Groundwater level modeling framework by combining the wavelet transform with a long short-term memory data-driven...

Area of Science:

• Hydrology and Water Resources
• Machine Learning Applications in Environmental Science
• Geospatial Modeling

Background:

• Accurate groundwater level simulation is crucial for effective water resource management and aquifer protection.
• Traditional hydrological models can be computationally intensive for long-term forecasting.
• Machine learning offers a promising alternative for efficient groundwater level prediction.

Purpose of the Study:

• To develop and evaluate a novel multistep modeling framework for simulating groundwater levels.
• To combine wavelet transform (WT) with a multivariate long short-term memory (LSTM) network for enhanced prediction accuracy.
• To assess the performance of the proposed WT-multivariate LSTM (WT-MLSTM) model against existing methods.

Main Methods:

• Decomposition of groundwater level time series using wavelet transform (WT) into intrinsic mode functions.
• Pearson correlation analysis to identify relationships between influencing factors (e.g., river stage) and groundwater levels.
• Development of a multivariate LSTM model incorporating external factors to simulate decomposed components, followed by reconstruction.

Main Results:

• The combined WT-MLSTM model demonstrated superior simulation accuracy compared to standard LSTM, MLSTM, and WT-LSTM models.
• The proposed model outperformed support vector machine (SVM) in groundwater level simulation.
• Model predictability was confirmed for short-term time series, with accuracy decreasing at greater distances from rivers or with increased complexity.

Conclusions:

• The WT-MLSTM method provides a rapid and accurate approach for simulating and predicting groundwater levels.
• This framework offers a computationally efficient alternative to complex hydrological models for water resource management.
• The study highlights the potential of integrated machine learning techniques for understanding and managing groundwater resources.