Assessing Water Quality Through Remote Sensing: A Regression-Based Approach with Sentinel-2 Data

Monitoring water quality is essential for ensuring human health and environmental sustainability. Traditional methods that rely on laboratory analysis and point-based sampling often lack sufficient spatial and temporal coverage. This study assessed the water quality along the Sabarmati Riverfront in Ahmedabad, India, using Google Earth Engine (GEE) and Sentinel-2 satellite imagery. Key parameters, including pH, turbidity (Tur), Electrical Conductivity (EC), Total Suspended Solids (TSS), Total Solids (TS), Dissolved Oxygen (DO), Biochemical Oxygen Demand (BOD), Total Phosphorus (TP), Fecal Coliform (FC), and ammonia (NH₃), were estimated using remote sensing. An empirical regression model was developed to relate in situ data to satellite-derived spectral indices. Results revealed significant seasonal and spatial variations, with some areas showing favorable levels of TSS, BOD, and FC. The model showed strong predictive accuracy for pH, TSS, and TP (R² = 0.80, R² = 0.76, R² = 0.75, respectively) and moderate performance for turbidity (R² = 0.62). Integrating remote sensing and GIS enables scalable, cost-effective, real-time water quality monitoring, providing critical insights for pollution control and water resource management. Future research should explore hyperspectral imaging and machine learning to enhance predictive accuracy and broaden the applicability of satellite-based monitoring models.