Good Abundances from Bad Spectra :

Stellar spectra derived from multiple-object bre-fed spectroscopic radial-velocity surveys, of the type feasible with, among other examples, AUTOFIB, 2dF, HY-DRA, NESSIE, and the Sloan survey, diier signiicantly from those traditionally used for determination of stellar abundances. The spectra tend to be of moderate resolution (around 1 A) and signal-to-noise ratio (around 10-20 per resolution element), and cannot usually have reliable continuum shapes determined over wavelength ranges in excess of a few tens of Angstrr oms. Nonetheless, with care and a calibration of stellar eeective temperature from photometry, independent of the spectroscopy, reliable iron abundances can be derived. We have developed techniques to extract true iron abundances and surface gravities from low signal-to-noise ratio, intermediate resolution spectra of G-type stars in the 4000{5000 A wavelength region. Spectroscopic indices sensitive to iron abundance and gravity are deened from a set of narrow (few { several A wide) wavelength intervals. The indices are calibrated theoretically using synthetic spectra. Given adequate data and a photometrically determined eeective temperature, one can derive estimates of the stellar iron abundance and surface gravity. We have also deened a single abundance indicator for the analysis of very low-signal spectra; with the further assumption of a value for the stellar surface gravity, this is able to provide useful iron abundance information from spectra having signal-to-noise ratios as low as 10 (1 A elements). The theoretical basis and calibration using synthetic spectra are described in this paper. The empirical calibration of these techniques by application to observational data is described in a separate paper 1995). The technique provides precise iron abundances, with zero-point correct to 0:1 dex, and is reliable, with typical uncertainties being < 0:2 dex. A derivation of the in situ thick disk metallicity distribution using these techniques is presented by Gilmore, Wyse and Jones (1995).