Tidjani et al. (2026) Applicability of the Lumped GR4J Model for Modeling the Hydrology of the Inland Valleys of the Sudanian Zones of Benin

Identification

• Journal: Water
• Year: 2026
• Date: 2026-01-29
• Authors: Akominon M. Tidjani, Quentin Fiacre Togbévi, Pierre G. Tovihoudji, P. B. Irénikatché Akponikpè, Marnik Vanclooster
• DOI: 10.3390/w18030340

Research Groups

• Earth and Life Institute (ELI), Environmental Sciences, Université Catholique de Louvain, Louvain-la-Neuve, Belgium
• Laboratory of Hydraulics and Environmental Modeling (HydroModE-Lab), Faculty of Agronomy, University of Parakou, Parakou, Benin
• Institut des Sciences et Technologies pour l’Innovation en Afrique (ISTI-Africa), Parakou, Benin

Short Summary

This study evaluates the applicability of the lumped GR4J model for simulating streamflow in three inland valleys of the Sudanian zone of Benin and assesses the reliability of satellite-based rainfall data. The GR4J model effectively simulates daily discharge, with CHIRPS emerging as the most consistent satellite rainfall product for reconstructing historical streamflow, providing valuable insights for water resource management.

Objective

• To evaluate the applicability of the lumped GR4J model for simulating daily streamflow in three inland valleys (Lower-Sowé, Bahounkpo, Nalohou) of the Sudanian zone of Benin.
• To test the reliability of satellite-based rainfall data (GPM-IMERG, CHIRPS, GSMAP) in modeling hydrological dynamics and reconstructing long-term discharge in these small catchments, particularly for understanding extreme hydrological events and supporting water management.

Study Configuration

• Spatial Scale: Three inland valleys in the Sudanian zone of Benin: Lower-Sowé (15.22 km²), Bahounkpo (5.38 km²), and Nalohou (0.16 km²).
• Temporal Scale: Daily time step. Calibration and validation periods: Lower-Sowé (2021–2022), Bahounkpo (2021–2023), Nalohou (2015–2019). Historical simulations for long-term analysis: CHIRPS (1981–2023) and GPM-IMERG (2000–2023).

Methodology and Data

• Models used:
  • GR4J (four-parameter, daily lumped rainfall–runoff model) implemented using the HydroGR package.
  • Generalized Extreme Value (GEV) distribution for high-flow return period analysis.
  • Mann–Kendall and Pettit tests for climatic and hydrological trend analysis.
  • Sobol's variance-based sensitivity analysis for model parameters.
• Data sources:
  • In situ data: Daily rainfall and streamflow measurements from the BAFONBE project (Lower-Sowé, Bahounkpo) and the AMMA-CATCH program (Nalohou).
  • Satellite-based rainfall data: GPM-IMERG, CHIRPS, and GSMAP.
  • Reference evapotranspiration: Estimated using the Hargreaves method with ERA5 temperature data.
  • Groundwater levels: Observed piezometric heads at Bahounkpo for correlation analysis.

Main Results

• The GR4J model effectively simulates daily discharge in the three inland valleys with in situ rainfall data (Kling-Gupta Efficiency (KGE) > 0.57 during calibration and validation), showing robust performance for mean-flow conditions.
• The model performs acceptably for high flows and runoff coefficients but shows consistent challenges in accurately capturing low and peak flow conditions.
• Sensitivity analysis indicates that parameter X2 (Groundwater exchange coefficient) is consistently critical across all catchments.
• Among satellite-based rainfall products, CHIRPS emerged as the most consistent, providing good capture of general trends and seasonal variations in discharge prediction (KGE > 0.47).
• GPM-IMERG and GSMAP showed mixed or less accurate results, particularly for high-flow prediction, with near-zero KGE scores indicating marginal improvement over a mean-flow benchmark for Bahounkpo and Lower-Sowé.
• The approach combining historical CHIRPS data and the GR4J model provides valuable insights for water resource management, especially for estimating mean water balance, resource capacity, and volume.
• Long-term analysis revealed an increase in annual rainfall, reference evapotranspiration (at Lower-Sowé and Nalohou), and total runoff across all three sites, with a slight increase in high discharge events.
• The GR4J routing-store relative storage (R/X3) showed high Spearman's rank correlations (rs) with observed groundwater levels at Bahounkpo, with normalized cross-correlation functions peaking at positive lags (τ ≈ +2 to +60 days), suggesting its potential as an operational proxy for groundwater state.

Contributions

• This study is the first to systematically evaluate the applicability of the lumped GR4J model for streamflow simulation in the small, complex inland valleys of the Sudanian zones of Benin.
• It provides a comprehensive assessment of the reliability of three prominent satellite-based rainfall products (GPM-IMERG, CHIRPS, GSMAP) for hydrological modeling in these data-scarce regions.
• The research demonstrates the practical utility of combining historical CHIRPS data with the GR4J model for reconstructing long-term discharge time series, enabling robust analysis of hydrological trends and extreme events crucial for water management.
• It identifies the groundwater exchange coefficient (X2) as a key driver of runoff response in these inland valleys, highlighting the importance of surface-groundwater interactions.
• The study proposes the GR4J routing-store relative storage as a potential operational proxy for groundwater levels, offering a novel tool for rule-based water management in data-limited contexts.

Funding

• Wallonie Bruxelles International (WBI) as part of the BAFONBE Project.

Citation

@article{Tidjani2026Applicability,
  author = {Tidjani, Akominon M. and Togbévi, Quentin Fiacre and Tovihoudji, Pierre G. and Akponikpè, P. B. Irénikatché and Vanclooster, Marnik},
  title = {Applicability of the Lumped GR4J Model for Modeling the Hydrology of the Inland Valleys of the Sudanian Zones of Benin},
  journal = {Water},
  year = {2026},
  doi = {10.3390/w18030340},
  url = {https://doi.org/10.3390/w18030340}
}