2007ApJ...655..179W

Spectra of galaxies contain an enormous amount of information about the relative mixture of ages and metallicities of constituent stars. We present a comprehensive study designed to extract the maximum information from spectra of data quality typical in large galaxy surveys. These techniques are not intended for detailed stellar population studies that use high-quality spectra. We test techniques on a sample of globular clusters, which should consist of single stellar populations and provide good test cases, using the Bruzual-Charlot high-resolution stellar population synthesis models to simultaneously estimate the ages and metallicities of 101 globular clusters in M31 and the Magellanic Clouds. The clusters cover a wide range of ages and metallicities, 4 Myr<t_age<20 Gyr and -1.6<[Fe/H]<+0.3, estimated by other methods in the literature. We compare results from model fits to both the spectra and photometry and find that fits to continuum-normalized spectra over the entire range available, typically 3500 Å to 1 µm for this sample, provide the best results. For clusters older than 1 Gyr we agree with literature ages to 0.16 dex (35%) and [Fe/H] to 0.12 dex. For younger clusters we agree with literature ages to 0.3 dex (63%) but cannot constrain the metallicity. It is particularly important to use the entire continuum-normalized spectrum to avoid problems with model continua for young objects and to break age-metallicity degeneracies of broadband photometry. Our required S/N is 15-30 Å^–1 for 20% age uncertainties and 30-55 Å^–1 for 10% uncertainties over spectral resolutions of Δλ=5-25 Å. This technique should work well for the age-metallicity parameter space expected for early-type galaxies at z∼1, although individual galaxy spectral S/N may require the co-addition of a few like objects.