2021MNRAS.508.5148C

We present a spectroscopic and photometric calibration to derive effective temperatures T_eff and metallicities [Fe/H] for M dwarfs, based on a principal component analysis (PCA) of 147 spectral indices measured off moderate resolution (R∼11 000), high S/N (>100) spectra in the λλ 8390-8834 region, plus the J-H colour. Internal uncertainties, estimated by the residuals, are 81 K and 0.12 dex, respectively, for T_eff and [Fe/H], the calibrations being valid for 3050 K < T_eff < 4100 K and -0.45 < [Fe/H] < +0.50 dex. The PCA calibration is a competitive model-independent method to derive T_eff and [Fe/H] for large samples of M dwarfs, well suited to the available database of far-red spectra. The median uncertainties are 105 K and 0.23 dex for T_eff and [Fe/H], respectively, estimated by Monte Carlo simulations. We compare our values to other works based on photometric and spectroscopic techniques and find median differences 75 ± 273 K and 0.02 ± 0.31 dex for T_eff and [Fe/H], respectively, achieving good accuracy but relatively low precision. We find considerable disagreement in the literature between atmospheric parameters for stars in common. We use the new calibration to derive T_eff and [Fe/H] for 178 K7-M5 dwarfs, many previously unstudied. Our metallicity distribution function for nearby M dwarfs peaks at [Fe/H]∼-0.10 dex, in good agreement with the RAVE distribution for GK dwarfs. We present radial velocities (internal precision 1.4 km s^–1) for 99 objects without previous measurements. The kinematics of the sample shows it to be fully dominated by thin/thick disc stars, excepting the well-known high-velocity Kapteyn's star.