A novel approach to precipitation prediction using a coupled CEEMDAN-GRU-Transformer model with permutation entropy...

Area of Science:

• Hydrology and Water Resources
• Climate Science
• Data Science and Machine Learning

Background:

• Accurate precipitation forecasting is vital for water resource management in the Yellow River basin.
• Existing precipitation prediction models often overlook the influence of different frequency components on accuracy.
• Understanding precipitation variability across different frequencies is key to improving predictive models.

Purpose of the Study:

• To develop and evaluate a novel coupled monthly precipitation prediction model for the upper reaches of the Yellow River.
• To investigate the impact of frequency partitioning on precipitation prediction accuracy.
• To enhance the reliability of regional medium and long-term precipitation forecasts.

Main Methods:

• Utilized adaptive noise complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) to decompose precipitation data.
• Employed permutation entropy (PE) to partition CEEMDAN-processed data into distinct frequency components.
• Developed a coupled model integrating CEEMDAN, gated recurrent unit neural network (GRU), and an attention mechanism-based transformer for frequency-specific predictions.

Main Results:

• The proposed CEEMDAN-GRU-Transformer model demonstrated superior performance in monthly precipitation prediction across four regions in the upper Yellow River.
• The model achieved a coefficient of determination (R²) greater than 0.8, indicating high prediction accuracy.
• Comparison with other models confirmed the effectiveness of the coupled approach in capturing precipitation patterns.

Conclusions:

• The CEEMDAN-GRU-Transformer model offers a significant advancement in regional medium and long-term precipitation forecasting.
• Integrating frequency analysis with advanced machine learning techniques improves prediction accuracy.
• This novel approach provides a valuable tool for water resource management and planning in river basins.