Water Quality Prediction of Puzhal Lake using Satellite Imagery and Machine Learning

Water quality is one of the most important components of environmental sustainability and human health, but conventional monitoring is often insufficient to provide the spatial and temporal resolution needed for effective management. This paper proposes a new approach to predicting water quality in Puzhal Lake based on the integration of highresolution Sentinel-2 satellite images and in-situ data provided by the Tamil Nadu Water Board from 2019 to 2023. The development of the proposed methodology entailed a series of steps that included pre-processing of the satellite data to remove noise, mask clouds, correct atmospheric effects and perform spatial-temporal integration to generate high definition and accurate images. Some spectral ratios like the blue to green ratio and the Normalized Difference Water Index (NDWI) were also extracted to see the subtle changes in water quality. These indices are combined with the ground data of DO, pH, Turbidity, and TDS to form a single dataset. Several machine learning algorithms are developed to predict the water quality parameters namely Random Forest, Gradient Boosting and Artificial Neural Networks. The models are trained and optimized to the best of their abilities through hyperparameter tuning and cross validation and give a good performance with RMSE as low as 0.078 and R² of 0.911. Besides contributing to the improvement of current practices in water resource management, this study also offers practical recommendations for policy making and environmental protection in the context of sustainable development.