An intelligent hybrid deep learning-machine learning model for monthly groundwater level prediction.

Area of Science:

• Environmental Science
• Artificial Intelligence
• Water Resource Management

Background:

• Accurate groundwater level (GWL) prediction is crucial for effective environmental protection and sustainable water resource management.
• Traditional methods often struggle with the complex, nonlinear dynamics inherent in hydrological systems.
• The need for advanced predictive models is driven by increasing environmental pressures and water scarcity concerns.

Purpose of the Study:

• To develop and evaluate a novel hybrid artificial intelligence model for precise monthly groundwater level forecasting.
• To integrate optimization algorithms with deep learning and fuzzy inference systems for enhanced predictive accuracy.
• To assess the performance of the proposed model against established benchmark methods using rigorous evaluation metrics.

Main Methods:

• A hybrid model, termed PSO-COO-GRU-ANFIS (PCGA), was developed, combining Particle Swarm Optimization (PSO), Coati Optimization (COO), Gated Recurrent Unit (GRU), and Adaptive Neuro-Fuzzy Inference System (ANFIS).
• The PSO-COO algorithm was employed for optimizing the parameters of both GRU and ANFIS components.
• GRU was utilized for extracting temporal dependencies, while ANFIS handled the final prediction generation based on these extracted patterns.

Main Results:

• The PCGA model demonstrated high accuracy, achieving a Mean Absolute Error (MAE) of 1.90 and a Nash-Sutcliffe Efficiency (NSE) of 0.90 for monthly GWL forecasting in the Ardabil Plain.
• PCGA outperformed benchmark models, improving MAE by 14-77% and NSE by 1-20%.
• The study confirmed the effectiveness of the PSO-COO optimization and GRU's capability in capturing long-term data dependencies, reducing error fluctuations during parameter tuning.

Conclusions:

• The PCGA model is a robust and superior tool for forecasting monthly groundwater levels, outperforming conventional and standalone models.
• The hybrid approach effectively captures complex, nonlinear relationships within GWL data.
• This research offers a significant advancement in hydrological forecasting, supporting better environmental and water resource management strategies.