Forecasting the daily evaporation by coupling the ensemble deep learning models with meta-heuristic algorithms and...

Area of Science:

• Environmental Science
• Hydrology
• Data Science
• Artificial Intelligence

Background:

• Accurate evaporation estimation is vital for managing scarce agricultural water resources.
• Existing data-driven models using time-series analysis often lack high accuracy and universality.
• Integration of meta-heuristic algorithms, ensemble deep learning, and advanced data preprocessing for evaporation prediction is underexplored.

Purpose of the Study:

• To develop a highly accurate and universal data-driven model for precise evaporation estimation using time-series analysis.
• To propose a novel hybrid deep learning model by integrating Variational Mode Decomposition (VMD), Convolutional Neural Network (CNN), Bidirectional Long Short-Term Memory (BiLSTM), and optimization algorithms.

Main Methods:

• Proposed a novel hybrid model: Whale Optimization Algorithm-Variational Mode Decomposition-Convolutional Neural Network-Sparrow Search Algorithm-Bidirectional Long Short-Term Memory (WOA-VMD-CNN-SSA-BiLSTM).
• Employed VMD optimized by WOA to extract multi-scale features from evaporation time series, overcoming subjectivity in parameter setting.
• Utilized CNN-BiLSTM as the core estimation module, with SSA optimizing its hyperparameters.

Main Results:

• Optimized VMD parameters (k=6, 0.1773) resulted in low Sample Entropy (SEn) of 0.0832, indicating effective amplitude-dependent feature extraction.
• The WOA-VMD preprocessing significantly enhanced the performance of the CNN-SSA-BiLSTM model.
• The final hybrid model (WOA-VMD-SSA-CNN-BiLSTM) demonstrated superior performance with MSE=0.1258, RMSE=0.3547, MAE=0.2833, and MAPE=6.17% in the testing stage.

Conclusions:

• The proposed WOA-VMD-CNN-SSA-BiLSTM model offers a highly accurate and robust approach for evaporation estimation.
• This hybrid deep learning model outperforms other existing hybrid and ensemble models.
• The model is recommended for practical application in estimating evaporation, aiding agricultural water resource management.