Prediction of the monthly river water level by using ensemble decomposition modeling

Area of Science:

• Hydrology and Water Resources
• Artificial Intelligence in Environmental Science
• Machine Learning for Predictive Modeling

Background:

• Accurate river basin prediction and forecasting are crucial for effective flood management and sustainable water resource development.
• Traditional hydrological models often struggle with the complex, non-linear dynamics of river water levels.
• The integration of advanced decomposition techniques and machine learning offers a promising approach to improve prediction accuracy.

Purpose of the Study:

• To develop and evaluate hybrid models for accurate monthly river water level prediction.
• To compare the performance of hybrid models incorporating Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) against standalone machine learning models.
• To identify the optimal modeling strategy for enhancing hydrological forecasting in the Sg Muar basin.

Main Methods:

• Hybrid modeling approach combining CEEMDAN for data decomposition with Support Vector Machine (SVM), Random Forest (RF), and Random Subspace (RS) algorithms.
• Utilized two variable combinations: lagged values and Intrinsic Mode Functions (IMFs) derived from CEEMDAN.
• Model performance evaluated using statistical metrics, including Coefficient of Determination (R²), Root Mean Square Error (RMSE), and Mean Square Error (MSE).

Main Results:

• Hybrid models, particularly CEEMDAN-RF (R²=0.98 training, R²=0.94 testing), significantly outperformed standalone models in predicting river water levels.
• The CEEMDAN decomposition technique effectively improved the predictive accuracy by separating data into sub-frequencies.
• The CEEMDAN-RF model achieved the best performance with the lowest RMSE (0.13) and MSE (0.02) during the testing phase.

Conclusions:

• The CEEMDAN-based hybrid modeling approach is highly effective for complex river water level prediction.
• CEEMDAN enhances model performance by enabling a better understanding of data trends, seasonality, and fluctuations.
• This novel hybrid modeling strategy contributes to the sustainable and optimized utilization of water resources, aligning with Sustainable Development Goals.