Spatially refined modeling for agricultural groundwater management in Northwestern China

Irrigation-driven surface water-groundwater (SW-GW) interactions are crucial for preventing salinization in arid and semi-arid irrigation districts. Traditional models face computational limitations across large spatial scales, hindering accurate representation of spatiotemporal heterogeneity. This study proposed a spatially refined coupled SW-GW flow model leveraging high-quality surface water diversion data, and applied it to the Qingtongxia Irrigation District, one of China’s oldest and largest irrigation areas. The model integrated key hydrologic components, including canal diversion, drainage, and groundwater depth, at the farmland scale to capture their spatiotemporal dynamics. Accuracy and efficiency were enhanced through a multi-scale validation framework and parallel computation. The spatially refined model outperformed a coarse-resolution model in groundwater level simulation, reducing RMSE by 0.37 m during calibration and 0.24 m during validation. From 2019 to 2022, groundwater levels rose in over-exploited areas, while depths exhibited an upward trend in shallow groundwater zones, with the proportion of areas with depths of 1.5–3.0 m expanding by 5.4%. The model’s ability to capture local-scale groundwater dynamic processes is highlighted through an analysis of northern farmland at high salinization risk. During summer-autumn irrigation, areas with groundwater depths below 2.2 m increased by 16.0%, demonstrating surface irrigation’s impact on groundwater levels. Quantified hydrological processes showed groundwater discharge dominated along the Yellow River, with an annual discharge of 2.4 × 107 m3. This study provides a quantitative, high-resolution method for analyzing SW-GW interactions, supporting sustainable groundwater management and salinization control.