Bayesian analysis of ages, masses and distances to cool stars with non-LTE spectroscopic parameters

For studies of Galactic evolution, the accurate characterization of stars in terms of their evolutionary stage and population membership is of fundamental importance. A standard approach relies on extracting this information from stellar evolution models but requires the effective temperature, surface gravity and metallicity of a star obtained by independent means. In previous work, we determined accurate effective temperatures and non-local thermodynamic equilibrium log g and [Fe/H] (NLTE-Opt) for a large sample of metal-poor stars, -3 < [Fe/H] < -0.5, selected from the Radial Velocity Experiment (RAVE) survey. As a continuation of that work, we derive here their masses, ages and distances using a Bayesian scheme and GARSTEC stellar tracks. For comparison, we also use stellar parameters determined from the widely used 1D LTE excitation-ionization balance of Fe (LTE-Fe). We find that the latter leads to systematically underestimated stellar ages, by 10-30 per cent, but overestimated masses and distances. Metal-poor giants suffer from the largest fractional distance biases of 70 per cent. Furthermore, we compare our results with those released by the RAVE collaboration (DR3) for the stars in common. This reveals -400 to +400K offsets in effective temperature, -0.5 to 1 dex offsets in surface gravity and 10 to 70 per cent in distances. The systematic trends strongly resemble the correlation we find between the NLTE-Opt and LTE-Fe parameters, indicating that the RAVE DR3 data may be affected by the physical limitations of the 1D LTE synthetic spectra. Our results bear on any study, where spectrophotometric distances underlie stellar kinematics. In particular, they shed new light on the debated controversy about the Galactic halo origin raised by the SDSS/SEGUE observations.