The impact of systematic uncertainties in stellar parameters on integrated spectra of stellar populations

Abstract

In this paper we investigate a hitherto unexplored source of potentially significant error in stellar population synthesis (SPS) models, caused by systematic uncertainties associated with the three fundamental stellar atmospheric parameters; effective temperature T_eff, surface gravity g, and iron abundance [Fe/H]. All SPS models rely on calibrations of T_eff, logg and [Fe/H] scales, which are implicit in stellar models, isochrones and synthetic spectra, and are explicitly adopted for empirical spectral libraries. We assess the effect of a mismatch in scales between isochrones and spectral libraries (the two key components of SPS models) and quantify the effects on 23 commonly used diagnostic line indices. We find that typical systematic offsets of 100K in T_eff, 0.15 dex in [Fe/H] and/or 0.25 dex in logg significantly alter inferred absolute ages of simple stellar populations (SSPs) and that in some circumstances, relative ages also change. Offsets in T_eff, logg and [Fe/H] scales for a scaled-solar SSP produce deviations from the model which can mimic the effects of altering abundance ratios to non-scaled-solar chemical compositions, and could also be spuriously interpreted as evidence for a more complex population, especially when multiple-index or full-SED fitting methods are used. We stress that the behavior we find can potentially affect any SPS models, whether using full integrated spectra or fitting functions to determine line strengths. We present measured offsets in 23 diagnostic line indices and urge caution in the over-interpretation of line-index data for stellar populations.