Artificial Intelligence Algorithms for Detecting Hazardous Substance Discharges into Water Bodies and Modeling Their Distribution

This study explores the application of Artificial Intelligence (AI) algorithms for detecting hazardous substance discharges into water bodies and modeling their distribution. River pollution remains a major environmental concern, while traditional monitoring methods are labor-intensive, costly, and lack operational efficiency. The paper analyzes key water quality parameters and modern Internet of Things (IoT) systems that enable real-time data collection on indicators such as pH, dissolved oxygen, temperature, and turbidity. Emphasis is placed on Machine Learning (ML) algorithms for anomaly detection and pollution forecasting. Methods including linear regression, decision trees, neural networks, and clustering are compared, with particular attention to ensemble models such as Random Forest (RF) and XGBoost. Using open environmental data on Ukrainian surface waters for 2022, RF and XGBoost were found to effectively predict nitrogen and dissolved oxygen concentrations and identify anomalies linked to anthropogenic pollution sources, including municipal sewage and agricultural runoff. The integration of AI and IoT technologies significantly improves the accuracy, speed, and efficiency of water quality monitoring, supporting early warning systems and sustainable water management while highlighting ongoing challenges related to data quality and implementation costs.