86$\pm$0.10 & 0.45 & 0.93$\pm$0.30 & 1.04 & 2.26 & 11\\[1mm]
939:
940: & & $I_c$ & 392 & 0.39$\pm$0.37 & 1.06$\pm$0.50 & 1.83 & 0.40$\pm$0.29 & 1.30 & 1.61 & 7 \\
941: 000911 & 1.058 & $R_c$ & 320 & 0.87$\pm$0.39 & 1.49$\pm$0.33 & 0.75 & 0.51$\pm$0.43 & 1.14 & 1.22 & 8 \\
942: & & $V$ & 267 & \nodata & \nodata &\nodata& 0.43$\pm$1.24 & 1.25 & 1.42 & 6 \\[1mm]
943:
944: & & $I_c$ & 556 & \nodata & \nodata &\nodata& 0.40$\pm$1.67 & 0.50 & 0.28 & 4 \\
945: 010921 & 0.450 & $R_c$ & 454 & 0.68$\pm$0.48 & 0.68$\pm$0.28 & 0.42 & 0.43$\pm$0.10 & 0.78 & 2.74 & 6 \\
946: & & $V$ & 380 & \nodata & \nodata &\nodata& \nodata &\nodata&\nodata& 2 \\[1mm]
947:
948: & & $I_c$ & 632 & \nodata & \nodata &\nodata& \nodata &\nodata&\nodata& 0 \\
949: 011121 & 0.360 & $R_c$ & 484 & 0.79$\pm$0.06 & 0.85$\pm$0.06 & 0.92 & 0.74$\pm$0.05 & 1.32 & $>$20 & 13\\
950: & & $V $ & 405 & 0.86$\pm$0.09 & 0.81$\pm$0.06 & 2.13 & 0.83$\pm$0.05 & 3.26 & $>$20 & 10\\[1mm]
951:
952: & & $I_c$ & 476 & 0.76$\pm$0.17 & 0.80$\pm$0.17 & 5.29 & 0.71$\pm$0.10 & 5.68 & $>$20 & 10\\
953: 020405 & 0.695 & $R_c$ & 389 & 0.74$\pm$0.17 & 0.98$\pm$0.17 & 5.26 & 0.72$\pm$0.11 & 4.86 & $>$20 & 18\\
954: & & $V$ & 325 & 0.69$\pm$0.22 & 0.74$\pm$0.13 & 6.79 & 0.53$\pm$0.16 & 6.91 & $>$20 & 14\\[1mm]
955:
956: & & $I_c$ & 428 & \nodata & \nodata &\nodata& \nodata &\nodata&\nodata& 0 \\
957: 021211 & 1.006 & $R_c$ & 328 & 0.97$\pm$0.87 & 0.74$\pm$0.23 & 2.68 & 0.52$\pm$0.34 & 2.65 & 2.79 & 35\\
958: & & $V$ & 274 & \nodata & \nodata &\nodata& \nodata &\nodata&\nodata& 0 \\
959: \noalign{\smallskip} \hline
960:
961: \end{tabular}
962: \tablenotetext{a}{Columns: (1) and (2): GRB and redshift; (3) photometric
963: band, in which the light curve was fitted; (4) central wavelength of the
964: photometric band in the host frame in units of nm, adopting for $V, R_c, I_c$
965: wavelengths of 550, 659, and 806 nm, respectively; (5) peak luminosity of the
966: fitted SN component in the corresponding wavelength band (observer frame) in
967: units of SN 1998bw, after correction for Galactic extinction; (6) stretch
968: factor $s$ (Eq.~\ref{ot}); (7) goodness of fit per degree of freedom; (8) and
969: (9) the same as (5) and (6) for $s=1$; (10) goodness of fit per degree of
970: freedom assuming that there is no underlying SN component; (11) total
971: number of data points used for the fit. Note that the low
972: $\chi^2$/d.o.f. for GRB 970228 is due to the small number of data points.}
973: \end{table*}
974:
975: \end{document}
976: