From erik Thu Dec 21 10:55:55 2000 Subject: GAIA-LL-33, photometry work in Lund To: gaia-sag@astro.estec.esa.nl (GAIA SAG), straizys@itpa.lt (Vytas Straizys), wladas@itpa.lt (Vladas Vansevicius), taut@itpa.lt (Grazina Tautvaisiene), indus (Jens Knude), cf (Claus Fabricius), makarov@astro.ku.dk (V.V. Makarov), ntk (Nadia Kaltcheva), helge (Helge J. Sorensen), cflynn@astro.utu.fi (C. Flynn), Frederic.Arenou@obspm.fr (F. Arenou), Daniel.Hestroffer@bdl.fr (D. Hestroffer), Carine.Babusiaux@obspm.fr (Carine Babusiaux), catherine.turon@obspm.fr (C. Turon), brown@bufadora.astrosen.unam.mx (A. Brown), janerik@phys.uit.no (Jan-Erik Solheim), michel.grenon@obs.unige.ch (M. Grenon), per.lilje@astro.uio.no (Per Lilje), kaare.aksnes@astro.uio.no (Kaare Aksnes), dominik@stsci.edu (Martin Dominik), egret@simbad.u-strasbg.fr (D. Egret), bienayme@leyla.u-strasbg.fr (Olivier Bienayme), mattila@cc.helsinki.fi, vaccari@pd.astro.it (Mattia Vaccari), munari@astras.pd.astro.it (Ulysses Munari), BERNACCA@ASTRAS.pd.astro.it (P.L. Bernacca), calj@mpia-hd.mpg.de (Coryn Bailer-Jones), csterken@vub.ac.be (Chris Sterken), weiss@astro.univie.ac.at (Werner W. Weiss), kerschbaum@astro.univie.ac.at (Franz Kerschbaum), jvc@bro.astro.ku.dk (J.V. Clausen), ja (Johannes Andersen), ohe (Ole Einicke), uffegj@astro.ku.dk, pen@ifa.au.dk (P.E. Nissen), nl@dsri.dk (N. Lund), JSLARSEN@tac.dk (J. Sommer-Larsen), hans@ifa.au.dk (Hans Kjeldsen), ddj@gfy.ku.dk (Dorthe Dahl-Jensen), sma (Soeren Madsen), fgj@ifa.au.dk (Frank Grundahl), srf@ifa.au.dk (S. Frandsen), birgitta (Birgitta Nordstrom), bodil (Bodil E. Helt), jcd@ifa.au.dk (J. Christensen-Dalsgaard), rosenkil (Bjarne R. Joergensen), jens (Jens Hjorth) Date: Thu, 21 Dec 2000 10:55:55 +0100 (MET) X-Mailer: ELM [version 2.5 PL2] Content-Length: 5637 Status: OR Report from GAIA photometry work in Lund, 4-8 Dec. 2000 ======================================================= GAIA-LL-33 (20 December 2000) Audrius Bridzius Institute of Physics, Vilnius, Lithuania Claus Fabricius Copenhagen University Observatory, Copenhagen, Denmark Lennart Lindegren Lund Observatory, Lund, Sweden During the week 4-8 December 2000, AB and CF visited LL at Lund Observatory. Two topics related to the GAIA photometry were discussed and investigated: A. Automatic classification of stars with BBP and MBP filters (AB) B. The Spectro Point Spread Function (CF) Summary reports of these investigations are included below. More extensive technical reports (including diagrams etc) are expected in due time. A. Automatic classification of stars with BBP and MBP filters ------------------------------------------------------------- by Audrius Bridzius The classification method developed at Vilnius was discussed, including some modifications identified for future testing. The main conclusions are summarised below. 1. Performance of the 4-D classification using all possible (61) colour indices combined from the 3G photometric bands was estimated. We found no improvement with respect to usage of only adjacent color indices; moreover the hottest and coolest stars are classified with lower precision. 2. A possible 4-D classification scheme based on the comparison of magnitudes rather than colour indices was discussed. In comparing the different magnitudes for a given star with the standard data base, a constant magnitude offset is determined based on a weighted least square fitting. The residuals of the fit then provide the classification criterion in the same way as for the colour indices. For practical reasons this method could however not be tested during the visit to Lund. 3. The 4-D classification capabilities for two of the photometric systems proposed for GAIA - 1F and 3G, using medium bands (MB) and MB plus broad bands (BB), were investigated. We found rather small improvement of the classification precision at V=17 mag if BB measurements are taken into account, compared with using only MB. The effect at V=19 mag of using only BB compared with MB+BB will be studied later, according to action #18 from the workshop at CUO. 4. The capability of the photometric systems 1F and 3G to determine the main stellar parameters - Teff, log g, [Fe/H], and E(B-V) - were estimated at V=17 mag, using models of solar metallicity without reddening. It was concluded that 3G is significantly better in all the parameters. Numerically, this is illustrated in Table 1, which gives the limiting deviations of the parameters (defined as max(|p05|,|p95|), where p05 and p95 are the 5th and 95th percentiles of the error distribution). At V=19 the 3G system is still better than the 1F, but only by 10-20%, as mentioned in the CUO-081 report. Table 1. Classification performance of 1F and 3G systems at V=17 mag. The table gives the maximum error for 90% of the cases. The corresponding robust standard error is about 0.6 times these values. --------------------------------- system 1F 3G --------------------------------- log Teff 0.032 0.024 log g 0.64 0.30 [Fe/H] 0.65 0.46 E(B-V) 0.07 0.04 --------------------------------- B. The Spectro Point Spread Function ------------------------------------ by Claus Fabricius A point spread function (PSF) for the Astro instruments was discussed by LL in the report SAG-LL-025 from 1998, based on an example of the wave front errors (WFE) computed by MMS. No similar estimate of typical WFE for Spectro exists, and instead the values for Astro were used, but only for the smaller aperture corresponding to Spectro. It is not clear if this approach is representative for as wide a field as in Spectro, but it gives a starting point for discussing the plausible effects of aberrations in that instrument. The program developed by LL in 1998 was adapted to the Spectro aperture and the pixel size of that detector. The transverse motion of the star was assumed to be one pixel also in Spectro because both the pixel size and integration time is about 3 times larger than in Astro. Instead of the triangular splines used in the original program to represent the spectral response (QE times filter curve), a rectangular response was assumed for the medium bands. Any central wavelength and width may be chosen. The first trials resulted in rather unexpected results. This was eventually traced mainly to the step size and partly to the number of grid points used in the FFT computations. Comparisons were made between the PSFs free of WFE and including the WFE at "point 15" (see SAG-LL-025). The rms WFE in the Spectro aperture at "point 15" is about 33 nm. At first glance the PSFs look very similar and no severe problems are expected for normal stars (fainter than 10 mag). For the brighter stars the image wings have been mentioned as a possible solution to work around the saturation of the CCD. At a distance of 10 pixels from the peak, where the intensity per pixel has dropped by 9 to 10 mag (depending on wavelength), the typical PSF difference depending on the WFE is of the order 0.02 mag, and the intensity gradient is 0.2 mag/pixel. This would make accurate photometric calibration difficult, but perhaps not impossible. The main result is perhaps that a Fortran program to calculate the Spectro PSF for abritrary spectral bands and WFE is now available for future simulations. (End of report)