Modelling CO emission from Mira's wind

We have modelled the circumstellar envelope of {\it o} Ceti (Mira) using new observational constraints. These are obtained from photospheric light scattered in near-IR vibrational-rotational lines of circumstellar CO molecules at 4.6 micron: absolute fluxes, the radial dependence of the scattered intensity, and two line ratios. Further observational constraints are provided by ISO observations of far-IR emission lines from highly excited rotational states of the ground vibrational state of CO, and radio observations of lines from rotational levels of low excitation of CO. A code based on the Monte-Carlo technique is used to model the circumstellar line emission. We find that it is possible to model the radio and ISO fluxes, as well as the highly asymmetric radio-line profiles, reasonably well with a spherically symmetric and smooth stellar wind model. However, it is not possible to reproduce the observed NIR line fluxes consistently with a `standard model' of the stellar wind. This is probably due to incorrectly specified conditions of the inner regions of the wind model, since the stellar flux needs to be larger than what is obtained from the standard model at the point of scattering, i.e., the intermediate regions at approximately 100-400 stellar radii (2"-7") away from the star. Thus, the optical depth in the vibrational-rotational lines from the star to the point of scattering has to be decreased. This can be accomplished in several ways. For instance, the gas close to the star (within approximately 2") could be in such a form that light is able to pass through, either due to the medium being clumpy or by the matter being in radial structures (which, further out, developes into more smooth or shell-like structures).