This dataset supports the study "On AI-based large-sample emulators for land/hydrology model calibration and regionalization", which presents a Large-Sample Emulator (LSE) approach for calibrating and regionalizing parameters in land/hydrology models. The dataset includes key files and resources necessary to reproduce and extend the LSE-based calibration experiments conducted with the Community Terrestrial Systems Model (CTSM) across 627 basins from the CAMELS dataset in the continental United States. Due to the large size of the complete CTSM forcing and output files, only essential components are included here. Full CTSM meteorological forcings and model outputs used in the experiments are available via the NCAR Research Data Archive. For questions about this dataset or related methods, please contact: Guoqiang Tang (guoqiang.tang@whu.edu.cn) and/or Andy Wood (andywood@ucar.edu). When using this dataset, please cite: Guoqiang Tang, Andy Wood, Sean Swenson (2025). On AI-based large-sample emulators for land/hydrology model calibration and regionalization. Water Resources Research.