Groundwater level forecasting with machine learning models: A review.

Area of Science:

• Hydrology
• Environmental Science
• Data Science

Background:

• Groundwater is the primary freshwater source but faces depletion globally.
• Accurate groundwater level (GWL) forecasting is vital for sustainable water management.
• Machine learning (ML) shows promise for improving GWL prediction accuracy.

Purpose of the Study:

• To comprehensively review ML applications in GWL forecasting.
• To identify key ML models and modeling concepts used in recent research (2017-2023).
• To provide insights and best practices for optimal GWL forecasting.

Main Methods:

• Systematic literature review of 142 articles from Web of Science (2017-2023).
• Focus on prominent ML models: ANN, ANFIS, SVR, EC, DL, EN, and HM.
• Analysis of critical modeling factors: dataset size, data splitting, input selection, time-step, performance metrics, study zones, and aquifers.

Main Results:

• Machine learning models demonstrate significant potential for accurate GWL forecasting.
• Specific models like artificial neural networks (ANN) and deep learning (DL) are frequently employed.
• Best practices in data handling and model selection are crucial for optimal performance.

Conclusions:

• This review synthesizes current ML trends in GWL forecasting.
• It offers recommendations for researchers and water management agencies.
• Effective ML implementation is key to addressing groundwater depletion challenges.