A Deep Learning Framework For Water Quality Analysis Using Multi-Index Sentinel-2 Imagery: A Novel Approach
DOI:
https://doi.org/10.29284/dsg1ya40Keywords:
NDWI, MNDWI, NDCI, TBI, WRI, AWEI, XGBoost, CatBoost, COD, ConductivityAbstract
Water quality monitoring is of critical importance in terms of environmental protection, public health, and sustainable management of water resources. However, traditional methods of in-situ measurement are cumbersome and spatially limited due to the time it takes to perform them. The aim of this paper is to propose a deep learning-based framework for large-scale water quality assessment using multi-index Sentinel-2A imagery in combination with ground-measured parameters taken from several locations along the Godavari River in Chhatrapati Sambhajinagar that include Paithan, Jayakwadi, Gangapur, Pathegaon, Kaigaon, and Naregaon. Satellite data were then preprocessed systematically: temporal filtering (from January 2021 to May 2025), cloud masking by QA60 indicators, and clipping by AOI to separate the river stretches of interest. Subsequently, a set of spectral water indices, which comprised NDWI, MNDWI, NDCI, TBI, WRI, and AWEI, was calculated as a way to capture variability associated with turbidity, suspended sediments, chlorophyll concentration, and general water clarity. These indices were therefore used as features for four predictive models: XGBoost, CatBoost, Multilayer Perceptron, and a novel Hybrid Autoencoder + XGBoost architecture. Model evaluation showed that XGBoost and CatBoost generated very stable baselines, while performance for the MLP model was relatively weaker. The proposed Hybrid Autoencoder + XGBoost model presented the best prediction performance among all models, with the lowest values of errors and highest explanatory power, especially for COD and conductivity. It indicates that integrating deep features extraction and boosted decision tree regression is effective in satellite-based water quality estimation. In general, this study demonstrates the value of multispectral remote sensing combined with advanced AI techniques to develop scalable, cost-effective, and continuous freshwater quality monitoring toward better environmental management and decision-making.
References
1) Adjovu, G. E., Stephen, H., James, D., & Ahmad, S. (2023). Overview of the application of remote sensing in effective monitoring of water quality parameters. Remote Sensing, 15(7), 1938.
2) Sun, X., Zhang, Y., Shi, K., Zhang, Y., Li, N., Wang, W., ... & Qin, B. (2022). Monitoring water quality using proximal remote sensing technology. Science of the Total Environment, 803, 149805.
3) Topp, S. N., Pavelsky, T. M., Jensen, D., Simard, M., & Ross, M. R. (2020). Research trends in the use of remote sensing for inland water quality science: Moving towards multidisciplinary applications. Water, 12(1), 169.
4) Batina, A., & Krtalić, A. (2024). Integrating remote sensing methods for monitoring lake water quality: A comprehensive review. Hydrology, 11(7), 92.
5) Torres-Bejarano, F., Arteaga-Hernández, F., Rodríguez-Ibarra, D., Mejía-Ávila, D., & González-Márquez, L. C. (2021). Water quality assessment in a wetland complex using Sentinel 2 satellite images. International Journal of Environmental Science and Technology, 18(8), 2345-2356.
6) Barzegar, R., Aalami, M. T., & Adamowski, J. (2020). Short-term water quality variable prediction using a hybrid CNN–LSTM deep learning model. Stochastic Environmental Research and Risk Assessment, 34(2), 415-433.
7) Yang, H., Kong, J., Hu, H., Du, Y., Gao, M., & Chen, F. (2022). A review of remote sensing for water quality retrieval: Progress and challenges. Remote Sensing, 14(8), 1770.
8) Singh, S., Bhardwaj, A., & Verma, V. K. (2020). Remote sensing and GIS based analysis of temporal land use/land cover and water quality changes in Harike wetland ecosystem, Punjab, India. Journal of environmental Management, 262, 110355.
