Extensive modifications to the non-LTE radiative transfer code of Hillier have been made in order to improve the spectroscopic analysis of stars with stellar winds. The main improvement to the code is the inclusion of blanketing due to thousands of overlapping lines. To implement this effect, we have used the idea of super levels first pioneered by Anderson. In our approach, levels with similar excitation energies and levels are grouped together. Within this group, we assume that the departure coefficients are identical. Only the population (or equivalently, the departure coefficient) of the super level need be solved in order to fully specify the populations of the levels within a super level. Our approach is a natural extension of the single-level LTE assumption, and thus LTE is recovered exactly at depth.

In addition to the line blanketing modifications, the code has been improved significantly in other regards. In particular, the new code incorporates the effect of level dissolution, the influence of resonances in the photoionization cross sections, and the effect of Auger ionization. Electron scattering with a thermal redistribution can be considered, although it is normally treated coherently in the comoving frame (which still leads to redistribution in the observer's frame).

Several example calculations are described to demonstrate the importance of line blanketing on spectroscopic analysis. We find that the inclusion of blanketing modifies the strengths of some optical CNO lines in Wolf-Rayet (W-R) stars by factors of 2-5. In particular, the strengths of the WC classification lines C III λ5696 and C IV λ5805 are both increased because of iron blanketing. This should help alleviate problems found with nonblanketed models, which were incapable of matching the strengths of these lines. We also find that, in the UV (1100-1800 Å), the influence of Fe is readily seen in both emission and absorption. The emission is sensitive to the iron abundance and should allow, for the first time, Fe abundances to be deduced in W-R stars.

The improvements made to our code should greatly facilitate the spectroscopic analysis of stars with stellar winds. We will be able to determine the importance and influence of line blanketing, as well as of several other effects that have been included in the new code. It will also allow us to better determine W-R star parameters, such as luminosity, elemental abundances, wind velocity, and mass-loss rate. With future application to related objects, such as novae and supernovae, our new code should also improve our understanding of these objects with extended outflowing atmospheres.