http://www.astro.ku.dk/~erik/gaia/73.hrs Subject: Work package SAG-13.2-HRS This is work package on a specific subject, as proposed at SAG-13 on 2/3 Feb. 2000. The text has been agreed with M. Perryman. High-resolution spectrophotometry (HRS) ======================================= E. Hoeg (draft no. 2 : 1.03.2000) 28 April 2000 SAG_CUO_73 ABSTRACT: The radial-velocity spectrometer (RVS) obtains spectra of the range 850-875 nm for determination of radial velocities mainly by means of the lines of the Calcium triplet. It has been proposed to use these spectral data at a nominal resolution of 0.75 A/pixel for "high-resolution spectrophotometry" (HRS) but no values for the obtainable accuracy have ever been given. A quantitative study of the use of the HRS measurements of the DIB at 862 nm is proposed. The possible accuracy (standard error versus magnitude) shall be given. Its role for determination of the interstellar reddening in connection with the photometric F-system shall be discussed. We define the tasks and give background information. It is necessary that the proposers of the DIB and the F-system carry out this study in order to verify and do justice to their own ideas, but a study by others is equally encouraged. ===== 1. Work Package Summary Quantify the HRS measurements of the DIB at 862 nm, including: - performances for nominal dispersion (0.75 A/pixel) - performance sensitivity for other dispersions - requirement on transverse samples for background: 3 (nominal) or 4 pixels - whether the DIB measurements are mandatory for determination of reddening with the photometric F-system Please inform me briefly if you start such a study. I will keep the SAG informed and also inform you of relevant contacts. ===== 2. Proposed task Estimate the precision of interstellar absorption measured by HRS of the diffuse interstellar band (DIB) at 862 nm with GAIA. Compare with the precision obtained by medium-band photometry with the proposed photometric systems for GAIA (see SAG_CUO_71 and _72). A discussion of the required spectral sampling should be included. 0.75 A/pixel along the spectrum should be the baseline since this is the current baseline for the radial-velocity sampling of the spectral range 850 - 875 nm, i.e. 310 pixels along scan. Perpendicular to the spectrum 3 pixels are sampled according to the Study Report, giving a total of 3*310=930 samples for transmission to ground. But are the 3 pixels sufficient to provide the sky background ? or are 4 pixels required ? Other issues of HRS than interstellar absorption measured with the DIB are also relevant, but it is proposed to keep the measurement of other spectral lines in separate reports for the sake of clarity. The measurement of the DIB are required for determination of the interstellar reddening if the photometric F system is used (see a section below.) ===== 3. Discussion of a report by U. Munari A report on the diffuse interstellar band (DIB) at 862 nm by U. Munari (UM) has been received on 12.01.2000 through the DMS system. (It has no date or number. It is written for the Proceedings of a meeting). A good correlation between the equivalent width and the interstellar reddening has been further confirmed by means of high-resolution spectra. Spectra of all types of stars have been published by Munari et al. (see the DMS). --- Comments to the report: 1. It would be interesting and necessary for a realistic discussion if an estimate were made of the precision of interstellar absorption measured by this method with GAIA. The necessary information about the GAIA instruments for such an estimate are available in the Study Report. The resulting precision should be compared with that obtained by medium-band photometry with the proposed photometric systems. 2. UM writes that the resolving power corresponds to 0.25 A/pixel. The correct value is however 0.75 A/pixel according to the final GAIA Study Report, or perhaps 1.0 A/pixel according to recent SAG discussions. 0.75 A/pixel has been the baseline value since two years. 3. UM writes that "photometry is accurate to 0.001 mag [1 millimag] for the majority of the 10^9 target stars." (ie. for more than 500 million stars.) The situation is however that we do not promiss 1 millimag for any stars, but we expect, e.g., 0.01 mag precision for MBP in the F57 band for V<18.5 mag. The photon noise in itself would promiss 0.001 mag precision at V<13.5 mag, ie. for 10 million stars. This may possibly be achieved due to the very good calibration possibilities with GAIA. That would be very valuable, but is still very far from 500 million stars. 4. The above issues have recently been discussed with the Padua group, ie. with P.L. Bernacca and U. Munari. UM writes on 21 Feb, on the accuracy estimation: "This is exactly what we are investigating in a massive paper for Astron.Astrophys. that is due for submission in a month or so. The dispersion has fluctuated a lot in the past. I have shown in UM-PWG-005 and UM-PWG-006 that going coarser than 0.5 Ang/pix will result in a serious loss of information, severely affecting the astrophysics on binary stars and much more. I prefer to work at the optimal compromise between better astrophysics vs. minimal pixel budget (i.e. 0.25 Ang/pixel). If it will be not the dispersion implemented on GAIA for technical reasons to emerge much later during the design finalization, it will be not much of a problem: my results will be easily degraded to a lower resolution. The reverse cannot be done." My answer is: Please, any publication should make a clear distinction between the nominal instrument design parameters and accuracies and the "wish list" of any worker. Otherwise a publication could create confusion about what GAIA is. Furthermore, a journal editor would probably more appreciate a manuscript for publication if he knows that it has a realistic connection with the ESA mission GAIA. ---------------------------------------------------------------- Comments on 17 April 2000 after correspondence ===== 4. Interstellar reddening - how crucial is the DIB ? The determination of reddening with the F-system is briefly described in the Study Report Section 2.3.2 : "The characteristics or early-type stars (OB, up to A2) will be derived using `reddening-free parameters' ... This (together with the DIB and HI maps) will allow local reddening to be determined, building the [three-dimensional] map to be used for the later-type stars." The Study Report Section 2.3.1 describes the use of Q-Q diagrams to determine reddening for OBA stars from F-photometry. This section does not mention the DIB. - It is thus not clear how crucial the DIB measurements are for the determination of reddening with the F-system. The F-system has been discussed by Knude and Hoeg (CUO-64 and -65, May 1999). CUO-64 on reddening says in summary: The proposed scheme for reddening determination from OBA stars and HI maps will not work well because of the size of spatial structure of the ISM and the scarcity of OBA tracers. The proposed application of a 3-D model of the ISM instead of individual observations of the reddening is liable to introduce biases in the metallicity, T_eff etc, and subsequently in the age computation. The 2G-system (CUO-72) will be able to determine reddening for individual stars of types OBAFGKM if the star can be classified and if the reddening law is normal. If the star density is sufficiently high the local reddening laws may be established directly by the photometric data. ===== 5. U Munari UM agrees to explain his assumptions precisely in the introduction to the paper under preparation for A&A in order to avoid confusion in the reader about the design and capabilities of GAIA. UM does not intend to investigate the accuracy of the DIB as proposed in the task description. ===== 6. M Perryman writes on 4 April: Anything which is requested in addition to what is baselined (for the radial velocity measurements) has to be justified in terms of expected accuracy and astrophysical returns based on that accuracy. Thus, if DIB is a technical driver, it's requirements must be substantiated. If it is just an extra, to be acquired "for free" with the radial velocity spectrum, I see no obvious need to do more at present. For the dispersion, the accuracies must be quantified for our baseline, which represents our global balance between what is scientifically desirable, and what is technically feasible. If there is a proven case to go request higher dispersion, the case must be presented, and the technical implications assessed. These comments correspond to our normal working procedures, of course. regards Michael ===== 7. Answer to Michael by EH on 25 April It is not clear whether the DIB is just an extra to be acquired "for free". We need an up-to-date justification from the proposers or supporters (Munari, Grenon, Favata, Gilmore et al.). A reference to the 3 documents by Grenon et al. of April 1999 is not sufficient. Much of these reports is, more or less obviously, out of date. --------------------------------------------========================