Assessing the accuracy of Gaia's stellar parameters using benchmark stars

Sammanfattning

The Gaia satellite maps out the Milky Way and its observations have been condensed into a gigantic data base. The quantity is unmatched by other data bases, therefore its quality must always be analysed and scrutinized to keep the data accurate. The purpose of this thesis is to assess the accuracy of the stellar parameters that Gaia provides through its data sets GSP-Phot and GSP-Spec. This was achieved through inspecting the so-called benchmark stars within these samples. These are stars which has been closely studied with accurate methods which gives supreme reliability in their values. The comparison resulted in some conclusions regarding the GSP-Phot and GSP-Spec data bases. GSP-Phot showed sensitivity to distance, its reliance on photometry and parallax leading to errors through reddening, extinction and parallax uncertainty when distance increases. GSP-Spec showed sensitivity to low metallicity, as well as uncertainty from the radial velocity spectrometer (RVS) flux. Lower metallicity coinciding with greater RVS flux uncertainty, was a common denominator for greater error offsets. Finally future approaches were discussed, such as setting up functions using methods like linear regression and machine learning, to use the results to correct Gaia stellar parameters and increase their accuracy.