Content-Type: text/plain; charset=us-ascii; x-mac-type="54455854"; x-mac-creator="4D4F5353"; name="Referee's report for \"Spectral " Content-Transfer-Encoding: 7bit Content-Description: Unknown Document Content-Disposition: inline; filename="Referee's report for \"Spectral " Subject: Referee's report for "Spectral homogeneity of SNe Ia" Date: Mon, 31 Mar 2003 17:36:42 -0800 From: Andy Howell To: Gaston Folatelli CC: Ariel Goobar , Tony Spadafora , scpexec@lbl.gov Gaston, I have reviewed your paper "Spectral homogeneity of SNe Ia". General comments are listed below. I will fax you a copy of the paper with my wording-related edits. Everything here refers to the draft from last week. My overall impression is that the paper is a good empirical investigation into changes in spectroscopic features in supernovae, although it needs work in a few places. The good news is that the problems are quite fixable with a little more work. What is good: EW's have never been investigated in depth before, so you are doing something new, and with a large enough sample to see trends in the data. What is not good: While there is nothing wrong in principle with doing an empirical investigation into correlations between features (indeed this must be done), the definition of some of your features mixes so many variables that one would not expect to find any correlation at all. Not surprisingly, you did not find much of any significance associated with these features. This can come off looking like you don't know what you are doing. To remedy this, I suggest: a) A more restricted definition of features that removes at least some of the ambiguity -- e.g. limiting the analysis to the photospheric phase b) New supplemental tests that are better motivated. c) More description of which ions are thought to exist over the wavelength interval you define as a feature. d) More weasel words that say "even though we know this mixes many variables, we are doing it anyway." e) More references to the literature. Actually, this last point deserves repeating. This paper does not show much familiarity with the literature, and that is a major (but easily corrected) weakness. I suggest several papers to read below. My comments are broken into several categories: I. General issues II. Text-specific issues III. New things to try IV. Style V. Papers you should read and possibly reference I. General problems / questions * My biggest problem is that your definitions of features is essentially physically meaningless for most features. You continue the definitions from the photospheric phase into the nebular phase. What starts out as an absorption feature with a well-defined EW becomes just an empty space in between two emission features at later times. Photospheric and nebular spectra are physically distinct -- considering them as part of the same definition of features muddles the interpretation of the results. On almost every plot, once you enter the nebular phase at 20-30 days the EW levels out. So you are not adding interesting information by including the nebular phase info. The nebular data can throw off your results in two ways: 1) Different SNe enter the nebular phase at different times, and this is correlated with stretch. So at say day 25 you might be comparing absorption from one SN to a gap in emission from another. 2) It adds noise that may affect the plots like Fig. 4, when you average delta EW. To avoid these problems and have a clean definition of EW, I think you should only use photospheric data. It may be possible to come up with different indicators for nebular phase features, such as the pseudo-EW of an emission feature. * You should mention that EW as you define it in SNe is very different than the EW over a single narrow line for a star. Many of the traditional properies associated with EW (e.g. a meaningful curve of growth) will not be applicable. * Fig. 1 is not sufficient to show your line definitions. You should show this for multiple epochs and types of spectra. * Your results may be dominated by emission near an absorption feature rather than than the true EW of the line itself. For example, as Ca IR emission grows at late times, this must affect the results. Can you comment on this or estimate the effects? Can you try drawing a line from say 7900 A to 9000 A, and use that as the continuum? This would leave out the emission and isolate the effects of absorption. * Similarly, does EW of Ca emission correlate with stretch? * How does reddening, or just a difference in the color of the SN affect the results? Redden a blue spectrum in IRAF by Av=1 and test the effects. * Your results for some lines are dominated by different ions coming into the spectrum. This happens at a different rate for different stretch SNe. You mention this in the text but not in much detail (it is a big effect!). Please add a table giving the rest and observed wavelength of various lines and some rough idea of when they affect the spectra of a normal Ia. A starting point is Wheeler's chapter on SNe in the 2000 edition of Allen's Astrophysical Quantities. Another good source is Dan Kasen's web page: http://panisse.lbl.gov/~dnkasen/tutorial/ * You should have a table of the resolution of each spectrum and comment on whether or not you expect this to affect the results. Since some features are defined by emission peaks, if you smooth out the sharpness of the emission peaks, this could have an effect. To test this, take your best resolution spectrum (and one with reasonably sharp features) and smooth it to get the resolution of your worst resolution spectrum. See if there are any differences in your measured features. It would be useful to see S/N in this table as well. II. Text-specific comments The Abstract should have some of your specific conclusions. When I read it now, I don't know why the paper is interesting. The first paragraph of the introduction should be rewritten. It is convoluted and it is not clear what you are trying to say. I would suggest saying that the only way to get information about the physical conditions inside a supernova (e.g. the elemental composition and its distribution) is through analyis of spectra. You should also summarize the work that has been done before in this area (see Section 5 below). p. 5 Sec. 4., par. 2. You say Fe