2011A&A...531A..83C

We attempt to derive the true effective temperature of a star from the spectroscopic observation of its Hα Balmer line profile. The method is possible thanks to advances in two respects. First there have been progresses in the theoretical treatment of the broadening mechanisms of Hα. Second, there has been a rapid increase in the number of stars with an apparent diameter measured with an accuracy of the order of 1 percent, enabling us to obtain an accurate effective temperature T_eff for a dozen of stars using the direct method by means of combining the apparent diameter and the bolometric flux. For the eleven stars with an accurate effective temperature derived from their apparent angular diameter we determined the effective temperature of the Kurucz Atlas9 model that provides the best fit of the computed theoretical Hα profile (using the recent theoretical advances) with the corresponding observed profile, extracted from the S4N spectroscopic database. The two sets of effective temperatures have a significant offset, but are tightly correlated, with a correlation coefficient of 0.9976. The regression straight line of T_eff(direct) versus T_eff(Hα) enables us to reach the true effective temperature from the spectroscopic observation of the Hα profile, with an rms error of only 30K. This provides a way of obtaining the true effective temperature of a reddened star. We succeeded in obtaining empirically the true stellar effective temperature from Hα profile using Kurucz's Atlas9 grid of 1D model atmospheres. Full understanding of the difference between T_eff(direct) and T_eff(Hα) would require a 3D approach, with radiative hydrodynamical models, which will be the subject of a future paper.