2005A&A...440.1133M

We investigate the line-profile variability (lpv) of Hα for a large sample of O-type supergiants (15 objects between O4 and O9.7), in an objective, statistically rigorous manner. We employed the Temporal Variance Spectrum (TVS) analysis, developed for the case of photospheric absorption lines and modified by us to take into account the effects of wind emission. By means of a comparative analysis we place constraints on the properties of this variability - quantified in terms of a mean and a newly defined fractional amplitude of deviations - as a function of stellar and wind parameters. The results of our analysis show that all the stars in the sample show evidence of significant lpv in Hα, mostly dominated by processes in the wind. The variations occur between zero and 0.3v_∞ (i.e., below ∼1.5R_*), in good agreement with results from similar studies. A comparison between the observations and corresponding line-profile simulations indicates that for stars with intermediate wind densities the properties of the Hα variability can be explained by simple models consisting of coherent or broken shells (blobs) uniformly distributed over the wind volume, with an intrinsic scatter in the maximum density contrast of about a factor of two. For stars at lower and higher wind densities, on the other hand, we found certain inconsistencies between the observations and our predictions, most importantly concerning the mean amplitude and the symmetry properties of the TVS. This disagreement might be explained by the presence of coherent large-scale structures, partly confined in a volume close to the star. Interpreted in terms of a variable mass-loss rate, the observed variations of Hα indicate changes of ±4% with respect to the mean value of {dot}(M) for stars with stronger winds and of±16% for stars with weaker winds. The effect of these variations on the corresponding wind momenta is rather insignificant (less than 0.16dex), increasing only the local scatter without affecting the Wind Momentum Luminosity Relationship.