Land-sea coordination is the core approach to addressing land-ocean pollution. However, existing measures focus on fragmented governance or the application of single models, and synergistic management technologies and quantitative schemes at the watershed scale remain insufficient. Pollution control therefore requires a shift from localized interventions to a synergistic strategy for the entire watershed. Here, OneBasin, a scalable framework for Multi-section Synergistic Simulation & Targeted Emission Reduction, was developed by integrating hydrological modeling, machine learning (ML), and multi-objective optimization. To improve the stability of the OneBasin framework, in ML model configuration, the non-cascaded configuration without GDP was identified as a superior approach, which can prevent mismatch between predictive variables and decision variables and reduce error accumulation. Based on the application practice of total nitrogen pollution control in the Yellow River Basin, this framework can generate emission reduction schemes that balance fairness and efficiency, and its core value lies in providing scientific support for the implementation of emission reduction measures such as agricultural upgrading and wastewater treatment standard improvement. These results aim to offer a valuable perspective for transitioning river basin water governance from fragmented "local standards" to integrated "systemic health".
🌊
