Forum Springer Astron. Astrophys.
Forum Whats New Search Orders

Astron. Astrophys. 333, 231-250 (1998)

Previous Section Next Section Title Page Table of Contents

2. ATLAS9 and NMARCS model atmospheres

2.1. The ATLAS9 models

Kurucz (1993, 1994) has made available models, fluxes, hydrogen line profiles, and colors for a large grid of temperatures, gravities, and abundances. The passbands and zeropoints used by Kurucz were slightly different to those used here, in particular for the Cousins R and I passbands.

Wood & Bessell (1994) computed colors on the UBVRIJHKL system for the Kurucz (1993) fluxes and these have been available via anonymous ftp as ubvrijhkl.dat.z from mso.anu.edu.au at /pub/bessell/. The isochrones of Bertelli et al. (1994) used these synthetic colors and bolometric corrections. These colors were normalised to the standard UBVRIJHKL system in a slightly different way to that described in Appendix A sect. 2. The colors in the present paper essentially supersede those of Wood & Bessell (1994).

In the 1993 flux data there was evidence of some discontinuities in the computed colors of A-G stars. Castelli (1996) explained how to eliminate these discontinuities which were related to a modification of the mixing-length convection adopted by Kurucz for computing the 1993 and 1994 models and called by him "approximate overshooting". The convective models before 1995 were recomputed by Kurucz by adopting the improvement suggested by Castelli for the "approximate overshooting". In Table 1 are presented the colors from the Kurucz 1995 grid of models computed for solar abundance, a microturbulent velocity of 2 km s-1 and a mixing length to the pressure scale height ratio l/H = 1.25. Because models with [FORMULA]   [FORMULA] 8750K were always computed without any convection, only models with [FORMULA]   [FORMULA] 8750K may be different.

Table 2 lists the colors and bolometric corrections for the same set of models as in Table 1 but computed by Castelli with "no-overshoot". Castelli, Gratton & Kurucz (1997) discussed the differences yielded by the overshoot and no overshoot models on some color indices and on Balmer profiles. They showed that the overshoot solar model fits the solar spectra better than the no-overshoot solar model, but that the no-overshoot models should be preferred mostly for stars hotter than the sun.

For completion, Table 3 lists the colors and bolometric corrections for the Kurucz (1994) solar abundance models between 8750K and 50000K and computed for [FORMULA] km s-1.

Castelli also computed no-overshoot models for lower metal abundances of -0.5, -1.0, -1.5, -2.0, -2.5 dex and work to extend these grids to more metallicities and microturbulent velocities is in progress. Both the revised 1995 Kurucz grids and the Castelli grids will be distributed on a forthcoming CD-ROM (Kurucz & Castelli, in preparation). In the meantime, the last computed ATLAS9 models and fluxes are available either from kurucz@cfaku3.harvard.edu or from castelli@astrts.oat.ts.astro.it.

2.2. The NMARCS models

The new revised MARCS program incorporating much improved line opacities in the opacity sampling scheme, spherical symmetry and a large number of species in the chemical equilibrium, has been used by Plez, Brett & Nordlund (1992) to model M giants and dwarfs. It has been traditionally divided into SOSMARCS - its spherical version and POSMARCS - the plane-parallel counterpart. There is no reason to separate these two cases for the present purpose and we will adopt NMARCS as a generic name. Initially, Plez, Brett & Nordlund (1992) and Plez (1992) presented a grid of solar composition models. Most of these models were computed for three values of atmospheric extension corresponding to 3 different masses: 1, 2 and 5  [FORMULA]. The higher the mass the less extended are the atmospheres. Another grid was calculated by Plez (1995). It contains mainly hotter, plane-parallel models for a range of scaled solar abundances [ [FORMULA] ] = 0.6, 0.3, 0.0, -0.3, -0.6. These models are available from plez@astro.uu.se. Some improvements were brought to the opacities between these 2 sets, mainly new measurements of some TiO band strengths by Doverstål & Weijnitz (1992) that replaced some of the earlier values. The 1995 grid was computed at this stage. Subsequently, Hedgecock et al. (1995) performed higher accuracy lifetime measurements for a number of electronic states. These are significantly different from older ones. Langhoff (1997) performed ab initio calculations in good agreement with the latter measurements, and provides electronic transition momenta for all transitions that were included in the present model spectra. The TiO a-f band, affecting the V magnitude, was also added recently. Karlsson et al. (1997) greatly improved the situation for VO by providing lifetime measurements for all 3 excited states A, B, and C. Finally, Jorgensen (1996) provided a line list for H2 O. Alvarez & Plez (1997) used the same data to compute narrow-band colors for M giants and miras and thoroughly discuss these improvements. We decided to recompute the colors of the models using these improved opacities, without, however, reconverging the models structures. Experiments with a few models show that differences in thermal structures are unimportant for the present work. Moreover further improvements in opacities will probably result in changes in both thermal structure and spectra. The next Plez et al. (1997) grid will provide models and spectra from a consistent data set. We found it therefore unnecessary to recompute the whole set of models at the present stage. An indication of the uncertainties/inconsistencies in the model structures is provided by a comparison of the new colors computed for models with identical parameters in the 1992 and 1995 grids (using the [FORMULA] models in the 1992 grid). At 4000K, log g = 1.5, the difference (1992-1995) in V-K amounts to 0.000, while it is -0.044 at 3800K; at 3600K, log g = 0.0 it is 0.027 and at 4000K, 0.076. In V-I the corresponding differences are 0.000, -0.025, 0.038 and 0.039.

The differences with the older colors (Plez et al. 1992) are small or nonexistent around 4000K and increase with decreasing [FORMULA], esp. for V-I and V-K. We find better agreement with observations using the present improved colors (esp. V-I and [FORMULA]).

Tables 4 and 5 list the colors and bolometric corrections for the Plez et al. (1992) and Plez (1995) models for giants using the newer set of opacities. Brett (1995a, b) computed NMARCS models for M dwarfs which are available from brett@SSMD.MRL.dsto.gov.au. Plez (1997) recomputed a few models with the new NMARCS opacity setup containing the improved opacities. These new models have solar composition, log g = 4.5 (one model for log g=5.5) and temperatures of 3800, 3200, 3000, 2800, 2600, 2400, 2200, and 2000K. Table 6 gives older colors for these Brett M dwarf models together with the improved solar composition dwarf model colors of Plez (1997).

Previous Section Next Section Title Page Table of Contents

© European Southern Observatory (ESO) 1998

Online publication: April 15, 1998