Stellar population models in the UV: I. Characterisation of the New Generation Stellar Library

TL;DR

This paper characterizes the New Generation Stellar Library (NGSL), improving UV stellar population models by assessing its spectral resolution, calibration, and stellar parameters, thus enhancing the accuracy of UV spectral predictions.

Contribution

It provides a detailed characterization and calibration of the NGSL, including atmospheric parameters and resolution, to improve UV stellar population modeling.

Findings

Wavelength calibration precise up to 0.1 px

Spectral resolution varies from 3Å to 10Å

Atmospheric parameters derived with high precision

Abstract

The spectral predictions of stellar population models are not as accurate in the ultra-violet (UV) as in the optical wavelength domain. One of the reasons is the lack of high-quality stellar libraries. The New Generation Stellar Library (NGSL), recently released,represents a significant step towards the improvement of this situation. To prepare NGSL for population synthesis, we determined the atmospheric parameters of its stars, we assessed the precision of the wavelength calibration and characterised its intrinsic resolution. We also measured the Galactic extinction for each of the NGSL stars. For our analyses we used ULySS, a full spectrum fitting package, fitting the NGSL spectra against the MILES interpolator. We find that the wavelength calibration is precise up to 0.1 px, after correcting a systematic effect in the optical range. The spectral resolution varies from 3{\AA} in the UV to 10{\AA} in the near-infrared (NIR), corresponding to a roughly constant reciprocal resolution R ~ 1000 and an instrumental velocity dispersion $\sigma_{ins}$ ~ 130 km/s. We derived the atmospheric parameters homogeneously. The precision for the FGK stars is 42K, 0.24 and 0.09 dex for Teff, logg and [Fe/H], respectively. The corresponding mean errors are 29K, 0.50 and 0.48 dex for theMstars, and for the OBA stars they are 4.5 percent, 0.44 and 0.18 dex. The comparison with the literature shows that our results are not biased.