Comparison of Stellar Population Model Predictions Using Optical and Infrared Spectroscopy

TL;DR

This study compares stellar population model predictions using optical and near-infrared spectroscopy of early-type galaxies, revealing significant model-dependent variations in the infrared and emphasizing the importance of spectral library quality.

Contribution

It demonstrates that stellar population models yield inconsistent star formation histories in the near-infrared, primarily due to differences in spectral libraries, highlighting the need for high-quality spectral data.

Findings

Models agree in optical spectra but diverge in near-infrared.

High-quality spectral libraries improve model fits and consistency.

Age information in the near-infrared is subtle and affected by continuum removal.

Abstract

We present Gemini/GNIRS cross-dispersed near-infrared spectra of 12 nearby early-type galaxies, with the aim of testing commonly used stellar population synthesis models. We select a subset of galaxies from the atlas/ sample which span a wide range of ages (SSP-equivalent ages of 1--15~Gyr) at approximately solar metallicity. We derive star formation histories using four different stellar population synthesis models, namely those of [bruzual_stellar_2003], Conroy, Gunn \& White (2009; 2010), [maraston_stellar_2011] and [vazdekis_uv-extended_2016]. We compare star formation histories derived from near-infrared spectra with those derived from optical spectra using the same models. We find that while all models agree in the optical, the derived star formation histories vary dramatically from model to model in the near-infrared. We find that this variation is largely driven by the choice of stellar spectral library, such that models including high quality spectral libraries provide the best fits to the data, and are the most self-consistent when comparing optically-derived properties with near-infrared ones. We also find the impact of age variation in the near-infrared to be subtle, and largely encoded in the shape of the continuum, meaning that the common approach of removing continuum information with a high-order polynomial greatly reduces our ability to constrain ages in the near-infrared.