Five-Year Spatiotemporal Water Quality Assessment of Horseshoe Lake Using Sentinel-2 Imagery and Google Earth Engine (GEE)
Session Number
ENVR 01
Advisor(s)
Dr. Anuj Tiwari, Discovery Partners Institute
Discipline
Environmental Science
Start Date
17-4-2025 10:15 AM
End Date
17-4-2025 10:30 AM
Abstract
Satellite-derived data analysis offers a cost-effective alternative to traditional in-situ measurements for monitoring water quality dynamics in critical water bodies. This study leverages remote sensing techniques to assess multiple zones within Horseshoe Lake, Illinois, identified based on physical barriers, hydrological flow, and proximity to water treatment facilities. To ensure scientific rigor, we applied unsupervised clustering techniques such as K-means, ISODATA, and DBSCAN to Sentinel-5P parameters. These techniques validated whether the selected zones are hydrologically distinct, allowing for a more precise characterization of spatial variations in water quality. Using Google Earth Engine (GEE), we retrieved cloud-free composite images spanning 2020-2024 and chose key water quality indicators for further analysis: chlorophyll-α concentration, turbidity, and phosphorus levels. We then applied statistical trend analysis and cross-indicator validation to explore parameter interdependence and assess how major meteorological events influence water quality trends. This study provides a comprehensive assessment of pollution dynamics within Horseshoe Lake, offering a robust methodology for long-term water quality monitoring. Future work will focus on integrating these analytical capabilities into a public dashboard, enabling users to monitor lake conditions in real time from any personal device.
Five-Year Spatiotemporal Water Quality Assessment of Horseshoe Lake Using Sentinel-2 Imagery and Google Earth Engine (GEE)
Satellite-derived data analysis offers a cost-effective alternative to traditional in-situ measurements for monitoring water quality dynamics in critical water bodies. This study leverages remote sensing techniques to assess multiple zones within Horseshoe Lake, Illinois, identified based on physical barriers, hydrological flow, and proximity to water treatment facilities. To ensure scientific rigor, we applied unsupervised clustering techniques such as K-means, ISODATA, and DBSCAN to Sentinel-5P parameters. These techniques validated whether the selected zones are hydrologically distinct, allowing for a more precise characterization of spatial variations in water quality. Using Google Earth Engine (GEE), we retrieved cloud-free composite images spanning 2020-2024 and chose key water quality indicators for further analysis: chlorophyll-α concentration, turbidity, and phosphorus levels. We then applied statistical trend analysis and cross-indicator validation to explore parameter interdependence and assess how major meteorological events influence water quality trends. This study provides a comprehensive assessment of pollution dynamics within Horseshoe Lake, offering a robust methodology for long-term water quality monitoring. Future work will focus on integrating these analytical capabilities into a public dashboard, enabling users to monitor lake conditions in real time from any personal device.