Abstract—Predicting nitrate levels in rivers is vital for
maintaining healthy ecosystems, as excessive nitrates can
disrupt aquatic life and lead to harmful algal blooms. Accurate
nitrate monitoring also ensures safe drinking water, protecting
communities from health risks. To keep our rivers healthy,
protect water resources, and ensure human safety, accurately
measuring nitrate levels—a key sign of water quality—is
essential. We developed three advanced models to predict
hourly nitrate levels in the Willamette River in Portland,
Oregon, using data from 2021–2022. These models were a Gated
Recurrent Unit (GRU), a Convolutional Neural Network (CNN),
and a new hybrid model combining both, called CNN-GRU. We
tested their performance using metrics like correlation
coefficient (R), Determination Coefficient (DC), and Root Mean
Squared Error (RMSE). The models used six key water quality
factors: river flow, turbidity (cloudiness), water velocity, specific
conductance, dissolved oxygen, and water temperature. The
hybrid CNN-GRU model performed best, with an R-value of
0.876, a DC of 0.761, and an RMSE of 0.086, outshining the
standalone models. Among the individual models, GRU did
better than CNN. We also found that turbidity was the most
critical factor for predicting nitrate levels.
Keywords—water quality, nitrate prediction, deep learning,
CNN, GRU
Cite: Ismail Farajpour and Sina Davoudi, "Predicting River Nitrate (NO₃) Concentrations Using Hybrid Deep Learning: Case Study of the Willamette River, Oregon," International Journal of Environmental Science and Development vol. 16, no. 5, pp. 345-352, 2025.
Copyright © 2025 by the authors. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited (CC BY 4.0).