9) Ye, S., Li, R., Bai, Z., Tuzikov, A., & Chen, C. (2024). Intelligent Classification of Water Bodies with Different Turbidity Levels Based on Gaofen-1 Multispectral Imagery. Optics Express, 32(20), 34929. https://doi.org/10.1364/oe.533540
10) Gautam, S., & Singhai, J. (2022). Fast and Accurate Water Region Classification from Remote Sensing Images Using Enhanced Convolutional Neural Network Classifier. 1–4. https://doi.org/10.1109/SCEECS54111.2022.9740993
11) DEROSIER, J. (2022). A Computer Vision Model for Detection of Water Pollutants Using Deep Learning Frameworks (pp. 543–553). https://doi.org/10.1007/978-981-19-2500-9_41
12) Yan, J., Zhang, Z., Lin, X., Hei, J., Wang, Y., & Zhang, A. (2025). AI-powered approaches for enhancing remote sensing-based water contamination detection in ecological systems. Frontiers in Environmental Science, 13. https://doi.org/10.3389/fenvs.2025.1612658
13) Phiri, D., Simwanda, M., Salekin, S., Nyirenda, V. R., Murayama, Y., & Ranagalage, M. (2020). Sentinel-2 data for land cover/use mapping: A review. Remote sensing, 12(14), 2291.
14) Kalpalathika, N., Salih, J., Lakshmi, P., Siddiqi, N., & Bhuvaneswari, A. (2025, August). Enhanced Short-Term GHI Prediction using Cloud-Aware Deep Autoencoder Network. In 2025 5th International Conference on Soft Computing for Security Applications (ICSCSA) (pp. 890-896). IEEE.
15) Ulatalita, N. A., Fatanaya, N., Ulfa, K., Robbaniyyah, N. A. I., & Alfian, M. R. (2025). Analysis of Changes in Agricultural Land Area in Central Lombok Regency Using Google Earth Engine. Semeton Mathematics Journal, 2(2), 97-109.
16) Laonamsai, J., Julphunthong, P., Saprathet, T., Kimmany, B., Ganchanasuragit, T., Chomcheawchan, P., & Tomun, N. (2023). Utilizing NDWI, MNDWI, SAVI, WRI, and AWEI for estimating erosion and deposition in Ping River in Thailand. Hydrology, 10(3), 70.
17) Rashid, M. B. (2023). Monitoring of drainage system and waterlogging area in the human-induced Ganges-Brahmaputra tidal delta plain of Bangladesh using MNDWI index. Heliyon, 9(6).
18) Rawat, K. S., Sahu, S. R., Singh, S. K., Chander, S., & Gujrati, A. (2023, December). Water Quality Analysis Using Normalized Difference Chlorophyll Index (NDCI) and Normalized Difference Turbidity Index (NDTI), Using Google Earth Engine Platform. In 2023 International Conference on Modeling, Simulation & Intelligent Computing (MoSICom) (pp. 408-413). IEEE.
19) Singh, R., Saritha, V., & Pande, C. B. (2024). Monitoring of wetland turbidity using multi-temporal Landsat-8 and Landsat-9 satellite imagery in the Bisalpur wetland, Rajasthan, India. Environmental Research, 241, 117638.
20) Kareem, H., Attaee, M., & Omran, A. (2024). Estimation the water ratio index (WRI) and automated water extraction index (AWEI) of bath in The United Kingdom using remote sensing technology of the multispectral data of landsat 8-oli. Water Conserv. Manag, 8, 125-132.
21) Yulianto, F., Kushardono, D., Budhiman, S., Nugroho, G., Chulafak, G. A., Dewi, E. K., & Pambudi, A. I. (2022). Evaluation of the threshold for an improved surface water extraction index using optical remote sensing data. The Scientific World Journal, 2022(1), 4894929.
22) Li, X., Dong, D., Liu, K., Zhao, Y., & Li, M. (2022). Identification of mine mixed water inrush source based on genetic algorithm and XGBoost algorithm: A case study of Huangyuchuan mine. Water, 14(14), 2150.
23) Singh, P., Ramkumar, K. R., & Hasija, T. (2024, July). Harnessing Machine Learning for Water Quality Evaluation: Comparative Analysis of XGBoost and CatBoost Algorithm. In 2024 Asia Pacific Conference on Innovation in Technology (APCIT) (pp. 1-6). IEEE.
24) Egbueri, J. C., & Agbasi, J. C. (2022). Performances of MLR, RBF-NN, and MLP-NN in the evaluation and prediction of water resources quality for irrigation purposes under two modeling scenarios. Geocarto International, 37(26), 14399-14431.
Downloads
Published
Issue
Section
License
This work is licensed under a Creative Commons Attribution 4.0 International License.
This work is licensed under a Creative Commons Attribution 4.0 International License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
