Shakib et al. (2026) MonoGRU: A theoretical and empirical evaluation of a streamlined gated recurrent unit for rainfall forecasting

Identification

• Journal: The Science of The Total Environment
• Year: 2026
• Date: 2026-01-01
• Authors: Md. Shadman Shakib, Md. Eyakub Ali, Nabil H. Bhuiyan, Kamaruzzaman, Badhan Kumar Dey, Md. Istihad Bhuiyan, Hasibul Hoque
• DOI: 10.1016/j.scitotenv.2025.181085

Research Groups

• Dept. of EEE, Metropolitan University, Sylhet, Bangladesh
• Dept. of EEE, RTM Al-Kabir Technical University, Sylhet, Bangladesh
• China Spallation Neutron Source, Dongguan, Guangdong, China
• Institute of High Energy Physics, Chinese Academy of Sciences, Beijing, China
• Dept. of EEE, Noakhali Science and Technology University, Bangladesh

Short Summary

This study proposes MonoGRU, a streamlined Gated Recurrent Unit model, for efficient and accurate rainfall prediction, demonstrating superior performance and reduced computational overhead compared to conventional GRU models across various evaluation metrics.

Objective

• To develop and empirically evaluate MonoGRU, an optimized Gated Recurrent Unit model, for accurate and computationally efficient rainfall prediction.

Study Configuration

• Spatial Scale: General, applicable to various regions for time-series rainfall data.
• Temporal Scale: Time-series based, specific resolution and duration not detailed but implied for forecasting.

Methodology and Data

• Models used: MonoGRU (proposed), Gated Recurrent Unit (GRU) (for comparison).
• Data sources: Rainfall time-series data (specific source not detailed).

Main Results

• MonoGRU consistently outperforms the conventional GRU model across key performance metrics, including Mean Absolute Error (MAE), Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and R2 Score.
• Q-Q plot analysis indicates MonoGRU's robustness and superior predictive reliability, particularly under high epoch settings.
• The proposed model demonstrates a reduced memory footprint and lower computational overhead compared to GRU.
• Architectural improvements in MonoGRU lead to enhanced computational efficiency and precision while demonstrating robust generalization and reduced overfitting.

Contributions

• Introduction of a novel MonoGRU model designed for improved rainfall prediction accuracy.
• Empirical validation demonstrating MonoGRU's superior performance, computational efficiency, and reduced memory footprint compared to conventional GRU architectures.
• Offers a scalable solution for rainfall prediction, particularly suitable for resource-constrained environments.
• Highlights the potential for extending MonoGRU's capabilities to other time-series forecasting domains.

Funding

• Not specified in the provided paper text.

Citation

@article{Shakib2026MonoGRU,
  author = {Shakib, Md. Shadman and Ali, Md. Eyakub and Bhuiyan, Nabil H. and Kamaruzzaman and Dey, Badhan Kumar and Bhuiyan, Md. Istihad and Hoque, Hasibul},
  title = {MonoGRU: A theoretical and empirical evaluation of a streamlined gated recurrent unit for rainfall forecasting},
  journal = {The Science of The Total Environment},
  year = {2026},
  doi = {10.1016/j.scitotenv.2025.181085},
  url = {https://doi.org/10.1016/j.scitotenv.2025.181085}
}