================================================================================ README for CONUS High-Resolution Ensemble Precipitation and Temperature Dataset ================================================================================ Dataset Title: High-Resolution Ensemble Precipitation and Temperature Datasets for CONUS based on a Probabilistic Geospatial Estimation Approach Authors: Guoqiang Tang, Andy W. Wood, Andrew J. Newman, Pierre E. Kirstetter, Chanel Mueller, Christopher Frans Release Date: 2025-12 ================================================================================ 1. Overview ================================================================================ This dataset provides daily high-resolution (0.02°, ~2 km) ensemble precipitation and temperature data for the Contiguous United States (CONUS), spanning from January 1, 1950, to December 31, 2023. The dataset includes 20 ensemble members and covers the following variables: - Daily precipitation (mm/day) - Daily mean temperature (Tmean, °C) - Daily temperature range (Trange, i.e., Tmax – Tmin, °C) The dataset is organized into five main components: 1. conus_gapless_station_data_noMADIS.nc - Serially complete station dataset. MADIS stations are not included due to its policy restriction. 2. ensembles/ - 20 ensemble member fields (member_001 to member_020) 3. ensembles_stats/ - Ensemble statistics (mean, std, min, max) 4. regression/ - Deterministic spatial regression estimates 5. regression_adj/ - PRISM climatology-adjusted regression estimates (optional) 6. config_example/ - Configurations used to run GPEP for 19500101-19500331 7. GPEP.zip - Software used to generate the ensemble dataset Data format: NetCDF (CF-compliant) ================================================================================ 2. Methodology ================================================================================ The dataset was generated through the following workflow: 2.1 Station Data Reconstruction: - Integration of high-density observations from GHCN-D and MADIS networks. - Rigorous quality control procedures (based on Durre et al., 2010; Hamada et al., 2011; Beck et al., 2019). - Application of 13 gap-filling and reconstruction strategies (including quantile mapping, interpolation, machine learning, and multi-strategy merging) to produce serially complete station records. 2.2 Ensemble Spatial Estimation: - Based on the probabilistic geospatial estimation framework implemented in the Geospatial Probabilistic Estimation Package (GPEP). - Spatial interpolation using locally weighted linear regression with topographic predictors (latitude, longitude, elevation, slope). - Estimation of errors via leave-one-out cross-validation (LOOCV) and generation of 20 ensemble members. - Optional climatology adjustment to match PRISM normals (provided in regression_adj/). For detailed methodology, refer to the accompanying manuscript. ================================================================================ 3. Data Structure and Files ================================================================================ The dataset contains the following files and directories: File/Directory Size Description ------------------------------------- --------- ----------------------------------- conus_gapless_station_data_noMADIS.nc 2.8 GB Serially complete station dataset ensembles/ 3.9 TB 20 ensemble members (member_001 to member_020) ensembles_stats/ 1.4 TB Ensemble statistics (mean, std, min, max, pctl25, pctl50, pctl75 where pctl is percentile) regression/ 264 GB Deterministic spatial regression estimates regression_adj/ 262 GB PRISM climatology-adjusted regression estimates config_example/ 2.5 KB Configuration files used to run GPEP for one example period GPEP.zip 18 MB Software used to generate the ensemble dataset ================================================================================ 4. Usage Recommendations ================================================================================ - This dataset is suitable for hydrologic modeling, climate analysis, trend detection, risk assessment, and other research applications. - The ensemble data in ensembles/ can be used for uncertainty quantification and probabilistic analysis. - The ensemble statistics in ensembles_stats/ provide summary measures across all 20 members. - The regression/ directory contains deterministic estimates suitable for applications requiring single-value inputs. - The regression_adj/ directory contains PRISM-anchored estimates that match the climatology of PRISM normals (1991-2020). Use this version if consistency with PRISM-based studies is required. - Caution is advised when using precipitation and Trange estimates in complex terrain regions (e.g., the Rocky Mountains) due to higher uncertainties. ================================================================================ 5. Acknowledgments and Citation ================================================================================ If you use this dataset, please cite the following publications: Tang, G., Wood, A. W., Newman, A. J., Kirstetter, P. E., Mueller, C., & Frans, C. (2025). High-resolution ensemble precipitation and temperature datasets for CONUS based on a probabilistic geospatial estimation approach. Journal of Hydrology, 134761. Tang, G., Wood, A. W., Newman, A. J., Clark, M. P., & Papalexiou, S. M. (2024). GPEP v1. 0: the Geospatial Probabilistic Estimation Package to support Earth science applications. Geoscientific Model Development, 17(3), 1153-1173. ================================================================================ 6. Contact ================================================================================ For questions or feedback, please contact: Guoqiang Tang: guoqiang.tang@whu.edu.cn Andy Wood: andywood@ucar.edu ================================================================================ 7. Version History ================================================================================ v1.0 (2025-12): Initial release covering 1950–2023. Data structure as described above.