1 / 22

ASPCAP Review Progress report and schedule now and ahead

October 4-5, 2012. ASPCAP Review Progress report and schedule now and ahead. Performance.

jubal
Télécharger la présentation

ASPCAP Review Progress report and schedule now and ahead

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. October 4-5, 2012 ASPCAP ReviewProgress report and schedulenow and ahead

  2. Performance Overall, good performance for high metallicity ([Fe/H]>-1) stars with good S/N (>70) spectra: Teff to about 3%, logg to about 0.2 dex, and [Fe/H] to better than 0.1 dex. Small systematic offsets (0.1-0.3 dex) in surface gravity. Significantly poorer performance at low metallicity, with a systematic offset in gravity of up to ~ 1 dex.

  3. Issues Systematic offsets in gravity Degeneracies for metal-poor stars Noding Speed

  4. Uncertainties • Raw mean (and std. deviation) for abundances in M13, M5 and M67 members (apstar results) • M67 M5 M13 • ________________________________ • <[Fe/H]> +0.01 (0.04) -1.26 (0.11) -1.51 (0.13) • <[C/Fe]> -0.06 (0.08) -0.23 (0.19) -0.02 (0.36) • <[N/Fe]> +0.20 (0.11) +0.50 (0.32) +0.58 (0.29) • <[α/Fe]> +0.02 (0.03) +0.11 (0.10) +0.20 (0.08)

  5. Uncertainties • Three types of sources • Random errors • Atmospheric/Instrumental distortions • Systematic errors • Model approximations

  6. Random

  7. random Random and systematic due to instrumental distortions may look random Random can be studied by adding purely random noise Instrumental distortions can be studied by looking at plate repeatability

  8. systematic Systematic errors in models identified looking at accurate reference values and trends

  9. [Fe/H] dispersion Includes three discussed sources of error σ ~ 0.04 dex for M67 σ ~ 0.12 dex for M5 and M13 Random noise should not be dominant source except for M13 at S/N<100

  10. Purely random noise Random errors are VERY small σ ~ 0.03 for S/N≥ 20 at [Fe/H]=0 σ ~ 0.1 at S/N=40 for M13 Need S/N≥80 for σ~0.04 in this cluster

  11. Purely random noise Random errors important for C, and especially N in metal-poor stars Small for α/Fe

  12. Purely random noise Random errors important for Teff at all [Fe/H] M13: 300, 150, 100 and 50K at S/N=10, 20, 40 and 80 M67: 300, 50, 35 and 30 K at S/N=10, 20, 40 and 80 for logg at low [Fe/H] M13: 1.3,0.5,0.2,0.1 dex at S/N=10, 20, 40 and 80 Teff logg

  13. Ferre random errors Random errors from inverting curvature matrix within a factor ~ 2 [Fe/H] logg Teff

  14. Model systematic errors 0.1 dex /1000 K for M67 <0.01 dex/1000 K for M5/M13 0.04 dex errors at [Fe/H]~0 but negligible in globulars (where random dominate) Note better results than those Shown by Szabolcs (apstar vs. <apvisit>)

  15. Repeatability

  16. Repeatability

  17. Repeatability Scatter, noding, and systematics get worse at low S/N At high S/N, repeatability to better than 0.1 dex in [Fe/H], but still seems to be dominant source of Error at S/N>100

  18. Error summary • We effectively have a S/N limited by systematics in the observations, probably around S/N~100. • Performance for metal-rich stars is currently limited by systematics in the models at about 0.04 dex, since required S/N is not as high for main parameters. • Performance at the metal-poor end is limited by random noise (dominated by systematics due to instrument/atmospheric distortion), since high S/N is required to reduce errors. Large model systematics for logg.

  19. Error summary • Currently random errors of about 100 K (100 K systematic) in Teff, 0.1 dex (0.2-1.0 dex systematic, depending on metallicity) in logg, and 0.1 dex (0.1 dex systematic) in [Fe/H]. • C and N require in general higher S/N, especially N and especially in metal-poor stars. [α/Fe] easier in terms of S/N. • ASPCAP does well within x2 estimated purely random errors, but need to consider two more contributions: systematics in the models for metal-rich stars (which limits, e.g. σ[Fe/H] to ~ 0.05 dex), and repeatability when photon noise is < 1% (which limits actual S/N<100-150).

  20. Models Currently using the original Kurucz & Castelli (2004) models. Newer, more consistent, better MARCS and Kurucz models in hand (Meszaros+ 2012). www.iac.es/proyecto/ATLAS-APOGEE Hotter Tlusty models in hand. Plans to provide cooler models by Garcia Hernandez & Zamora. Main issue now to figure out how to fill the holes for synthesis.

  21. Linelists Long evolution that culminated early this year. A paper is in preparation (Shetrone+ 2012). Only limited evolution expected in the future (somenonetheless): solar f-values consistent with models, Arcturus/solar f-values.

  22. DR10 • Teff,logg,[Fe/H] solid for start doing science now, with the caveat that systematic errors in logg need to be corrected. • C and N require higher S/N than Teff,logg,[Fe/H] and have therefore larger error bars, especially for metal-poor stars, and exhibit (unless S/N is fairly high) noding issues. [α/Fe] easier. • Our proposal for DR10 is to release Teff,logg,[Fe/H], and possibly [α/Fe], with error bars from internal FERRE calculated random errors (capping S/N~100 in the input error bars), inflated to account for known systematics as a function of parameters.

More Related