Applications of MESH, SUMMA, HYPE, FUSE, and MARRMoT over 304 Headwater Basins of the CAMELS-SPAT dataset

This dataset contains harmonized hydrological model configurations, forcing inputs, and output data for 304 headwater basins included in the CAMELS-SPAT dataset. It was generated using the Model-Agnostic Framework (MAF), a high-performance computing (HPC)-ready system that streamlines the configuration, data preprocessing, and orchestration of diverse hydrological models. The dataset supports reproducible, large-sample modeling across multiple process-based models: MESH, SUMMA, HYPE, FUSE, and MARRMoT. Each archive contains: static attributes (elevation, land cover, and soil class), meteorological forcing data derived from the CaSR dataset (1980–1984, hourly resolution), model-specific configuration files, and geospatial shapefiles representing basin and river geometries. Preprocessing was carried out using open-source components of MAF including `datatool`, `gistool`, `easymore`, and model-specific flows (e.g., `SUMMAFlow`, `meshflow`, `fuseflow`). These tools ensured standardized data handling and simulation setup, enabling consistent benchmarking across models and study domains. This dataset enables hydrologists to compare model behavior across scales, evaluate model performance in ungauged or nonstationary conditions, and test the impact of structural diversity in hydrological modeling. It supports studies in model evaluation, benchmarking, uncertainty analysis, and hydrological forecasting, especially in large-sample or high-resolution contexts.