I've been investigating the PS1 astrometry and stumbled across some odd features. When comparing PS1 to the SDSS, there was a ~.1 arcsecond systematic offset in the astrometry. I then compared PS1 to 2MASS, and recovered the same ~.1 arcsecond offset. This is particularly troubling as the PS1 astrometry is tied to 2MASS via the synthetic catalog, which is based on 2MASS data combined with USNO data as available.

I then compared 2MASS to the synthetic catalog matched references, which I expected to be themselves 2MASS objects. The result was the following:

PS1 synthetic reference / 2MASS astrometry comparison

More later, but the synthetic catalog is actually a combination of 2MASS positions and USNO positions, which could be the cause.