Al-Jahwari et al. (2025) Robust Rainfall Gap-Filling in Coastal Arid Regions Using Ensemble Fusion Models

⚠️ Warning: This summary was generated from the abstract only, as the full text was not available.

Identification

• Journal: Hydrology
• Year: 2025
• Date: 2025-12-20
• Authors: Badar Al-Jahwari, Ghazi Al-Rawas, Mohammad Reza Nikoo, Talal Etri, Jens Grundmann
• DOI: 10.3390/hydrology13010001

Research Groups

Not explicitly stated in the provided text.

Short Summary

This study implemented and evaluated multiple machine learning and novel ensemble fusion techniques to fill daily rainfall data gaps across 88 stations in Oman from 1993 to 2024, finding that the Gradient-Boosting Trees (GBT) model performed best and ensemble fusion significantly improved prediction accuracy.

Objective

• To explore and implement multiple machine learning techniques, including a novel Ensemble Fusion Model, to address the complexity of filling daily rainfall data gaps for 88 rainfall stations in the Al-Batinah region of Oman.

Study Configuration

• Spatial Scale: 88 rainfall stations in the Al-Batinah region of Oman; nearby stations within a 50 km radius were considered.
• Temporal Scale: Daily rainfall data covering the period from 1993 to 2024 (31 years).

Methodology and Data

• Models used: Multiple Linear Regression (MLR), Random Forest (RF), K-Nearest Neighbors (KNN), Support Vector Regression (SVR), Gradient-Boosting Trees (GBT). K-means clustering for optimal cluster number calculation. Novel Ensemble Fusion Model (RF Fusion Model, Multi-Model Super Ensemble Fusion Model (MMSE)). Bayesian optimization for hyperparameter tuning, Multiple Imputation by Chained Equations (MICE) for dataset imputation, and Leave-One-Year-Out (LOYO) cross-validation for evaluation.
• Data sources: Daily rainfall observations from 88 rainfall stations.

Main Results

• The Gradient-Boosting Trees (GBT) model demonstrated the best performance among all tested models in both clustered and non-clustered approaches for rainfall gap-filling.
• The application of Ensemble Fusion Models significantly improved the efficacy of the predictive models, increasing the coefficient of determination (R²) by 22.5% for the clustering approach and 22.2% for the non-clustering approach.

Contributions

• Comprehensive evaluation of multiple machine learning techniques for daily rainfall gap-filling in an arid region (Al-Batinah, Oman).
• Introduction and assessment of a novel Ensemble Fusion Model (RF Fusion Model and Multi-Model Super Ensemble Fusion Model) to enhance prediction accuracy in rainfall imputation.
• Integration of advanced techniques such as Bayesian optimization, MICE, and LOYO cross-validation for robust model enhancement and evaluation.

Funding

Not explicitly stated in the provided text.

Citation

@article{AlJahwari2025Robust,
  author = {Al-Jahwari, Badar and Al-Rawas, Ghazi and Nikoo, Mohammad Reza and Etri, Talal and Grundmann, Jens},
  title = {Robust Rainfall Gap-Filling in Coastal Arid Regions Using Ensemble Fusion Models},
  journal = {Hydrology},
  year = {2025},
  doi = {10.3390/hydrology13010001},
  url = {https://doi.org/10.3390/hydrology13010001}
}