Model outputs, observations, and analysis code supporting " Sensitivity of Nitrogen Cycling to Future Nitrogen Wet Deposition in West African Savannas"

This record contains the model outputs, observational data, MATLAB analysis code, metadata and documentation supporting the manuscript “Sensitivity of Nitrogen Cycling to Future Nitrogen Wet Deposition in West African Savannas ”. The deposited model outputs were generated for three African savanna sites: Dahra in Senegal, and Korhogo and Lamto in Côte d’Ivoire. The package contains NetCDF files representing multi-model ensemble simulations and individual GCM simulations. The ensemble simulations include five atmospheric nitrogen-deposition chemistry experiments, while the individual GCM simulations provide the Case 1 runs used for model evaluation. The five experiments are: native site-specific deposition chemistry under RCP4.5; 0% NH4 and 100% NO3; 100% NH4 and 0% NO3; 70% NH4 and 30% NO3; and 85% NH4 and 15% NO3. The experiment analysis covers 2009–2050. The baseline period is 2009–2020, and treatment responses are evaluated for 2021–2050. The observational datasets support model validation for Dahra in 2017, Lamto in 2022 and Korhogo in 2023. Validation variables include air temperature, relative humidity, rainfall, soil temperature and soil moisture. The MATLAB scripts included in the archive reproduce the site-level daily and monthly validation, Target and Taylor diagrams, composite forcing-model rankings, nitrogen-cycle experiment analyses, derived tables and manuscript figures. The package also includes site and experiment metadata, a variable inventory, workflow documentation and dependency information. The simulations were generated using the modified eCLM source code archived separately under DOI 10.5281/zenodo.21490378. The archive does not include raw atmospheric forcing products, restart files, compiled executables or the complete unprocessed model workflow. Third-party forcing products must be obtained from their original providers. The observational datasets are not publicly redistributed because permission for public release has not been granted. Permission should be requested before use.