Assessment of [Fe/H] determinations for FGK stars in spectroscopic surveys

Abstract

Context. The iron abundance [Fe/H] in the atmosphere of FGK-type stars is crucial in stellar and galactic physics. The number of stars with a measured value of [Fe/H] is increasing considerably thanks to spectroscopic surveys. However, different methodologies, inputs, and assumptions used in spectral analyses lead to different precisions in [Fe/H] and possibly to systematic differences, which need to be evaluated. It is essential to understand the characteristics of each survey to fully exploit their potential, in particular if the surveys are combined to probe a larger galactic volume and to improve statistics. Aims. The purpose of this study is to compare [Fe/H] determinations from the largest spectroscopic surveys to other catalogues taken as reference. Offsets and dispersions of the residuals are examined, as are their trends with other parameters. The investigated surveys are the latest public releases of APOGEE, GALAH, RAVE, LAMOST, SEGUE, and the Gaia-ESO Survey. Methods. We use reference samples that provide independent determinations of [Fe/H], which are compared to those from the surveys for common stars. The distribution of the residuals is assessed through simple statistics that measure the offset between two catalogues and the dispersion representative of the precision of both catalogues. When relevant, linear fits are performed. A large sample of FGK-type stars with [Fe/H] based on high-resolution, high-signal-to-noise spectroscopy was built from the PASTEL catalogue to provide a reference sample. We also use FGK members in open and globular clusters to assess the internal consistency of [Fe/H] of each survey. The agreement of median [Fe/H] values for clusters observed by different surveys is discussed. Results. All the surveys overestimate the low metallicities, and some of them also underestimate the high metallicities. They perform well in the most populated intermediate metallicity range whatever the resolution. In most cases, the typical precision that we deduce from the comparisons is in good agreement with the uncertainties quoted in the catalogues. Some exceptions to this general behaviour are discussed.