Forecasting groundwater level changes using machine learning techniques in Tazerbo area, Al Kufra Basin, southeast
Osama A El Fallah1, Lobna M Abou El-Magd2, Mohamed M El Kammar3
1Department of Earth Sciences, Faculty of Science, Benghazi University, P.O. Box 1308, Benghazi, Libya. osama.elfallah@uob.edu.ly.
Groundwater management is critical in arid regions. A Nonlinear Autoregressive Exogenous Neural Network (NARX-NN) model accurately predicts groundwater levels, forecasting significant declines by 2040 under current and higher pumping rates.
Area of Science:
- Hydrogeology and Water Resource Management
- Artificial Intelligence in Environmental Science
Background:
- Increasing global demand for consumable water necessitates robust groundwater management, particularly in arid and semi-arid regions.
- Sustainable water supply and environmental health depend on effective strategies to monitor and predict groundwater fluctuations.
- Machine learning offers advanced tools for identifying patterns and forecasting groundwater level changes, crucial for preventing water scarcity.
Purpose of the Study:
- To develop and apply a Nonlinear Autoregressive Exogenous Neural Network (NARX-NN) model for predicting groundwater levels in Tazerbo, Al Kufra Basin, Libya.
- To forecast future groundwater levels under various pumping scenarios to inform sustainable water resource management.
- To assess the spatial variability of groundwater decline trends in the study area.
Main Methods:
- Utilized annual groundwater level data from 2004-2024 across 14 piezometric wells in the Tazerbo region.
- Implemented a time series neural network, specifically the Nonlinear Autoregressive Exogenous Neural Network (NARX-NN), for predictive modeling.
- Trained and validated the NARX-NN model using statistical metrics (R², MSE, RMSE) to ensure high predictive accuracy.
Main Results:
- The NARX-NN model demonstrated excellent performance during training and testing, achieving high predictive accuracy for all wells.
- Scenario-based forecasts indicate a projected groundwater decline of approximately 2 m by 2030 and 1.6 m by 2040 at current pumping rates.
- Under higher pumping rates (255,000 m³/day and 400,000 m³/day), drawdowns could exceed 50 m by 2030 and 2040, with significant drops in northern and eastern zones.
Conclusions:
- The NARX-NN model is a reliable tool for forecasting groundwater level changes in water-stressed basins.
- Findings highlight the urgent need for sustainable groundwater management strategies to mitigate severe drawdowns.
- The study provides critical insights for long-term water resource planning in the Al Kufra Basin and similar arid regions.
Related Concept Videos
Application of Linearization and Approximation
Design Example: Analyzing Capacity Contours for Flood Risk Assessment
Design Example: Maintaining Level of an Embankment
Precipitation Gravimetry
In determining nickel by gravimetric analysis, a precipitant of ethanolic dimethylglyoxime is added to a hot nickel salt solution. This is quickly followed by the dropwise addition of dilute ammonia solution until precipitation occurs. A...
Taping Over Different Ground Profiles
Levels of Use of a GIS


