Almeida et al. (2026) Enhancing water management in data-scarce watersheds using satellite and reanalysis precipitation: a combined SWAT+ and SMOGN machine learning approach

Identification

• Journal: Journal of Hydrology
• Year: 2026
• Date: 2026-06-22
• Authors: Manuel Almeida, Narciso Ambrósio, Senlin Zhu, Pedro Coelho
• DOI: 10.1016/j.jhydrol.2026.135895

Research Groups

MARE – Marine and Environmental Sciences Centre, ARNET – Aquatic Research Network Associate Laboratory, NOVA School of Science and Technology, NOVA University Lisbon, Caparica, Portugal; Instituto Nacional de Recursos Hídricos de Angola, Talatona, Luanda, Angola; College of Hydraulic Science and Engineering, Yangzhou University, Yangzhou, China

Short Summary

This study combines satellite and reanalysis precipitation products with machine learning models to enhance water management in data-scarce watersheds. The results show that model performance rankings are sensitive to the choice of evaluation metric.

Objective

• Investigate the effectiveness of a combined SWAT+ and SMOGN machine learning approach for reliable streamflow simulation in data-scarce regions.

Study Configuration

• Spatial Scale: Two contrasting Angolan watersheds (2716–94,976 km2) representing temperate and tropical savanna climates.
• Temporal Scale: Long-term runoff data simulation

Methodology and Data

• Models used: Soil and Water Assessment Tool Plus (SWAT+), Random Forest, Artificial Neural Network, Bayesian optimization with SMOGN
• Data sources: Satellite precipitation products (MSWEP, IMERG, ERA5-Land) and reanalysis precipitation datasets

Main Results

• Model performance rankings are sensitive to the choice of evaluation metric.
• MSWEP emerges as the most robust precipitation dataset across models and watersheds.
• Higher agreement with observed rainfall does not necessarily translate into superior hydrological model performance.

Contributions

• This study provides a comprehensive assessment of hydrological behavior across flow regimes using a dual evaluation framework.
• The results highlight the importance of multi-metric evaluation and context-specific modeling strategies for reliable streamflow predictions in data-scarce watersheds.

Funding

• This research was funded by [project/program reference codes, e.g., "MARE-ARNET Associate Laboratory" or "Angola's National Water Resources Institute"].

Citation

@article{Almeida2026Enhancing,
  author = {Almeida, Manuel and Ambrósio, Narciso and Zhu, Senlin and Coelho, Pedro},
  title = {Enhancing water management in data-scarce watersheds using satellite and reanalysis precipitation: a combined SWAT+ and SMOGN machine learning approach},
  journal = {Journal of Hydrology},
  year = {2026},
  doi = {10.1016/j.jhydrol.2026.135895},
  url = {https://doi.org/10.1016/j.jhydrol.2026.135895}
}