Machine learning (ML), a branch of artificial intelligence (AI), has been increasingly used in environmental engineering due to the ability to analyze complex nonlinear problems (such as ones connected with water quality management) through a data-driven approach. This study provides an overview of different ML algorithms applied for monitoring and predicting river water quality. Different parameters could be monitored or predicted, such as dissolved oxygen (DO), biological and chemical oxygen demand (BOD and COD), turbidity levels, the concentration of different ions (such as Mg2+ and Ca2+), heavy metal or other pollutant's concentration, pH, temperature, and many more. Although many algorithms have been investigated for the prediction of river water quality, there are several which are most commonly used in engineering practice. These models mostly include so-called supervised learning algorithms, such as artificial neural network (ANN), support vector machine (SVM), random forest (RF), decision tree (DT), and deep learning (DL). To further enhance prediction power, novel hybrid algorithms, could be used. However, the quality of prediction is not only dependent on the applied algorithm but also on the availability of previously mentioned water quality parameters, their selection, and the combination of input data used to train the ML model.

Water quality prediction is crucial for effective river stream management. Dissolved oxygen, conductivity and chemical oxygen demand are vital chemical parameters for water quality. Development of machine learning (ML) and deep learning (DL) methods made them widely used in this domain. Sophisticated DL techniques, especially long short-term memory (LSTM) networks, are required for accurate, real-time multistep prediction. LSTM networks are effective in predicting water quality due to their ability to handle long-term dependencies in sequential data. We propose a novel hybrid approach for water quality parameters prediction combining DL with data smoothing method. The Sava river at the Jamena hydrological station serves as a case study. Our workflow uses LSTM networks alongside LOcally WEighted Scatterplot Smoothing (LOWESS) technique for data filtering. For comparison, Support Vector Regressor (SVR) is used as the baseline method. Performance is evaluated using Root Mean Squared Error (RMSE) and Coefficient of Determination R2 metrics. Results demonstrate that LSTM outperforms the baseline method, with an R2 up to 0.9998 and RMSE of 0.0230 on the test set for dissolved oxygen. Over a 5-day prediction period, our approach achieves R2 of 0.9912 and RMSE of 0.1610 confirming it as a reliable method for water quality multistep parameters prediction.

Investigating the impact of climate change on the management of a complex multipurpose water system is a critical issue. The presented study focuses on different steps of the climate change impact analysis process: (i) Use of three regional climate models (RCMs), (ii) use of four bias correction methods (BCMs), (iii) use of three concentration scenarios (CSs), (iv) use of two model averaging procedures, (v) use of the hydrological model and (vi) use of the system dynamics simulation model (SDSM). The analyses are performed for a future period, from 2006 to 2055 and the reference period, from 1971 to 2000. As a case study area, the Lim water system in Serbia (southeast Europe) is used. The Lim river system consists of four hydraulically connected reservoirs (Uvac, Kokin Brod, Radojnja, Potpec) with a primary purpose of hydropower generation. The results of the climate change impact analyses indicate change in the future hydropower generation at the annual level from −3.5% to +17.9%. The change has a seasonal variation with an increase for the winter season up to +20.3% and decrease for the summer season up to −33.6%. Furthermore, the study analyzes the uncertainty in the SDSM outputs introduced by different steps of the modelling process. The most dominant source of uncertainty in power production is the choice of BCMs (54%), followed by the selection of RCMs (41%). The least significant source of uncertainty is the choice of CSs (6%). The uncertainty in the inflows and outflows is equally dominated by the choice of BCM (49%) and RCM (45%).

In the present paper we propose a new model of monthly river flow rate as a simple nonlinear function of air temperature and rainfall. Response surface methodology is used to analyze the observed monthly flow rates from 1950 to 1990 for Great Morava River, as the largest domestic river in Serbia. Obtained results indicate significant linear and quadratic effect of both individual factors, while two-factor interactions show significantly smaller influence, indicating occurrence of maximum flow rate for low temperature and high rainfall regime. Statistical reliability of the proposed model is verified by internal and external validation, the latter of which included comparison of predicted and observed values from 1991 to 2012. It is shown that predicted flow rates exhibit a similar statistical pattern as observed ones, with a satisfying value of Nash–Sutcliffe coefficient (NSE = 0.73), although the derived model cannot capture well the highest flow rates. Obtained results further indicate the sequence of residuals represents random time series, which is confirmed by appropriate test statistics and surrogate data testing. The advantage of using the derived model for hydrological simulations in river basins instead of existing ones lies in its explicit mathematical form, making it suitable for quick and reliable estimation and prediction of monthly flow rates.