next up previous
Next: Final parameters Up: Atmospheric parameters Previous: Atmospheric parameters

Input parameters

The vast majority of the reference stars which form the library were selected in the catalogue of [Fe/H] determinations (Cayrel de Strobel et al. 1997). A few others are from the list of proper motion stars of Carney et al. (1994). The [Fe/H] catalogue provides near 6000 determinations of ($T_\mathrm{eff}$, log g, [Fe/H]) from detailed analyses of high resolution, high S/N spectra for 3248 stars. For each star, several different values are listed for their atmospheric parameters. The difficulty of finding the "true" parameters for a given star can be illustrated by the case of the well-known deficient sub-giant HD 140283. Its first detailed analysis was performed by Chamberlain and Aller (1951), and since then 28 detailed analyses have been reported, quoting $T_\mathrm{eff}$ from 5362 K to 6300 K, $\log g$ from 3.2 to 4.8 and [Fe/H] from -1.04 to -3.06, with standard deviations of 111 K, 0.39 and 0.23. A simple average is not correct since all the analyses do not have the same weight. It is also worth noting that $T_\mathrm{eff}$ and $\log g$ quoted in detailed analyses are often the values of the model atmosphere chosen to deduce the iron abundance and are not obtained directly from spectroscopy. It is thus quite difficult to know which parameters should be adopted for a given star. The [Fe/H] catalogue is complete up to december 1995, and several recent references including new atmospheric parameters have been added to our list of determinations: Carney et al. (1994), Alonso et al. (1996b), Pilachowski (1996), Gratton et al. (1997), Nissen & Schuster (1997), Nissen et al. (1997), Thévenin (1998). The sample of Carney et al. is not in the [Fe/H] catalogue because the metallicity estimations rely on low S/N spectra. The sample of Nissen et al. includes photometric metallicities and surface gravities derived from Hipparcos parallaxes. The study of Alonso et al. is not based on spectroscopy but concerns only effective temperatures calibrated with the InfraRed Flux Method. As temperature is the parameter which shows the largest scatter between the authors, two independant calibrations of $T_\mathrm{eff}$ were also used : $T_\mathrm{eff}$ versus V-K and [Fe/H] and $T_\mathrm{eff}$ versus b-y, c1, [Fe/H] (Alonso et al. 1996a). The V-K colour indices were found for 70 stars in the catalogue of Morel & Magnenat (1978). Both b-y and c1 were found for 143 stars in the catalogue of Hauck & Mermilliod (1998). The references which appear the most often for this sample are Thévenin (1998) for 128 stars, Gratton et al. (1997) for 70 stars, McWilliam (1990) for 62 stars, Alonso et al. (1996b) for 64 stars, Axer et al. (1994) for 32 stars, Pilachowski (1993) for 27 stars, Edvardsson et al. (1993) for 24 stars, Tomkin et al. (1992) and Luck & Challener (1995) for 19 stars. The determinations of these different authors were compared to uncover any systematic offsets in the various photometric and spectroscopic data sets. There were none: values for individual stars present scatter , but there are no systematic trends between these authors.


  
Figure: Histogram of the rms of the determinations of $T_\mathrm{eff}$ from the literature and photometry for the 211 standards.
\begin{figure}
\resizebox {8cm}{!}{\includegraphics{bib_fig1.ps}}\end{figure}

We first computed a weighted mean of the different determinations with higher weight for recent determinations (after 1990) and lower weight for old ones (before 1980 when solid state detectors were not yet available). The weighted root-mean square (rms) of the different determinations measures their agreement. The histogram of the rms for each parameter is shown respectively on Fig. 1 to 3. Figure 4 shows the histogram of the number of [Fe/H] determinations per star. For the majority of the standards the determinations are in a reasonable agreement, but large discrepancies can also be seen for several stars. The situation is particularly worrisome for temperatures which appear to be that parameter the most delicate to determine. As it is difficult to find an objective criterion to keep some determinations and to eliminate others, we had to think of another way to find the best parameters for the reference stars. The availability of high S/N spectra at the same resolution for all these stars provided a useful support to perform an internal check of the library by comparing standards of the same type, and to adjust their parameters until reaching the self-consistency of the library. The method TGMET developed for the on-line determination of atmospheric parameters of anonymous stars, presented in paper I, offered the opportunity to realise the comparison between the spectra in a quantitative way.


  
Figure: Histogram of the rms of the determinations of $\log g$ from the literature for the 211 standards.
\begin{figure}
\resizebox {8cm}{!}{\includegraphics{bib_fig2.ps}}\end{figure}


  
Figure: Histogram of the rms of the determinations of [Fe/H] from the literature for the 211 standards.
\begin{figure}
\resizebox {8cm}{!}{\includegraphics{bib_fig3.ps}}\end{figure}


  
Figure: Histogram of the number of [Fe/H] determinations per standard.
\begin{figure}
\resizebox {8cm}{!}{\includegraphics{bib_fig4.ps}}\end{figure}


next up previous
Next: Final parameters Up: Atmospheric parameters Previous: Atmospheric parameters

9/11/1998