Author(s): Lennart Lindegren
The various quality assessment and validation procedures give confidence in the overall correctness of Gaia EDR3 primary and secondary solutions, but also indicate many weaknesses and problems that have to be addressed and removed in subsequent releases. Compared with Gaia DR2 (Gaia Collaboration et al. 2018a) the second release contains about 10% more sources with full astrometric data, but the formal uncertainties have decreased roughly by 20% for the positions and parallaxes and by 50% for the proper motions. Most of this improvement can be ascribed to the longer stretch of data included, but the level of systematic errors appear to have decreased by similar factors.
In spite of these improvements it should be remembered that the astrometric results in Gaia EDR3 are based on less than three years of observations and preliminary calibrations. At the time of the release of Gaia EDR3, Gaia has accumulated more than twice as much data as was used for Gaia EDR3 and the work to improve calibration models and processing methods will continue for a number of years. As a consequence, it is reasonable to expect an even lower level of random and systematic errors in subsequent data releases, which, in addition, will contain many more data products, for example for various kinds of binaries, which did not get any special treatment in Gaia EDR3.