Intelligent water environment diagnosis and treatment center: A machine learning framework for automated surface...

Surface water quality is essential for public health and ecosystem sustainability, but pollution events arise from distinct biogeochemical mechanisms that require differentiated responses. Conventional management systems nevertheless treat exceedance events as homogeneous emergencies, resulting in inefficient interventions and limited preparedness for true crises. Machine learning has improved water quality prediction and pollution source identification, yet most studies target isolated analytical tasks and remain poorly connected to environmental mechanisms. This disconnect creates a persistent gap between statistical accuracy and management applicability. To bridge this gap, we developed the Intelligent Water Environment Diagnosis and Treatment Center, an automated framework that translates medical diagnostic principles into surface water pollution management. The framework links mechanistic syndrome recognition with knowledge-guided intervention to enable management-relevant diagnosis and response. Comparative analysis demonstrates that domain knowledge optimization transforms feature selection from statistical correlation to ecological relevance, with 58.4% of events misclassified by purely data-driven methods. These findings show that effective environmental decision-making requires mechanistic interpretability in addition to statistical performance. This framework supports a transition from uniform reactive control to differentiated adaptive management, offering a scalable strategy for sustainable water governance.