A bi-level game model for dynamic pollutant load allocation integrating pollutant transport processes and stakeholder interaction

Basin-wide pollutant load allocation is essential for integrated pollution control, but current approaches often overlook future pollutant discharge, rely on single policy instruments, and neglect multi-stakeholder interactions. To address these limitations, this study proposes a Bi-Level Game Dynamic Allocation Model (BLG-DAM) that synergistically integrates pollutant transport processes, stakeholder interactions, and environmental tax responses. Driven by meteorological data fusion from the China Meteorological Assimilation Driving Datasets (CMADS) and the European Centre for Medium-Range Weather Forecasts Reanalysis (ERA5) datasets, the SWAT model achieves high-fidelity simulation of pollutant transport. Incorporating system dynamics enables the bi-level game to capture strategic stakeholder interactions across governance levels while facilitating dynamic information transfer between decision-makers and implementers. Applied to the ecologically sensitive Meishan reach of China’s Minjiang River Basin, which subject to intense urban-rural pollution pressures, the model quantified 2020 annual loads at 7690.25 tons NH3N, 8759.11 tons TN, and 9387.29 tons TP. Optimization results demonstrate that the proposed BLG-DAM can achieve reductions in NH₃-N, TN, and TP loads by 43.77 %, 32.48 %, and 35.79 %, respectively, by 2025. Under an environmental tax rate of 14 CNY per pollution equivalent, the model is projected to generate a total basin revenue of 2.15 billion CNY, while also increasing cooperation among control units from 25 % to 90 %. These outcomes indicate a successful alignment of economic incentives with water-quality targets, underscoring the model’s utility as a practical and scalable tool for supporting sustainable watershed management.