Star/Galaxy Separation in Hyper Suprime-Cam and Mapping the Milky Way with Star Counts

Abstract

We study the problem of separating stars and galaxies in the Hyper Suprime-Cam (HSC) multi-band imaging data at high galactic latitudes. We show that the current separation technique implemented in the HSC pipeline is unable to produce samples of stars with $i \gtrsim 24$ without a significant contamination from galaxies ($\gtrsim 50\%$). We study various methods for measuring extendedness in HSC with simulated and real data and find that there are a number of available techniques that give nearly optimal results; the extendedness measure HSC is currently using is among these. We develop a star/galaxy separation method for HSC based on the Extreme Deconvolution (XD) algorithm that uses colors and extendedness simultaneously, and show that with it we can generate samples of faint stars keeping contamination from galaxies under control to $i \leq 25$. We apply our star/galaxy separation method to carry out a preliminary study of the structure of the Milky Way (MW) with main sequence (MS) stars using photometric parallax relations derived for the HSC photometric system. We show that

it will be possible to generate a tomography of the MW stellar halo to galactocentric radii $\sim 100 \textrm{ kpc}$ with $\sim 10^6$ MS stars in the HSC Wide layer once the survey has been completed. We report two potential detections of the Sagittarius tidal stream with MS stars in the XMM and GAMA15 fields at $\approx 20 \textrm{ kpc}$ and $\approx 40 \textrm{ kpc}$ respectively.