Red clump stars (RCs) are useful tracers of distances, extinction, chemical abundances, and Galactic structures and kinematics. Accurate estimation of the RC parameters -- absolute magnitude and intrinsic color -- is the basis for obtaining high-precision RC distances. By combining astrometric data from Gaia, spectroscopic data from APOGEE and LAMOST, and multi-band photometric data from Gaia, APASS, Pan-STARRS1, 2MASS, and WISE surveys, we use the Gaussian process regression to train machine learners to derive the multi-band absolute magnitudes $M_lambda$ and intrinsic colors $(lambda_1-lambda_2)_0$ for each spectral RC. The dependence of $M_lambda$ on metallicity decreases from optical to infrared bands, while the dependence of $M_lambda$ on age is relatively similar in each band. $(lambda_1-lambda_2)_0$ are more affected by metallicity than age. The RC parameters are not suitable to be represented by simple constants but are related to the Galactic stellar population structure. By analyzing the variation of $M_lambda$ and $(lambda_1-lambda_2)_0$ in the spatial distribution, we construct $(R, z)$ dependent maps of mean absolute magnitudes and mean intrinsic colors of the Galactic RCs. Through external and internal validation, we find that using three-dimensional (3D) parameter maps to determine RC parameters avoids systematic bias and reduces dispersion by about 20% compared to using constant parameters. Based on Gaias EDR3 parallax, our 3D parameter maps, and extinction-distance profile selection, we obtain a photometric RC sample containing 11 million stars with distance and extinction measurements.
Download