Teng et al. (2026) Coupling GSFLOW with a river hydrodynamic model for flow simulation in a mountain river basin

Identification

• Journal: Modeling Earth Systems and Environment
• Year: 2026
• Date: 2026-03-25
• Authors: Fei Teng, Wenrui Huang, Yi Cai
• DOI: 10.1007/s40808-026-02756-1

Research Groups

• Shanghai Water Engineering Design & Research Institute Co., Ltd., Shanghai, China
• Tongji University, Shanghai, China
• Florida Agricultural and Mechanical University, Tallahassee, USA

Short Summary

This study developed a numerical algorithm to dynamically couple the GSFLOW hydrological model with a river hydrodynamic model (RHM-SG) to improve streamflow and flood simulation accuracy in mountain rivers. The integrated GSFLOW–RHM-SG model significantly enhanced predictions of extreme flows during flood events and better represented stream–groundwater interactions in the Zamask–Yingluoxia subbasin of the Heihe River Basin, China.

Objective

• To develop and evaluate a dynamically coupled GSFLOW-RHM_SG model to enhance the accuracy of streamflow and flood simulations and better represent stream–groundwater interactions in mountain river basins, addressing the limitations of GSFLOW's kinematic wave-based stream routing.

Study Configuration

• Spatial Scale: Zamask–Yingluoxia subbasin, Heihe River Basin, China (area: 2934 km²; river reach length: 104 km). GSFLOW utilized 50 Hydrologic Response Units (HRUs) and 24 stream segments. MODFLOW grids were 743 active finite-difference cells (2 km x 2 km horizontally, 5 layers vertically with thicknesses of 150 m, 100 m, 100 m, 100 m, and 600 m). RHM-SG used a space interval of 1000 m.
• Temporal Scale: GSFLOW operated with daily or hourly time steps, while RHM-SG used a 20-second time step. The model was calibrated using daily data from 2006 and validated with daily data from 2007. Further application simulations were conducted for July to October 2009.

Methodology and Data

• Models used:
  • GSFLOW (version 1.1.5): A coupled groundwater and surface-water flow model, integrating PRMS (Precipitation Runoff Modeling System) for rainfall-runoff and MODFLOW-2005 (Modular Ground-water Flow Model) for groundwater.
  • RHM-SG (River Hydrodynamic Model for Surface and Groundwater interactions, version 1.0): A one-dimensional river hydrodynamic model based on the Saint-Venant Equations, modified to include a water exchange term between channels and aquifers. It was solved numerically using an implicit finite difference method (Preissmann scheme).
  • Coupling: A numerical algorithm was developed for dynamic coupling, managing different time and spatial intervals between GSFLOW and RHM-SG.
• Data sources:
  • Observed daily discharge data from Zamask, Qilian, and Yingluoxia hydrological stations.
  • Observed daily meteorological data (maximum/minimum temperature, precipitation, evapotranspiration, solar radiation) from observation stations (e.g., Qilian Station).
  • Digital Elevation Model (DEM) and GIS data for the Zamask–Yingluoxia Subbasin.
  • Soil distribution data for hydraulic conductivity determination.
  • Data provided by the Center for Water Research of Beijing University and the Cold and Arid Regions Science Data Center at Lanzhou.

Main Results

• The coupled GSFLOW–RHM-SG model significantly improved streamflow simulation accuracy, particularly for extreme flows during flood events, compared to the original GSFLOW model.
  • During calibration (2006), the coupled model reduced the peak discharge relative error from 7.6% (GSFLOW) to 4.1%, improved RMSE from 24.41 m³/s to 20.85 m³/s, and increased NSE from 0.82 to 0.86, and KGE from 0.72 to 0.87.
  • During verification (2007), the coupled model reduced the peak discharge relative error from 7.7% (GSFLOW) to -2.1%, improved RMSE from 24.5 m³/s to 20.0 m³/s, and increased NSE from 0.79 to 0.86, and KGE from 0.84 to 0.89.
• The coupled model better represented stream–groundwater interactions, especially during low-flow periods, by dynamically accounting for water levels and wetted perimeters, reducing the underestimation of discharge.
• Simulations for July–October 2009 indicated that groundwater recharge was a stable and significant component, contributing an average of 24.6 m³/s, accounting for 54.6% of the total streamflow, followed by surface runoff (40.9%) and subsurface flow (4.5%). Similar proportions were found for 2006 (59%) and 2007 (56%).
• The dynamic-wave algorithm in RHM-SG accurately captured flood wave attenuation and deformation, improving the reliability of flux calculations between river flow and groundwater.

Contributions

• Developed a novel numerical algorithm for dynamically coupling the hydrological model GSFLOW with a river hydrodynamic model (RHM-SG), addressing the limitations of GSFLOW's kinematic wave-based stream routing in mountainous regions.
• Demonstrated a method for integrating models with disparate spatial and temporal scales, enabling simultaneous simulation of large-scale hydrological processes and fine-scale river hydrodynamics.
• Quantitatively showed significant improvements in simulating flood peaks, low-flow characteristics, and the dynamic interaction between surface water and groundwater.
• Provided a more accurate and flexible modeling tool for water resource management and flood mitigation strategies in complex mountain river basins.

Funding

Not explicitly stated in the paper. Data for this study were provided by the Center for Water Research of Beijing University and the Cold and Arid Regions Science Data Center at Lanzhou.

Citation

@article{Teng2026Coupling,
  author = {Teng, Fei and Huang, Wenrui and Cai, Yi},
  title = {Coupling GSFLOW with a river hydrodynamic model for flow simulation in a mountain river basin},
  journal = {Modeling Earth Systems and Environment},
  year = {2026},
  doi = {10.1007/s40808-026-02756-1},
  url = {https://doi.org/10.1007/s40808-026-02756-1}
}