Goudiaby et al. (2025) Hydrological evaluation of gridded rainfall products for streamflow simulation in West Africa

Identification

• Journal: Journal of Hydrology Regional Studies
• Year: 2025
• Date: 2025-11-27
• Authors: Omar Goudiaby, Ansoumana Bodian, Alain Dezetter, Andrew Ogilvie, Ibrahima Diouf, Papa Ndiaye
• DOI: 10.1016/j.ejrh.2025.102984

Research Groups

• Laboratoire Leïdi « Dynamique des Territoires et D´eveloppement », Universit´e Gaston Berger (UGB), Saint-Louis, Senegal
• HydroSciences Montpellier, Univ. Montpellier, IRD, CNRS, UFR Pharmacie, Montpellier, France
• UMR G-EAU, AgroParisTech, Cirad, Universit´e de Montpellier, IRD, INRAE, Montpellier SupAgro, Montpellier, France
• Laboratoire de Physique de l’Atmosph`ere et de l’Oc´ean-Simon Fongang, Ecole Sup´erieure Polytechnique de l’Universit´e Cheikh Anta Diop (UCAD), Dakar, Senegal

Short Summary

This study evaluates the hydrological performance of 23 gridded precipitation products for streamflow simulation in eight West African river basins using GR2M and GR4J models, finding that multi-source products like IMERGDF, MSWEP, GPCP, and TAMSAT are the most reliable alternatives in data-scarce regions.

Objective

• To assess the ability of twenty-three gridded precipitation products to simulate streamflows using monthly (GR2M) and daily (GR4J) rainfall-runoff models in eight sub-catchments of the Senegal, Gambia, and Casamance river basins in West Africa.

Study Configuration

• Spatial Scale: Eight sub-catchments within the Casamance, Gambia, and Senegal river basins in West Africa (Bafing Makana, Daka Saïdou, Kidira, Gourbassi, Oualia, Mako, K´edougou, and Kolda). Watershed boundaries were delineated using Shuttle Radar Topography Mission (SRTM) data with a 30 meter spatial resolution.
• Temporal Scale: Data series for observed rainfall and streamflow cover varying periods, generally between 1952 and 2019. Gridded precipitation products span from 1979 to the present (or specific end dates). Hydrological evaluations were performed at daily and monthly time steps, with specific calibration/validation periods defined for each basin and product group (e.g., 1984–2019, 1998–2019).

Methodology and Data

• Models used:
  • GR2M (G´enie Rural 2-parameter Monthly model): A conceptual rainfall-runoff model for monthly time steps.
  • GR4J (G´enie Rural 4-parameter Daily model): A conceptual rainfall-runoff model for daily time steps.
• Data sources:
  • Gridded Precipitation: Twenty-three products, including satellite-based (e.g., IMERGDF, TAMSAT, PERSIANN), reanalysis (e.g., ERA5, MERRA-2), and in situ-interpolated or multi-source composite datasets (e.g., MSWEP, CHIRPS, GPCP, CPC). Spatial resolutions range from 0.0375° to 1°.
  • Observed Rainfall: Daily records from 137 rain gauge stations collected from national meteorological services of Senegal, Mali, Guinea, and Gambia.
  • Observed Streamflow: Daily discharge data from 8 hydrometric stations, obtained from the Directorate of Water Resources Management and Planning (DGPRE) and the Organization for the Development of the Senegal River (OMVS), covering periods between 1952 and 2015.
  • Potential Evapotranspiration (PET): Calculated using the temperature-based method of Droogers and Allen (2002) from daily maximum and minimum temperature data sourced from the National Centers for Environmental Prediction (NCEP) reanalysis (1948-present).
  • Topography: Shuttle Radar Topography Mission (SRTM) data (30 m resolution) for watershed delineation.
  • Interpolation: Inverse Distance Weighted (IDW) method used to calculate mean observed rainfall and PET per catchment. For gridded precipitation, two approaches were used: Gridded Products Extracted from Ground Station Coordinates (GPEGSC) and Gridded Products Extracted from Grid Centroid Coordinates (GPEGCC).
  • Evaluation Metrics: Kling-Gupta Efficiency (KGE) and Percentage Bias (PBIAS).

Main Results

• Multi-source gridded precipitation products, which integrate satellite, reanalysis, and in situ observations, generally provided the most accurate streamflow simulations at both monthly and daily time steps.
• IMERGDF emerged as the most reliable product, consistently demonstrating high KGE values across basins and time steps, followed closely by MSWEP, GPCP, and TAMSAT.
• The CPC product, which relies on interpolated ground data, exhibited unexpectedly poor performance.
• Re-calibrating hydrological models with gridded precipitation data (GPEGCC) yielded more accurate simulations with gridded datasets compared to using parameters derived from observed rainfall data.
• Model performance varied across basins: strong performance was observed in Bafing Makana, Daka Saïdou, Kidira, Gourbassi, Mako, and K´edougou (KGE generally > 0.75 in calibration, > 0.5 in validation for several products).
• Weaker performance was noted in the Kolda and Oualia basins, attributed to issues such as hydrometric data quality, large catchment size (Oualia: 87,000 km²), complex hydrological responses spanning multiple climatic zones, and low annual runoff.
• The highest-performing products successfully reproduced the general shape, amplitude, and timing of observed hydrographs, although some discrepancies were observed, such as peak flows occurring up to 1 month earlier and occasional overestimation of low-flow conditions or underestimation of peak flows.
• Spatial resolution (ranging from 0.0375° to 1°) did not appear to be a decisive factor in the hydrological performance of the gridded products; rather, the quality, density, and spatial distribution of the underlying observation network were more critical.

Contributions

• Provides a comprehensive hydrological evaluation of 23 gridded precipitation products, a larger number than many previous studies, across eight diverse West African catchments.
• Identifies specific multi-source gridded products (IMERGDF, MSWEP, GPCP, TAMSAT) as valuable and reliable alternatives for hydrological simulations in data-scarce regions of West Africa.
• Emphasizes the critical importance of re-calibrating hydrological models with gridded precipitation datasets for optimal simulation performance when using these products.
• Offers insights into the factors influencing the performance of gridded precipitation products in hydrological modeling, including data quality, catchment characteristics, and climatic complexity.
• Supports the use of selected gridded precipitation datasets for reconstructing streamflow time series in poorly gauged or ungauged basins, enhancing water resource management in the region.

Funding

Funding information for this research is not explicitly provided in the paper.

Citation

@article{Goudiaby2025Hydrological,
  author = {Goudiaby, Omar and Bodian, Ansoumana and Dezetter, Alain and Ogilvie, Andrew and Diouf, Ibrahima and Ndiaye, Papa},
  title = {Hydrological evaluation of gridded rainfall products for streamflow simulation in West Africa},
  journal = {Journal of Hydrology Regional Studies},
  year = {2025},
  doi = {10.1016/j.ejrh.2025.102984},
  url = {https://doi.org/10.1016/j.ejrh.2025.102984}
}