Characterising the spatio-temporal patterns of water quality parameters in the cradle of humankind world heritage site using Sentinel-2 and random forest regressor

A combination of Sentinel-2 satellite data and random forest regression models successfully predicted water quality parameters with high accuracy, particularly dissolved oxygen (DO) under low-flow conditions (R² = 0.88) and electrical conductivity (EC) (R² = 0.63). The study, conducted in the Cradle of Humankind World Heritage Site, revealed distinct seasonal variations in optically active (e.g., suspended solids) and non-optically active parameters (e.g., DO, EC), influenced by runoff dynamics and upstream pollution. Model 2, incorporating spectral bands and indices, outperformed Model 1 for most parameters, demonstrating the potential of remote sensing and machine learning to monitor freshwater ecosystems dynamically. These findings highlight the importance of high-resolution data in assessing water quality in ecologically significant areas.

Ngamile, S., Kganyago, M., Madonsela, S., & Mvandaba, V. (2025). Characterising the spatio-temporal patterns of water quality parameters in the cradle of humankind world heritage site using Sentinel-2 and random forest regressor. Frontiers in Remote Sensing, 6. https://doi.org/10.3389/frsen.2025.1631403