The Andromeda Optical and Infrared Disk Survey
Astronomy , Stellar Populations , Extragalactic Astronomy
The spectral energy distributions of galaxies inform us about a galaxy’s stellar populations and interstellar medium, revealing stories of galaxy formation and evolution. How we interpret this light depends in part on our proximity to the galaxy. For nearby galaxies, detailed star formation histories can be extracted from the resolved stellar populations, while more distant galaxies feature the contributions of entire stellar populations within their integrated spectral energy distribution (SED). This thesis aims to resolve whether the techniques used to investigate stellar populations in distant galaxies are consistent with those available for nearby galaxies. As the nearest spiral galaxy, the Andromeda Galaxy (M31) is the ideal testbed for the joint study of resolved stellar populations and panchromatic SEDs. We present the Andromeda Optical and Infrared Disk Survey (ANDROIDS), which adds new near-UV to near-IR (ugriJKs) imaging using the MegaCam and WIRCam cameras at the Canada-France-Hawaii telescope to the available M31 panchromatic dataset. To accurately subtract photometric background from our extremely wide-field (14 square degree) mosaics, we present observing and data reduction techniques with sky-target nodding, optimization of image-to-image surface brightness, and a novel hierarchical Bayesian model to trace the background signal while modelling the astrophysical SED. We model the spectral energy distributions of M31 pixels with MAGPHYS (da Cunha et al. 2008) and compare those results to resolved stellar population models of the same pixels from the Panchromatic Hubble Andromeda Treasury (PHAT) survey (Williams et al. 2017). We find substantial (0.3 dex) differences in stellar mass estimates despite a common use of the Chabrier (2003) initial mass function. Stellar mass estimated from the resolved stellar population is larger than any mass estimate from SED models or colour-M/L relations (CMLRs). There is also considerable diversity among CMLR estimators, largely driven by differences in the star formation history prior distribution. We find broad consistency between the star formation history estimated by integrated spectral energy distributions and resolved stars. Generally, spectral energy distribution models yield a stronger inside-out radial metallicity gradient and bias towards younger mean ages than resolved stellar population models.