II 4800, Ca II IR, and Mg II 4300 "are believed to be formed by lines from single elements and suffer limited contamination at all epochs." This is *not true* for Fe II 4800 and Mg II 4300. The 4000-5000 A region is dominated by many overlapping lines of Fe, Ni, and Co of various ionization states. See: http://panisse.lbl.gov/~dnkasen/tutorial/ You even say this later, for Mg II, in section 4.3 p. 5-6 I don't believe the trend shown for Fe II 4800 in Fig. 4. You had to throw out 1/6 of the data to get the trend, and the trend is driven by 3 SNe. You need to at least have a reason to throw out the SNe that you did. p. 7 Sec. 4.3 You say Mg II 4300 is formed by Mg II, Co II, Fe II, Fe III, and Si III, but you leave off Ti II, which is *huge* in underluminous SNe. In fact, Ti II can completely change the morphology around 4300 A when it exists in a SN. Leaving this off of your description of what happens to this feature in section 4.3 is a huge oversight. p. 10 Somewhere in sec. 4.4 include a paragraph noting that high-z SNe are often classified based on the presence of the Si 4000 A feature. This could cause a selection effect at high-z (though there is not much we can do about it). p. 11, sec 4.5, par 2. SN 1991T is known for having essentially no Ca II. Are your numbers measuring something significant? See the literature below on 91T. p.11 sec. 4.5 3rd par. About SN 1999bm: "... an unusually broad Ca II H&K feature. After examining the reduction and calibration for these spectra this feature is claimed to be real." There is no need for this statement -- this SN apparently belongs to a category of high-velocity SNe which have many broad features. It may be similar to SN 2001el: astro-ph/0303397, 0301312, 0210327. III. New things to try: * You did a good job of investigating the equivalent widths of features, but this can be difficult to interpret physically. This is because several effects contribute to the EW, such as the depth of the feature and the FWHM of the feature. For example, a shallow, wide feature can have the same EW as a deep, narrow one, yet arise from completely different physical conditions. There are two more indicators that you can measure trivially from the spectra, and are easier to interpret, so they may make a good complement to the EWs you have measured. These are only applicable to well-defined, single features such as Si, Ca IR, Ca H & K, etc. They are: (1) Blueshift of the absorption minima in km/s. This has been investigated in other studies, but you have a better set of SNe now. This tells you the velocity at which the ion is centered in the SN. As the photosphere recedes, the absorption minima gets redder. Both the beginning velocity and its rate of change should be different in high and low stretch SNe. (2) The Full Width at Half Maximum (FWHM) of the feature. This will give you some information about the distribution in velocity of the feature at the photosphere. You can measure this by fitting a gaussian to the feature in IRAF. * Are there linear combinations of parameters that produce a stonger correlation than a single parameter alone? * How do the EWs that you calculated compare to EWs determined by fitting a gaussian in IRAF? Maybe you could do this for a few lines where the process is well defined as a sanity check, since this is how most people measure EWs. * Is it possible to make any plots that show absolute magnitude rather than delta-m15 (I don't know how many have a reliable distance determination)? * You may be able to find more SNe to add to your sample at the SuSPECT archive. I know they have data for SN 1996X and SN 1999ee. http://tor.nhn.ou.edu/~suspect/ IV. Style: * Equating overluminous with high-stretch may not be correct. SN 1991T may not be overluminous (Saha et al. 2001, ApJ, 551, 973). At any rate it is an unnecessary assumption. You should talk about high delta-m15 and low delta-m15 (or low stretch and high stretch) SNe. * Don't use the words explosion and SN interchangeably. The explosion happens in a few seconds. What we see as the SN is the aftermath of the explosion, when the SN is in free expansion and is powered by Ni decay. * Figures of EW vs. epoch are hard to follow. Please experiment with the following: List the SNe in delta-m15 order, and possibly color them with different colors along a sequence or differently sized points along a sequence, or group them into three groups (high, low, normal, and peculiar). I see that you did this on the newest version. This makes the trends easier to see, but the grayscale is too faint to see. You might also try connecting the points for a given SN with lines. * In general the grayscale on some figures is too faint to see. * I would change the title to "Spectroscopic Homogeneity of Type Ia Supernovae" V. A sample of papers to investigate. This is not meant to be a comprehensive list, only a sample of some important papers that came to mind that you should at least read if you are going to write a paper on spectroscopic features. You should certainly cite the papers that the data was taken from in your study. General: Nuget et al. 1995, ApJ, 455L, 147 (spectroscopic sequence) Branch & van den Bergh, 1993, AJ, 105, 2231 (blueshift of Si). SN 1981B: Branch et al. 1983, ApJ, 270, 123 SN 1986G: Phillips et al. 1987, PASP, 99, 592 SN 1989B: Wells et al. 1994, AJ, 108, 2233 SN 1990N: Jeffrey, et al. 1992, ApJ, 397, 304 (theoretical) Fisher et al. 1997, ApJ, 481L, 89 (theoretical) SN 1991bg: Filippenko et al. 1992, AJ, 104, 1543 Turatto et al. 1996, MNRAS, 283, 1 Leibundguit et al. 1993, AJ, 105, 301 Mazzali, et al. 1997, MNRAS, 284, 151 (theoretical) SN 1991T: (but watch out because early estimates of the luminosity were wrong) Fisher, et al. 1999, MNRAS, 304, 67 Meikle, et al. 1996, MNRAS, 281, 263 Mazzali, et al. 1995, A&A, 297, 509 (theoretical) Jeffrey, et al. 1992, ApJ, 397, 304 (theoretical) Filippenko et al. 1992, ApJ, 384L, 15 SN 1992A: Kirshner et al. 1993, ApJ, 415, 589 SN 1994D: Meikle et al. 1996, MNRAS, 281, 263 Hatano et al. 1999, ApJ, 525, 881 (theoretical, but has line IDs) SN 1999by: Garnavich et al. 2001 astro-ph/0105490 (I don't know why it hasn't come out yet in the journal; first ref to 5800 as Ti) SN 1999aw: Strolger et al. 2002, AJ, 124, 2905 -Andy --------------E49247B825D8619468C20B82--