Gaia DR2: orbit improvement based on new debiasing of asteroid astrometry

Gaia DR2, validating the debiasing of asteroid astrometry by orbit improvement

The optimal exploitation of asteroid astrometry is seminal at many tasks such as the monitoring of impact risks by potentially hazardous asteroids, and the measurement of subtle dynamical effects. These can include, most notably, the Yarkovsky thermal recoil force or perturbations due to other asteroids.

The Gaia mission has published astrometry with very high accuracy for 14.099 asteroids in the Data Release 2 (DR2), and about 10 times more are coming in DR3 (end 2021). The level of accuracy of Gaia is unprecedented, reaching 1 mas or better for each epoch, but it deserves unprecedented care to be exploited.
 
In particular, most archival data (astrometry available at the Minor Planet Center) are the result of a calibration with respect to pre-Gaia catalogues, that are often affected by local systematic errors. Such errors have different possible sources. They can be the result of the tiling of the celestial sphere by a imaging device, whose field of view presents some residual distortion in its astrometric reduction. There can also be effects related to the coupling of two different catalogs, distant in time, used to derive proper motions. Eventually, the adopted reference frame can also introduce other effects.

As it has been documented several times in literature such systematic bias, that can vary on spatial scales of a few degrees or less, can also be function of other parameters, such as the magnitude range considered (different bias affect stars of different brightness). 

To take into account these effects and apply the required corrections, we developed a completely new bias correction computation around on the position of single asteroid observations, instead of the classical approach of computing corrections on fixed grid for each catalogue. Despite being much more time-consuming, our approach allows us to reach a full flexibility on effects related to the field of view size of single surveys, magnitude limit and also epoch-dependent variations. We also implement corrections to the reference frame rotation detected for bright stars (V<12) in Gaia DR2 (Lindegren 2020) necessary to obtain a full consistency.

After having completed the debiasing of astrometry archived at MPC for all asteroids in Gaia DR2, we have run an orbit improvement procedure for all of them, that also exploits a refined error model. We illustrate here the results of our processing, in particular investigating the improvement in the ephemeris uncertainty, and the perfomance of the debiasing.