1) [Name of sender and manuscript number (if you have)] Michael Richmond 2) [Contact address for the sender (Address, Telephone, Fax, E-mail address)] Address: Physics Department Rochester Institute of Technology 85 Lomb Memorial Drive Rochester, New York 14623-5603 USA Telephone: +0041 - 585 - 475 - 2538 Fax: +0041 - 585 - 475 - 5988 E-Mail: mwrsps@rit.edu 3) Title of manuscript and list of authors Proper motions with Subaru I. Methods and a first sample in the Subaru Deep Field Michael Richmond Tomoki Morokuma Yutaka Komiyama Mamoru Doi Naoki Yasuda Sadanori Okamura Avishay Gal-Yam 4) List of files that will be sent (Note name of each files accurately) README (1/12) body.tex (2/12) Figure1.eps (3/12) Figure2a.eps (4/12) Figure2b.eps (5/12) Figure3.eps (6/12) Figure4a.eps (7/12) Figure4b.eps (8/12) Figure5.eps (9/12) Figure6.eps (10/12) Figure7.eps (11/12) Figure8.eps (12/12) 5) Comment for editorial office, if any. This is the revised version of manuscript No. 3401. We have replied to the referee's many suggestions. Our detailed comments to the referee appear below. ----------------- response to referee follows --------------------------- Reply to referee's comments on paper No. 3401 "Proper Motions with Subaru I. Methods and a sample in the Subaru Deep Field" Michael Richmond Aug 25, 2008 In this reply, the referee's comments are preceded by "> " characters. The referee wrote: > If the purpose of the paper > is to serve as a "master" reference point for future work on proper > motions by this team or others, I believe more details are needed in > the main body. If the paper serves specifically for the present > science case (studying galaxy structure and populations), then the > authors should provide a more in-depth analysis of the results. We have decided that this paper serves the first purpose: as a "master" reference point for future work. Therefore, we have carefully followed the referee's suggestions for changes in this area. However, we have not produced "a more in-depth analysis of the results", leaving that job for future papers. There are two major changes to the paper: 1) In order to follow the referee's suggestion that all analysis act on the entire sample of possible stellar objects, we have completely re-analyzed our measurements. 2) Following the referee's suggestions that we investigate incompleteness, we have used two sets of artificial stars (one stationary, one moving) to quantify the completeness of our catalogs. Below are our replies to each of the referee's comments, item-by-item. ------------------------------------------------------------------------- > No. 3401 > Referee Report Form #2 [to the author(s)] > > 1.) There are strange formatting subscripts in the keywords and also > separating various paragraphs...I'm not sure if the Journal put these > in or if its a problem with the latex. These subscripts were present in the LaTeX template provided by the PASJ. I have removed them from the current version. > 2.) Table 1: add in the Kalirai et al. (2004) survey numbers here. Done. > 3.) In Section 1: I think the authors are underestimating the > advantage of Subaru. Rather than refer to the present work as a > "pencil-beam" survey, it really is the best of both worlds. The > Subaru approach offers both a deeper survey than the shallower 4-m or > smaller approaches, and at the same time offers a much larger field > of view than something like HST. We have modified the text to emphasize the wider area of Subaru, compared to most pencil-beam surveys. > 4.) In section 3, I'd like to see a plot showing the source > morphology. It becomes increasingly difficult to measure the > position of a galaxy as the source becomes extended. A diagnostic > similar to "SHARP" in DAOPHOT or "STELLARITY" in Source Extractor can > be used to weed out galaxies based on their morphology. It would be > interesting to comment on the agreement of things that are > morphologically found to be galaxies, and also confirmed to be such > with proper motions. This can be very important for future work > trying to eliminate galaxies that doesn't have access to proper motions. We sympathize with the referee on this topic, but we do not agree that proper motions can be an important method for separating stars from galaxies. The problem is numbers: in our field of the SDF, we counted 79,000 objects which were compact with sharp cores. Some of these objects are stars, and some galaxies. We were able to measure significant proper motions for only 110 objects. That is only 0.14 percent of all compact objects. It is not practical to use proper motions to separate stars from galaxies, since so very, very few objects will show proper motion. A better choice is to use color to separate stars from compact galaxies. See, for example, Richmond, PASJ, 57, 969 (2005) Finlator et al., AJ, 120, 2615 (2000) Strateva et al., AJ, 122, 1861 (2001) In Richmond (2005), for example, found that of 6348 objects with very compact, star-like shapes in the SDF, 1061 could be classified as stars based on their colors. That success rate of about 16% is large enough to be useful. > 5.) More details are also needed on how the different frames were > registered. For example, what type of transformation was used in > each sector to register the frames? Was this tweaked through > successive iterations after the initial solution was derived. What > magnitude stars were used for the matching? A plot showing the > positional difference of matched sources as a function of magnitude > would help illustrate the accuracy of this method for different types > of stars. We have added detail to the discussion of how frames were registered and coordinates transformed to the fiducial frame. The referee requests a plot of positional difference as a function of magnitude; we already provide this information in Table 3 of the paper. > 6.) Several cuts are made in Section 3 that are not justified. If > this paper is to stand as a "master" reference point as discussed > earlier, the catalogs should not be cut down purely for reasons > "manageable size" reasons. Additionally, the 1.0" search radius is > chosen arbitrarily and some discussion should be given on how the > results > change if this radius is made larger/smaller. The referee is exactly right. We made big changes to our procedures and re-reduced all the data from the beginning to follow his suggestions. We now a) keep all objects in a "master" catalog b) do NOT apply a preliminary cut c) compute proper motions and uncertainties for all objects in the master catalog We have also added a figure showing the distribution of proper motions. We mark on this figure the limit imposed by the 1.0" search radius, and show that this limit is far larger than any detected proper motion. Therefore, the 1.0" search radius does not affect our sample in a significant manner. > 7.) The Besancon model, and its assumptions, should be summarized in > this paper. Also, can the authors comment on findings that the model > is inconsistent with previous observations along certain lines of > sight (e.g., M31 work by Ibata et al. 2007). We have added a paragraph which describes the Besancon model to the Introduction of our paper. This paragraph also mentions briefly the good points and bad points of the model as pointed out by Ibata et al. 2007. > 8.) Figure 2: It would be nice to also show some snapshot images > illustrating several moving sources. We agree with the referee that this would be nice, and we have made illustrations of this type (indeed, we made these for our own use in checking for spurious motions). However, even the largest motions are so much smaller than the FWHM of the images that they do not appear perceptible to the eye when printed on paper. We can barely see the motion when we blink two perfectly registered images on our computer screens. Therefore, we have not added snapshot images. > 9.) The authors may want to consider their comment on page 9 > regarding the use of (V-I) color. This color offers a > wide baseline and thus is often more useful than V-R or B-V. We have modified our discussion of (V-I) to follow this suggestion. > 10.) I am specifically concerned that the methods described above to > yield proper motions is particularly relevant > for the subsequent discussion on populations. I would suggest to > generalize the catalogs to create a master file from which authors > can make cuts for their specific studies. As mentioned above, we have now done this: we keep a master catalog of ~80,000 sources and always start with it when performing some piece of analysis. > 11.) Section 4 presents a nice preliminary study of the stellar > populations along this sightline, and interprets the results through > a comparison with the Besancon models. I would first like to see a > discussion of incompleteness effects, both in the photometry and > proper motions before the comparison to Besancon models. We have now added a significant component to our analysis: artificial stars, added to the original Subaru images. We have made two sets of artificial stars, one stationary and one moving with known motions, added these stars to the images, and re-analyzed the images. We checked the output of our procedures against the input set of artificial stars to quantify the efficiency of our survey as a function of magnitude alone, and as a function of magnitude and proper motion together. The paper includes two new figures showing the results. We also include incompleteness when comparing the number of white dwarfs in the Besancon model to the number of white dwarf candidates in our observations. Before we added this completeness factor, the number of WDs in the model was roughly twice as large as the number we observed. After adding completeness, we find the two numbers agree. Thanks to the referee for suggesting this important test. > I would > then suggest that the authors use these results to constrain some > problems and compare their values to previous studies. For example, > > - do these results rule suggest that the microlensing events detected > in MACHO are not caused by white dwarfs? > - what is the new local WD spheroid and halo density based on these > numbers? > - can the normalizations of the thin and thick disks be estimated > from these numbers? As stated at the start of this letter, we have decided to make the current paper more of a "master reference" for future work on the technical aspects of measuring proper motion, than an in-depth analysis of the results. We wish to postpone a detailed comparison of the results to galactic models for a future paper, as we state in this paper.