Deep learning-based optimal adaptive regulation pathway of algal blooms in urban rivers under long-term uncertainties

Algal bloom control remains a critical challenge in urban water resource management, particularly under the intensifying impacts of global climate change. Effectively addressing this issue requires strategies that account for long-term climatic uncertainties. This study applies the Dynamic Adaptive Policy Pathways (DAPP) framework to enhance urban algal bloom management through the integration of system failure signals, a deep learning-based surrogate model, and optimization algorithm. Using Suzhou, China, as a case study, developed 8 climate change scenarios and 19 regulatory scenarios to identify adaptive strategies. A surrogate model incorporating AdaBoost and Bagging techniques was developed to simulate algal bloom dynamics, achieving simulation times of under 10 s per scenario while maintaining high accuracy (R²>0.95). The NSGA-II algorithm was applied to optimize trade-offs between environmental performance and economic costs, accounting for infrastructure lock-in effects through incremental constraints. Results show that temperature rise poses the greatest threat to urban aquatic ecosystems, while precipitation and irradiance changes have marginal impacts. Among the control measures, green infrastructure (GI) was identified as the most cost-effective strategy to mitigate thermal impacts. This research provides a scalable decision-support framework for enhancing urban water ecosystems resilience under climate uncertainty.