3.2. Stellar atmospheres
The ionization structure of nebulae obviously depends on the spectral distribution of the stellar radiation field. The theory of stellar atmospheres has made enormous progress these last years, due to advanced computing facilities. Several sets of models for massive O stars and for PNe nuclei are now available. The most detailed stellar atmosphere computations now include non-LTE effects and blanketing for numerous elements (e.g. Dreizler & Werner 1993, Hubeny & Lanz 1995, Rauch et al. 2000) and supersede previous works. The effect of winds, which is especially important for evolved stars such as Wolf-Rayet stars, is included in several codes, although with different assumptions (Schaerer & de Koter 1997, Hillier & Miller 1998, Koesterke et al. 2000, Pauldrach et al. 2001).
The resulting model atmospheres differ considerably between each other in the extreme UV. This has a strong impact on the predicted nebular ionization structure (see e.g. Stasinska & Schaerer 1997 for the Ne and the N+ / O+problems). Actually, the confrontation of photoionization models with observations of nebulae is expected to provide tests of the ionizing fluxes from model atmospheres (see Oey et al. 2000, Schaerer 2000, Giveon et al. 2002, Morisset et al. 2002). This is especially rewarding with the ISO data which provide accurate measurements for many fine-structure lines of adjacent ions.
For exploration purposes, it is sometimes sufficient to assume that the ionizing stars radiate as blackbodies, e.g. when interested in a general description of the temporal evolution of PNe spectra as their nuclei evolve from the AGB to the white dwarf stage (e.g. Schmidt Voigt & Köppen 1987a, b, Stasinska et al. 1998). On the other hand, for a detailed model analysis of specific objects, the black body approximation is generally not well suited. For example, the emission of [Ne V] lines in PNe cannot be understood when using blackbodies of reasonable temperatures.