Extensive efforts to adaptively manage nutrient pollution rely on Chesapeake Bay Program's (Phase 6) Watershed Model, called Chesapeake Assessment Scenario Tool (CAST), which helps decision-makers plan and track implementation of Best Management Practices (BMPs). We describe mathematical characteristics of CAST and develop a constrained nonlinear BMP-subset model, software, and visualization framework. This represents the first publicly available optimization framework for exploring least-cost strategies of pollutant load control for the United States' largest estuary. The optimization identifies implementation options for a BMP subset modeled with load reduction effectiveness factors, and the web interface facilitates interactive exploration of >30,000 solutions organized by objective, nutrient control level, and for ~200 counties. We assess framework performance and demonstrate modeled cost improvements when comparing optimization-suggested proposals with proposals inspired by jurisdiction plans. Stakeholder feedback highlights the framework's current utility for investigating cost-effective tradeoffs and its usefulness as a foundation for future analysis of restoration strategies.

适应性管理养分污染的广泛努力依赖于切萨皮克湾计划（第 6 阶段）流域模型，称为切萨皮克评估情景工具 (CAST)，它帮助决策者规划和跟踪最佳管理实践 (BMP) 的实施。我们描述了 CAST 的数学特征并开发了一个受约束的非线性 BMP 子集模型、软件和可视化框架。这是第一个公开可用的优化框架，用于探索美国最大河口污染物负荷控制的最低成本策略。优化确定了 BMP 子集的实施选项，该子集以减少负荷的有效性因素为模型，网络界面促进了对超过 30,000 个按目标、营养控制水平和约 200 个县组织的解决方案的交互式探索。在将优化建议的提案与受管辖计划启发的提案进行比较时，我们评估框架性能并展示建模成本改进。利益相关者的反馈强调了该框架当前用于研究成本效益权衡的效用及其作为未来分析恢复策略基础的有用性。