Bayesian Investigation of Isochrone Consistency Using the Old Open Cluster NGC 188

Abstract

This paper provides a detailed comparison of the differences in parameters derived for a star cluster from its color-magnitude diagrams depending on the filters and models used. We examine the consistency and reliability of fitting three widely-used stellar evolution models to fifteen combinations of optical and near-IR photometry for the old open cluster NGC 188. The optical filter response curves match those of the theoretical systems and are thus not the source of fit inconsistencies. NGC 188 is ideally suited to the present study thanks to a wide variety of high-quality photometry and available proper motions and radial velocities which enable us to remove non-cluster members and many binaries. Our Bayesian fitting technique yields inferred values of age, metallicity, distance modulus, and absorption as a function of the photometric band combinations and stellar models. We show that the historically-favored three band combinations of UBV and VRI can be meaningfully inconsistent with each other and with longer baseline datasets such as UBVRIJHKs. Differences among model sets can also be substantial. For instance, fitting Yi et al. (2001) and Dotter et al. (2008) models to UBVRIJHKs photometry for NGC 188 yields the following cluster parameters: age={5.78+ 0.03, 6.45+-0.04} Gyr, [Fe/H]={+0.125+-0.003, -0.077+-0.003} dex, m-M={11.441+-0.007, 11.525+-0.005} mag, and Av={0.162+-0.003, 0.236+-0.003} mag, respectively. Within the formal fitting errors, these two fits are substantially and statistically different. Such differences amongst fits using different filters and models are a cautionary tale regarding our current ability to fit star cluster color-magnitude diagrams. Additional modeling of this kind, with more models and star clusters, and future GAIA parallaxes are critical for isolating and quantifying the most relevant uncertainties in stellar evolutionary models.