10.1 Introduction

Author(s): Frédéric Arenou

Before each data release, dedicated validation processes are being undertaken to check, and possibly improve the quality of the Gaia Catalogue. In a first step, every Gaia DPAC Coordination Unit (CU) implement verification and validation tests before producing their data, and these tests are described in the previous chapters, Section 3.5 for the astrometry, Section 5.5 for the photometry, Section 6.5 for the spectroscopy, Section 8.4 for the astrophysical parameters, Section 7.3.4 for the variability, and Section 4.5 for the solar system objects.

A second and last step, which is developed below, is the CU9 Catalogue validation, being done after collecting the various CU data and building the Catalogue before publication. A large effort has been dedicated to this Catalogue validation, with the following main purposes for the validation tests:

  • implement general sanity checks on the fields of the Catalogue

  • check the accuracy and precision of the Catalogue parameters

  • verify the correct distribution of these parameters, in particular the absence of large numbers of outliers

  • study the completeness of the Catalogue

  • detect as much as possible instrumental or data processing problems

  • and more generally check what would not be covered by the internal CU verifications, i.e., in particular, cross-CU checks.

While the DPAC is mostly organised in terms of astronomical fields (astrometry, photometry, spectroscopy, etc.), the validation areas have been split in terms of various methods which can be applied, with the purpose of being able to make a cross-check of the different results obtained with the different methods.

These methods are mentioned in turn in the sections below and each have their pros and cons. Internal methods, using only the Gaia data, can in general be applied to any source without any cross-match ambiguity. On the other hand, external catalogues provide independent results which can be compared to the Gaia data, but they have their own problems. And, where external data are not available, galaxy models may sometimes help to explain whether observed features are, or not, artefacts. Also, statistical methods for multidimensional analysis can find data properties inside the observational noise. Finally, while this validation development model is transversal, some special objects require a special treatment, such as clusters, and those having an important time dependence such as variability analysis, and also solar system objects.

In the following sections, we describe the tests which have been applied, together with the results mostly obtained on preliminary data (named ‘pre-DR2’ in what follows), and which contributed to improve the Gaia DR2 Catalogue. Main validation results are being published as part of the A&A Gaia Special Issue (Arenou et al. 2018) and this chapter refers to this publication and gives complementary details. The user may also refer to the Gaia DR1 documentation for further details.