2017A&A...606A...6L

Context. Wind-driving in asymptotic giant branch (AGB) stars is commonly attributed to a two-step process. First, matter in the stellar atmosphere is levitated by shock waves, induced by stellar pulsation, and second, this matter is accelerated by radiation pressure on dust, resulting in a wind. In dynamical atmosphere and wind models the effects of the stellar pulsation are often simulated by a simplistic prescription at the inner boundary.
Aims. We test a sample of dynamical models for M-type AGB stars, for which we kept the stellar parameters fixed to values characteristic of a typical Mira variable but varied the inner boundary condition. The aim was to evaluate the effect on the resulting atmosphere structure and wind properties. The results of the models are compared to observed mass-loss rates and wind velocities, photometry, and radial velocity curves, and to results from 1D radial pulsation models. The goal is to find boundary conditions which give realistic atmosphere and wind properties.
Methods. Dynamical atmosphere models are calculated, using the DARWIN code for different combinations of photospheric velocities and luminosity variations. The inner boundary is changed by introducing an offset between maximum expansion of the stellar surface and the luminosity and/or by using an asymmetric shape for the luminosity variation. Ninety-nine different combinations of theses two changes are tested.
Results. The model atmospheres are very sensitive to the inner boundary. Models that resulted in realistic wind velocities and mass-loss rates, when compared to observations, also produced realistic photometric variations. For the models to also reproduce the characteristic radial velocity curve present in Mira stars (derived from CO Δv=3 lines), an overall phase shift of 0.2 between the maxima of the luminosity and radial variation had to be introduced. This is a larger phase shift than is found by 1D radial pulsation models.
Conclusions. We find that a group of models with different boundary conditions (29 models, including the model with standard boundary conditions) results in realistic velocities and mass-loss rates, and in photometric variations. To achieve the correct line splitting time variation a phase shift is needed.