Survey of Surveys: expanding homogenized stellar parameters catalogue in the new release of SoS-Spectro

Method To build a unified stellar parameter catalogue, we: Merged APOGEE and GALAH using weighted averages (weights from re-calibrated uncertainties). Calibrated LAMOST [Fe/H] and log g to the APOGEE/GALAH scale. Aligned effective temperatures: GALAH and LAMOST each calibrated to APOGEE. Merged all surveys together again using weighted averages. Results After calibration, parameters are consistent across surveys even outside of their overlap regions. For example, GALAH and LAMOST effective temperatures, both calibrated to APOGEE, agree well when compared directly. The final catalogue contains homogenized stellar parameters (Teff, logg, [Fe/H]) with robust error estimates for 3.7 million unique stars. It provides a common reference dataset for large-scale stellar studies. Survey of Surveys: expanding homogenized stellar parameters catalogue in the new release of SoS-Spectro Authors: Aleksandra Avdeeva, Elena Pancino, Alessio Turchi, StarDance Team INAF - Osservatorio Astrofisico di Arcetri Introduction Spectroscopic surveys are key sources of stellar parameters, but their reported values and uncertainties often differ between datasets. These discrepancies hinder direct comparisons and the detection of subtle astrophysical trends. We examine the internal consistency of APOGEE, LAMOST, and GALAH, focusing on effective temperature, surface gravity, and metallicity. By using repeated observations and cross-survey comparisons, we derive correction factors that re-scale uncertainties and minimize systematic biases. This creates a unified framework for consistent stellar parameters across surveys. The approach can be extended to other datasets, and applying it to radial velocities will be especially useful for identifying binary stars in large samples. Differences in [Fe/H] between LAMOST and the reference scale as functions of log g (top) and differences in logg as a function of [Fe/H] (bottom). Error re-calibration Repeated observations in LAMOST and APOGEE allow internal checks of parameter uncertainties. Normalized pairwise differences should be centered at zero with unit standard deviation. We filter problematic stars and compute normalized differences in temperature, gravity, and metallicity. The distributions deviate from Gaussian, showing that uncertainties are often misestimated in different directions.To correct this, we bin stars in the Kiel diagram and derive a scaling factor kappa under Gaussianity tests. Maps of kappa reveal systematic underand over-estimations, which we propagate to all stars using KNN. For GALAH, we re-scale uncertainties through cross-comparisons with LAMOST and APOGEE. This produces consistent, re-normalized errors across all three surveys. This step ensures consistent errors across surveys and will be crucial for robust binary detection. An example of distributions in a 2D-bin (top) and a resulting coefficient map for APOGEEs logg (bottom) Conclusion and next steps We have shown that cross-calibration and error re-normalization significantly reduce discrepancies between spectroscopic surveys. These intermediate results demonstrate that APOGEE, GALAH, and LAMOST can be brought onto a common scale with consistent uncertainties. The next steps are to extend the methodology to additional surveys, refine calibrations using high-resolution reference datasets such as PASTEL, and apply the same framework to radial velocities. This will further improve parameter reliability and open the way for future applications such as binary detection and various Galactic studies. Funded by the European Union (ERC-2022-AdG, "StarDance: the non-canonical evolution of stars in clusters", Grant Agreement 101093572, PI: E. Pancino). Kiel diagram of the resulted union of APOGEE, LAMOST and GALAH Our latest catalogue of 23 million stars (SoS-Spectro and SoS-ML) is already available online and can be found here →