EventsThe 10th International Electronic Conference on Water Sciences
Published
This submission belongs to the session S7. Remote Sensing, Artificial Intelligence, and New Technologies in Water Sciences of the event The 10th International Electronic Conference on Water Sciences
Published date
19 Nov, 2026
Academic Editor
author-avatarNikiforos Samarinas
Citation
Lakachew Yihunie Alemneh, Dagnachew Aklog yihun, Ann Van Griensven, Minychl G Dersseh, Goraw Goshu Yimer, Seleshi G Yalew, Demesew A. Mhiret, Sisay Asress, Claire I. Michailovsky, Tigistu Wassie Agegnehu, Samuel Berihun Kassa, Shawl Abebe Desta, Integrating Sentinel-2 Remote Sensing and Machine Learning for Spatiotemporal Monitoring of Water Quality Dynamics in Lake Tana, Ethiopia, in Proceedings of The 10th International Electronic Conference on Water Sciences, 24 November–27 November 2026, MDPI: Basel, Switzerland
Share
Email
Facebook
Twitter
LinkedIn

Integrating Sentinel-2 Remote Sensing and Machine Learning for Spatiotemporal Monitoring of Water Quality Dynamics in Lake Tana, Ethiopia

image
Seleshi G Yalew 5
Samuel Berihun Kassa 8
1. Faculty of Civil and Water Resources Engineering, Bahir Dar Institute of Technology, Bahir Dar University, Bahir Dar P.O. Box 79, Ethiopia
2. Department of Water and Climate Engineering, Vrije Universiteit Brussel, Boulevard de la Plaine 2, 1050 Ixelles, Belgium.
3. llege of Agriculture and Environmental Sciences, Bahir Dar University, Bahir Dar P.O. Box 79, Ethiopia
4. Blue Nile Water Institute, Bahir Dar University, Bahir Dar P.O. Box 79, Ethiopia
5. Department of Land and Water Management, IHE Delft Institute for Water Education, Westvest 7, 2611 AX Delft, The Netherlands
6. Department of Hydro informatics and Socio-Technical Innovation, IHE Delft Institute for Water Education, Westvest 2611 AX Delft, The Netherlands
7. Bahir Dar City Water Supply, Sewerage and Sanitation Services, Amhara Water and Energy Bureau P.O. Box 35, Ethiopia.
8. The United Graduate School of Agricultural Sciences, Tottori University, 4-101 Koyama-cho Minami, Tottori 680-8553, Japan
9. Department of Hydrolics and water resource engineering, Wollo University, Kombolcha Instutute of technolgy, p.O.BOX,208, Kombolcha, Ethiopia
Abstract

Spatiotemporal monitoring of water quality is essential for assessing ecosystem health, detecting environmental change, and supporting sustainable management of large freshwater lakes. Lake Tana, Ethiopia’s largest freshwater lake, is increasingly affected by watershed degradation and anthropogenic pressures, resulting in water quality deterioration. Conventional monitoring approaches are costly, labor-intensive, and insufficient for capturing spatial and temporal variability in large lake systems. This study integrates Sentinel-2 Multispectral Instrument (MSI) imagery and machine learning (ML) algorithms to estimate chlorophyll-a (Chl-a), turbidity, total nitrogen (TN), and total phosphorus (TP). A total of 572 in situ observations (2016–2025) were combined with Sentinel-2 spectral bands and derived indices processed in Google Earth Engine. Four ML algorithms Random Forest (RF), Extreme Gradient Boosting (XGBoost), Artificial Neural Network (ANN), and Support Vector Regression (SVR) were developed and validated. Ensemble models consistently outperformed other approaches, with RF achieving the highest accuracy for Chl-a (R² = 0.974, RMSE = 0.0161) and TP (R² = 0.977, RMSE = 0.0238), while XGBoost best predicted TN (R² = 0.975, RMSE = 0.0220). ANN yielded the highest performance for turbidity (R² = 0.815, RMSE = 0.0660). Long-term Mann–Kendall trend analysis indicated no significant monotonic trends in TN (p = 0.144) and Chl-a (p = 0.217), while TP showed a significant decreasing trend (p = 0.038) and turbidity exhibited a significant increasing trend (p < 0.001). However, pronounced short-term and inter-seasonal variability was observed, particularly near river inflows, where episodic nutrient and sediment inputs caused localized fluctuations. Persistent hotspots were identified near the Rib, Megech, and Gilgel Abay river mouths, highlighting strong watershed–lake interactions. Overall, the integration of Sentinel-2 MSI and machine learning provides an accurate, cost-effective, and scalable framework for long-term water quality monitoring in data-limited freshwater ecosystems, supporting adaptive lake management and eutrophication control

Keywords
Sentinel 2 MSI
Lake Tana
Water Quality
Machine Learning
Non-Optically active
Optically Active
Remote Sensing of Water Quality Using Sentinel-2: A Case Study of the Roxo Reservoir, Alentejo, Portugal
Multi-Index Assessment of Hydroclimatic and Vegetation Drought Trends in the Lake İznik Basin