Prediction of the monthly river water level by using ensemble decomposition modeling
Open Access
- 24 July 2025
- journal article
- research article
- Published by Springer Nature in Scientific Reports
- Vol. 15 (1) , 1-25
Abstract
The decomposition, artificial intelligence (AI) and machine learning (ML) modeling have been important role in hydrological and river basin related prediction and forecasting to help the flood management and sustainable water resources development. In this paper, developed the hybrid modeling combined with complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN), along with standalone models support vector machine (SVM-linear), and Random Forest (RF), Random Subspace (RS) for accurate prediction of monthly river water level in Sg Muar at Buloh Kasap, Johor station during 2014 to 2023. In this paper, two combinations of variables such as Lags and IMFS are used for development of different models for river water level prediction. Hence, these models are compared and measured the performance of models based on the various statistics metrics. Therefore hybrid models performance is measured based on the coefficient of determination (R2), hence all models results are shown the CEEMDAN-SVM-LINEAR (R2 = 0.87), CEEMDAN-SVM-RBF (R2 = 0.91), CEEMDAN-RF (R2 = 0.98), and CEEMDAN-RS (R2 = 0.88) in the second combination variables, while standalone models performance are shown SVM-Linear (R2 = 0.84), SVM-RBF (R2 = 0.87), RF (R2 = 0.97), and RS (R2 = 0.86) during the training phase stage in the first combination variables. Similarly, in the testing phase, the best two models performances are very well as a CEEMDAN-RF (R2:0.94) and CEEMDAN-RS (R2:0.90) in second combination variables, and the first combination variables based SVM- Linear (R2:0.93) and RF (R2:0.89) models are performance higher compared with other models. Finally, the CEEMDAN-RF hybrid model is best model based on the lowest observed errors of Root mean square error (RMSE): 0.13, Mean square error (MSE): 0.02 and high R2: 0.94, hence this model is appropriate for prediction of river water level. Hence, the best hybrid model has been concluded that the CEEMDAN data decomposition technique is very useful for improve performance of the prediction model, the complex river water level predictions by separating the data sets into various sub-frequencies, allowing a better understanding of trends, seasonality and fluctuations in the data. Therefore, the CEEMDAN based novel hybrid modeling is effective decomposition modeling for complex field utilized in the sustainable and optimized utilization of the water resources for sustainable development goal (SDG).Keywords
This publication has 70 references indexed in Scilit:
- Research on short-term precipitation forecasting method based on CEEMDAN-GRU algorithmScientific Reports, 2024
- Daily flow prediction of the Huayuankou hydrometeorological station based on the coupled CEEMDAN–SE–BiLSTM modelScientific Reports, 2023
- Carbonation inhibitor by polyethylene glycol encapsulation of calcium hydroxide fine particles to improve antimicrobial and root canal penetration propertiesHeliyon, 2023
- Runoff prediction of lower Yellow River based on CEEMDAN–LSSVM–GM(1,1) modelScientific Reports, 2023
- New double decomposition deep learning methods for river water level forecastingScience of The Total Environment, 2022
- Predicting polycyclic aromatic hydrocarbons in surface water by a multiscale feature extraction-based deep learning approachScience of The Total Environment, 2021
- Modeling streamflow driven by climate change in data-scarce mountainous basinsScience of The Total Environment, 2021
- Design and implementation of a hybrid model based on two-layer decomposition method coupled with extreme learning machines to support real-time environmental monitoring of water quality parametersScience of The Total Environment, 2019
- Predicting motor vehicle crashes using Support Vector Machine modelsAccident Analysis & Prevention, 2008
- Asymmetric bagging and random subspace for support vector machines-based relevance feedback in image retrievalIEEE Transactions on Pattern Analysis and Machine Intelligence, 2006