Accurate forecasting of river flows is essential for effective water resource management, flood risk reduction and environmental protection. The ongoing effects of climate change, in particular the shift in precipitation patterns and the increasing frequency of extreme weather events, necessitate the development of advanced forecasting models. This study investigates the application of long short-term memory (LSTM) neural networks in predicting river runoff in the Velika Morava catchment in Serbia, representing a pioneering application of LSTM in this region. The study uses daily runoff, precipitation and temperature data from 1961 to 2020, interpolated using the inverse distance weighting method. The LSTM model, which was optimized using a trial-and-error approach, showed a high prediction accuracy. For the Velika Morava station, the model showed a mean square error (MSE) of 2936.55 and an R² of 0.85 in the test phase. The findings highlight the effectiveness of LSTM networks in capturing nonlinear hydrological dynamics, temporal dependencies and regional variations. This study underlines the potential of LSTM models to improve river forecasting and water management strategies in the Western Balkans.

In hydrological practice, individual elements of the hydrological cycle are most often estimated and evaluated separately. Uncertainty in the size estimation of extrema discharges and their return period can affect the statistical assessment of the significance of floods. One example is the simultaneous occurrence and joining of extremes at the confluence of rivers. The paper dealt with the statistical evaluation of the occurrence of two independent variables and their joint probabilities of occurrence. Bivariate joint analysis is a statistical approach for the assessment of flood threats at the confluence of rivers. In our study, the annual maximum discharges monitored on four selected Slovak rivers and their tributaries represent the analyzed variables. The Archimedean class of copula functions was used as a set of mathematical tools for the determination and evaluation of the joint probability of annual maximal discharges at river confluences. The results of such analysis can contribute to a more reliable assessment of flood threats, especially in cases where extreme discharges occur simultaneously, increasing the risk of devastating effects. Finally, the designed discharges of the different return periods calculated by using the univariate approach and the bivariate approach for the gauging station below the confluence of the rivers was evaluated and compared.

This study aimed at the analysis and modelling of the groundwater temperature at the water table in different regions of Slovakia. In the first part, the analysis of the long-term trends of air and soil/ground temperature to a depth of 10 m is presented. The average annual soil/groundwater temperatures at different depths were the same but lower than the annual average air temperature by about 0.8 °C. The long-term trend analysis of the air temperature and soil temperature at a depth of up to 10 m in Slovakia showed that the air and soil/ground water temperature have risen by 0.6 and 0.5 °C, respectively, per decade over the past 30 years. The second part of the study aimed at modelling the daily groundwater temperatures at depths of 0.6–15 m below the surface. The simple groundwater temperature model was constructed based on a one-dimensional differential Fourier heat conduction equation. The given model can be used to estimate future groundwater temperature trends using regional air temperature projections calculated for different greenhouse gas emission scenarios.

This paper assesses the vulnerability to erosion of the soils in the upper Torysa catchment down to the planned drinking water reservoir at Tichy Potok (Slovakia). Experimental observations in the Torysa catchment, together with the outputs of the AGricultural Non-Point Source pollution model (AGNPS), have been used to select optimal sites for small sediment-trapping dams on the Torysa tributaries. The rainfall scenario for the upper Torysa was determined from records at the Plavec climate station from 1980–1998. In addition, an extreme rainfall event scenario was envisaged (100 mm rainfall in three hours); AGNPS model simulations suggest that such an event might produce as much as half or even the total amount of sediment generated by all the rainfalls over the whole period of simulation, 1980–1998. Keywords: erosion, sediment transport, simulation, AGNPS model, Torysa basin