5 Conclusion

We have constructed a comprehensive library of theoretical stellar energy distributions from a combination of different basic grids of blanketed model atmosphere spectra. This new grid complements the preliminary version described in LCB97 by extending it to the M dwarfs models of Allard & Hauschildt (1995). It provides synthetic stellar spectra with useful resolution on a homogeneous wavelength grid, from 9.1 nm to 160 $\mu$m, over large ranges of fundamental parameters. This standard library should therefore be particularly suitable for spectral evolutionary synthesis studies of stellar systems and other synthetic photometry applications.

Comparison of synthetic photometry with empirical $T_{\mathrm{eff}}$-color sequences, established for the first time down to 2000 K, have shown important discrepancies for the coolest dwarf models. The correction procedure designed in the previous paper to provide color-calibrated fluxes has been extended and applied to the original dwarf spectra in the range 4500 K to 2000 K. Although this seems to result in more realistic colors, the method induces significant changes in the original differential color properties. This stems from too strongly variable correction functions at low temperatures, resulting from the fact that the pseudo-continuum cannot be adequately defined for these stars. The empirical colors for dwarfs below 2500 K remain uncertain due to the lack of reliable observations. At the lowest temperatures, the corrected models should therefore be used with caution.

Despite these limitations, we expect that the corrected spectra provide at present a valuable option for deriving realistic stellar colors over extensive ranges of temperatures, luminosities, and metallicities which are required for reliable population synthesis modelling. Grids of model spectra and (UBVRIJHKLM) colors for both the original and the corrected versions of the present stellar library, as well as the semi-empirical calibrations presented in Tables 1 to 10, are fully available by electronic form at the Strasbourg data center (CDS).

Acknowledgements

We wish to thank France Allard and Peter Hauschildt for making their extensive grids of models available on public ftp. The anonymous referee is also acknowledged for his helpful comments. This work was supported by the Swiss National Science Foundation, and this research has made use of the Simbad database, operated at CDS, Strasbourg, France.