2017ApJ...844..166Z

We present evidence for a spatially dependent systematic error in the first data release of Gaia parallaxes based on comparisons to asteroseismic parallaxes in the Kepler field and provide a parameterized model of the angular dependence of these systematics. We report an error of 0.059_–0.004^+0.004 mas on scales of 0.°3, which decreases for larger scales to become 0.011_–0.004^+0.006 mas at 8°. This is consistent with the ∼2% zero-point offset for the whole sample discussed by Huber et al. and is compatible with the effect predicted by the Gaia team. Our results are robust to dust prescriptions and choices in temperature scales used to calculate asteroseismic parallaxes. We also do not find evidence for significant differences in the signal when using red clump versus red giant stars. Our approach allows us to quantify and map the correlations in an astrophysically interesting field, resulting in a parameterized model of the spatial systematics that can be used to construct a covariance matrix for any work that relies on TGAS parallaxes.