Known issues with the Gaia EDR3 data

     

    This page lists any issues found in Gaia Early Data Release 3 which could not be incorporated into the official data release, or which were discovered after the release of the data and publication of the release documentation. The Gaia EDR3 contents page contains a summary of limitations that are known, and documented, at the release. Further information can be found from the data release documentation, as well as from the data release processing papers.

     

    Overview:

     

    Photometry - Sources with missing G Fluxes

     

    Poor input colour recovery

    The sources in this issue were originally found to be much fainter than expected during the validation phase and subsequently their photometry was removed from the Gaia EDR3 as well as from the Gaia DR3 catalogues. The cause of this was found to be poor input Spectral Shape Coefficient, SSC, (colour) values. Following this, the transits for these sources were all processed as Bronze sources i.e. using default SSCs, for all three flux types (G, GBP and GRP) and generating weighted mean flux values (and their errors) for each source. This will make the source consistently Bronze. See Section 8.2 of Riello et al. 2020 for more details.

    In this file, magnitudes have been generated from the fluxes using the formula:

    zp-2.5*log10(flux)

    where the zeropoints are the Gaia EDR3 VEGAMAG ones which can be found in Riello et al. 2021 (Section 7, Table 3).

    This issue affects 5,401,215 sources.

    Note that the GBP magnitudes will still be affected by the bias mentioned in Section 8.1 of Riello et al. 2021. Care should be taken when using GBP magnitudes fainter than 20.3. See also Section 9.2 for some suggestions on possible filtering criteria.

     

    Download the file

    Please connect with an FTP client of your choice or using the terminal to ftp.cosmos.esa.int and navigate to the folder GAIA_PUBLIC_DATA. Here you can find a folder PhotometryEDR3. The file with details on these sources can be found there along with a readme.txt file.

     

    Insufficient valid transits

    During the validation of the Gaia EDR3 catalogue, 54,125 sources were found not to have a mean G flux available in the Gaia EDR3 and Gaia DR3 catalogues even though astrometric information was available for them. Investigation showed that these sources had too few valid transits for photometric processing. The validity criteria are different in the photometric and astrometric processing chains.

    As a convenience to the user, a median Gvpu on-board magnitude (see Gaia EDR3 Documentation Section 1.1.3.5.3) is provided for these sources, but stored with a reduced precision (10%). This is to stop users misguidedly trying to use these values for science. They are purely a general indicator of the brightness of the source. It must be stressed that these values are uncalibrated and on a different photometric system to that of the G values in Gaia EDR3.

    Download the file

    The file with details on these sources can be downloaded through this link.

     

    G-band corrections

     

    Sections 6.3.2, 7.2, and Appendix A in the Gaia EDR3 release paper erroneously state that the correction to the G-band fluxes and magnitudes presented in Riello et al. (2021; their Table 5) should be applied to sources in Gaia EDR3 with six-parameter astrometric solutions. In fact, the corrections should be applied to sources with two-parameter or six-parameter astrometric solutions. The corrected ADQL query and Python source code are presented in a corrigendum to the paper. The Python code has also been corrected on Github.

    Following the discovery of the above error, a more detailed investigation was done for the sources with two-parameter (2-p) astrometric solutions. Out of the 344 million 2-p sources present in Gaia EDR3, about 20 million have an astrometric solution in which the actual source colour was used instead of a default colour. This means that for these 20 million 2-p sources the G-band correction should actually not be applied. These sources are mostly faint, with 96% at magnitudes G>20, and for 75% of these 20 million sources the correction that is (wrongly) applied amounts to less than ~4 milli-magnitudes. It was thus decided not to make a special effort to exclude these sources from the correction.

    Should a user of the Gaia EDR3 data wish to undo the wrong correction for one or more of these 20 million sources, the list of source IDs and applied corrections can be provided on request.

    NOTE: the corrections to the G-band photometry have been included in the G-band photometry that is published in Gaia DR3.