DGp.1.04 — Accuracy of the atmospheric parameters determination in FGK stars based on spectral fitting

Date & Time

Aug 4th at 6:00 PM until 6:00 PM

Track

Presentations 

Location

Rating ( votes)

Author(s): Lyudmila Mashonkina1

Institution(s): 1. Institute of Astronomy, RAS

We perform an extensive testing of the accuracy of atmospheric parameters determination in FGK stars based on the spectral fitting procedure SME (Spectroscopy Made Easy). Our set of stars consists of 13 objects, including the Sun, in temperature range 5000-6600 K and metallicity range -1.4 to +0.4. For these stars the parameters derived by means of interferometry are known. For each star we use spectra obtained with different echelle spectrographs (42000 <= R <= 110000) and different signal-to-noise ratios. We also test how the values of the derived parameters depend on the spectral regions used in the fitting procedure. We tested three different constraints on the spectral masks; four regions of 100 A, each, 4485-4590, 5100-5200, 5600-5700, and 6100-6200 A, the same regions plus Halpha and Hbeta, and the mask employed by Valenti & Fischer (2005). We propose a new method for estimating the uncertainties of the free parameters in SME, based on the fit residuals, partial derivatives and data uncertainties. To construct the distribution for a given free parameter (Teff, log g, etc.) we estimate the change required to match the observations in every pixel involved in the fit. The main difficulty in estimating uncertainties is a non-Gaussian shape of the resulting distribution. This difficulty is alleviated by constructing a cumulative distribution. For stars in the 5700-6600 K temperature range the closest agreement with the effective temperatures derived by interferometry is achieved when the spectral fitting includes the Halpha and Hbeta lines, while for cooler stars the choice of the mask does not effect the results. The derived values of atmospheric parameters do not strongly depend on the spectral resolution and S/N ratio, while the uncertainty both in temperature and surface gravity grows with the effective temperature never being smaller than 50 K in Teff and 0.1 in log g, for typical S/N = 150-200. Better accuracy Delta Teff = 20-30 K and Delta log g = 0.06-0.07 may be achieved only for S/N > 500.