Re-calibration of SSPP Parameters – SSPP
The parameters derived for the calibration sample of 126 stars with SDSS/SEGUE and high-resolution spectra were used to re-calibrate the SSPP’s estimates of effective temperature, surface gravity and metallicity. Here, the re-calibration of each method in the SSPP is briefly described.
In DR9-DR12 three color-based temperature estimates used previously, TK, TG, and Teff are no longer used and are replaced by the new IRFM estimate of Teff. As a result, the total number of Teff estimators is 9, including the auxiliary estimates from HA24 and HD24. For each of these nine temperature estimates, we compute the difference between the estimates of each method and the IRFM temperature estimates from the SDSS/SEGUE high-resolution sample. We then fit a quadratic function to the offset as a function of Teff and use this relation to apply a correction to all the Teff estimates reported by the SSPP. One thing to note is that there is a smaller number of 107 stars available for the temperature comparison, as only those stars have available J, H, and K band photometry.
The gravity estimates delivered by MgH and CaI2 line index methods have been removed from the SSPP. In addition, due to large deviation from the high-resolution results, the gravity determined by k24 is also not used in any of the final, adopted surface gravity estimates from the SSPP. Therefore, there are seven estimators of log g considered when computing the final adopted log g. For each of the seven methods, we fit a linear relation to the offset between the SSPP estimate and the value of log g determined from the high resolution spectra. This relation is used to correct the log g estimates reported by the SSPP.
First of all, the metallicity estimates from ACF and CaIIT are not reported in this version of the SSPP. The metallicity estimates from the new version of the ANNSR estimator proved to have significant offsets relative to the high-resolution sample. As a result, it is not considered in the determination of the adopted SSPP metallicity. Hence, there are nine metallicity estimators used to compute the final adopted metallicity.
Because the high-resolution sample does not include the stars with [Fe/H]~-1.5 and [Fe/H] > 0.0, we attempt to fill in the hole in the abundance space by making use of members of some open and globular clusters with known metallicity. Therefore, we added some of the cluster members with SEGUE spectra of S/N > 50 to the high resolution sample to calibrate each metallicity estimate in the SSPP. For each SSPP metallicity estimator, we fit a quadratic relation to the difference between the SSPP value and the metallicity determined from the high resolution spectrum or, for cluster stars, the known literature value. We use this quadratic relation to correct the SSPP metallicity estimates.
The introduction of new synthetic spectra with variable micro-turbulence velocities with surface gravities improves the metallicity estimates of NGS1 and CaIIK1 methods in a sense that there is no need to correct the offset for metal-rich stars ([Fe/H] > -0.5). This figure well illustrates the effect of micro-turbulence velocity on the metallicity estimate. The figure shows metallicity distribution as a function of g-r for the members of M67 and NGC 6791 (Top two panels) and difference in metallicity between NGS1 and the high-resolution analysis. The black dots indicate the estimates from the grid of synthetic spectra with micro-turbulence velocity of 2 km/s, while the filled squares are the determinations from the grid with the adopted relation between the gravity and micro-turbulence velocity. It is obvious to notice the big improvement in estimating [Fe/H], especially [Fe/H] > -0.5.