The 35th annual conference on Astronomical Data Analysis Software & Systems 2025 in Görlitz, Germany

Large spectroscopic surveys such as GALAH, APOGEE, LAMOST and others, provide fundamental measurements of stellar parameters and chemical abundances for millions of stars. These data are essential for addressing a wide range of astrophysical questions, from understanding stellar evolution to reconstructing the formation history of the Milky Way. However, systematic differences in instrumentation, analysis methods, and calibration strategies across surveys introduce offsets and trends in the reported parameters. Such inconsistencies limit the potential of combined datasets for tackling scientific problems.

The Survey of Surveys (SoS) project was initiated to address these challenges by homogenizing data across heterogeneous catalogues. In our first public release, we focused on the homogenization of radial velocities, producing the largest consistent compilation of spectroscopic velocity measurements to date. In our second release, we extended this effort to stellar parameters, publishing a catalogue that combined homogenized spectroscopic parameters with machine-learning-based estimates derived from broadband photometry, thereby providing consistent measurements for stars observed in spectroscopy as well as reliable parameter estimates for a much larger photometric sample not covered by spectroscopy.

In this work, we present the continuation of the SoS project, incorporating updates from the most recent spectroscopic survey data releases. Our approach applies data-driven methods to identify and correct systematic differences between surveys, and we have also developed a pipeline to re-calibrate the reported uncertainties, ensuring more realistic error distributions.