Name in original language
Predikcija dnevnih dotoka u slivu reke Drine upotrebom veštačke inteligencije
Initiative overview
The initiative uses artificial intelligence to improve the prediction of daily water inflows into reservoirs in the Drina River basin. It was developed to address limitations of traditional hydrological forecasting, which often struggles with missing data, changing weather patterns, and the increasing variability of water conditions. By integrating machine learning with historical inflow patterns, meteorological variables, and reservoir levels, the initiative enhances accuracy and supports more informed planning in water and energy management.
Its main objectives are to provide more reliable inflow forecasts, enable better operation of reservoirs, and strengthen decision making for hydropower, flood preparedness, and drought management. The system applies time series analysis, regression techniques, and neural networks, combined with robust data preprocessing and feature engineering. Validation using real-world data from multiple reservoirs shows that the approach outperforms traditional methods and adapts effectively to varying hydrological conditions.
The initiative also creates opportunities to expand AI based forecasting to other river systems facing similar challenges. Since many basins require improved predictive tools, the underlying framework can be adapted to additional datasets and institutional settings. As a result, the initiative has the potential to contribute to broader, cross-basin improvements in environmental planning and resource management.
In the future, the initiative is expected to evolve through ongoing refinement of predictive models, deeper integration with hydrological monitoring infrastructure, and sustained collaboration with sector experts such as the Jaroslav Černi Water Institute. These partnerships support long term institutional adoption and scale up. Over time, the system could be embedded into routine operational processes across Serbia’s water sector and extended to regional partners, helping to strengthen resilience and promote evidence-based management of rivers and reservoirs.
Results, outcomes and impacts
The initiative has achieved more accurate daily inflow forecasts than traditional methods, confirmed through validation using real reservoir data and comparative performance metrics. This has improved planning for water and energy management. Future impacts include more reliable reservoir operation, better flood and drought preparedness, and the ability to scale the AI model to other river basins.
