LEVERAGING MACHINE LEARNING FOR REAL-TIME WATER QUALITY MONITORING IN SURFACE WATER BODIES

Abstract

The rapid degradation of surface water quality due to urbanization, industrial discharge, and agricultural runoff poses a significant threat to both environmental sustainability and public health. This study presents a machine learning-driven framework for real-time monitoring and prediction of water quality using IoT-enabled sensor networks. The proposed methodology integrates multi-parameter sensors deployed across river systems to collect key indicators such as pH, dissolved oxygen, turbidity, nitrates, and phosphates. These inputs are processed through supervised learning models—including Random Forest, XGBoost, and LSTM—to classify pollution levels and forecast contaminant trends. Experimental results demonstrate that XGBoost achieves the highest classification accuracy of 95.3%, followed by Random Forest at 94.5%, highlighting the robustness of ensemble methods. Feature importance analysis identifies pH and dissolved oxygen as critical determinants of water quality. A real-time dashboard visualizes the data and model predictions, enabling authorities to detect pollution hotspots and take timely interventions. This framework paves the way for smart, data-driven water resource management systems that ensure ecological protection and informed policy-making.