Combining CMB datasets with consistent foreground modelling
M. Tristram, M. Douspis, A. Gorce, S. Henrot-Versillé, L. T. Hergt, S. Ilic, L. McBride, M. Muñoz-Echeverría, E. Pointecouteau, L. Salvati
TL;DR
This work demonstrates the feasibility and benefits of a fully joint CMB analysis of Planck, ACT, and SPT data using a single, coherent foreground and instrumental model. By sharing foreground templates and accounting for instrument systematics within a unified likelihood, the study achieves robust ΛCDM parameter constraints and reveals the reduced impact of foreground choices on these parameters, while exposing greater sensitivity for certain cosmological extensions. The analysis finds no significant deviations from ΛCDM, and tensions such as nonzero curvature or elevated lensing amplitude are mitigated when foreground modelling is treated consistently; however, foreground parameters—especially the CIB, kSZ, and their correlations—remain highly model-dependent. The framework also provides a path forward for next-generation high-sensitivity CMB surveys, emphasizing consistent foreground treatment and the potential to incorporate halo-model descriptions of extragalactic components to propagate astrophysical uncertainties. Overall, this work confirms the value of joint multi-dataset analyses for robust cosmology and foreground astrophysics, guiding future survey design and analysis strategies.
Abstract
We present a joint cosmological analysis combining data from the Planck satellite, the Atacama Cosmology Telescope, and the South Pole Telescope, constructing a unified likelihood that reproduces the measured temperature and polarisation power spectra by jointly modelling the cosmic microwave background (CMB) signal, galactic and extragalactic foregrounds, and instrumental systematics across all datasets. This approach reduces reliance on external priors and improves the robustness of parameter estimation. Within this joint analysis, $Λ$CDM parameters exhibit remarkable stability with respect to variations in foreground modelling. Extended cosmological parameters are more sensitive to these assumptions, with uncertainties increasing by up to 35%. Despite this, the combined constraints show no significant deviation from $Λ$CDM expectations, and several previously reported tensions -- such as the preference for non-zero curvature or the excess of lensing amplitude A_L -- are significantly reduced or resolved. In contrast, the determination of foreground parameters more severely depends on the assumptions made about the underlying models. Overall, this work demonstrates the feasibility and reliability of a fully joint analysis of current CMB experiments, and emphasizes the importance of consistent and accurate foreground modelling for the scientific goals of next-generation, high-sensitivity CMB surveys.